IBM ILOG CPLEX Optimization Studio

CPLEX Parameters Reference
Version 12 Release 6
Describes general use restrictions and trademarks related to this document and the software described in this document.

© Copyright IBM Corp. 1987, 2014

US Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM Corp.

Trademarks

IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of International Business Machines Corp., registered in
many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of
IBM trademarks is available on the Web at "Copyright and trademark information" at www.ibm.com/legal/copytrade.shtml.

Adobe, the Adobe logo, PostScript, and the PostScript logo are either registered trademarks or trademarks of Adobe Systems
Incorporated in the United States, and/or other countries.

Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both.

UNIX is a registered trademark of The Open Group in the United States and other countries.

Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other
countries, or both.

Java and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its affiliates.

Other company, product, or service names may be trademarks or service marks of others.

IBM ILOG CPLEX states these additional registered trademarks, copyrights, and acknowledgments.

Additional registered trademarks, copyrights, licenses

Python is a registered trademark of the Python Software Foundation.

MATLAB is a registered trademark of The MathWorks, Inc.

OpenMPI is distributed by The Open MPI Project under the New BSD license and copyright 2004 - 2012.

MPICH2 is copyright 2002 by the University of Chicago and Argonne National Laboratory.
Chapter 1. Parameters of CPLEX
Parameters, accessible and manageable by users, control the behavior of CPLEX.

Accessing parameters
Users access and modify parameters by means of methods in the various APIs.

Documentation about CPLEX parameters specific to the Python API is available as

online help inside a Python session. A brief introduction to CPLEX parameters is
available in the topic Using CPLEX parameters in the CPLEX Python API in the
tutorial about Python in Getting Stated with CPLEX.

Documentation about CPLEX parameters specific to the CPLEX for MATLAB

connector is available in the topic Using parameters in the Getting Started with
CPLEX for MATLAB manual. This information is also available as online help
inside a MATLAB session.

The following methods set and access parameters for objects of the class IloCplex
in C++ and Java:
setParam
getParam
getMin
getMax
getDefault
setDefaults

The names of the corresponding accessors in the class Cplex in the .NET API
follow the usual conventions of names and capitalization of languages in that
framework. For example, the class Cplex and its method Solve are denoted
Cplex.Solve. Likewise, the methods Cplex.GetParam and SetParam access and set
parameters in the .NET API.

C applications and applications written in other languages callable from C access

and set parameters with the following routines:

CPXgetdblparam Accesses a parameter of type double

CPXsetdblparam Changes a parameter of type double
CPXinfodblparam Gets the default value and range of a
parameter of type double
CPXgetintparam Accesses a parameter of type integer
CPXsetintparam Changes a parameter of type integer
CPXinfointparam Gets the default value and range of a
parameter of type integer
CPXgetlongparam Accesses a parameter of type long
CPXsetlongparam Changes a parameter of type long
CPXinfolongparam Gets the default value and range of a
parameter of type long
CPXgetstrparam Accesses a parameter of type string
CPXsetstrparam Changes a parameter of type string

© Copyright IBM Corp. 1987, 2014 1

 CPXinfostrparam Gets the default value of a parameter of
type string
CPXsetdefaults Resets all parameters to their standard
default values
CPXgetparamname Accesses the name of a parameter
CPXgetparamnum Access the identifying number assigned to a
parameter
CPXgetchgparams Accesses all parameters not currently at their
default value

Managing sets of parameters

Users may groups parameters into sets in the object-oriented APIs.

The object-oriented APIs of CPLEX also allow you to group parameters into a set
and then manage that set of parameters together.
v In the C++ API, use the member functions of an instance of the class
IloCplex::ParameterSet.
v In the Java API, use the methods of an object of the class
IloCplex.ParameterSet.
v In the .NET API, use the methods of the class Cplex.ParameterSet.
v In the Python API, use the methods of the class Cplex.ParameterSet.
v In the CPLEX for MATLAB Toolbox API, use the function cplexoptimset.
v In the MATLAB Cplex class API, use the structure Cplex.Param.

Parameter names
Names of CPLEX parameters follow the coding conventions of each API.

In the parameter table, each parameter has a name (that is, a symbolic constant) to
refer to it within an application.
v For the Callable Library (C API), these constants start with CPXPARAM_; for
example, CPXPARAM_Simplex_Limits_Iterations. They are used as the second
argument in all parameter routines (except CPXXsetdefaults, which does not
require them).
Legacy names
For the C API, these constants are capitalized and start with CPX_PARAM_; for
example, CPX_PARAM_ITLIM. They are used as the second argument in all
parameter routines (except CPXsetdefaults, which does not require them).
v For C++ applications, the parameters are defined in classes that specify a
hierarchy of applicability of the parameter; for example,
IloCplex::Param::Simplex::Limits::Iterations.
Legacy names
For C++ applications, the parameters are defined in nested enumeration types
for Boolean, integer, floating-point, and string parameters. The enum names use
mixed (lower and upper) case letters and must be prefixed with the class name
IloCplex:: for scope. For example, IloCplex::ItLim is the IloCplex equivalent
of CPX_PARAM_ITLIM.

2 CPLEX Parameters Reference

 v For Java applications, the parameters are defined as final static objects in nested
classes that specify a hierarchy of applicability of the parameter; for example,
IloCplex.Param.Simplex.Limits.Iterations.
Legacy names
For Java applications, the parameters are defined as final static objects in nested
classes called IloCplex.BooleanParam, IloCplex.IntParam, IloCplex.LongParam,
IloCplex.DoubleParam, and IloCplex.StringParam for Boolean, integer, long,
floating-point, and string parameters, respectively. The parameter object names
use mixed (lower and upper) case letters and must be prefixed with the
appropriate class for scope. For example, IloCplex.IntParam.ItLim is the object
representing the parameter CPX_PARAM_ITLIM.
v For .NET applications, the parameters follow the usual conventions for
capitalizing attributes and defining scope within a namespace.
v For Python applications, the names of parameters resemble the names in the
CPLEX Interactive Optimizer, modified for the syntax of a Python application.
For example, the command in the Interactive Optimizer set mip cuts mcfcut
looks like this in a Python application:cplex.parameters.mip.cuts.set(mcfcut).
v For MATLAB Cplex Class API applications, the names of parameters resemble
the names in the CPLEX Interactive Optimizer. For example, setting the MIP
variable selection strategy looks like this in a MATLAB Cplex Class API
application: cplex.Param.mip.strategy.variableselect = 3; where 3 indicates
strong branching. The parameters in CPLEX for MATLAB Toolbox applications
are named similarly. For example, setting the MIP variable selection strategy
looks like this in a toolbox application:
options.mip.strategy.variableselect.Cur = 3; where options is a structure
created with the function cplexoptimset(’cplex’). In addition, in order
maintain compatibility with the MATLAB Optimization Toolbox, a number of
parameters may be set using the MATLAB Optimization Toolbox parameter
names; for example, the maximum nodes can be set with: options =
cplexoptimset(’MaxNodes’, 400);.

The reference page of each parameter documents an integer that serves as a

reference number or unique identifier for each parameter. That integer reference
number corresponds to the value that each symbolic constant represents, as found
in the cplex.h , cplexx.h , and cpxconst.h header files of the Callable
Library (C API).

Correspondence of parameters between APIs

Certain specialized parameters of one API may not have an equivalent in another
API.

Some parameters available for the C API are not supported as parameters for the
object-oriented APIs or have a slightly different name there. In particular:
v “epsilon (degree of tolerance) used in linearization” on page 52 (EpLin), the
parameter specifying the tolerance to use in linearization in the object oriented
APIs (C++, Java, .NET), is not applicable in the C API, nor in the Python API.
v “MIP callback switch between original model and reduced, presolved model” on
page 71(CPX_PARAM_MIPCBREDLP), the parameter specifying whether to use the
reduced or original model in MIP callbacks, has no equivalent in the
object-oriented APIs (C++, Java, .NET) nor in the Python API, nor in the
MATLAB connector.

Chapter 1. Parameters of CPLEX 3

 v Logging output is controlled by a parameter in the C API (CPX_PARAM_SCRIND),
but when using the object-oriented APIs, you control logging by configuring the
output channel:
– IloCplex::out in C++
For example, to turn off output to the screen, use
cplex.setOut(env.getNullStream()) .
– IloCplex.output in Java
For example, to turn off output to the screen, use cplex.setOut(null).
– Cplex.Out in .NET
For example, to turn off output to the screen, use Cplex.SetOut(Null).
– cplex.set_results_stream in Python
For example, to turn off output to the screen, use
cplex.set_results_stream(None).
– DisplayFunc in the MATLAB Cplex Class API
For example, to turn off output to the screen, usecplex.DisplayFunc();.
– display and Display in the CPLEX for MATLAB Toolbox API
For example, to turn off output to the screen, use
optimset(’Display’,’off’); or options.display = ’off’;.
v The parameter IloCplex::RootAlg in the C++ API corresponds to these
parameters in the C API:
– “algorithm for initial MIP relaxation” on page 123: CPX_PARAM_STARTALG
– “algorithm for continuous problems” on page 120: CPX_PARAM_LPMETHOD
– “algorithm for continuous quadratic optimization” on page 122:
CPX_PARAM_QPMETHOD
v The parameter IloCplex::NodeAlg in the C++ API corresponds to the parameter
“MIP subproblem algorithm” on page 86 CPX_PARAM_SUBALG in the C API.

Saving parameter settings to a file in the C API

Users of the Callable Library (C API) can save current parameter settings in a PRM
file.

You can read and write a file of parameter settings with the C API. The file
extension is .prm . The C routine CPXreadcopyparam reads parameter values from a
file with the .prm extension. The routine CPXwriteparam writes a file of the current
non-default parameter settings to a file with the .prm extension. Here is the format
of such a file:
CPLEX Parameter File Version number
parameter_name parameter_value

Tip:

The heading with a version number in the first line of a PRM file is significant to
CPLEX. An easy way to produce a correctly formatted PRM file with a proper
heading is to have CPLEX write the file for you.

CPLEX reads the entire file before changing any of the parameter settings. After
successfully reading a parameter file, the C API first sets all parameters to their
default value. Then it applies the settings it read in the parameter file. No changes
are made if the parameter file contains errors, such as missing or illegal values.
There is no checking for duplicate entries in the file. In the case of duplicate
entries, the last setting in the file is applied.

4 CPLEX Parameters Reference

When you create a parameter file from the C API, only the non-default values are
written to the file. You can double-quote string values or not when you create a
PRM file, but CPLEX always writes string-valued parameters with double
quotation marks.

The comment character in a parameter file is #. After that character, CPLEX ignores
the rest of the line.

The C API issues a warning if the version recorded in the parameter file does not
match the version of the product. A warning is also issued if a nonintegral value is
given for an integer-valued parameter.

Here is an example of a correct CPLEX parameter file:

CPLEX Parameter File Version 11.0
CPX_PARAM_EPPER 3.45000000000000e-06
CPX_PARAM_OBJULIM 1.23456789012345e+05
CPX_PARAM_PERIND 1
CPX_PARAM_SCRIND 1
CPX_PARAM_WORKDIR "tmp"

Chapter 1. Parameters of CPLEX 5

6 CPLEX Parameters Reference
Chapter 2. Topical list of parameters
Users can browse CPLEX parameters, organized by topics.

Simplex
Here are links to parameters of interest to users of the simplex optimizers.

Selects the “algorithm for continuous problems” on page 120

“advanced start switch” on page 17

“lower objective value limit” on page 92

“upper objective value limit” on page 92

“dual simplex pricing algorithm” on page 48

“primal simplex pricing algorithm” on page 103

“simplex crash ordering” on page 40

“Markowitz tolerance” on page 53

“optimality tolerance” on page 53

“perturbation constant” on page 54

“simplex perturbation switch” on page 95

“simplex perturbation limit” on page 96

“relaxation for FeasOpt” on page 55

“feasibility tolerance” on page 56

“simplex maximum iteration limit” on page 68

“memory reduction switch” on page 71

“numerical precision emphasis” on page 89

“simplex pricing candidate list size” on page 108

“sifting subproblem algorithm” on page 127

“simplex iteration information display” on page 129

“simplex singularity repair limit” on page 129

© Copyright IBM Corp. 1987, 2014 7

Barrier
Here are links to parameters of interest to users of the barrier optimizer.

“advanced start switch” on page 17

“barrier algorithm” on page 23

“barrier starting point algorithm” on page 30

“barrier crossover algorithm” on page 24

“sifting subproblem algorithm” on page 127

“barrier ordering algorithm” on page 28

“barrier display information” on page 25

“barrier growth limit” on page 26

“barrier column nonzeros” on page 23

“barrier iteration limit” on page 26

“barrier maximum correction limit” on page 27

“barrier objective range” on page 28

“convergence tolerance for LP and QP problems” on page 25

“convergence tolerance for QC problems” on page 29

“memory reduction switch” on page 71

“numerical precision emphasis” on page 89

MIP
Here are topics of interest to users of the MIP optimizer.

The parameters controlling MIP behavior are accessible through the following
topics:
v “MIP general” on page 9
v “MIP strategies” on page 9
v “MIP cuts” on page 10
v “MIP tolerances” on page 11
v “MIP limits” on page 11
v Distributed MIP

8 CPLEX Parameters Reference

MIP general
Here are links to parameters of general interest to users of the MIP optimizer.

“advanced start switch” on page 17

“MIP emphasis switch” on page 74

“MIP repeat presolve switch” on page 119

“relaxed LP presolve switch” on page 117

“indefinite MIQP switch” on page 110

“solution target type” on page 135

“bound strengthening switch” on page 31

“memory reduction switch” on page 71

“numerical precision emphasis” on page 89

“MIP callback switch between original model and reduced, presolved model” on
page 71

“MIP node log display information” on page 72

“MIP node log interval” on page 75

“node storage file switch” on page 87

MIP strategies
Here are links to parameters controlling MIP strategies.

“algorithm for initial MIP relaxation” on page 123

“MIP subproblem algorithm” on page 86

“MIP variable selection strategy” on page 148

“MIP strategy best bound interval” on page 30

“MIP branching direction” on page 32

“backtracking tolerance” on page 32

“MIP dive strategy” on page 47

“MIP heuristic frequency” on page 64

“local branching heuristic” on page 68

“MIP priority order switch” on page 78

“MIP priority order generation” on page 78

Chapter 2. Topical list of parameters 9

 “MIP node selection strategy” on page 89

“node presolve switch” on page 107

“MIP probing level” on page 108

“RINS heuristic frequency” on page 119

“feasibility pump switch” on page 60

MIP cuts
Here are links to parameters controlling cuts.

“constraint aggregation limit for cut generation” on page 18

“row multiplier factor for cuts” on page 42

“MIP cliques switch” on page 35

“MIP covers switch” on page 39

“MIP disjunctive cuts switch” on page 46

“MIP flow cover cuts switch” on page 59

“MIP flow path cut switch” on page 60

“MIP Gomory fractional cuts switch” on page 62

“MIP GUB cuts switch” on page 63

“MIP globally valid implied bound cuts switch” on page 65

MIP locally valid implied bound cuts switch

“MIP lift-and-project cuts switch” on page 69

“MCF cut switch” on page 70

“MIP MIR (mixed integer rounding) cut switch” on page 81

“MIP zero-half cuts switch” on page 152

“pass limit for generating Gomory fractional cuts” on page 63

“candidate limit for generating Gomory fractional cuts” on page 61

“type of cut limit” on page 48

“number of cutting plane passes” on page 42

10 CPLEX Parameters Reference

MIP tolerances
Here are links to parameters setting MIP tolerances.

“backtracking tolerance” on page 32

“lower cutoff” on page 41

“upper cutoff” on page 43

“absolute objective difference cutoff” on page 91

“relative objective difference cutoff” on page 117

“absolute MIP gap tolerance” on page 49

“relative MIP gap tolerance” on page 50

“integrality tolerance” on page 51

“relaxation for FeasOpt” on page 55

MIP limits
Here are links to parameters setting MIP limits.

“MIP integer solution-file switch and prefix” on page 66

“pass limit for generating Gomory fractional cuts” on page 63

“candidate limit for generating Gomory fractional cuts” on page 61

“constraint aggregation limit for cut generation” on page 18

“type of cut limit” on page 48

“row multiplier factor for cuts” on page 42

“number of cutting plane passes” on page 42

“MIP node limit” on page 88

“time spent probing” on page 109

“number of attempts to repair infeasible MIP start” on page 118

“MIP strong branching candidate list limit” on page 136

“MIP strong branching iterations limit” on page 137

“limit on nodes explored when a subMIP is being solved” on page 138

“tree memory limit” on page 142

“time spent in ramp up during distributed parallel optimization” on page 112

Chapter 2. Topical list of parameters 11

 “deterministic time spent in ramp up during distributed parallel optimization” on
page 111

“ramp up duration” on page 113

Solution polishing
Here are links to parameters controlling starting conditions for solution polishing

“absolute MIP gap before starting to polish a feasible solution” on page 97

“relative MIP gap before starting to polish a feasible solution” on page 98

“MIP integer solutions to find before starting to polish a feasible solution” on page
99

“nodes to process before starting to polish a feasible solution” on page 100

“time before starting to polish a feasible solution” on page 101

Solution pool
Here are links to parameters controlling the solution pool.

“solution pool intensity” on page 132

“solution pool replacement strategy” on page 134

“maximum number of solutions generated for solution pool by populate” on page

102

“maximum number of solutions kept in solution pool” on page 131

“absolute gap for solution pool” on page 130

“relative gap for solution pool” on page 132

Network
Here are links to parameters of interest to users of the network flow optimizer.

“network optimality tolerance” on page 83

“network primal feasibility tolerance” on page 84

“simplex network extraction level” on page 84

“network simplex iteration limit” on page 85

“network simplex pricing algorithm” on page 85

“network logging display switch” on page 82

12 CPLEX Parameters Reference

Parallel optimization
Here are links to parameters controlling parallel optimization.

“parallel mode switch” on page 93

“global default thread count” on page 139

Sifting
Here are links to parameters of interest to users of the sifting optimizer.

“sifting subproblem algorithm” on page 127

“sifting information display” on page 127

“upper limit on sifting iterations” on page 128

Preprocessing: aggregator, presolver

Here are links to parameters related to preprocessing.

“symmetry breaking” on page 139

“preprocessing aggregator fill” on page 18

“preprocessing aggregator application limit” on page 19

“bound strengthening switch” on page 31

“coefficient reduction setting” on page 37

“dependency switch” on page 44

“presolve dual setting” on page 104

“presolve switch” on page 105

“linear reduction switch” on page 105

“limit on the number of presolve passes made” on page 106

“node presolve switch” on page 107

“relaxed LP presolve switch” on page 117

“MIP repeat presolve switch” on page 119

“primal and dual reduction type” on page 115

Chapter 2. Topical list of parameters 13

Tolerances
Here are links to parameters setting tolerances.

“convergence tolerance for LP and QP problems” on page 25

“convergence tolerance for QC problems” on page 29

“backtracking tolerance” on page 32

“lower cutoff” on page 41

“upper cutoff” on page 43

“absolute MIP gap tolerance” on page 49

“absolute MIP gap before starting to polish a feasible solution” on page 97

“relative MIP gap tolerance” on page 50

“relative MIP gap before starting to polish a feasible solution” on page 98

“integrality tolerance” on page 51

“epsilon (degree of tolerance) used in linearization” on page 52

“Markowitz tolerance” on page 53

“optimality tolerance” on page 53

“network optimality tolerance” on page 83

“feasibility tolerance” on page 56

“relaxation for FeasOpt” on page 55

“absolute objective difference cutoff” on page 91

“relative objective difference cutoff” on page 117

“perturbation constant” on page 54

“absolute gap for solution pool” on page 130

“relative gap for solution pool” on page 132

Limits
Here are links to parameters setting general limits.

“memory available for working storage” on page 150

“global default thread count” on page 139

“optimizer time limit in seconds” on page 141

14 CPLEX Parameters Reference

 “variable (column) read limit” on page 38

“constraint (row) read limit” on page 125

“nonzero element read limit” on page 90

“QP Q-matrix nonzero read limit” on page 111

Display and output

Here are links to parameters controlling screen displays, logs, and files.

“messages to screen switch” on page 126

“tuning information display” on page 144

“barrier display information” on page 25

“simplex iteration information display” on page 129

“sifting information display” on page 127

“MIP node log display information” on page 72

“MIP node log interval” on page 75

“network logging display switch” on page 82

“clock type for computation time” on page 35

“conflict information display” on page 38

“data consistency checking switch” on page 44

“precision of numerical output in MPS and REW file formats” on page 82

“directory for working files” on page 149

“write level for MST, SOL files” on page 151

Chapter 2. Topical list of parameters 15

16 CPLEX Parameters Reference
Chapter 3. List of CPLEX parameters
CPLEX parameters, documented here alphabetically by name in the Callable
Library (C API), are available in the C++, Java, .NET, and Python APIs, as well as
in the Interactive Optimizer, the MathWorks MATLAB connector, and the Excel
Connector.

advanced start switch

If set to 1 or 2, this parameter specifies that CPLEX should use advanced starting
information when it initiates optimization.

Purpose

Advanced start switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Advance CPX_PARAM_ADVIND (int)
C++ IloCplex::Param::Advance AdvInd (int)
Java IloCplex.Param.Advance AdvInd (int)
.NET Cplex.Param.Advance AdvInd (int)
OPL advind
Python parameters.advance advance
MATLAB Cplex.Param.advance advance
Interactive advance advance
Identifier 1001 1001

Description

If set to 1 or 2, this parameter specifies that CPLEX should use advanced starting
information when optimization is initiated.

For MIP models, setting 1 (one) will cause CPLEX to continue with a partially
explored MIP tree if one is available. If tree exploration has not yet begun, setting
1 (one) specifies that CPLEX should use a loaded MIP start, if available. Setting 2
retains the current incumbent (if there is one), re-applies presolve, and starts a new
search from a new root.

Setting 2 is useful for continuous models. Consequently, it can be particularly

useful for solving fixed MIP models, where a start vector but no corresponding
basis is available.

For continuous models solved with simplex, setting 1 (one) will use the currently
loaded basis. If a basis is available only for the original, unpresolved model, or if
CPLEX has a start vector rather than a simplex basis, then the simplex algorithm
will proceed on the unpresolved model. With setting 2, CPLEX will first perform
presolve on the model and on the basis or start vector, and then proceed with
optimization on the presolved problem.

For continuous models solved with the barrier algorithm, settings 1 or 2 will
continue simplex optimization from the last available barrier iterate.

© Copyright IBM Corp. 1987, 2014 17

 Tip:

If you optimize your MIP model, then change a tolerance (such as “upper cutoff”
on page 43, “lower cutoff” on page 41, “integrality tolerance” on page 51), and
then re-optimize, the change in tolerance may not be taken into account in certain
circumstances, depending on characteristics of your model and parameter settings.
In order for CPLEX to take into account your change in tolerance, you must restart
the second optimization from the beginning. That is, you must set
CPX_PARAM_ADVIND, AdvInd to 0 (zero).
Table 1. Values.
Value Meaning
0 Do not use advanced start information
1 Use an advanced basis supplied by the user;
default
2 Crush an advanced basis or starting vector
supplied by the user

constraint aggregation limit for cut generation

Limits the number of constraints that can be aggregated for generating flow cover
and mixed integer rounding (MIR) cuts.

Purpose

Constraint aggregation limit for cut generation

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_AggForCut CPX_PARAM_AGGCUTLIM
C++ IloCplex::Param::MIP::Limits::AggForCut AggCutLim (int)
Java IloCplex.Param.MIP.Limits.AggForCut AggCutLim (int)
.NET Cplex.Param.MIP.Limits.AggForCut AggCutLim (int)
OPL aggcutlim
Python parameters.mip.limits.aggforcut mip.limits.aggforcut
MATLAB Cplex.Param.mip.limits.aggforcut mip.limits.aggforcut
Interactive mip limits aggforcut mip limits aggforcut
Identifier 2054 2054

Description

Limits the number of constraints that can be aggregated for generating flow cover
and mixed integer rounding (MIR) cuts.

Values

Any nonnegative integer; default: 3

preprocessing aggregator fill

Limits variable substitutions by the aggregator.

18 CPLEX Parameters Reference

 Purpose

Preprocessing aggregator fill

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Fill CPX_PARAM_AGGFILL
C++ IloCplex::Param::Preprocessing::Fill AggFill (CPXLONG)
Java IloCplex.Param.Preprocessing.Fill AggFill (CPXLONG)
.NET Cplex.Param.Preprocessing.Fill AggFill (CPXLONG)
OPL aggfill
Python parameters.preprocessing.fill preprocessing.fill
MATLAB Cplex.Param.preprocessing.fill preprocessing.fill
Interactive preprocessing fill preprocessing fill
Identifier 1002 1002

Description

Limits number of variable substitutions by the aggregator. If the net result of a

single substitution is more nonzeros than this value, the substitution is not made.

Tip:

The symbols CPXINT and CPXLONG declare a type of integer appropriate to your
specification of a relatively small or large model by means of the symbol
CPX_APIMODEL.

Values

Any nonnegative integer; default: 10

preprocessing aggregator application limit

Invokes the aggregator to use substitution where possible to reduce the number of
rows and columns before the problem is solved.

Purpose

Preprocessing aggregator application limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Aggregator CPX_PARAM_AGGIND (int)
C++ IloCplex::Param::Preprocessing::Aggregator AggInd (int)
Java IloCplex.Param.Preprocessing.Aggregator AggInd (int)
.NET Cplex.Param.Preprocessing.Aggregator AggInd (int)
OPL aggind
Python parameters.preprocessing.aggregator preprocessing.aggregator
MATLAB Cplex.Param.preprocessing.aggregator preprocessing.aggregator
Interactive preprocessing aggregator preprocessing aggregator
Identifier 1003 1003

Chapter 3. List of CPLEX parameters 19

 Description

Invokes the aggregator to use substitution where possible to reduce the number of
rows and columns before the problem is solved. If set to a positive value, the
aggregator is applied the specified number of times or until no more reductions
are possible.
Table 2. Values
Value Meaning
-1 Automatic (1 for LP, infinite for MIP)
default
0 Do not use any aggregator
Any positive integer Number of times to apply aggregator

API string encoding switch

API string encoding switch

Purpose

API string encoding switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Read_APIEncoding CPX_PARAM_APIENCODING
C++ IloCplex::Param::Read::APIEncoding APIEncoding
Java not in this API not in this API
.NET not in this API not in this API
OPL not in this API not in this API
Python parameters.read.apiencoding read.apiencoding
MATLAB Cplex.Param.read.apiencoding read.apiencoding
Interactive not in this component not in this component
Identifier 1130 1130

Description

Specifies which encoding (also known as the code page) that CPLEX uses for
strings passed to and from routines of the Callable Library (C API) or methods of
the C++ application programming interface (API) or methods of the Python API or
methods of the CPLEX connector for MATLAB. That is, this parameter tells CPLEX
which characters to expect as input and how to represent as output such strings as
the name of a model, of a variable, of a constraint. If, for example, your C or C++
application uses an accent in the name of a model, an umlaut in the name of a
variable, or a Chinese character for the name of a constraint, then this parameter is
of interest to you.

Note:

This parameter has no effect on IBM CPLEX Optimizer for z/OS, where only
EBCDIC IBM-1047 encoding is available.

Which features does this parameter govern?

20 CPLEX Parameters Reference

In the Callable Library (C API), this parameter specifies the encoding in which
CPLEX passes a string to a function destination added to a channel by means of
the routine CPXaddfuncdest.

In the C++ API, this parameter specifies the encoding of streams accessed by the
methods setWarning and setOut. CPLEX also encodes exceptions according to the
value of this parameter.

In the Python API, this parameter specifies the encoding of streams accessed by
methods such as Cplex.set_log_stream, Cplex.set_warning_stream, or
Cplex.set_error_stream.

In the CPLEX for MATLAB APIs, the default value is the empty string ("").

Tip:

This parameter is not relevant to the Java API because CPLEX respects the
encoding-conventions of Java. In fact, CPLEX relies on the encoding UTF-8 in Java
applications. For a brief description of the advantages of UTF-8, see the topic
Selecting an encoding in the CPLEX User’s Manual.

Which values does this parameter accept?

This parameter accepts a string specifying the user’s choice of encoding, such as
UTF-8, ISO-8859-1, US-ASCII, and so forth.

An asterisk (*) is an acceptable value in some situations. This wildcard instructs

CPLEX to use the default converter according to the protocols and standards of
International Components for Unicode (ICU). In other words, CPLEX calls ICU
with a NULL encoding. When ICU is called with a NULL encoding (as in this
case), ICU takes the name of the encoding from the terminal where the application
started.

Other acceptable values of this parameter depend on the API.

v In the Callable Library (C API), this parameter accepts any string that is the
name of a valid code page. For example, UTF-8 is a multi-byte encoding that is
an acceptable value for this parameter; it encompasses the ASCII character set; it
does not allow valid characters to include a NULL byte. If you use another
multi-byte encoding, such as UTF-32 or UTF-16, for example, be sure to specify
the encoding fully by also including the byte order, like this: UTF-32BE or
UTF-32LE.
For a complete list of valid strings that are the name of an encoding (that is, the
name of a code page), consult the web site of a standards organization such as:
– A brief introduction to code pages
– ICU: International Components for Unicode
– International Components for Unicode at IBM
v In the C++ API, the value of this parameter must be the name of an encoding
that is a superset of ASCII. For example, ASCII is a subset of the encoding or
code page UTF-8, so UTF-8 is an acceptable value for this parameter. Likewise,
ASCII is a subset of ISO-8859-1, also known as Latin-1, so ISO-8859-1 is an
acceptable value for this parameter. However, the code page UTF-16 is not
acceptable, nor is UTF-32 because both allow valid characters that contain a
NULL byte.

Chapter 3. List of CPLEX parameters 21

 v In the Python API, the value of this parameter cannot be the name of an
encoding that allows a NULL byte within a valid character. In practice, this
stricture means that UTF-16 and UTF-32 are not acceptable values of this
parameter. Further restrictions depend on the version of Python that you are
using. Early versions of Python accepted a limited range of code pages. Recent
versions of Python accept a greater variety of code pages. For more information
about those choices dependent on Python, consult the documentation of your
Python installation and observe the stricture documented here about avoiding an
encoding that contains NULL bytes within the representation of a character.
v In the .NET API, the only acceptable value of this parameter is its default value
ISO-8859-1.

What is the default value of this parameter?

The default value of this parameter depends on the API.

v In the Python API of CPLEX, the default value is the empty string (" ") for
consistency with Python conventions.
v In the C API, the default value is the string ISO-8859-1 (also known as Latin-1).
v In the C++ API, the default value is the string ISO-8859-1 (also known as
Latin-1).
v In the .NET API, the only acceptable value of this parameter is its default value
ISO-8859-1.

The encoding ISO-8859-1 is a superset of the familiar ASCII encoding, so it

supports many widely used character sets. However, this default encoding cannot
represent multi-byte character sets, such as Chinese, Japanese, Korean, Vietnamese,
or Indian characters, for example. If you want to represent a character set that
requires multiple bytes per character, then a better choice for the value of this
parameter is UTF-8.

If you change the value of this parameter, you must verify that your choice is
compatible with the “file encoding switch” on page 57 (CPX_PARAM_FILEENCODING,
FileEncoding). In fact, the encoding or code page specified by the API encoding
parameter must be a subset of the encoding or code page specified by the file
encoding parameter. For example, if the value of the file encoding parameter is
UTF-8, then US-ASCII is an acceptable value of the API encoding parameter because
US-ASCII is a subset of UTF-8. For examples of the hazards of incompatible choices
of the encoding parameters, see the topic Selecting an encoding in the CPLEX
User’s Manual.

What about errors?

In situations where CPLEX encounters a string, such as content in a file, that is not
compatible with the specified encoding, the behavior is not defined. Because of the
incompatibility, CPLEX silently converts the string to an inappropriate character of
the specified encoding, or CPLEX raises the error CPXERR_ENCODING_CONVERSION. For
an example of why such unpredictable behavior occurs, see the topic Selecting an
encoding in the CPLEX User’s Manual.

Values

valid string for the name of an encoding (code page) that is a superset of ASCII;
default: ISO-8859-1 or empty string.

See also

22 CPLEX Parameters Reference

 “file encoding switch” on page 57

barrier algorithm
The default setting 0 uses the "infeasibility - estimate start" algorithm (setting 1)
when solving subproblems in a MIP problem, and the standard barrier algorithm
(setting 3) in other cases.

Purpose

Barrier algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM Barrier Algorithm CPX_PARAM_BARALG
C++ IloCplex::Param::Barrier::Algorithm BarAlg (int)
Java IloCplex.Param.Barrier.Algorithm BarAlg (int)
.NET Cplex.Param.Barrier.Algorithm BarAlg (int)
OPL baralg
Python parameters.barrier.algorithm barrier.algorithm
MATLAB Cplex.Param.barrier.algorithm barrier.algorithm
Interactive barrier algorithm barrier algorithm
Identifier 3007 3007

Description

The default setting 0 uses the "infeasibility - estimate start" algorithm (setting 1)
when solving subproblems in a MIP problem, and the standard barrier algorithm
(setting 3) in other cases. The standard barrier algorithm is almost always fastest.
However, on problems that are primal or dual infeasible (common for MIP
subproblems), the standard algorithm may not work as well as the alternatives.
The two alternative algorithms (settings 1 and 2) may eliminate numerical
difficulties related to infeasibility, but are generally slower.

Value Meaning
0 Default setting
1 Infeasibility-estimate start
2 Infeasibility-constant start
3 Standard barrier

barrier column nonzeros

Used in the recognition of dense columns.

Purpose

Barrier column nonzeros

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_ColNonzeros CPX_PARAM_BARCOLNZ
C++ IloCplex::Param::Barrier::ColNonzeros BarColNz (int)
Java IloCplex.Param.Barrier.ColNonzeros BarColNz (int)
.NET Cplex.Param.Barrier.ColNonzeros BarColNz (int)
OPL barcolnz

Chapter 3. List of CPLEX parameters 23

 API Parameter Name Name prior to V12.6.0
Python parameters.barrier.colnonzeros barrier.colnonzeros
MATLAB Cplex.Param.barrier.colnonzeros barrier.colnonzeros
Interactive barrier colnonzeros barrier colnonzeros
Identifier 3009 3009

Description

Used in the recognition of dense columns. If columns in the presolved and

aggregated problem exist with more entries than this value, such columns are
considered dense and are treated specially by the CPLEX barrier optimizer to
reduce their effect.

Value Meaning
0 Dynamically calculated; default
Any positive integer Number of nonzero entries that make a
column dense

barrier crossover algorithm

Decides which, if any, crossover is performed at the end of a barrier optimization.

Purpose

Barrier crossover algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_Crossover CPX_PARAM_BARCROSSALG
C++ IloCplex::Param::Barrier::Crossover BarCrossAlg (int)
Java IloCplex.Param.Barrier.Crossover BarCrossAlg (int)
.NET Cplex.Param.Barrier.Crossover BarCrossAlg (int)
OPL barcrossalg
Python parameters.barrier.crossover barrier.crossover
MATLAB Cplex.Param.barrier.crossover barrier.crossover
Interactive barrier crossover barrier crossover
Identifier 3018 3018

Description

Decides which, if any, crossover is performed at the end of a barrier optimization.

This parameter applies when CPLEX uses the Barrier Optimizer to solve an LP or
QP problem, or when it is used to solve the continuous relaxation of an MILP or
MIQP at a node in a MIP.

By default, CPLEX does not invoke crossover on a QP problem. On an LP problem,

it invokes primal and dual crossover in parallel when multiple threads are
available.

Value Meaning
-1 No crossover
0 Automatic: let CPLEX choose; default
1 Primal crossover

24 CPLEX Parameters Reference

 Value Meaning
2 Dual crossover

barrier display information

Sets the level of barrier progress information to be displayed.

Purpose

Barrier display information

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_Display CPX_PARAM_BARDISPLAY
C++ IloCplex::Param::Barrier::Display BarDisplay (int)
Java IloCplex.Param.Barrier.Display BarDisplay (int)
.NET Cplex.Param.Barrier.Display BarDisplay (int)
OPL bardisplay
Python parameters.barrier.display barrier.display
MATLAB Cplex.Param.barrier.display barrier.display
Interactive barrier display barrier display
Identifier 3010 3010

Description

Sets the level of barrier progress information to be displayed.

Value Meaning
0 No progress information
1 Normal setup and iteration information;
default
2 Diagnostic information

convergence tolerance for LP and QP problems

Sets the tolerance on complementarity for convergence.

Purpose

Convergence tolerance for LP and QP problems

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_ConvergeTol CPX_PARAM_BAREPCOMP
C++ IloCplex::Param::Barrier::ConvergeTol BarEpComp (double)
Java IloCplex.Param.Barrier.ConvergeTol BarEpComp (double)
.NET Cplex.Param.Barrier.ConvergeTol BarEpComp (double)
OPL barepcomp
Python parameters.barrier.convergetol barrier.convergetol
MATLAB Cplex.Param.barrier.convergetol barrier.convergetol
Interactive barrier convergetol barrier convergetol
Identifier 3002 3002

Chapter 3. List of CPLEX parameters 25

 Description

Sets the tolerance on complementarity for convergence. The barrier algorithm

terminates with an optimal solution if the relative complementarity is smaller than
this value.

Changing this tolerance to a smaller value may result in greater numerical

precision of the solution, but also increases the chance of failure to converge in the
algorithm and consequently may result in no solution at all. Therefore, caution is
advised in deviating from the default setting.

Values

Any positive number greater than or equal to 1e-12; default: 1e-8.

See also

For problems with quadratic constraints (QCP), see “convergence tolerance for QC
problems” on page 29

barrier growth limit

Used to detect unbounded optimal faces.

Purpose

Barrier growth limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_Limits_Growth CPX_PARAM_BARGROWTH
C++ IloCplex::Param::Barrier::Limits::Growth BarGrowth (double)
Java IloCplex.Param.Barrier.Limits.Growth BarGrowth (double)
.NET Cplex.Param.Barrier.Limits.Growth BarGrowth (double)
OPL bargrowth
Python parameters.barrier.limits.growth barrier.limit.growth
MATLAB Cplex.Param.barrier.limits.growth barrier.limits.growth
Interactive barrier limits growth barrier limits growth
Identifier 3003 3003

Description

Used to detect unbounded optimal faces. At higher values, the barrier algorithm is
less likely to conclude that the problem has an unbounded optimal face, but more
likely to have numerical difficulties if the problem has an unbounded face.

Values

1.0 or greater; default: 1e12.

barrier iteration limit

Sets the number of barrier iterations before termination.

26 CPLEX Parameters Reference

 Purpose

Barrier iteration limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_Limits_Iteration CPX_PARAM_BARITLIM
C++ IloCplex::Param::Barrier::Limits::Iteration BarItLim (long)
Java IloCplex.Param.Barrier.Limits.Iteration BarItLim (long)
.NET Cplex.Param.Barrier.Limits.Iteration BarItLim (long)
OPL baritlim
Python parameters.barrier.limits.iteration barrier.limit.iteration
MATLAB Cplex.Param.barrier.limits.iteration barrier.limits.iteration
Interactive barrier limits iteration barrier limits iteration
Identifier 3012 3012

Description

Sets the number of barrier iterations before termination. When this parameter is set
to 0 (zero), no barrier iterations occur, but problem setup occurs and information
about the setup is displayed (such as Cholesky factor statistics).
Table 3. Values
Value Meaning
0 No barrier iterations
9223372036800000000 default
Any positive integer Number of barrier iterations before
termination

barrier maximum correction limit

Sets the maximum number of centering corrections done on each iteration.

Purpose
Barrier maximum correction limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_Limits_Corrections CPX_PARAM_BARMAXCOR
C++ IloCplex::Param::Barrier::Limits::Corrections BarMaxCor (long)
Java IloCplex.Param.Barrier.Limits.Corrections BarMaxCor (long)
.NET Cplex.Param.Barrier.Limits.Corrections BarMaxCor (long)
OPL barmaxcor
Python parameters.barrier.limits.corrections barrier.limit.corrections
MATLAB Cplex.Param.barrier.limits.corrections barrier.limits.corrections
Interactive barrier limits corrections barrier limits corrections
Identifier 3013 3013

Description

Sets the maximum number of centering corrections done on each iteration. An

explicit value greater than 0 (zero) may improve the numerical performance of the
algorithm at the expense of computation time.

Chapter 3. List of CPLEX parameters 27

 Table 4. Values
Value Meaning
-1 Automatic; let CPLEX choose; default
0 None
Any positive integer Maximum number of centering corrections
per iteration

barrier objective range

Sets the maximum absolute value of the objective function.

Purpose

Barrier objective range

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_Limits_ObjRange CPX_PARAM_BAROBJRNG
C++ IloCplex::Param::Barrier::Limits::ObjRange BarObjRng (double)
Java IloCplex.Param.Barrier.Limits.ObjRange BarObjRng (double)
.NET Cplex.Param.Barrier.Limits.ObjRange BarObjRng (double)
OPL barobjrng
Python parameters.barrier.limits.objrange barrier.limit.objrange
MATLAB Cplex.Param.barrier.limits.objrange barrier.limits.objrange
Interactive barrier limits objrange barrier limits objrange
Identifier 3004 3004

Description

Sets the maximum absolute value of the objective function. The barrier algorithm
looks at this limit to detect unbounded problems.

Values

Any nonnegative number; default: 1e20

barrier ordering algorithm

Sets the algorithm to be used to permute the rows of the constraint matrix in order
to reduce fill in the Cholesky factor.

Purpose

Barrier ordering algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_Ordering CPX_PARAM_BARORDER
C++ IloCplex::Param::Barrier::Ordering BarOrder (int)
Java IloCplex.Param.Barrier.Ordering BarOrder (int)
.NET Cplex.Param.Barrier.Ordering BarOrder (int)
OPL barorder
Python parameters.barrier.ordering barrier.ordering
MATLAB Cplex.Param.barrier.ordering barrier.ordering
Interactive barrier ordering barrier ordering

28 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Identifier 3014 3014

Description
Sets the algorithm to be used to permute the rows of the constraint matrix in order
to reduce fill in the Cholesky factor.
Table 5. Values
Value Meaning
0 Automatic: let CPLEX choose; default
1 Approximate minimum degree (AMD)
2 Approximate minimum fill (AMF)
3 Nested dissection (ND)

convergence tolerance for QC problems

Sets the tolerance on complementarity for convergence in quadratically constrained
problems (QCPs).

Purpose

Convergence tolerance for quadratically constrained problems

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_QCPConvergeTol CPX_PARAM_BARQCPEPCOMP
C++ IloCplex::Param::Barrier::QCPConvergeTol BarQCPEpComp (double)
Java IloCplex.Param.Barrier.QCPConvergeTol BarQCPEpComp (double)
.NET Cplex.Param.Barrier.QCPConvergeTol BarQCPEpComp (double)
OPL barqcpepcomp
Python parameters.barrier.qcpconvergetol barrier.qcpconvergetol
MATLAB Cplex.Param.barrier.qcpconvergetol barrier.qcpconvergetol
Interactive barrier qcpconvergetol barrier qcpconvergetol
Identifier 3020 3020

Description

Sets the tolerance on complementarity for convergence in quadratically constrained

problems (QCPs). The barrier algorithm terminates with an optimal solution if the
relative complementarity is smaller than this value.

Changing this tolerance to a smaller value may result in greater numerical

precision of the solution, but also increases the chance of a convergence failure in
the algorithm and consequently may result in no solution at all. Therefore, caution
is advised in deviating from the default setting.

Values

Any positive number greater than or equal to 1e-12; default: 1e-7.

For LPs and for QPs (that is, when all the constraints are linear) see “convergence
tolerance for LP and QP problems” on page 25.

Chapter 3. List of CPLEX parameters 29

barrier starting point algorithm
Sets the algorithm to be used to compute the initial starting point for the barrier
optimizer.

Purpose

Barrier starting point algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Barrier_StartAlg CPX_PARAM_BARSTARTALG
C++ IloCplex::Param::Barrier::StartAlg BarStartAlg (int)
Java IloCplex.Param.Barrier.StartAlg BarStartAlg (int)
.NET Cplex.Param.Barrier.StartAlg BarStartAlg (int)
OPL barstartalg
Python parameters.barrier.startalg barrier.startalg
MATLAB Cplex.Param.barrier.startalg barrier.startalg
Interactive barrier startalg barrier startalg
Identifier 3017 3017

Description

Sets the algorithm to be used to compute the initial starting point for the barrier
optimizer.

Value Meaning
1 Dual is 0 (zero); default
2 Estimate dual
3 Average of primal estimate, dual 0 (zero)
4 Average of primal estimate, estimate dual

MIP strategy best bound interval

Sets the best bound interval for MIP strategy.

Purpose

MIP strategy best bound interval

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_BBInterval CPX_PARAM_BBINTERVAL
C++ IloCplex::Param::MIP::Strategy::BBInterval BBInterval (long)
Java IloCplex.Param.MIP.Strategy.BBInterval BBInterval (long)
.NET Cplex.Param.MIP.Strategy.BBInterval BBInterval (long)
OPL bbinterval
Python parameters.mip.strategy.bbinterval mip.strategy.bbinterval
MATLAB Cplex.Param.mip.strategy.bbinterval mip.strategy.bbinterval
Interactive mip strategy bbinterval mip strategy bbinterval
Identifier 2039 2039

Description

Sets the best bound interval for MIP strategy.

30 CPLEX Parameters Reference

 When you set this parameter to best estimate node selection, the best bound
interval is the interval at which the best bound node, instead of the best estimate
node, is selected from the tree. A best bound interval of 0 (zero) means “never
select the best bound node.” A best bound interval of 1 (one) means “always select
the best bound node,” and is thus equivalent to node select 1 (one).

Higher values of this parameter mean that the best bound node will be selected
less frequently; experience has shown it to be beneficial to select the best bound
node occasionally, and therefore the default value of this parameter is 7.
Table 6. Values
Value Meaning
0 Never select best bound node; always select
best estimate
1 Always select best bound node
7 Select best bound node occasionally; default
Any positive integer Select best bound node less frequently than
best estimate node

See also

“MIP node selection strategy” on page 89

bound strengthening switch

Decides whether to apply bound strengthening in mixed integer programs (MIPs).

Purpose

Bound strengthening switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_BoundStrength CPX_PARAM_BNDSTRENIND
C++ IloCplex::Param::Preprocessing::BoundStrength BndStrenInd (int)
Java IloCplex.Param.Preprocessing.BoundStrength BndStrenInd (int)
.NET Cplex.Param.Preprocessing.BoundStrength BndStrenInd (int)
OPL bndstrenind
Python parameters.preprocessing.boundstrength preprocessing.boundstrength
MATLAB Cplex.Param.preprocessing.boundstrength preprocessing.boundstrength
Interactive preprocessing boundstrength preprocessing boundstrength
Identifier 2029 2029

Description

Decides whether to apply bound strengthening in mixed integer programs (MIPs).

Bound strengthening tightens the bounds on variables, perhaps to the point where
the variable can be fixed and thus removed from consideration during branch and
cut.

Tip:

Strengthening means to replace one row of a model with another such that an
integer vector is feasible in the new row if and only if the integer vector was

Chapter 3. List of CPLEX parameters 31

 feasible in the original row. Strengthening improves the LP relaxation of the row
by finding a dominating row. In other words, the LP region defined by the
strengthened row plus the bounds on the variables will be strictly contained in the
LP region defined by the original row plus bounds on the variables.

Value Meaning
-1 Automatic: let CPLEX choose; default
0 Do not apply bound strengthening
1 Apply bound strengthening

MIP branching direction

Decides which branch, the up or the down branch, should be taken first at each
node.

Purpose

MIP branching direction

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_Branch CPX_PARAM_BRDIR
C++ IloCplex::Param::MIP::Strategy::Branch BrDir (int)
Java IloCplex.Param.MIP.Strategy.Branch BrDir (int)
.NET Cplex.Param.MIP.Strategy.Branch BrDir (int)
OPL brdir
Python parameters.mip.strategy.branch mip.strategy.branch
MATLAB Cplex.Param.mip.strategy.branch mip.strategy.branch
Interactive mip strategy branch mip strategy branch
Identifier 2001 2001

Description

Decides which branch, the up or the down branch, should be taken first at each
node.

Value Symbol Meaning

-1 CPX_BRDIR_DOWN Down branch selected first
0 CPX_BRDIR_AUTO Automatic: let CPLEX
choose; default
1 CPX_BRDIR_UP Up branch selected first

backtracking tolerance
Controls how often backtracking is done during the branching process.

Purpose

Backtracking tolerance

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_Backtrack CPX_PARAM_BTTOL

32 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
C++ IloCplex::Param::MIP::Strategy::Backtrack BtTol (double)
Java IloCplex.Param.MIP.Strategy.Backtrack BtTol (double)
.NET Cplex.Param.MIP.Strategy.Backtrack BtTol (double)
OPL bttol
Python parameters.mip.strategy.backtrack mip.strategy.backtrack
MATLAB Cplex.Param.mip.strategy.backtrack mip.strategy.backtrack
Interactive mip strategy backtrack mip strategy backtrack
Identifier 2002 2002

Description

Controls how often backtracking is done during the branching process. The
decision when to backtrack depends on three values that change during the course
of the optimization:
v the objective function value of the best integer feasible solution (incumbent);
v the best remaining objective function value of any unexplored node (best node);
v the objective function value of the most recently solved node (current objective).

If a cutoff tolerance (“upper cutoff” on page 43 or “lower cutoff” on page 41) has
been set by the user, then that value is used as the incumbent until an integer
feasible solution is found.

The target gap is defined to be the absolute value of the difference between the
incumbent and the best node, multiplied by this backtracking parameter. CPLEX
does not backtrack until the absolute value of the difference between the objective
of the current node and the best node is at least as large as the target gap.

Low values of this backtracking parameter thus tend to increase the amount of
backtracking, which makes the search process more of a pure best-bound search.
Higher parameter values tend to decrease backtracking, making the search more of
a pure depth-first search.

The backtracking value has effect only after an integer feasible solution is found or
when a cutoff has been specified. Note that this backtracking value merely permits
backtracking but does not force it; CPLEX may choose to continue searching a limb
of the tree if that limb seems a promising candidate for finding an integer feasible
solution.

Values

Any number from 0.0 to 1.0; default: 0.9999

See also

“upper cutoff” on page 43, “lower cutoff” on page 41

calculate QCP dual values

Instructs CPLEX to calculate the dual values of a quadratically constrained
problem

Chapter 3. List of CPLEX parameters 33

 Purpose

calculating QCP dual values

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_QCPDuals CPX_PARAM_CALCQCPDUALS
C++ IloCplex::Param::Preprocessing::QCPDuals CalcQCPDuals (int)
Java IloCplex.Param.Preprocessing.QCPDuals CalcQCPDuals (int)
.NET Cplex.Param.Preprocessing.QCPDuals CalcQCPDuals (int)
OPL not available
Python parameters.preprocessing.qcpduals preprocessing.qcpduals
MATLAB Cplex.Param.preprocessing.qcpduals preprocessing.qcpduals
Interactive preprocessing qcpduals preprocessing qcpduals
Identifier 4003 4003

Description

This parameter determines whether CPLEX preprocesses a quadratically

constrained program (QCP) so that the user can access dual values for the QCP.

If this parameter is set to 0 (zero), then CPLEX does not calculate dual values for
the QCP.

If this parameter is set to 1 (one), its default value, then CPLEX calculates dual
values for the QCP as long as the calculations do not interfere with presolve
reductions.

If this parameter is set to 2, then CPLEX calculates dual values and moreover,
CPLEX disables any presolve reductions that interfere with these dual-value
calculations.

For more information about accessing dual values of a QCP, see the topic
Accessing dual values and reduced costs of QCP solutions in the CPLEX User's
Manual.

For more information about presolve reductions, see the topic Advanced presolve
routines in the CPLEX User's Manual.

Values
Table 7. Values.
Value Meaning Symbol in Callable Library (C Symbol in C++, Java, Symbol in Python
API) .NET APIs API
0 Do not calculate dual CPX_QCPDUALS_NO QCPDualsNo no
values for the QCP
1 Calculate dual values CPX_QCPDUALS_IFPOSSIBLE QCPDualsIfPossible if_possible
for the QCP as long
as the calculations do
not interfere with
presolve reductions.
default

34 CPLEX Parameters Reference

Table 7. Values (continued).
Value Meaning Symbol in Callable Library (C Symbol in C++, Java, Symbol in Python
API) .NET APIs API
2 Calculate dual values CPX_QCPDUALS_FORCE QCPDualsForce force
and disable any
presolve reductions
that interfere with
these calculations.

MIP cliques switch

Decides whether or not clique cuts should be generated for the problem.

Purpose

MIP cliques switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_Cliques CPX_PARAM_CLIQUES
C++ IloCplex::Param::MIP::Cuts::Cliques Cliques (int)
Java IloCplex.Param.MIP.Cuts.Cliques Cliques (int)
.NET Cplex.Param.MIP.Cuts.Cliques Cliques (int)
OPL cliques
Python parameters.mip.cuts.cliques mip.cuts.cliques
MATLAB Cplex.Param.mip.cuts.cliques mip.cuts.cliques
Interactive mip cuts cliques mip cuts cliques
Identifier 2003 2003

Description

Decides whether or not clique cuts should be generated for the problem. Setting
the value to 0 (zero), the default, specifies that the attempt to generate cliques
should continue only if it seems to be helping.

For a definition of a clique cut, see the topic Clique cuts in the general topic Cuts
in the CPLEX User’s Manual. The table Parameters for controlling cuts, also in the
user’s manual, includes links to the documentation of other parameters affecting
other types of cuts.

Value Meaning
-1 Do not generate clique cuts
0 Automatic: let CPLEX choose; default
1 Generate clique cuts moderately
2 Generate clique cuts aggressively
3 Generate clique cuts very aggressively

clock type for computation time

Decides how computation times are measured for both reporting performance and
terminating optimization when a time limit has been set.

Chapter 3. List of CPLEX parameters 35

 Purpose

Clock type for computation time

API Parameter Name Name prior to V12.6.0

C CPXPARAM_ClockType CPX_PARAM_CLOCKTYPE
C++ IloCplex::Param::ClockType ClockType (int)
Java IloCplex.Param.ClockType ClockType (int)
.NET Cplex.Param.ClockType ClockType (int)
OPL clocktype
Python parameters.clocktype clocktype
MATLAB Cplex.Param.clocktype clocktype
Interactive clocktype clocktype
Identifier 1006 1006

Description

Decides how computation times are measured for both reporting performance and
terminating optimization when a time limit has been set. Small variations in
measured time on identical runs may be expected on any computer system with
any setting of this parameter.

The default setting 2 supports wall clock time.

Value Meaning
0 Automatic: let CPLEX choose
1 CPU time
2 Wall clock time (total physical time elapsed);
default

clone log in parallel optimization

Specifies whether to create clone log files during parallel optimization.

Purpose

Creates clone log files in parallel optimization

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Output_CloneLog CPX_PARAM_CLONELOG
C++ IloCplex::Param::Output::CloneLog CloneLog
Java IloCplex.Param.Output.CloneLog CloneLog
.NET Cplex.Param.Output.CloneLog CloneLog
OPL not available
Python parameters.output.clonelog output.clonelog
MATLAB Cplex.Param.output.clonelog output.clonelog
Interactive output clonelog output clonelog
Identifier

Description

Specifies whether CPLEX clones the log files of nodes during parallel or concurrent
optimization. When you use parallel or concurrent CPLEX, this feature makes it

36 CPLEX Parameters Reference

 more convenient to check node logs when you use more than one thread to solve
models. For parallel optimization on N threads, for example, turning on this
parameter creates N logs,clone[0].log through clone[N-1].log. This feature is
available only during concurrent optimization and mixed integer programming
(MIP) optimization.

Value Meaning
-1 CPLEX does not clone log files. (off)
0 Automatic: CPLEX clones log files if log file
is specified. default
1 CPLEX clones log files. (on)

coefficient reduction setting

Decides how coefficient reduction is used.

Purpose

Coefficient reduction setting

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_CoeffReduce CPX_PARAM_COEREDIND
C++ IloCplex::Param::Preprocessing::CoeffReduce CoeRedInd (int)
Java IloCplex.Param.Preprocessing.CoeffReduce CoeRedInd (int)
.NET Cplex.Param.Preprocessing.CoeffReduce CoeRedInd (int)
OPL coeredind
Python parameters.preprocessing.coeffreduce preprocessing.coeffreduce
MATLAB Cplex.Param.preprocessing.coeffreduce preprocessing.coeffreduce
Interactive preprocessing coeffreduce preprocessing coeffreduce
Identifier 2004 2004

Description

Decides how coefficient reduction is used. Coefficient reduction improves the

objective value of the initial (and subsequent) LP relaxations solved during branch
and cut by reducing the number of non-integral vertices. By default, CPLEX
applies coefficient reductions during preprocessing of a model.

The value 0 (zero) turns off coefficient reduction during preprocessing.

The value 1 (one) applies limited coefficient reduction to achieve only integral
coefficients.

The value 2, applies coefficient reduction somewhat more aggressively, reducing all
coefficients that can be reduced.

The value 3, the most aggressive setting of this parameter, applies a technique
known as tilting. Tilting can cut off additional fractional solutions in some models.
Cutting off these fractional solutions potentially yields more progress in both the
best node and best integer solution in those particular models.

Tip:

Chapter 3. List of CPLEX parameters 37

 Tilting means to replace one row of a model with another such that an integer
vector is feasible in the new row if and only if the integer vector was feasible in
the original row. In contrast to a dominating row, the LP region for a tilted row can
contain fractional points that are excluded by the LP region of the original row and
the bounds of the variables. But, the aim is to tilt the row in such a way that it
gets tighter in those areas of the LP polyhedron that are not already covered by
other constraints.

Value Meaning
-1 Automatic: let CPLEX decide; default
0 Do not use coefficient reduction
1 Reduce only to integral coefficients
2 Reduce all potential coefficients
3 Reduce aggressively with tilting

variable (column) read limit

Specifies a limit for the number of columns (variables) to read for an allocation of
memory.

Purpose

Limit the number of variables (columns) read for memory allocation

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Read_Variables CPX_PARAM_COLREADLIM
C++ IloCplex::Param::Read::Variables ColReadLim (int)
Java IloCplex.Param.Read.Variables ColReadLim (int)
.NET Cplex.Param.Read.Variables ColReadLim (int)
OPL not available
Python parameters.read.variables read.variables
MATLAB Cplex.Param.read.variables read.variables
Interactive read variables read variables
Identifier 1023 1023

Description

Specifies a limit for the number of columns (variables) to read for an allocation of
memory.

This parameter does not restrict the size of a problem. Rather, it indirectly specifies
the default amount of memory that will be pre-allocated before a problem is read
from a file. If the limit is exceeded, more memory is automatically allocated.

Values

Any integer from 0 (zero) to CPX_BIGINT; default: 60 000.

conflict information display

Decides how much information CPLEX reports when the conflict refiner is
working.

38 CPLEX Parameters Reference

 Purpose

Conflict information display

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Conflict_Display CPX_PARAM_CONFLICTDISPLAY
C++ IloCplex::Param::Conflict::Display ConflictDisplay (int)
Java IloCplex.Param.Conflict.Display ConflictDisplay (int)
.NET Cplex.Param.Conflict.Display ConflictDisplay (int)
OPL conflictdisplay
Python parameters.conflict.display conflict.display
MATLAB Cplex.Param.conflict.display conflict.display
Interactive conflict display i conflict display i
Identifier 1074 1074

Description

Decides how much information CPLEX reports when the conflict refiner is
working.
Table 8. Values
Value Meaning
0 No display
1 Summary display; default
2 Detailed display

MIP covers switch

Decides whether or not cover cuts should be generated for the problem.

Purpose

MIP covers switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_Covers CPX_PARAM_COVERS
C++ IloCplex::Param::MIP::Cuts::Covers Covers (int)
Java IloCplex.Param.MIP.Cuts.Covers Covers (int)
.NET Cplex.Param.MIP.Cuts.Covers Covers (int)
OPL covers
Python parameters.mip.cuts.covers mip.cuts.covers
MATLAB Cplex.Param.mip.cuts.covers mip.cuts.covers
Interactive mip cuts covers mip cuts covers
Identifier 2005 2005

Description

Decides whether or not cover cuts should be generated for the problem. Setting the
value to 0 (zero), the default, indicates that the attempt to generate covers should
continue only if it seems to be helping.

Chapter 3. List of CPLEX parameters 39

 For a definition of a cover cut, see the topic Cover cuts in the general topic Cuts,
in the CPLEX User’s Manual. The table Parameters for controlling cuts, also in the
user’s manual, includes links to the documentation of other parameters affecting
other types of cuts
Table 9. Values
Value Meaning
-1 Do not generate cover cuts
0 Automatic: let CPLEX choose; default
1 Generate cover cuts moderately
2 Generate cover cuts aggressively
3 Generate cover cuts very aggressively

simplex crash ordering

Decides how CPLEX orders variables relative to the objective function when
selecting an initial basis.

Purpose

Simplex crash ordering

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Crash CPX_PARAM_CRAIND
C++ IloCplex::Param::Simplex::Crash CraInd (int)
Java IloCplex.Param.Simplex.Crash CraInd (int)
.NET Cplex.Param.Simplex.Crash CraInd (int)
OPL craind
Python parameters.simplex.crash simplex.crash
MATLAB Cplex.Param.simplex.crash simplex.crash
Interactive simplex crash simplex crash
Identifier 1007 1007

Description

Decides how CPLEX orders variables relative to the objective function when
selecting an initial basis.
Table 10. Values
Value Meaning
LP Primal
-1 Alternate ways of using objective coefficients
0 Ignore objective coefficients during crash
1 Alternate ways of using objective
coefficients; default
LP Dual
-1 Aggressive starting basis
0 Aggressive starting basis
1 Default starting basis; default
QP Primal

40 CPLEX Parameters Reference

 Table 10. Values (continued)
Value Meaning
-1 Slack basis
0 Ignore Q terms and use LP solver for crash
1 Ignore objective and use LP solver for crash;
default
QP Dual
-1 Slack basis
0 Use Q terms for crash
1 Use Q terms for crash; default

lower cutoff
Sets lower cutoff tolerance.

Purpose

Lower cutoff

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Tolerances_LowerCutoff CPX_PARAM_CUTLO
C++ IloCplex::Param::MIP::Tolerances::LowerCutoff CutLo (double)
Java IloCplex.Param.MIP.Tolerances.LowerCutoff CutLo (double)
.NET Cplex.Param.MIP.Tolerances.LowerCutoff CutLo (double)
OPL cutlo
Python parameters.mip.tolerances.lowercutoff mip.tolerances.lowercutoff
MATLAB Cplex.Param.mip.tolerances.lowercutoff mip.tolerances.lowercutoff
Interactive mip tolerances lowercutoff mip tolerances lowercutoff
Identifier 2006 2006

Description

Sets the lower cutoff tolerance in a MIP. When the problem is a maximization
problem, CPLEX cuts off or discards solutions that are less than the specified cutoff
value. If the model has no solution with an objective value greater than or equal to
the cutoff value, then CPLEX declares the model infeasible. In other words, setting
the lower cutoff value c for a maximization problem is similar to adding this
constraint to the objective function of the model: obj >= c.

Tip:

This parameter is effective only when the branch and bound algorithm is invoked,
for example, in a mixed integer program (MIP). It does not have the expected
effect when branch and bound is not invoked.

This parameter is not effective with the conflict refiner nor with FeasOpt. That is,
neither of those tools can analyze an infeasibility introduced by this parameter. If
you want to analyze such a condition, add an explicit objective constraint to your
model instead before you invoke either of those tools.

Chapter 3. List of CPLEX parameters 41

 Values

Any number; default: -1e+75.

number of cutting plane passes

Sets the upper limit on the number of cutting plane passes CPLEX performs when
solving the root node of a MIP model.

Purpose

Number of cutting plane passes

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_CutPasses CPX_PARAM_CUTPASS
C++ IloCplex::Param::MIP::Limits::CutPasses CutPass (long)
Java IloCplex.Param.MIP.Limits.CutPasses CutPass (long)
.NET Cplex.Param.MIP.Limits.CutPasses CutPass (long)
OPL cutpass
Python parameters.mip.limits.cutpasses mip.limits.cutpasses
MATLAB Cplex.Param.mip.limits.cutpasses mip.limits.cutpasses
Interactive mip limits cutpasses mip limits cutpasses
Identifier 2056 2056

Description

Sets the upper limit on the number of cutting plane passes CPLEX performs when
solving the root node of a MIP model.
Table 11. Values
Value Meaning
-1 None
0 Automatic: let CPLEX choose; default
Any positive integer Number of passes to perform

row multiplier factor for cuts

Limits the number of cuts that can be added.

Purpose

Row multiplier factor for cuts

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_CutsFactor CPX_PARAM_CUTSFACTOR
C++ IloCplex::Param::MIP::Limits::CutsFactor CutsFactor (double)
Java IloCplex.Param.MIP.Limits.CutsFactor CutsFactor (double)
.NET Cplex.Param.MIP.Limits.CutsFactor CutsFactor (double)
OPL cutsfactor
Python parameters.mip.limits.cutsfactor mip.limits.cutsfactor
MATLAB Cplex.Param.mip.limits.cutsfactor mip.limits.cutsfactor
Interactive mip limits cutsfactor mip limits cutsfactor
Identifier 2033 2033

42 CPLEX Parameters Reference

 Description

Limits the number of cuts that can be added. The number of rows in the problem
with cuts added is limited to CutsFactor times the original number of rows. If the
problem is presolved, the original number of rows is that from the presolved
problem.

A CutsFactor of 1.0 or less means that no cuts will be generated.

Because cuts can be added and removed during the course of optimization,
CutsFactor may not correspond directly to the number of cuts seen in the node log
or in the summary table at the end of optimization.

Values

Any nonnegative number; default: 4.0

upper cutoff
Sets the upper cutoff tolerance.

Purpose

Upper cutoff

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Tolerances_UpperCutoff CPX_PARAM_CUTUP
C++ IloCplex::Param::MIP::Tolerances::UpperCutoff CutUp (double)
Java IloCplex.Param.MIP.Tolerances.UpperCutoff CutUp (double)
.NET Cplex.Param.MIP.Tolerances.UpperCutoff CutUp (double)
OPL cutup
Python parameters.mip.tolerances.uppercutoff mip.tolerances.uppercutoff
MATLAB Cplex.Param.mip.tolerances.uppercutoff mip.tolerances.uppercutoff
Interactive mip tolerances uppercutoff mip tolerances uppercutoff
Identifier 2007 2007

Description

Sets the upper cutoff tolerance. When the problem is a minimization problem,
CPLEX cuts off or discards any solutions that are greater than the specified upper
cutoff value. If the model has no solution with an objective value less than or
equal to the cutoff value, CPLEX declares the model infeasible. In other words,
setting an upper cutoff value c for a minimization problem is similar to adding this
constraint to the objective function of the model: obj <= c.

Tip:

This parameter is effective only in the branch and bound algorithm, for example,
in a mixed integer program (MIP). It does not have the expected effect when
branch and bound is not invoked.

This parameter is not effective with the conflict refiner nor with FeasOpt. That is,
neither of those tools can analyze an infeasibility introduced by this parameter. If
you want to analyze such a condition, add an explicit objective constraint to your
model instead before you invoke either of those tools.

Chapter 3. List of CPLEX parameters 43

 Values

Any number; default: 1e+75.

data consistency checking switch

Decides whether data should be checked for consistency.

Purpose

Data consistency checking switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Read_DataCheck CPX_PARAM_DATACHECK
C++ IloCplex::Param::Read::DataCheck DataCheck (bool)
Java IloCplex.Param.Read.DataCheck DataCheck (bool)
.NET Cplex.Param.Read.DataCheck DataCheck (bool)
OPL datacheck
Python parameters.read.datacheck read.datacheck
MATLAB Cplex.Param.read.datacheck read.datacheck
Interactive read datacheck read datacheck
Identifier 1056 1056

Description

Decides whether data should be checked for consistency. When this parameter is
on, the routines CPXcopy____, CPXread____, and CPXchg____ of the C API
perform extensive checking of data in their array arguments, such as checking that
indices are within range, that there are no duplicate entries, and that values are
valid for the type of data or are valid numbers. This checking is useful for
debugging applications. When this checking identifies trouble, you can gather
more specific detail by calling one of the routines in check.c, as described in the
CPLEX User's Manual in the topic Checking and debugging problem data.
Table 12. Values
int bool Symbol Meaning
0 false CPX_OFF Data checking off; do
not check; default
1 true CPX_ON Data checking on

dependency switch
Decides whether to activate the dependency checker.

Purpose

Dependency switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Dependency CPX_PARAM_DEPIND
C++ IloCplex::Param::Preprocessing::Dependency DepInd (int)
Java IloCplex.Param.Preprocessing.Dependency DepInd (int)
.NET Cplex.Param.Preprocessing.Dependency DepInd (int)
OPL depind

44 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Python parameters.preprocessing.dependency preprocessing.dependency
MATLAB Cplex.Param.preprocessing.dependency preprocessing.dependency
Interactive preprocessing dependency preprocessing dependency
Identifier 1008 1008

Description

Decides whether to activate the dependency checker. If on, the dependency checker
searches for dependent rows during preprocessing. If off, dependent rows are not
identified.
Table 13. Values
Value Meaning
-1 Automatic: let CPLEX choose; default
0 Off: do not use dependency checker
1 Turn on only at the beginning of
preprocessing
2 Turn on only at the end of preprocessing
3 Turn on at the beginning and at the end of
preprocessing

deterministic time limit

Deterministic time limit

Purpose

Deterministic time limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_DetTimeLimit CPX_PARAM_DETTILIM
C++ IloCplex::Param::DetTimeLimit DetTiLim (double)
Java IloCplex.Param.DetTimeLimit DetTiLim (double)
.NET Cplex.Param.DetTimeLimit DetTiLim (double)
OPL not available
Python parameters.dettimelimit dettimelimit
MATLAB Cplex.Param.dettimelimit dettimelimit
Interactive dettimelimit dettimelimit
Identifier 1127 1127

Description

Sets a time limit expressed in ticks, a unit to measure work done deterministically.

The length of a deterministic tick may vary by platform. Nevertheless, ticks are
normally consistent measures for a given platform (combination of hardware and
software) carrying the same load. In other words, the correspondence of ticks to
clock time depends on the hardware, software, and the current load of the
machine. For the same platform and same load, the ratio of ticks per second stays
roughly constant, independent of the model solved. However, for very short
optimization runs, the variation of this ratio is typically high.

Chapter 3. List of CPLEX parameters 45

 CPLEX measures deterministic time only for work inside CPLEX. In other words,
deterministic time does not include user algorithms implemented by means of
callbacks. However, each application programming interface (API) of CPLEX offers
routines or methods that access the deterministic clock and provide deterministic
time stamps. In the APIs that support callbacks, you can use such deterministic
time stamps in your application to mark time even from callbacks. For more detail
about these deterministic time stamps, see the reference manual of the API that
you use.
v In the Callable Library (C API), see the documentation of CPXgetdettime and
CPXgetcallbackinfo.
v In the C++ API, see the documentation of IloCplex::CallbackI::getStartDetTime
and IloCplex::CallbackI::getEndDetTime.
v In the Java API, see the documentation of IloCplex.Callback.getStartDetTime and
IloCplex.Callback.getEndDetTime.
v In the .NET API, see the documentation of Cplex.ICallback.GetStartDetTime and
GetEndDetTime.
v In the Python API, see the documentation of Callback.get_start_dettime and
Callback.get_end_dettime.
v In the MATLAB connector, see the documentation of cplex.Solution.dettime.

At the end of optimization, the Interactive Optimizer displays the deterministic

time spent to optimize the model as well as the ratio of ticks per second. For
example, consider these lines, typical of output from the Interactive Optimizer:
MIP - Integer optimal solution: Objective = 1.1580000000e+03
Solution time = 2.81 sec. Iterations = 72793 Nodes = 2666
Deterministic time = 1996.47 ticks (709.54 ticks/sec)

See also

For a nondeterministic time limit measured in seconds, see “optimizer time limit in
seconds” on page 141 (CPX_PARAM_TILIM, TiLim).

For more detail about use of time limits, see the topic Timing interface in the
CPLEX User's Manual.

Value
Any nonnegative double value in deterministic ticks; default:1.0E+75

MIP disjunctive cuts switch

Decides whether or not disjunctive cuts should be generated for the problem.

Purpose

MIP disjunctive cuts switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_Disjunctive CPX_PARAM_DISJCUTS
C++ IloCplex::Param::MIP::Cuts::Disjunctive DisjCuts (int)
Java IloCplex.Param.MIP.Cuts.Disjunctive DisjCuts (int)
.NET Cplex.Param.MIP.Cuts.Disjunctive DisjCuts (int)
OPL disjcuts
Python parameters.mip.cuts.disjunctive mip.cuts.disjunctive

46 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
MATLAB Cplex.Param.mip.cuts.disjunctive mip.cuts.disjunctive
Interactive mip cuts disjunctive mip cuts disjunctive
Identifier 2053 2053

Description

Decides whether or not disjunctive cuts should be generated for the problem.
Setting the value to 0 (zero), the default, specifies that the attempt to generate
disjunctive cuts should continue only if it seems to be helping.

For a definition of a disjunctive cut, see the topic Disjunctive cuts in the general
topic Cuts in the CPLEX User’s Manual. The table Parameters for controlling cuts,
also in the user’s manual, includes links to the documentation of other parameters
affecting other types of cuts.
Table 14. Values
Value Meaning
-1 Do not generate disjunctive cuts
0 Automatic: let CPLEX choose; default
1 Generate disjunctive cuts moderately
2 Generate disjunctive cuts aggressively
3 Generate disjunctive cuts very aggressively

MIP dive strategy

Controls the MIP dive strategy.

Purpose

MIP dive strategy

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_Dive CPX_PARAM_DIVETYPE
C++ IloCplex::Param::MIP::Strategy::Dive DiveType (int)
Java IloCplex.Param.MIP.Strategy.Dive DiveType (int)
.NET Cplex.Param.MIP.Strategy.Dive DiveType (int)
OPL divetype
Python parameters.mip.strategy.dive mip.strategy.dive
MATLAB Cplex.Param.mip.strategy.dive mip.strategy.dive
Interactive mip strategy dive mip strategy dive
Identifier 2060 2060

Description

Controls the MIP dive strategy. The MIP traversal strategy occasionally performs
probing dives, where it looks ahead at both children nodes before deciding which
node to choose. The default (automatic) setting lets CPLEX choose when to
perform a probing dive, 1 (one) directs CPLEX never to perform probing dives, 2
always to probe, 3 to spend more time exploring potential solutions that are
similar to the current incumbent. Setting 2, always to probe, is helpful for finding
integer solutions.

Chapter 3. List of CPLEX parameters 47

 Table 15. Values
Value Meaning
0 Automatic: let CPLEX choose; default
1 Traditional dive
2 Probing dive
3 Guided dive

dual simplex pricing algorithm

Decides the type of pricing applied in the dual simplex algorithm.

Purpose

Dual simplex pricing algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_DGradient CPX_PARAM_DPRIIND
C++ IloCplex::Param::Simplex::DGradient DPriInd (int)
Java IloCplex.Param.Simplex.DGradient DPriInd (int)
.NET Cplex.Param.Simplex.DGradient DPriInd (int)
OPL dpriind
Python parameters.simplex.dgradient simplex.dgradient
MATLAB Cplex.Param.simplex.dgradient simplex.dgradient
Interactive simplex dgradient simplex dgradient
Identifier 1009 1009

Description

Decides the type of pricing applied in the dual simplex algorithm. The default
pricing (0) usually provides the fastest solution time, but many problems benefit
from alternate settings.
Table 16. Values
Value Symbol Meaning
0 CPX_DPRIIND_AUTO Automatic: let CPLEX choose; default
1 CPX_DPRIIND_FULL Standard dual pricing
2 CPX_DPRIIND_STEEP Steepest-edge pricing
3 CPX_DPRIIND_FULL_STEEP Steepest-edge pricing in slack space
4 CPX_DPRIIND_STEEPQSTART Steepest-edge pricing, unit initial norms
5 CPX_DPRIIND_DEVEX devex pricing

See also

“candidate limit for generating Gomory fractional cuts” on page 61, “MIP Gomory
fractional cuts switch” on page 62, “pass limit for generating Gomory fractional
cuts” on page 63

type of cut limit

Sets a limit for each type of cut.

48 CPLEX Parameters Reference

 Purpose

Type of cut limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_EachCutLimit CPX_PARAM_EACHCUTLIM
C++ IloCplex::Param::MIP::Limits::EachCutLimit EachCutLim (int)
Java IloCplex.Param.MIP.Limits.EachCutLimit EachCutLim (int)
.NET Cplex.Param.MIP.Limits.EachCutLimit EachCutLim (int)
OPL eachcutlim
Python parameters.mip.limits.eachcutlimit mip.limits.eachcutlimit
MATLAB Cplex.Param.mip.limits.eachcutlimit mip.limits.eachcutlimit
Interactive mip limits eachcutlimit mip limits eachcutlimit
Identifier 2102 2102

Description

Sets a limit for each type of cut.

This parameter allows you to set a uniform limit on the number of cuts of each
type that CPLEX generates. By default, the limit is the largest integer supported by
a given platform; that is, there is no effective limit by default.

Tighter limits on the number of cuts of each type may benefit certain models. For
example, a limit on each type of cut will prevent any one type of cut from being
created in such large number that the limit on the total number of all types of cuts
is reached before other types of cuts have an opportunity to be created.

A setting of 0 (zero) means no cuts.

This parameter does not influence the number of Gomory cuts. For means to
control the number of Gomory cuts, see also the fractional cut parameters:
v “candidate limit for generating Gomory fractional cuts” on page 61:
CPX_PARAM_FRACCAND, FracCand;
v “MIP Gomory fractional cuts switch” on page 62: CPX_PARAM_FRACCUTS, FracCuts;
v “pass limit for generating Gomory fractional cuts” on page 63:
CPX_PARAM_FRACPASS, FracPass.
Table 17. Values
Value Meaning
0 No cuts
Any positive number Limit each type of cut
2100000000 default

absolute MIP gap tolerance

Sets an absolute tolerance on the gap between the best integer objective and the
objective of the best node remaining.

Chapter 3. List of CPLEX parameters 49

 Purpose

Absolute MIP gap tolerance

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Tolerances_AbsMIPGap CPX_PARAM_EPAGAP
C++ IloCplex::Param::MIP::Tolerances::AbsMIPGap EpAGap (double)
Java IloCplex.Param.MIP.Tolerances.AbsMIPGap EpAGap (double)
.NET Cplex.Param.MIP.Tolerances.AbsMIPGap EpAGap (double)
OPL epagap
Python parameters.mip.tolerances.absmipgap mip.tolerances.absmipgap
MATLAB Cplex.Param.mip.tolerances.absmipgap mip.tolerances.absmipgap
Interactive mip tolerances absmipgap mip tolerances absmipgap
Identifier 2008 2008

Description

Sets an absolute tolerance on the gap between the best integer objective and the
objective of the best node remaining. When this difference falls below the value of
this parameter, the mixed integer optimization is stopped.

Values

Any nonnegative number; default: 1e-06.

relative MIP gap tolerance

Sets a relative tolerance on the gap between the best integer objective and the
objective of the best node remaining.

Purpose

Relative MIP gap tolerance

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Tolerances_MIPGap CPX_PARAM_EPGAP
C++ IloCplex::Param::MIP::Tolerances::MIPGap EpGap (double)
Java IloCplex.Param.MIP.Tolerances.MIPGap EpGap (double)
.NET Cplex.Param.MIP.Tolerances.MIPGap EpGap (double)
OPL epgap
Python parameters.mip.tolerances.mipgap mip.tolerances.mipgap
MATLAB Cplex.Param.mip.tolerances.mipgap mip.tolerances.mipgap
Interactive mip tolerances mipgap mip tolerances mipgap
Identifier 2009 2009

Description

When the value

|bestbound-bestinteger|/(1e-10+|bestinteger|)

falls below the value of this parameter, the mixed integer optimization is stopped.

50 CPLEX Parameters Reference

 For example, to instruct CPLEX to stop as soon as it has found a feasible integer
solution proved to be within five percent of optimal, set the relative MIP gap
tolerance to 0.05.

Values

Any number from 0.0 to 1.0; default: 1e-04.

integrality tolerance
Specifies the amount by which an integer variable can be different from an integer
and still be considered feasible.

Purpose

Integrality tolerance

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Tolerances_Integrality CPX_PARAM_EPINT
C++ IloCplex::Param::MIP::Tolerances::Integrality EpInt (double)
Java IloCplex.Param.MIP.Tolerances.Integrality EpInt (double)
.NET Cplex.Param.MIP.Tolerances.Integrality EpInt (double)
OPL epint
Python parameters.mip.tolerances.integrality mip.tolerances.integrality
MATLAB Cplex.Param.mip.tolerances.integrality mip.tolerances.integrality
Cplex class
API
MATLAB CPLEX Toolbox compatible mip.tolerances.integrality
MATLAB Optimization Toolbox compatible TolXInteger
Interactive mip tolerances integrality mip tolerances integrality
Identifier 2010 2010

Description

Specifies the amount by which an integer variable can be different from an integer
and still be considered feasible.

A value of zero is permitted, and the optimizer will attempt to meet this tolerance.

However, in some models, computer round-off may still result in small, nonzero
deviations from integrality. If any of these deviations exceed the value of this
parameter, or exceed 1e-10 in the case where this parameter has been set to a value
less than that, a solution status of CPX_STAT_OPTIMAL_INFEAS will be returned
instead of the usual CPX_STAT_OPTIMAL.

Tip: This parameter sets the amount by which a computed solution value for an
integer variable can violate integrality; it does not specify an amount by which
CPLEX relaxes integrality.

Values

Any number from 0.0 to 0.5; default: 1e-05.

Chapter 3. List of CPLEX parameters 51

epsilon (degree of tolerance) used in linearization
Sets the epsilon (degree of tolerance) used in linearization in the object-oriented
APIs.

Purpose

Epsilon used in linearization

API Parameter Name Name prior to V12.6.0

C
C++ IloCplex::Param::MIP::Tolerances::Linearization EpLin (double)
Java IloCplex::Param::MIP::Tolerances::Linearization EpLin (double)
.NET IloCplex::Param::MIP::Tolerances::Linearization EpLin (double)
OPL
Python
MATLAB
Interactive
Identifier 2068

Description

Sets the epsilon (degree of tolerance) used in linearization in the object-oriented

APIs.

Not applicable in the C API.

Not applicable in the Python API.

Not applicable in the CPLEX connector for MATLAB.

Not available in the Interactive Optimizer.

This parameter controls how strict inequalities are managed during linearization.
In other words, it provides an epsilon for deciding when two values are not equal
during linearization. For example, when x is a numeric variable (that is, an
instance of IloNumVar ),

x < a

becomes

x <= a-eplin .

Similarly, x!=a

becomes

{(x < a) || (x > a)}

which is linearized automatically for you in the object-oriented APIs as

{( x <= a-eplin) || (x >= a+eplin)} .

52 CPLEX Parameters Reference

 Exercise caution in changing this parameter from its default value: the smaller the
epsilon, the more numerically unstable the model will tend to become. If you are
not getting an expected solution for an object-oriented model that uses
linearization, it might be that this solution is cut off because of the relatively high
EpLin value. In such a case, carefully try reducing it.

Values

Any positive value greater than zero; default: 1e-3.

Markowitz tolerance
Influences pivot selection during basis factoring.

Purpose

Markowitz tolerance

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Tolerances_Markowitz CPX_PARAM_EPMRK
C++ IloCplex::Param::Simplex::Tolerances::Markowitz EpMrk (double)
Java IloCplex.Param.Simplex.Tolerances.Markowitz EpMrk (double)
.NET Cplex.Param.Simplex.Tolerances.Markowitz EpMrk (double)
OPL epmrk
Python parameters.simplex.tolerances.markowitz simplex.tolerances.markowitz
MATLAB Cplex.Param.simplex.tolerances.markowitz simplex.tolerances.markowitz
Interactive simplex tolerances markowitz simplex tolerances markowitz
Identifier 1013 1013

Description

Influences pivot selection during basis factoring. Increasing the Markowitz

threshold may improve the numerical properties of the solution.

Values

Any number from 0.0001 to 0.99999; default: 0.01.

optimality tolerance
Influences the reduced-cost tolerance for optimality.

Purpose

Optimality tolerance

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Tolerances_Optimality CPX_PARAM_EPOPT
C++ IloCplex::Param::Simplex::Tolerances::Optimality EpOpt (double)
Java IloCplex.Param.Simplex.Tolerances.Optimality EpOpt (double)
.NET Cplex.Param.Simplex.Tolerances.Optimality EpOpt (double)
OPL epopt
Python parameters.simplex.tolerances.optimality simplex.tolerances.optimality

Chapter 3. List of CPLEX parameters 53

 API Parameter Name Name prior to V12.6.0
MATLAB Cplex.Param.simplex.tolerances.optimality simplex.tolerances.optimality
(Cplex
class API)
MATLAB CPLEX Toolbox compatible simplex.tolerances.optimality
MATLAB Optimization Toolbox compatible TolFun and TolRLPFun
Interactive simplex tolerances optimality simplex tolerances optimality
Identifier 1014 1014

Description

Influences the reduced-cost tolerance for optimality. This parameter governs how
closely CPLEX must approach the theoretically optimal solution.

The simplex algorithm halts when it has found a basic feasible solution with all
reduced costs nonnegative. CPLEX uses this optimality tolerance to make the
decision of whether or not a given reduced cost should be considered nonnegative.
CPLEX considers "nonnegative" a negative reduced cost having absolute value less
than the optimality tolerance. For example, if your optimality tolerance is set to
1e-6, then CPLEX considers a reduced cost of -1e-9 as nonnegative for the purpose
of deciding whether the solution is optimal.

Values

Any number from 1e-9 to 1e-1; default: 1e-06.

perturbation constant
Sets the amount by which CPLEX perturbs the upper and lower bounds or
objective coefficients on the variables when a problem is perturbed in the simplex
algorithm.

Purpose

Perturbation constant

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Perturbation_Constant CPX_PARAM_EPPER
C++ IloCplex::Param::Simplex::Perturbation::Constant EpPer (double)
Java IloCplex.Param.Simplex.Perturbation.Constant EpPer (double)
.NET Cplex.Param.Simplex.Perturbation.Constant EpPer (double)
OPL epper
Python parameters.simplex.perturbation.constant simplex.perturbation.constant
MATLAB Cplex.Param.simplex.perturbation.constant simplex.perturbation.constant
Interactive simplex perturbationlimit 0/1 C simplex perturbationlimit 0/1 C
Identifier 1015 1015

Description

Sets the amount by which CPLEX perturbs the upper and lower bounds or
objective coefficients on the variables when a problem is perturbed in the simplex
algorithm. This parameter can be set to a smaller value if the default value creates
too large a change in the problem.

54 CPLEX Parameters Reference

 In the Interactive Optimizer, the command
set simplex perturbationlimit

accepts two arguments and actually sets two parameters simultaneously. The first
argument is a switch or indicator; its value is 1 (one) to turn on perturbation or 0
(zero) to turn off perturbation. See the parameter “simplex perturbation switch” on
page 95 for more detail about this effect. The second argument is a constant value
to set an amount of perturbation.

Values

Any positive number greater than or equal to 1e-8; default: 1e-6.

relaxation for FeasOpt

Controls the amount of relaxation for the routine CPXfeasopt in the C API or for
the method feasOpt in the object-oriented APIs.

Purpose

Relaxation for feasOpt

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Feasopt_Tolerance CPX_PARAM_EPRELAX
C++ IloCplex::Param::Feasopt::Tolerance EpRelax (double)
Java IloCplex.Param.Feasopt.Tolerance EpRelax (double)
.NET Cplex.Param.Feasopt.Tolerance EpRelax (double)
OPL eprelax
Python parameters.feasopt.tolerance feasopt.tolerance
MATLAB Cplex.Param.feasopt.tolerance feasopt.tolerance
Interactive feasopt tolerance feasopt tolerance
Identifier 2073 2073

Description
Controls the amount of relaxation for the routine CPXfeasopt in the C API or for
the method feasOpt in the object-oriented APIs.

In the case of a MIP, it serves the purpose of the absolute gap for the feasOpt
model in Phase I (the phase to minimize relaxation).

Using this parameter, you can implement other stopping criteria as well. To do so,
first call feasOpt with the stopping criteria that you prefer; then set this parameter
to the resulting objective of the Phase I model; unset the other stopping criteria,
and call feasOpt again. Since the solution from the first call already matches this
parameter, Phase I will terminate immediately in this second call to feasOpt, and
Phase II will start.

In the case of an LP, this parameter controls the lower objective limit for Phase I of
feasOpt and is thus relevant only when the primal optimizer is in use.

Values

Any nonnegative value; default: 1e-6.

Chapter 3. List of CPLEX parameters 55

 See also

“lower objective value limit” on page 92

feasibility tolerance
Specifies the feasibility tolerance, that is, the degree to which the basic variables of
a model may violate their bounds.

Purpose

Feasibility tolerance

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Tolerances_Feasibility CPX_PARAM_EPRHS
C++ IloCplex::Param::Simplex::Tolerances::Feasibility EpRHS (double)
Java IloCplex.Param.Simplex.Tolerances.Feasibility EpRHS (double)
.NET Cplex.Param.Simplex.Tolerances.Feasibility EpRHS (double)
OPL eprhs
Python parameters.simplex.tolerances.feasibility simplex.tolerances.feasibility
MATLAB Cplex.Param.simplex.tolerances.feasibility simplex.tolerances.feasibility
Interactive simplex tolerances feasibility simplex tolerances feasibility
Identifier 1016 1016

Description

Specifies the feasibility tolerance, that is, the degree to which values of the basic
variables calculated by the simplex method may violate their bounds. CPLEX®
does not use this tolerance to relax the variable bounds nor to relax right hand
side values. This parameter specifies an allowable violation. Feasibility influences
the selection of an optimal basis and can be reset to a higher value when a
problem is having difficulty maintaining feasibility during optimization. You can
also lower this tolerance after finding an optimal solution if there is any doubt that
the solution is truly optimal. If the feasibility tolerance is set too low, CPLEX may
falsely conclude that a problem is infeasible. If you encounter reports of
infeasibility during Phase II of the optimization, a small adjustment in the
feasibility tolerance may improve performance.

Values

Any number from 1e-9 to 1e-1; default: 1e-06.

mode of FeasOpt
Decides how FeasOpt measures the relaxation when finding a minimal relaxation
in an infeasible model.

Purpose

Mode of FeasOpt

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Feasopt_Mode CPX_PARAM_FEASOPTMODE
C++ IloCplex::Param::Feasopt::Mode FeasOptMode (int)
Java IloCplex.Param.Feasopt.Mode FeasOptMode (int)

56 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
.NET Cplex.Param.Feasopt.Mode FeasOptMode (int)
OPL feasoptmode
Python parameters.feasopt.mode feasopt.mode
MATLAB Cplex.Param.feasopt.mode feasopt.mode
Interactive feasopt mode feasopt mode
Identifier 1084 1084

Description

Decides how FeasOpt measures the relaxation when finding a minimal relaxation
in an infeasible model. FeasOpt works in two phases. In its first phase, it attempts
to minimize its relaxation of the infeasible model. That is, it attempts to find a
feasible solution that requires minimal change. In its second phase, it finds an
optimal solution among those that require only as much relaxation as it found
necessary in the first phase. Values of this parameter indicate two aspects to
CPLEX:
v whether to stop in phase one or continue to phase two and
v how to measure the relaxation, according to one of the following criteria:
– as a sum of required relaxations;
– as the number of constraints and bounds required to be relaxed;
– as a sum of the squares of required relaxations.
Table 18. Values
Value Symbol Symbol (C API) Meaning
0 MinSum CPX_FEASOPT_MIN_SUM Minimize the sum of all required
relaxations in first phase only;
default
1 OptSum CPX_FEASOPT_OPT_SUM Minimize the sum of all required
relaxations in first phase and execute
second phase to find optimum
among minimal relaxations
2 MinInf CPX_FEASOPT_MIN_INF Minimize the number of constraints
and bounds requiring relaxation in
first phase only
3 OptInf CPX_FEASOPT_OPT_INF Minimize the number of constraints
and bounds requiring relaxation in
first phase and execute second phase
to find optimum among minimal
relaxations
4 MinQuad CPX_FEASOPT_MIN_QUAD Minimize the sum of squares of
required relaxations in first phase
only
5 OptQuad CPX_FEASOPT_OPT_QUAD Minimize the sum of squares of
required relaxations in first phase
and execute second phase to find
optimum among minimal relaxations

file encoding switch

file encoding switch

Chapter 3. List of CPLEX parameters 57

 Purpose

File encoding switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Read_FileEncoding CPX_PARAM_FILEENCODING
C++ IloCplex::Param::Read::FileEncoding FileEncoding (string)
Java IloCplex.Param.Read.FileEncoding FileEncoding (string)
.NET Cplex.Param.Read.FileEncoding FileEncoding (string)
OPL not available
Python parameters.read.fileencoding read.fileencoding
MATLAB Cplex.Param.read.fileencoding read.fileencoding
Interactive read fileencoding read fileencoding
Identifier 1129 1129

Description

Specifies which encoding (also known as the code page) that CPLEX uses for
reading and writing files. This parameter accepts a string, such as UTF-8, UTF-16LE,
ISO-8859-1, US-ASCII, and so forth, specifying the user’s choice of encoding for
reading and writing files.

Note:

This parameter has no effect on IBM CPLEX Optimizer for z/OS, where only
EBCDIC IBM-1047 encoding is available.

The default value of this parameter depends on the CPLEX component.

v In the CPLEX connector for MATLAB, the default value of the file encoding
parameter is the empty string (" ") for consistency with MATLAB conventions.
v In the Python API, the default value of the file encoding parameter is the empty
string (" ") for consistency with Python conventions.
v In other APIs of CPLEX, such as the C, C++, Java, .NET API, the default value
of the file encoding parameter is the string ISO-8859-1 (also known as Latin-1).

The encoding ISO-8859-1 is a superset of the familiar ASCII encoding, so it

supports many widely used character sets. However, this default encoding cannot
represent multi-byte character sets, such as Chinese, Japanese, or Korean
characters, for example. If you want CPLEX to represent a character set that
requires multiple bytes per character, then a better choice for the value of this
parameter is UTF-8. The encoding UTF-8 is compatible with ASCII encoding; it
represents every character in Unicode; it does not include a NULL byte in a valid
character; it does not require specification of big-end or little-end byte order; it
does not require a byte-order mark. If you use another multi-byte encoding, such
as UTF-32 or UTF-16, for example, be sure to specify the encoding fully by
including the byte order, like this: UTF-32LE or UTF-32BE.

When you change the value of this parameter, you also need to verify that the
“API string encoding switch” on page 20 (CPX_PARAM_APIENCODING, APIEncoding) is
compatible. The encoding specified by the API encoding parameter must be a
subset of the encoding specified by the file encoding parameter. For example, if the
API encoding parameter specifies US-ASCII, then UTF-8 is a reasonable choice for
the file encoding parameter because the code page US-ASCII is a subset of UTF-8.

58 CPLEX Parameters Reference

 For a complete list of valid strings that are the name of an encoding (that is, the
name of a code page), consult the web site of a standards organization such as:
v A brief introduction to code pages
v ICU: International Components for Unicode
v International Components for Unicode at IBM

In situations where CPLEX encounters a string, such as content in a file, that is not
compatible with the specified encoding, the behavior is not defined. Because of the
incompatibility, CPLEX silently converts the string to an inappropriate character of
the specified encoding, or CPLEX raises the error CPXERR_ENCODING_CONVERSION.

Values

valid string for the name of an encoding (code page); default: ISO-8859-1 or the
empty string (“ “)

See also

“API string encoding switch” on page 20

MIP flow cover cuts switch

Decides whether or not to generate flow cover cuts for the problem.

Purpose

MIP flow cover cuts switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_FlowCovers CPX_PARAM_FLOWCOVERS
C++ IloCplex::Param::MIP::Cuts::FlowCovers FlowCovers (int)
Java IloCplex.Param.MIP.Cuts.FlowCovers FlowCovers (int)
.NET Cplex.Param.MIP.Cuts.FlowCovers FlowCovers (int)
OPL flowcovers
Python parameters.mip.cuts.flowcovers mip.cuts.flowcovers
MATLAB Cplex.Param.mip.cuts.flowcovers mip.cuts.flowcovers
Interactive mip cuts flowcovers mip cuts flowcovers
Identifier 2040 2040

Description

Decides whether or not to generate flow cover cuts for the problem. Setting the
value to 0 (zero), the default, indicates that the attempt to generate flow cover cuts
should continue only if it seems to be helping.

For a definition of a flow cover cut, see the topic Flow cover cuts in the general
topic Cuts in the CPLEX User’s Manual. The table Parameters for controlling cuts,
also in the user’s manual, includes links to the documentation of other parameters
affecting other types of cuts.
Table 19. Values
Value Meaning
-1 Do not generate flow cover cuts
0 Automatic: let CPLEX choose; default

Chapter 3. List of CPLEX parameters 59

 Table 19. Values (continued)
Value Meaning
1 Generate flow cover cuts moderately
2 Generate flow cover cuts aggressively

MIP flow path cut switch

Decides whether or not flow path cuts should be generated for the problem.

Purpose

MIP flow path cut switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_PathCut CPX_PARAM_FLOWPATHS
C++ IloCplex::Param::MIP::Cuts::PathCut FlowPaths (int)
Java IloCplex.Param.MIP.Cuts.PathCut FlowPaths (int)
.NET Cplex.Param.MIP.Cuts.PathCut FlowPaths (int)
OPL flowpaths
Python parameters.mip.cuts.pathcut mip.cuts.pathcut
MATLAB Cplex.Param.mip.cuts.pathcut mip.cuts.pathcut
Interactive mip cuts pathcut mip cuts pathcut
Identifier 2051 2051

Description

Decides whether or not flow path cuts should be generated for the problem.
Setting the value to 0 (zero), the default, indicates that the attempt to generate flow
path cuts should continue only if it seems to be helping.

For a definition of a flow path cut, see the topic Flow path cuts in the general topic
Cuts in the CPLEX User’s Manual. The table Parameters for controlling cuts, also in
the user’s manual, includes links to the documentation of other parameters
affecting other types of cuts.
Table 20. Values
Value Meaning
-1 Do not generate flow path cuts
0 Automatic: let CPLEX choose; default
1 Generate flow path cuts moderately
2 Generate flow path cuts aggressively

feasibility pump switch

Turns on or off the feasibility pump heuristic for mixed integer programming
(MIP) models.

60 CPLEX Parameters Reference

 Purpose

Feasibility pump switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_FPHeur CPX_PARAM_FPHEUR
C++ IloCplex::Param::MIP::Strategy::FPHeur FPHeur (int)
Java IloCplex.Param.MIP.Strategy.FPHeur FPHeur (int)
.NET Cplex.Param.MIP.Strategy.FPHeur FPHeur (int)
OPL fpheur
Python parameters.mip.strategy.fpheur mip.strategy.fpheur
MATLAB Cplex.Param.mip.strategy.fpheur mip.strategy.fpheur
Interactive mip strategy fpheur mip strategy fpheur
Identifier 2098 2098

Description

Turns on or off the feasibility pump heuristic for mixed integer programming
(MIP) models.

At the default setting 0 (zero), CPLEX automatically chooses whether or not to

apply the feasibility pump heuristic on the basis of characteristics of the model.
The feasibility pump does not apply to models of the type mixed integer
quadratically constrained programs (MIQCP).

To turn off the feasibility pump heuristic, set the parameter to -1 (minus one).

To turn on the feasibility pump heuristic, set the parameter to 1 (one) or 2.

If the parameter is set to 1 (one), the feasibility pump tries to find a feasible
solution without taking the objective function into account.

If the parameter is set to 2, the heuristic usually finds solutions of better objective
value, but is more likely to fail to find a feasible solution.

For more detail about the feasibility pump heuristic, see research by Fischetti,
Glover, and Lodi (2003, 2005), by Bertacco, Fischetti, and Lodi (2005), and by
Achterberg and Berthold (2005, 2007).
Table 21. Values
Value Meaning
-1 Do not apply the feasibility pump heuristic
0 Automatic: let CPLEX choose; default
1 Apply the feasibility pump heuristic with an
emphasis on finding a feasible solution
2 Apply the feasibility pump heuristic with an
emphasis on finding a feasible solution with
a good objective value

candidate limit for generating Gomory fractional cuts

Limits the number of candidate variables for generating Gomory fractional cuts.

Chapter 3. List of CPLEX parameters 61

 Purpose

Candidate limit for generating Gomory fractional cuts

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_GomoryCand CPX_PARAM_FRACCAND
C++ IloCplex::Param::MIP::Limits::GomoryCand FracCand (int)
Java IloCplex.Param.MIP.Limits.GomoryCand FracCand (int)
.NET Cplex.Param.MIP.Limits.GomoryCand FracCand (int)
OPL fraccand
Python parameters.mip.limits.gomorycand mip.limits.gomorycand
MATLAB Cplex.Param.mip.limits.gomorycand mip.limits.gomorycand
Interactive mip limits gomorycand mip limits gomorycand
Identifier 2048 2048

Description

Limits the number of candidate variables for generating Gomory fractional cuts.

Values

Any positive integer; default: 200.

MIP Gomory fractional cuts switch

Decides whether or not Gomory fractional cuts should be generated for the
problem.

Purpose

MIP Gomory fractional cuts switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_Gomory CPX_PARAM_FRACCUTS
C++ IloCplex::Param::MIP::Cuts::Gomory FracCuts (int)
Java IloCplex.Param.MIP.Cuts.Gomory FracCuts (int)
.NET Cplex.Param.MIP.Cuts.Gomory FracCuts (int)
OPL fraccuts
Python parameters.mip.cuts.gomory mip.cuts.gomory
MATLAB Cplex.Param.mip.cuts.gomory mip.cuts.gomory
Interactive mip cuts gomory mip cuts gomory
Identifier 2049 2049

Description

Decides whether or not Gomory fractional cuts should be generated for the
problem. Setting the value to 0 (zero), the default, indicates that the attempt to
generate Gomory fractional cuts should continue only if it seems to be helping.

For a definition of a Gomory fractional cut, see the topic Gomory fractional cuts in
the general topic Cuts in the CPLEX User’s Manual. The table Parameters for
controlling cuts, also in the user’s manual, includes links to the documentation of
other parameters affecting other types of cuts.

62 CPLEX Parameters Reference

 Table 22. Values
Value Meaning
-1 Do not generate Gomory fractional cuts
0 Automatic: let CPLEX choose; default
1 Generate Gomory fractional cuts moderately
2 Generate Gomory fractional cuts
aggressively

pass limit for generating Gomory fractional cuts

Limits the number of passes for generating Gomory fractional cuts.

Purpose

Pass limit for generating Gomory fractional cuts

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_GomoryPass CPX_PARAM_FRACPASS
C++ IloCplex::Param::MIP::Limits::GomoryPass FracPass (long)
Java IloCplex.Param.MIP.Limits.GomoryPass FracPass (long)
.NET Cplex.Param.MIP.Limits.GomoryPass FracPass (long)
OPL fracpass
Python parameters.mip.limits.gomorypass mip.limits.gomorypass
MATLAB Cplex.Param.mip.limits.gomorypass mip.limits.gomorypass
Interactive mip limits gomorypass mip limits gomorypass
Identifier 2050 2050

Description

Limits the number of passes for generating Gomory fractional cuts. At the default
setting of 0 (zero), CPLEX decides the number of passes to make. The parameter is
ignored if the Gomory fractional cut parameter (“MIP Gomory fractional cuts
switch” on page 62: CPX_PARAM_FRACCUTS, FracCuts) is set to a nonzero value.
Table 23. Values
Value Meaning
0 Automatic: let CPLEX choose; default
Any positive integer Number of passes to generate Gomory
fractional cuts

MIP GUB cuts switch

Decides whether or not to generate GUB cuts for the problem.

Purpose

MIP GUB cuts switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_GUBCovers CPX_PARAM_GUBCOVERS
C++ IloCplex::Param::MIP::Cuts::GUBCovers GUBCovers (int)

Chapter 3. List of CPLEX parameters 63

 API Parameter Name Name prior to V12.6.0
Java IloCplex.Param.MIP.Cuts.GUBCovers GUBCovers (int)
.NET Cplex.Param.MIP.Cuts.GUBCovers GUBCovers (int)
OPL gubcovers
Python parameters.mip.cuts.gubcovers mip.cuts.gubcovers
MATLAB Cplex.Param.mip.cuts.gubcovers mip.cuts.gubcovers
Interactive mip cuts gubcovers mip cuts gubcovers
Identifier 2044 2044

Description

Decides whether or not to generate generalized upper bound (GUB) cover cuts for
the problem. Setting the value to 0 (zero), the default, indicates that the attempt to
generate GUB cuts should continue only if it seems to be helping.

For a definition of a GUB cover cut, see the topic Generalized upper bound (GUB)
cover cutsin the general topic Cuts in the CPLEX User’s Manual. The table
Parameters for controlling cuts, also in the user’s manual, includes links to the
documentation of other parameters affecting other types of cuts.
Table 24. Values
Value Meaning
-1 Do not generate GUB cuts
0 Automatic: let CPLEX choose; default
1 Generate GUB cuts moderately
2 Generate GUB cuts aggressively

MIP heuristic frequency

Decides how often to apply the periodic heuristic.

Purpose

MIP heuristic frequency

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_HeuristicFreq CPX_PARAM_HEURFREQ
C++ IloCplex::Param::MIP::Strategy::HeuristicFreq HeurFreq (long)
Java IloCplex.Param.MIP.Strategy.HeuristicFreq HeurFreq (long)
.NET Cplex.Param.MIP.Strategy.HeuristicFreq HeurFreq (long)
OPL heurfreq
Python parameters.mip.strategy.heuristicfreq mip.strategy.heuristicfreq
MATLAB Cplex.Param.mip.strategy.heuristicfreq mip.strategy.heuristicfreq
Interactive mip strategy heuristicfreq mip strategy heuristicfreq
Identifier 2031 2031

Description

Decides how often to apply the periodic heuristic. Setting the value to -1 turns off
the periodic heuristic. Setting the value to 0 (zero), the default, applies the periodic
heuristic at an interval chosen automatically. Setting the value to a positive number

64 CPLEX Parameters Reference

 applies the heuristic at the requested node interval. For example, setting this
parameter to 20 dictates that the heuristic be called at node 0, 20, 40, 60, etc.

For an introduction to heuristics in CPLEX, see the topic Applying heuristicsamong

the topics in Tuning performance features of the mixed integer optimizerin the
CPLEX User’s Manual. For more about other heuristics, see the topics in Heuristics
(also in the CPLEX User’s Manual). There, the topic Node heuristic refers
specifically to this parameter.
Table 25. Values
Value Meaning
-1 None
0 Automatic: let CPLEX choose; default
Any positive integer Apply the periodic heuristic at this
frequency

MIP globally valid implied bound cuts switch

Decides whether or not to generate globally valid implied bound cuts for the
model.

Purpose

MIP globally valid implied bound cuts switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_Implied CPX_PARAM_IMPLBD
C++ IloCplex::Param::MIP::Cuts::Implied ImplBd (int)
Java IloCplex.Param.MIP.Cuts.Implied ImplBd (int)
.NET Cplex.Param.MIP.Cuts.Implied ImplBd (int)
OPL implbd
Python parameters.mip.cuts.implied mip.cuts.implied
MATLAB Cplex.Param.mip.cuts.implied mip.cuts.implied
Interactive mip cuts implied mip cuts implied
Identifier 2041 2041

Description

Decides whether or not to generate globally valid implied bound cuts for the
model. Setting the value to 0 (zero), the default, specifies that the attempt to
generate globally valid implied bound cuts should continue only if it seems to be
helping.

For a definition of an implied bound cut, see the topic Implied bound cuts: global
and local in the general topic Cuts in the CPLEX User’s Manual. The table
Parameters for controlling cuts, also in the user’s manual, includes links to the
documentation of other parameters affecting other types of cuts.
Table 26. Values
Value Meaning
-1 Do not generate implied bound cuts
0 Automatic: let CPLEX choose; default

Chapter 3. List of CPLEX parameters 65

 Table 26. Values (continued)
Value Meaning
1 Generate implied bound cuts moderately
2 Generate implied bound cuts aggressively

See also

MIP locally valid implied bound cuts switch

MIP integer solution-file switch and prefix

MIP integer solution file switch and filename prefix.

Purpose

MIP integer solution-file switch and filename prefix.

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Output_IntSolFilePrefix CPX_PARAM_INTSOLFILEPREFIX
C++ IloCplex::Param::Output::IntSolFilePrefix IntSolFilePrefix (string)
Java IloCplex.Param.Output.IntSolFilePrefix IntSolFilePrefix (string)
.NET Cplex.Param.Output.IntSolFilePrefix IntSolFilePrefix (string)
OPL not available
Python parameters.output.intsolfileprefix output.intsolfileprefix
MATLAB Cplex.Param.output.intsolfileprefix output.intsolfileprefix
Interactive output intsolfileprefix output intsolfileprefix
Identifier 2143 2143

Description

Decides whether CPLEX writes the current MIP incumbent integer solution to a
file and (if so) sets a prefix for the name of that file.

By default, the value of this parameter is the empty string, and file-writing is
turned off. When this parameter is set to a non empty string, CPLEX writes each
new incumbent to a file at the time the MIP integer solution is found.

In addition to switching on the writing of a file of solutions, this parameter also

specifies the prefix of the name of the file to use. The prefix can contain a relative
or absolute path. If the prefix does not contain a relative or absolute path, CPLEX
writes to a file in the “directory for working files” on page 149.

The complete file name of the file that CPLEX writes is PREFIX-NNNNN.sol, where:
v PREFIX is the prefix specified by this parameter;
v NNNNN is the sequence number of the solution; the sequence starts at 00001;
v sol represents the solution file format, documented in the topic SOL file format:
solution files in the reference manual, File formats supported by CPLEX.

Note:

Existing files of the same name will be overwritten.

66 CPLEX Parameters Reference

 If the specified file cannot be written (for example, in case of lack of disk space, or
no write access to the specified location), optimization stops with an error status
code.

This parameter accepts a string as its value. If you change either the “API string
encoding switch” on page 20 or the “file encoding switch” on page 57 from their
default value to a multi-byte encoding where a NULL byte can occur within the
encoding of a character, you must take into account the issues documented in the
topic Selecting an encoding in the CPLEX User's Manual. Especially consider the
possibility that a NULL byte occurring in the encoding of a character can
inadvertently signal the termination of a string, such as a filename or directory
path, and thus provoke surprising or incorrect results.

Values

valid string for the prefix of a file name; default: ” “ (the empty string; that is, the
switch is off)

See also

“directory for working files” on page 149

MIP integer solution limit

Sets the number of MIP solutions to be found before stopping.

Purpose

MIP integer solution limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_Solutions CPX_PARAM_INTSOLLIM
C++ IloCplex::Param::MIP::Limits::Solutions IntSolLim (long)
Java IloCplex.Param.MIP.Limits.Solutions IntSolLim (long)
.NET Cplex.Param.MIP.Limits.Solutions IntSolLim (long)
OPL intsollim
Python parameters.mip.limits.solutions mip.limits.solutions
MATLAB Cplex.Param.mip.limits.solutions mip.limits.solutions
Interactive mip limits solutions mip limits solutions
Identifier 2015 2015

Description

Sets the number of MIP solutions to be found before stopping.

This integer solution limit does not apply to the populate procedure, which
generates solutions to store in the solution pool. For a limit on the number of
solutions generated by populate, see the populate limit parameter: “maximum
number of solutions generated for solution pool by populate” on page 102.

Values

Any positive integer strictly greater than zero; zero is not allowed; default:
9223372036800000000.

Chapter 3. List of CPLEX parameters 67

 See also

“maximum number of solutions generated for solution pool by populate” on page

102

simplex maximum iteration limit

Sets the maximum number of simplex iterations to be performed before the
algorithm terminates without reaching optimality.

Purpose

Simplex maximum iteration limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Limits_Iterations CPX_PARAM_ITLIM
C++ IloCplex::Param::Simplex::Limits::Iterations ItLim (long)
Java IloCplex.Param.Simplex.Limits.Iterations ItLim (long)
.NET Cplex.Param.Simplex.Limits.Iterations ItLim (long)
OPL itlim
Python parameters.simplex.limits.iterations simplex.limits.iterations
MATLAB Cplex.Param.simplex.limits.iterations simplex.limits.iterations
Cplex class
API
MATLAB CPLEX Toolbox compatibility simplex.limits.iterations
MATLAB Optimization Toolbox compatibility MaxIter
Interactive simplex limits iterations simplex limits iterations
Identifier 1020 1020

Description

Sets the maximum number of simplex iterations to be performed before the

algorithm terminates without reaching optimality. When set to 0 (zero), no simplex
method iteration occurs. However, CPLEX factors the initial basis from which
solution routines provide information about the associated initial solution.

Values

Any nonnegative integer; default: 9223372036800000000.

local branching heuristic

Controls whether CPLEX applies a local branching heuristic to try to improve new
incumbents found during a MIP search.

Purpose

Local branching heuristic

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_LBHeur CPX_PARAM_LBHEUR
C++ IloCplex::Param::MIP::Strategy::LBHeur LBHeur (bool)
Java IloCplex.Param.MIP.Strategy.LBHeur LBHeur (bool)
.NET Cplex.Param.MIP.Strategy.LBHeur LBHeur (bool)
OPL lbheur

68 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Python parameters.mip.strategy.lbheur mip.strategy.lbheur
MATLAB Cplex.Param.mip.strategy.lbheur mip.strategy.lbheur
Interactive mip strategy lbheur mip strategy lbheur
Identifier 2063 2063

Description

Controls whether CPLEX applies a local branching heuristic to try to improve new
incumbents found during a MIP search. By default, this parameter is off. If you
turn it on, CPLEX will invoke a local branching heuristic only when it finds a new
incumbent. If CPLEX finds multiple incumbents at a single node, the local
branching heuristic will be applied only to the last one found.
Table 27. Values
Value bool Symbol Meaning
0 false CPX_OFF Local branching
heuristic is off;
default
1 true CPX_ON Apply local
branching heuristic to
new incumbent

MIP lift-and-project cuts switch

Decides whether or not lift-and-project cuts are generated for the problem.

Purpose

MIP lift-and-project cuts switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_LiftProj CPX_PARAM_LANDPCUTS
C++ IloCplex::Param::MIP::Cuts::LiftProj LiftProjCuts (int)
Java IloCplex.Param.MIP.Cuts.LiftProj LiftProjCuts (int)
.NET Cplex.Param.MIP.Cuts.LiftProj LiftProjCuts (int)
OPL not available
Python parameters.mip.cuts.liftproj mip.cuts.liftproj
MATLAB Cplex.Param.mip.cuts.liftproj mip.cuts.liftproj
Interactive mip cuts liftproj mip cuts liftproj
Identifier 2152 2152

Description

Decides whether or not lift-and-project cuts are generated for the problem. Setting
the value of this parameter to 0 (zero), the default, specifies that the attempt to
generate lift-and-project cuts should continue only if it seems to be helping.

For a brief definition of lift-and-project cuts, see the topic MIP lift-and-project cuts
in the general topic Cuts in the CPLEX User’s Manual. That same topic also
includes a bibliography for further reading about lift-and-project cuts.

Chapter 3. List of CPLEX parameters 69

 The table Parameters for controlling cuts, also in the user’s manual, includes links
to the documentation of other parameters affecting other types of cuts.
Table 28. Values
Value Meaning
-1 Do not generate lift-and-project cuts
0 Automatic: let CPLEX choose; default
1 Generate lift-and-project cuts moderately
2 Generate lift-and-project cuts aggressively
3 Generate lift-and-project cuts very
aggressively

MCF cut switch

Switches on or off generation of multi-commodity flow cuts in a MIP.

Purpose

Switches on or off generation of multi-commodity flow cuts in a MIP.

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_MCFCut CPX_PARAM_MCFCUTS
C++ IloCplex::Param::MIP::Cuts::MCFCut MCFCuts (int)
Java IloCplex.Param.MIP.Cuts.MCFCut MCFCuts (int)
.NET Cplex.Param.MIP.Cuts.MCFCut MCFCuts (int)
OPL mcfcuts
Python parameters.mip.cuts.mcfcut mip.cuts.mcfcut
MATLAB Cplex.Param.mip.cuts.mcfcut mip.cuts.mcfcut
Interactive mip cuts mcfcut mip cuts mcfcut
Identifier 2134 2134

Description

Specifies whether CPLEX should generate multi-commodity flow cuts in a

problem where CPLEX detects the characteristics of a multi-commodity flow
network with arc capacities. By default, CPLEX decides whether or not to generate
such cuts.

To turn off generation of such cuts, set this parameter to -1 (minus one).

CPLEX is able to recognize the structure of a network as represented in many

real-world models. When it recognizes such a network structure, CPLEX is able to
generate cutting planes that usually help solve such problems. In this case, the cuts
that CPLEX generates state that the capacities installed on arcs pointing into a
component of the network must be at least as large as the total flow demand of the
component that cannot be satisfied by flow sources within the component.

For a definition of a multi-commodity flow cut, see the topic Multi-commodity

flow (MCF) cuts in the general topic Cuts in the CPLEX User’s Manual. The table
Parameters for controlling cuts, also in the user’s manual, includes links to the
documentation of other parameters affecting other types of cuts.

70 CPLEX Parameters Reference

 Table 29. Values
Value Meaning
-1 Turn off MCF cuts
0 Automatic: let CPLEX decide whether to
generate MCF cuts; default
1 Generate a moderate number of MCF cuts
2 Generate MCF cuts aggressively

memory reduction switch

Directs CPLEX that it should conserve memory where possible.

Purpose

Reduces use of memory

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Emphasis_Memory CPX_PARAM_MEMORYEMPHASIS
C++ IloCplex::Param::Emphasis::Memory MemoryEmphasis (bool)
Java IloCplex.Param.Emphasis.Memory MemoryEmphasis (bool)
.NET Cplex.Param.Emphasis.Memory MemoryEmphasis (bool)
OPL memoryemphasis
Python parameters.emphasis.memory emphasis.memory
MATLAB Cplex.Param.emphasis.memory emphasis.memory
Interactive emphasis memory emphasis memory
Identifier 1082 1082

Description

Directs CPLEX that it should conserve memory where possible. When you set this
parameter to its nondefault value, CPLEX will choose tactics, such as data
compression or disk storage, for some of the data computed by the simplex,
barrier, and MIP optimizers. Of course, conserving memory may impact
performance in some models. Also, while solution information will be available
after optimization, certain computations that require a basis that has been factored
(for example, for the computation of the condition number Kappa) may be
unavailable.
Table 30. Values
Value bool Symbol Meaning
0 false CPX_OFF Off; do not conserve
memory; default
1 true CPX_ON On; conserve
memory where
possible

MIP callback switch between original model and reduced, presolved

model
Controls whether your callback accesses node information of the original model
(off) or node information of the reduced, presolved model (on, default).

Chapter 3. List of CPLEX parameters 71

 Purpose

MIP callback switch between original model and reduced, presolved model

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_CallbackReducedLP CPX_PARAM_MIPCBREDLP (int)
C++
Java
.NET
OPL
Python
MATLAB
Interactive
Identifier 2055 2055

Description

Controls whether your callback accesses node information of the original model
(off) or node information of the reduced, presolved model (on, default); also
known as the MIP callback reduced LP parameter.

Advanced routines to control MIP callbacks (such as CPXgetcallbacklp ,

CPXsetheuristiccallbackfunc , CPXsetbranchcallbackfunc ,
CPXgetbranchcallbackfunc , CPXsetcutcallbackfunc ,
CPXsetincumbentcallbackfunc , CPXgetcallbacksosinfo , CPXcutcallbackadd ,
CPXcutcallbackaddlocal , and others) consider the setting of this parameter and
access the original model or the reduced, presolved model accordingly.

The routine CPXgetcallbacknodelp is an exception: it always accesses the current

node LP associated with the presolved model, regardless of the setting of this
parameter.

For certain routines, such as CPXcutcallbackadd , when you set the parameter
CPX_PARAM_MIPCBREDLP to zero, you should also set CPX_PARAM_PRELINEAR to zero as
well.

In the C++, Java, .NET, Python, and MATLAB APIs of CPLEX, only the original
model is available to callbacks. In other words, this parameter is effective only for
certain advanced routines of the C API.
Table 31. Values.
Value Symbol Meaning
0 CPX_OFF Off: use original model
1 CPX_ON On: use reduced, presolved
model; default

MIP node log display information

Decides what CPLEX reports to the screen during mixed integer optimization
(MIP).

72 CPLEX Parameters Reference

Purpose

MIP node log display information

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Display CPX_PARAM_MIPDISPLAY
C++ IloCplex::Param::MIP::Display MIPDisplay (int)
Java IloCplex.Param.MIP.Display MIPDisplay (int)
.NET Cplex.Param.MIP.Display MIPDisplay (int)
OPL mipdisplay
Python parameters.mip.display mip.display
MATLAB Cplex.Param.mip.display mip.display
Interactive mip display mip display
Identifier 2012 2012

Description

Decides what CPLEX reports to the screen and records in a log during mixed
integer optimization (MIP).

The amount of information displayed increases with increasing values of this

parameter.
v A setting of 0 (zero) causes no node log to be displayed until the optimal
solution is found.
v A setting of 1 (one) displays an entry for each integer feasible solution found.
Each entry contains:
– the value of the objective function;
– the node count;
– the number of unexplored nodes in the tree;
– the current optimality gap.
v A setting of 2 also generates an entry at a frequency determined by the “MIP
node log interval” on page 75 parameter. At a lower frequency, the log
additionally displays elapsed time in seconds and deterministic time in ticks.
v A setting of 3 gives all the information of option 2 plus additional information:
– At the same frequency as option 2, the node log adds a line specifying the
number of cutting planes added to the problem since the last node log line
was displayed; this additional line is omitted if the number of cuts added
since the last log line is 0 (zero).
– Whenever a MIP start was successfully used to find a new incumbent
solution, that success is recorded in the node log. (This information about
MIP starts is independent of the MIP interval frequency in option 2.)
– For each new incumbent that is found, the node log displays how much time
in seconds and how many deterministic ticks elapsed since the beginning of
optimization. (This information about elapsed time between new incumbents
is independent of the MIP interval frequency in option 2.)
v A setting of 4 additionally generates entries for the LP root relaxation according
to the setting of the parameter to control the “simplex iteration information
display” on page 129 (SimDisplay, CPX_PARAM_SIMDISPLAY).
v A setting of 5 additionally generates entries for the LP subproblems, also
according to the setting of the parameter to control the “simplex iteration
information display” on page 129 (SimDisplay, CPX_PARAM_SIMDISPLAY).

Chapter 3. List of CPLEX parameters 73

 Table 32. Values
Value Meaning
0 No display until optimal solution has been found
1 Display integer feasible solutions
2 Display integer feasible solutions plus an entry at a frequency set
by “MIP node log interval” on page 75; default
3 Display the number of cuts added since previous display;
information about the processing of each successful MIP start;
elapsed time in seconds and elapsed time in deterministic ticks for
integer feasible solutions
4 Display information available from previous options and
information about the LP subproblem at root
5 Display information available from previous options and
information about the LP subproblems at root and at nodes

See also

“MIP node log interval” on page 75, “simplex iteration information display” on
page 129, “network logging display switch” on page 82, and “messages to screen
switch” on page 126

MIP emphasis switch

Controls trade-offs between speed, feasibility, optimality, and moving bounds in
MIP.

Purpose

MIP emphasis switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Emphasis_MIP CPX_PARAM_MIPEMPHASIS
C++ IloCplex::Param::Emphasis::MIP MIPEmphasis (int)
Java IloCplex.Param.Emphasis.MIP MIPEmphasis (int)
.NET Cplex.Param.Emphasis.MIP MIPEmphasis (int)
OPL mipemphasis
Python parameters.emphasis.mip emphasis.mip
MATLAB Cplex.Param.emphasis.mip emphasis.mip
Interactive emphasis mip emphasis mip
Identifier 2058 2058

Description

Controls trade-offs between speed, feasibility, optimality, and moving bounds in

MIP.

With the default setting of BALANCED, CPLEX works toward a rapid proof of an
optimal solution, but balances that with effort toward finding high quality feasible
solutions early in the optimization.

74 CPLEX Parameters Reference

 When this parameter is set to FEASIBILITY, CPLEX frequently will generate more
feasible solutions as it optimizes the problem, at some sacrifice in the speed to the
proof of optimality.

When set to OPTIMALITY, less effort may be applied to finding feasible solutions
early.

With the setting BESTBOUND, even greater emphasis is placed on proving

optimality through moving the best bound value, so that the detection of feasible
solutions along the way becomes almost incidental.

When the parameter is set to HIDDENFEAS, the MIP optimizer works hard to find
high quality feasible solutions that are otherwise very difficult to find, so consider
this setting when the FEASIBILITY setting has difficulty finding solutions of
acceptable quality.
Table 33. Values
Value Symbol Meaning
0 CPX_MIPEMPHASIS_BALANCED Balance optimality and feasibility;
default
1 CPX_MIPEMPHASIS_FEASIBILITY Emphasize feasibility over
optimality
2 CPX_MIPEMPHASIS_OPTIMALITY Emphasize optimality over
feasibility
3 CPX_MIPEMPHASIS_BESTBOUND Emphasize moving best bound
4 CPX_MIPEMPHASIS_HIDDENFEAS Emphasize finding hidden feasible
solutions

MIP node log interval

Controls the frequency of node logging when the MIP display parameter is set
higher than 1 (one).

Purpose

MIP node log interval

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Interval CPX_PARAM_MIPINTERVAL
C++ IloCplex::Param::MIP::Interval MIPInterval (long)
Java IloCplex.Param.MIP.Interval MIPInterval (long)
.NET Cplex.Param.MIP.Interval MIPInterval (long)
OPL mipinterval
Python parameters.mip.interval mip.interval
MATLAB Cplex.Param.mip.interval mip.interval
MATLAB CPLEX Toolbox compatibility mip.interval
MATLAB Optimization Toolbox compatibility NodeDisplayInterval
Interactive mip interval mip interval
Identifier 2013 2013

Chapter 3. List of CPLEX parameters 75

 Description

Controls the frequency of node logging when the MIP display parameter (“MIP
node log display information” on page 72) is set higher than 1 (one). Frequency
must be an integer; it may be 0 (zero), positive, or negative.

By default, CPLEX displays new information in the node log during a MIP solve
at relatively high frequency during the early stages of solving a MIP model, and
adds lines to the log at progressively longer intervals as solving continues. In other
words, CPLEX logs information frequently in the beginning and progressively less
often as it works.

When the value is a positive integer n, CPLEX displays new incumbents, plus it
displays a new line in the log every n nodes.

When the value is a negative integer n, CPLEX displays new incumbents, and the
negative value determines how much processing CPLEX does before it displays a
new line in the node log. A negative value close to zero means that CPLEX
displays new lines in the log frequently. A negative value far from zero means that
CPLEX displays new lines in the log less frequently. In other words, a negative
value of this parameter contracts or dilates the interval at which CPLEX displays
information in the node log.
Table 34. Values
Value Meaning
n<0 Display new incumbents, and display a log line frequently at the
beginning of solving and less frequently as solving progresses
0 (zero) automatic: let CPLEX decide the frequency to log nodes (default)
n>0 Display new incumbents, and display a log line every n nodes

See also

“MIP node log display information” on page 72

MIP kappa computation

Sets the strategy for computing statistics about MIP kappa

Purpose

MIP kappa computation

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_KappaStats CPX_PARAM_MIPKAPPASTATS
C++ IloCplex::Param::MIP::Strategy::KappaStats MIPKappaStats (int)
Java IloCplex.Param.MIP.Strategy.KappaStats MIPKappaStats (int)
.NET Cplex.Param.MIP.Strategy.KappaStats MIPKappaStats (int)
OPL mipkappastats
Python parameters.mip.strategy.kappastats mip.strategy.kappastats
MATLAB Cplex.Param.mip.strategy.kappastats mip.strategy.kappastats
Interactive mip strategy kappastats mip strategy kappastats
Identifier 2137 2137

76 CPLEX Parameters Reference

Description

Sets the strategy for CPLEX to gather statistics about the MIP kappa of
subproblems of a MIP.

What is MIP kappa?

MIP kappa summarizes the distribution of the condition number of the optimal
bases CPLEX encountered during the solution of a MIP model. That summary may
let you know more about the numerical difficulties of your MIP model.

When can you compute MIP kappa?

Because MIP kappa (as a statistical distribution) requires CPLEX to compute the
condition number of the optimal bases of the subproblems during branch-and-cut
search, you can compute the MIP kappa only when CPLEX solves the subproblem
with its simplex optimizer. In other words, in order to obtain results with this
parameter, you can not use the sifting optimizer nor the barrier without crossover
to solve the subproblems. See the parameters “MIP subproblem algorithm” on
page 86 (CPX_PARAM_SUBALG, NodeAlg) and “algorithm for initial MIP relaxation” on
page 123 (CPX_PARAM_STARTALG, RootAlg) for more details about those choices.

What are the performance trade-offs for computing MIP kappa?

Computing the kappa of a subproblem has a cost. In fact, computing MIP kappa
for the basis matrices can be computationally expensive and thus generally slows
down the solution of a problem. Therefore,

the automatic setting CPX_MIPKAPPA_AUTO tells CPLEX generally not to compute

MIP kappa, but in cases where the parameter “numerical precision emphasis” on
page 89 (CPX_PARAM_NUMERICALEMPHASIS, NumericalEmphasis) is turned on, that is,
set to 1 (one), CPLEX computes MIP kappa for a sample of subproblems.

The value CPX_MIPKAPPA_SAMPLE leads to a negligible performance degradation on

average, but can slow down the branch-and-cut exploration by as much as 10% on
certain models.

The value CPX_MIPKAPPA_FULL leads to a 2% performance degradation on average,

but can significantly slow the branch-and-cut exploration on certain models.

In practice, the value CPX_MIPKAPPA_SAMPLE is a good trade-off between

performance and accuracy of statistics.

If you need very accurate statistics, then use value CPX_MIPKAPPA_FULL.

Table 35. Values
Value Symbol Meaning
–1 CPX_MIPKAPPA_OFF No MIP kappa statistics
0 CPX_MIPKAPPA_AUTO Automatic: let CPLEX
decide; default
1 CPX_MIPKAPPA_SAMPLE Compute MIP kappa for a
sample of subproblems
2 CPX_MIPKAPPA_FULL Compute MIP kappa for all
subproblems

Chapter 3. List of CPLEX parameters 77

MIP priority order switch
Decides whether to use the priority order, if one exists, for the next mixed integer
optimization.

Purpose

MIP priority order switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_Order CPX_PARAM_MIPORDIND
C++ IloCplex::Param::MIP::Strategy::Order MIPOrdInd (bool)
Java IloCplex.Param.MIP.Strategy.Order MIPOrdInd (bool)
.NET Cplex.Param.MIP.Strategy.Order MIPOrdInd (bool)
OPL mipordind
Python parameters.mip.strategy.order mip.strategy.order
MATLAB Cplex.Param.mip.strategy.order mip.strategy.order
Interactive mip strategy order mip strategy order
Identifier 2020 2020

Description

Decides whether to use the priority order, if one exists, for the next mixed integer
optimization.
Table 36. Values
Value bool Symbol Meaning
false CPX_OFF Off: do not use
priority order
true CPX_ON On: use priority
order, if it exists;
default

MIP priority order generation

Selects the type of generic priority order to generate when no priority order is
present.

Purpose

MIP priority order generation

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_OrderType CPX_PARAM_MIPORDTYPE
C++ IloCplex::Param::MIP::OrderType MIPOrdType (int)
Java IloCplex.Param.MIP.OrderType MIPOrdType (int)
.NET Cplex.Param.MIP.OrderType MIPOrdType (int)
OPL mipordtype
Python parameters.mip.ordertype mip.ordertype
MATLAB Cplex.Param.mip.ordertype mip.ordertype
Interactive mip ordertype mip ordertype
Identifier 2032 2032

78 CPLEX Parameters Reference

 Description

Selects the type of generic priority order to generate when no priority order is
present.
Table 37. Values
Value Symbol Meaning
0 default Do not generate a priority
order
1 CPX_MIPORDER_COST Use decreasing cost
2 CPX_MIPORDER_BOUNDS Use increasing bound range
3 CPX_MIPORDER_SCALEDCOST Use increasing cost per
coefficient count

MIP dynamic search switch

Sets the search strategy for a mixed integer program (MIP).

Purpose

MIP dynamic search switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_Search CPX_PARAM_MIPSEARCH
C++ IloCplex::Param::MIP::Strategy::Search MIPSearch (int)
Java IloCplex.Param.MIP.Strategy.Search MIPSearch (int)
.NET Cplex.Param.MIP.Strategy.Search MIPSearch (int)
OPL mipsearch
Python parameters.mip.strategy.search mip.strategy.search
MATLAB Cplex.Param.mip.strategy.search mip.strategy.search
Interactive mip strategy search mip strategy search
Identifier 2109 2109

Description

Sets the search strategy for a mixed integer program (MIP). By default, CPLEX
chooses whether to apply dynamic search or conventional branch and cut based on
characteristics of the model and the presence (or absence) of callbacks.

Only informational callbacks are compatible with dynamic search. For more detail
about informational callbacks and how to create and install them in your
application, see Informational callbacks in the CPLEX User’s Manual.

To benefit from dynamic search, a MIP must not include query callbacks. In other
words, query callbacks are not compatible with dynamic search. For a more
detailed definition of query or diagnostic callbacks, see Query or diagnostic
callbacks in the CPLEX User’s Manual.

To benefit from dynamic search, a MIP must not include control callbacks (that is,
callbacks that alter the search path through the solution space). In other words,
control callbacks are not compatible with dynamic search. These control callbacks
are identified as advanced in the reference manuals of the APIs. If control callbacks
are present in your application, CPLEX will disable dynamic search, issue a

Chapter 3. List of CPLEX parameters 79

 warning, and apply only static branch and cut. If you want to control the search
yourself, for example, through advanced control callbacks, then you should set this
parameter to 1 (one) to disable dynamic search and to apply conventional branch
and cut.
Table 38. Values
Value Symbolic Name Meaning
0 CPX_MIPSEARCH_AUTO Automatic: let CPLEX
choose; default
1 CPX_MIPSEARCH_TRADITIONAL Apply traditional branch and
cut strategy; disable dynamic
search
2 CPX_MIPSEARCH_DYNAMIC Apply dynamic search

MIQCP strategy switch

Sets the strategy that CPLEX uses to solve a quadratically constrained mixed
integer program (MIQCP).

Purpose

MIQCP strategy switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_MIQCPStrat CPX_PARAM_MIQCPSTRAT
C++ IloCplex::Param::MIP::Strategy::MIQCPStrat MIQCPStrat (int)
Java IloCplex.Param.MIP.Strategy.MIQCPStrat MIQCPStrat (int)
.NET Cplex.Param.MIP.Strategy.MIQCPStrat MIQCPStrat (int)
OPL miqcpstrat
Python parameters.mip.strategy.miqcpstrat mip.strategy.miqcpstrat
MATLAB Cplex.Param.mip.strategy.miqcpstrat mip.strategy.miqcpstrat
Interactive mip strategy miqcpstrat mip strategy miqcpstrat
Identifier 2110 2110

Description

Sets the strategy that CPLEX uses to solve a quadratically constrained mixed
integer program (MIQCP).

This parameter controls how MIQCPs ( that is, mixed integer programs with one
or more constraints including quadratic terms) are solved. For more detail about
the types of quadratically constrained models that CPLEX solves, see Identifying a
quadratically constrained program (QCP) in the CPLEX User’s Manual.

At the default setting of 0 (zero), CPLEX automatically chooses a strategy.

When you set this parameter to the value 1 (one), you tell CPLEX to solve a QCP
relaxation of the model at each node.

When you set this parameter to the value 2, you tell CPLEX to attempt to solve an
LP relaxation of the model at each node.

CPLEX uses a linear approximation of the quadratic constraints, adding cone cuts
as it proceeds. This approach has advantages that can yield better overall

80 CPLEX Parameters Reference

 performance, despite the disadvantage of approximating the quadratic constraints.
First advantage: when you solve a QCP relaxation at each node, the barrier method
used to do the solve cannot take advantage of advanced start information. In
contrast, solving LP relaxations can use advanced starts, potentially making the
node relaxations run faster. Second advantage: the second order cone
transformations used to solve the QCP relaxation occasionally create numerical
instabilities that make the QCP relaxation difficult to solve. LP relaxations require
fewer transformations. Also, LP relaxations can use either the barrier or simplex
methods, so they are less likely to have such issues.

For some models, the setting 2 may be more effective than 1 (one). You may need
to experiment with this parameter to determine the best setting for your model.

Specifically, if the node log indicates long solve times for a QCP relaxation,
consider setting this parameter to the value 2. Conversely, if you see that the best
node value appears to move very slowly, the linear approximation may not be
particularly accurate; in such cases, setting the parameter to value 1 (one) may
improve performance.
Table 39. Values
Value Meaning
0 Automatic: let CPLEX choose; default
1 Solve a QCP node relaxation at each node
2 Solve an LP node relaxation at each node

MIP MIR (mixed integer rounding) cut switch

Decides whether or not to generate MIR cuts (mixed integer rounding cuts) for the
problem.

Purpose

MIP MIR (mixed integer rounding) cut switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_MIRCut CPX_PARAM_MIRCUTS
C++ IloCplex::Param::MIP::Cuts::MIRCut MIRCuts (int)
Java IloCplex.Param.MIP.Cuts.MIRCut MIRCuts (int)
.NET Cplex.Param.MIP.Cuts.MIRCut MIRCuts (int)
OPL mircuts
Python parameters.mip.cuts.mircut mip.cuts.mircut
MATLAB Cplex.Param.mip.cuts.mircut mip.cuts.mircut
Interactive mip cuts mircut mip cuts mircut
Identifier 2052 2052

Description

Decides whether or not to generate MIR cuts (mixed integer rounding cuts) for the
problem. The value 0 (zero), the default, specifies that the attempt to generate MIR
cuts should continue only if it seems to be helping.

For a definition of a MIR cut, see the topic Mixed integer rounding (MIR) cuts in
the general topic Cuts in the CPLEX User’s Manual. The table Parameters for

Chapter 3. List of CPLEX parameters 81

 controlling cuts, also in the user’s manual, includes links to the documentation of
other parameters affecting other types of cuts.

Value Meaning
-1 Do not generate MIR cuts
0 Automatic: let CPLEX choose; default
1 Generate MIR cuts moderately
2 Generate MIR cuts aggressively

precision of numerical output in MPS and REW file formats

Decides the precision of numerical output in the MPS and REW file formats.

Purpose

Precision of numerical output in MPS and REW file formats

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Output_MPSLong CPX_PARAM_MPSLONGNUM
C++ IloCplex::Param::Output::MPSLong MPSLongNum (bool)
Java IloCplex.Param.Output.MPSLong MPSLongNum (bool)
.NET Cplex.Param.Output.MPSLong MPSLongNum (bool)
OPL not available
Python parameters.output.mpslong output.mpslong
MATLAB Cplex.Param.output.mpslong output.mpslong
Interactive output mpslong output mpslong
Identifier 1081 1081

Description

Decides the precision of numerical output in the MPS and REW file formats. When
this parameter is set to its default value 1 (one), numbers are written to MPS files
in full-precision; that is, up to 15 significant digits may be written. The setting 0
(zero) writes files that correspond to the standard MPS format, where at most 12
characters can be used to represent a value. This limit may result in loss of
precision.

Value bool Symbol Meaning

0 false CPX_OFF Off: use limited MPS
precision
1 true CPX_ON On: use
full-precision; default

See also

MPS file format: industry standard

network logging display switch

Decides what CPLEX reports to the screen during network optimization.

82 CPLEX Parameters Reference

 Purpose

Network logging display switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Network_Display CPX_PARAM_NETDISPLAY
C++ IloCplex::Param::Network::Display NetDisplay (int)
Java IloCplex.Param.Network.Display NetDisplay (int)
.NET Cplex.Param.Network.Display NetDisplay (int)
OPL netdisplay
Python not available
MATLAB Cplex.Param.network.display not available
Interactive network display network display
Identifier 5005 5005

Description

Decides what CPLEX reports to the screen during network optimization. Settings 1
and 2 differ only during Phase I. Setting 2 shows monotonic values, whereas 1
usually does not.

Value Symbol Meaning

0 CPXNET_NO_DISPLAY_OBJECTIVE No display
1 CPXNET_TRUE_OBJECTIVE Display true objective values
2 CPXNET_PENALIZE_OBJECTIVE Display penalized objective
values; default

network optimality tolerance

Specifies the optimality tolerance for network optimization.

Purpose

Optimality tolerance for network optimization

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Network_Tolerances_Optimality CPX_PARAM_NETEPOPT
C++ IloCplex::Param::Network::Tolerances::Optimality NetEpOpt (double)
Java IloCplex.Param.Network.Tolerances.Optimality NetEpOpt (double)
.NET Cplex.Param.Network.Tolerances.Optimality NetEpOpt (double)
OPL netepopt
Python not available
MATLAB Cplex.Param.network.tolerances.optimality not available
Interactive network tolerances optimality network tolerances optimality
Identifier 5002 5002

Description

Specifies the optimality tolerance for network optimization; that is, the amount a
reduced cost may violate the criterion for an optimal solution.

Values

Chapter 3. List of CPLEX parameters 83

 Any number from 1e-11 to 1e-1; default: 1e-6.

network primal feasibility tolerance

Specifies feasibility tolerance for network primal optimization. The feasibility
tolerance specifies the degree to which the flow value of a model may violate its
bounds.

Purpose

Feasibility tolerance for network primal optimization

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Network_Tolerances_Feasibility CPX_PARAM_NETEPRHS
C++ IloCplex::Param::Network::Tolerances::Feasibility
NetEpRHS (double)
Java IloCplex.Param.Network.Tolerances.Feasibility NetEpRHS (double)
.NET Cplex.Param.Network.Tolerances.Feasibility NetEpRHS (double)
OPL neteprhs
Python not available not available
MATLAB Cplex.Param.network.tolerances.feasibility not available
Interactive network tolerances feasibility network tolerances feasibility
Identifier 5003 5003

Description

Specifies feasibility tolerance for network primal optimization. The feasibility

tolerance specifies the degree to which the flow value of a model may violate its
bounds. This tolerance influences the selection of an optimal basis and can be reset
to a higher value when a problem is having difficulty maintaining feasibility
during optimization. You may also wish to lower this tolerance after finding an
optimal solution if there is any doubt that the solution is truly optimal. If the
feasibility tolerance is set too low, CPLEX may falsely conclude that a problem is
infeasible. If you encounter reports of infeasibility during Phase II of the
optimization, a small adjustment in the feasibility tolerance may improve
performance.

Values

Any number from 1e-11 to 1e-1; default: 1e-6.

simplex network extraction level

Establishes the level of network extraction for network simplex optimization.

Purpose
Simplex network extraction level

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Network_NetFind CPX_PARAM_NETFIND
C++ IloCplex::Param::Network::NetFind NetFind (int)
Java IloCplex.Param.Network.NetFind NetFind (int)
.NET Cplex.Param.Network.NetFind NetFind (int)
OPL netfind
Python not available not available

84 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
MATLAB Cplex.Param.network.netfind not available
Interactive network netfind network netfind
Identifier 1022 1022

Description

Establishes the level of network extraction for network simplex optimization. The
default value is suitable for recognizing commonly used modeling approaches
when representing a network problem within an LP formulation.
Table 40. Values
Value Symbol Meaning
1 CPX_NETFIND_PURE Extract pure network only
2 CPX_NETFIND_REFLECT Try reflection scaling; default
3 CPX_NETFIND_SCALE Try general scaling

network simplex iteration limit

Sets the maximum number of iterations to be performed before the algorithm
terminates without reaching optimality.

Purpose

Network simplex iteration limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Network_Iterations CPX_PARAM_NETITLIM
C++ IloCplex::Param::Network::Iterations NetItLim (long)
Java IloCplex::Param::Network::Iterations NetItLim (long)
.NET Cplex::Param::Network::Iterations NetItLim (long)
OPL netitlim
Python not available not available
MATLAB Cplex.Param.network.iterations not available
Interactive network iterations network iterations
Identifier 5001 5001

Description

Sets the maximum number of iterations to be performed before the algorithm

terminates without reaching optimality.

Values

Any nonnegative integer; default: 9223372036800000000.

network simplex pricing algorithm

Specifies the pricing algorithm for network simplex optimization.

Chapter 3. List of CPLEX parameters 85

 Purpose

Network simplex pricing algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Network_Pricing CPX_PARAM_NETPPRIIND
C++ IloCplex::Param::Network::Pricing NetPPriInd (int)
Java IloCplex.Param.Network.Pricing NetPPriInd (int)
.NET Cplex.Param.Network.Pricing NetPPriInd (int)
OPL netppriind
Python not available not available
MATLAB Cplex.Param.network.pricing not available
Interactive network pricing network pricing
Identifier 5004 5004

Description

Specifies the pricing algorithm for network simplex optimization. The default (0)
shows best performance for most problems, and currently is equivalent to 3.
Table 41. Values
Value Symbol Meaning
0 CPXNET_PRICE_AUTO Automatic: let CPLEX choose;
default
1 CPXNET_PRICE_PARTIAL Partial pricing
2 CPXNET_PRICE_MULT_PART Multiple partial pricing
3 CPXNET_PRICE_SORT_MULT_PART Multiple partial pricing with
sorting

MIP subproblem algorithm

Decides which continuous optimizer will be used to solve the subproblems in a
MIP, after the initial relaxation.

Purpose

MIP subproblem algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_SubAlgorithm CPX_PARAM_SUBALG
C++ IloCplex::Param::NodeAlgorithm NodeAlg (int)
Java IloCplex.Param.NodeAlgorithm NodeAlg (int)
.NET Cplex.Param.NodeAlgorithm NodeAlg (int)
OPL nodealg
Python parameters.mip.strategy.subalgorithm mip.strategy.subalgorithm
MATLAB Cplex.Param.mip.strategy.subalgorithm mip.strategy.subalgorithm
Interactive mip strategy subalgorithm mip strategy subalgorithm
Identifier 2026 2026

Description

Decides which continuous optimizer will be used to solve the subproblems in a

MIP, after the initial relaxation.

86 CPLEX Parameters Reference

 The default Automatic setting (0 zero) of this parameter currently selects the dual
simplex optimizer for subproblem solution for MILP and MIQP. The Automatic
setting may be expanded in the future so that CPLEX chooses the algorithm based
on additional characteristics of the model.

For MILP (integer constraints and otherwise continuous variable), all settings are
permitted.

For MIQP (integer constraints and positive semi-definite quadratic terms in

objective), setting 3 (Network) is not permitted, and setting 5 (Sifting) reverts to 0
(Automatic).

For MIQCP (integer constraints and positive semi-definite quadratic terms among
the constraints), only the Barrier optimizer is implemented, and therefore no
settings other than 0 (Automatic) and 4 (Barrier) are permitted.
Table 42. Values
Value Symbol Meaning
0 CPX_ALG_AUTOMATIC Automatic: let CPLEX choose;
default
1 CPX_ALG_PRIMAL Primal simplex
2 CPX_ALG_DUAL Dual simplex
3 CPX_ALG_NET Network simplex
4 CPX_ALG_BARRIER Barrier
5 CPX_ALG_SIFTING Sifting

node storage file switch

Used when working memory (CPX_PARAM_WORKMEM, WorkMem) has been exceeded by
the size of the tree.

Purpose

Node storage file switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_File CPX_PARAM_NODEFILEIND
C++ IloCplex::Param::MIP::Strategy::File NodeFileInd (int)
Java IloCplex.Param.MIP.Strategy.File NodeFileInd (int)
.NET Cplex.Param.MIP.Strategy.File NodeFileInd (int)
OPL nodefileind
Python parameters.mip.strategy.file mip.strategy.file
MATLAB Cplex.Param.mip.strategy.file mip.strategy.file
Interactive mip strategy file mip strategy file
Identifier 2016 2016

Description

Used when working memory (CPX_PARAM_WORKMEM, WorkMem) has been exceeded by

the size of the tree. If the node file parameter is set to zero when the tree memory
limit is reached, optimization is terminated. Otherwise, a group of nodes is
removed from the in-memory set as needed. By default, CPLEX transfers nodes to
node files when the in-memory set is larger than 128 MBytes, and it keeps the
Chapter 3. List of CPLEX parameters 87
 resulting node files in compressed form in memory. At settings 2 and 3, the node
files are transferred to disk, in uncompressed and compressed form respectively,
into a directory named by the working directory parameter (CPX_PARAM_WORKDIR,
WorkDir), and CPLEX actively manages which nodes remain in memory for
processing.

Value Meaning
0 No node file
1 Node file in memory and compressed;
default
2 Node file on disk
3 Node file on disk and compressed

Related reference:
“directory for working files” on page 149
Specifies the name of an existing directory into which CPLEX may store temporary
working files.
“memory available for working storage” on page 150
Specifies an upper limit on the amount of central memory, in megabytes, that
CPLEX is permitted to use for working memory.

MIP node limit

Sets the maximum number of nodes solved before the algorithm terminates
without reaching optimality.

Purpose

MIP node limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_Nodes CPX_PARAM_NODELIM
C++ IloCplex::Param::MIP::Limits::Nodes NodeLim (long)
Java IloCplex.Param.MIP.Limits.Nodes NodeLim (long)
.NET Cplex.Param.MIP.Limits.Nodes NodeLim (long)
OPL nodelim
Python parameters.mip.limits.nodes mip.limits.nodes
MATLAB Cplex.Param.mip.limits.nodes mip.limits.nodes
Cplex class
compatible
MATLAB CPLEX Toolbox compatible mip.limits.nodes
MATLAB Optimization Toolbox compatible MaxNodes
Cplex class
compatible
Interactive mip limits nodes mip limits nodes
Identifier 2017 2017

Description

Sets the maximum number of nodes solved before the algorithm terminates
without reaching optimality. When this parameter is set to 0 (zero), CPLEX
completes processing at the root; that is, it creates cuts and applies heuristics at the
root. When this parameter is set to 1 (one), it allows branching from the root; that
is, nodes are created but not solved.

88 CPLEX Parameters Reference

 Values

Any nonnegative integer; default: 9223372036800000000.

MIP node selection strategy

Used to set the rule for selecting the next node to process when backtracking.

Purpose

MIP node selection strategy

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_NodeSelect CPX_PARAM_NODESEL
C++ IloCplex::Param::MIP::Strategy::NodeSelect NodeSel (int)
Java IloCplex.Param.MIP.Strategy.NodeSelect NodeSel (int)
.NET Cplex.Param.MIP.Strategy.NodeSelect NodeSel (int)
OPL nodesel
Python parameters.mip.strategy.nodeselect mip.strategy.nodeselect
MATLAB Cplex.Param.mip.strategy.nodeselect mip.strategy.nodeselect
Cplex class
compatibity
MATLAB CPLEX Toolbox compatibility mip.strategy.nodeselect
MATLAB Optimization Toolbox compatibility NodeSearchStrategy
Interactive mip strategy nodeselect mip strategy nodeselect
Identifier 2018 2018

Description

Sets the rule for selecting the next node to process when the search is backtracking.
The depth-first search strategy chooses the most recently created node. The
best-bound strategy chooses the node with the best objective function for the
associated LP relaxation. The best-estimate strategy selects the node with the best
estimate of the integer objective value that would be obtained from a node once all
integer infeasibilities are removed. An alternative best-estimate search is also
available.
Table 43. Values
Value Symbol Meaning
0 CPX_NODESEL_DFS Depth-first search
1 CPX_NODESEL_BESTBOUND Best-bound search; default
2 CPX_NODESEL_BESTEST Best-estimate search
3 CPX_NODESEL_BESTEST_ALT Alternative best-estimate
search

numerical precision emphasis

Emphasizes precision in numerically unstable or difficult problems.

Chapter 3. List of CPLEX parameters 89

 Purpose

Numerical precision emphasis

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Emphasis_Numerical CPX_PARAM_NUMERICALEMPHASIS
C++ IloCplex::Param::Emphasis::Numerical NumericalEmphasis (bool)
Java IloCplex.Param.Emphasis.Numerical NumericalEmphasis (bool)
.NET Cplex.Param.Emphasis.Numerical NumericalEmphasis (bool)
OPL numericalemphasis
Python parameters.emphasis.numerical emphasis.numerical
MATLAB Cplex.Param.emphasis.numerical emphasis.numerical
Interactive emphasis numerical emphasis numerical
Identifier 1083 1083

Description

Emphasizes precision in numerically unstable or difficult problems. This parameter

lets you specify to CPLEX that it should emphasize precision in numerically
difficult or unstable problems, with consequent performance trade-offs in time and
memory.
Table 44. Values
Value bool Symbol Meaning
0 false CPX_OFF Do not emphasize
numerical precision;
default
1 true CPX_ON Exercise extreme
caution in
computation

nonzero element read limit

Specifies a limit for the number of nonzero elements to read for an allocation of
memory.

Purpose

Nonzero element read limit

API Parameter Name Name prior to V12.6.0

C 64-bit CPXPARAM_Read_Nonzeros CPX_PARAM_NZREADLIM
(CPXLONG)
C 32-bit CPXPARAM_Read_Nonzeros CPX_PARAM_NZREADLIM
(CPXINT)
C++ IloCplex::Param::Read::Nonzeros NzReadLim (CPXLONG)
Java IloCplex.Param.Read.Nonzeros NzReadLim (CPXLONG)
.NET Cplex.Param.Read.Nonzeros NzReadLim (CPXLONG)
OPL not available
Python parameters.read.nonzeros read.nonzeros
MATLAB Cplex.Param.read.nonzeros read.nonzeros
Interactive read nonzeros read nonzeros
Identifier 1024 1024

90 CPLEX Parameters Reference

 Description

Specifies a limit for the number of nonzero elements to read for an allocation of
memory. This parameter does not restrict the size of a problem. Rather, it indirectly
specifies the default amount of memory that will be pre-allocated before a problem
is read from a file. If the limit is exceeded, more memory is automatically
allocated.

Values

Any integer from 0 to CPX_BIGINT or CPX_BIGLONG, depending on integer type;

default: 250 000.

absolute objective difference cutoff

Used to update the cutoff each time a mixed integer solution is found.

Purpose
Absolute objective difference cutoff

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Tolerances_ObjDifference CPX_PARAM_OBJDIF
C++ IloCplex::Param::MIP::Tolerances::ObjDifference ObjDif (double)
Java IloCplex.Param.MIP.Tolerances.ObjDifference ObjDif (double)
.NET Cplex.Param.MIP.Tolerances.ObjDifference ObjDif (double)
OPL objdif
Python parameters.mip.tolerances.objdifference mip.tolerances.objdifference
MATLAB Cplex.Param.mip.tolerances.objdifference mip.tolerances.objdifference
Interactive mip tolerances objdifference mip tolerances objdifference
Identifier 2019 2019

Description

Used to update the cutoff each time a mixed integer solution is found. This
absolute value is subtracted from (added to) the newly found integer objective
value when minimizing (maximizing). This forces the mixed integer optimization
to ignore integer solutions that are not at least this amount better than the best one
found so far.

The objective difference parameter can be adjusted to improve problem solving

efficiency by limiting the number of nodes; however, setting this parameter at a
value other than zero (the default) can cause some integer solutions, including the
true integer optimum, to be missed.

Negative values for this parameter can result in some integer solutions that are
worse than or the same as those previously generated, but does not necessarily
result in the generation of all possible integer solutions.

Values

Any number; default: 0.0.

Related reference:
“relative objective difference cutoff” on page 117
Used to update the cutoff each time a mixed integer solution is found.

Chapter 3. List of CPLEX parameters 91

lower objective value limit
Sets a lower limit on the value of the objective function in the simplex algorithms.

Purpose

Lower objective value limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Limits_LowerObj CPX_PARAM_OBJLLIM
C++ IloCplex::Param::Simplex::Limits::LowerObj ObjLLim (double)
Java IloCplex.Param.Simplex.Limits.LowerObj ObjLLim (double)
.NET Cplex.Param.Simplex.Limits.LowerObj ObjLLim (double)
OPL objllim
Python parameters.simplex.limits.lowerobj simplex.limits.lowerobj
MATLAB Cplex.Param.simplex.limits.lowerobj simplex.limits.lowerobj
Interactive simplex limits lowerobj simplex limits lowerobj
Identifier 1025 1025

Description

Sets a lower limit on the value of the objective function in the simplex algorithms.
Setting a lower objective function limit causes CPLEX to halt the optimization
process when the minimum objective function value limit has been reached. This
limit applies only during Phase II of the simplex algorithm in minimization
problems.

Tip:

This parameter is not effective with the conflict refiner nor with FeasOpt. That is,
neither of those tools can analyze an infeasibility introduced by this parameter. If
you want to analyze such a condition, add an explicit objective constraint, such as
obj >= c, to your model instead before you invoke either of those tools.

Values

Any number; default: -1e+75.

upper objective value limit

Sets an upper limit on the value of the objective function in the simplex
algorithms.

Purpose

Upper objective value limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Limits_UpperObj CPX_PARAM_OBJULIM
C++ IloCplex::Param::Simplex::Limits::UpperObj ObjULim (double)
Java IloCplex.Param.Simplex.Limits.UpperObj ObjULim (double)
.NET Cplex.Param.Simplex.Limits.UpperObj ObjULim (double)
OPL objulim
Python parameters.simplex.limits.upperobj simplex.limits.upperobj
MATLAB Cplex.Param.simplex.limits.upperobj simplex.limits.upperobj

92 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Interactive simplex limits upperobj simplex limits upperobj
Identifier 1026 1026

Description

Sets an upper limit on the value of the objective function in the simplex
algorithms. Setting an upper objective function limit causes CPLEX to halt the
optimization process when the maximum objective function value limit has been
reached. This limit applies only during Phase II of the simplex algorithm in
maximization problems.

Tip:

This parameter is not effective with the conflict refiner nor with FeasOpt. That is,
neither of those tools can analyze an infeasibility introduced by this parameter. If
you want to analyze such a condition, add an explicit objective constraint, such as
obj <= c. to your model instead before you invoke either of those tools.

Values

Any number; default: 1e+75.

parallel mode switch

Sets the parallel optimization mode. Possible modes are automatic, deterministic,
and opportunistic.

Purpose

Parallel mode switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Parallel CPX_PARAM_PARALLELMODE
C++ IloCplex::Param::Parallel ParallelMode (int)
Java IloCplex.Param.Parallel ParallelMode (int)
.NET Cplex.Param.Parallel ParallelMode (int)
OPL parallelmode
Python parameters.parallel parallel
MATLAB Cplex.Param.parallel parallel
Interactive parallel parallel
Identifier 1109 1109

Description

Sets the parallel optimization mode. Possible modes are automatic, deterministic,
and opportunistic.

In this context, deterministic means that multiple runs with the same model at the
same parameter settings on the same platform will reproduce the same solution
path and results. In contrast, opportunistic implies that even slight differences in
timing among threads or in the order in which tasks are executed in different
threads may produce a different solution path and consequently different timings
or different solution vectors during optimization executed in parallel threads. In

Chapter 3. List of CPLEX parameters 93

 multithreaded applications, the opportunistic setting entails less synchronization
between threads and consequently may provide better performance.

By default, CPLEX applies as much parallelism as possible while still achieving

deterministic results. That is, when you run the same model twice on the same
platform with the same parameter settings, you will see the same solution and
optimization run. This condition is referred to as the deterministic mode.

More opportunities to exploit parallelism are available if you do not require

determinism. In other words, CPLEX can find more possibilities for parallelism if
you do not require an invariant, repeatable solution path and precisely the same
solution vector. To use all available parallelism, you need to select the opportunistic
parallel mode. In this mode, CPLEX will use all opportunities for parallelism in
order to achieve best performance.

However, in opportunistic mode, the actual optimization may differ from run to
run, including the solution time itself and the path traveled in the search.

Deterministic and sequential optimization

Parallel MIP optimization can be opportunistic or deterministic.

Parallel barrier optimization is only deterministic.

Concurrent optimization can be opportunistic or deterministic. In either mode,

when six or more threads are available, concurrent optimization launches primal
simplex, dual simplex, and barrier optimizers by default. If ten threads are
available to CPLEX, concurrent optimization launches primal simplex, dual
simplex, barrier, and sifting (if sifting is applicable to the current type of problem).

Tip: This parallel mode parameter also applies to distributed parallel MIP
optimization. That is, you can instruct CPLEX to execute distributed parallel MIP
optimization either opportunistically or deterministically.

Callbacks and MIP optimization

If callbacks other than informational callbacks are used for solving a MIP, the order
in which the callbacks are called cannot be guaranteed to remain deterministic, not
even in deterministic mode. Thus, to make sure of deterministic runs when the
parallel mode parameter and the global default thread count parameter are at their
default settings, CPLEX reverts to sequential solving of the MIP in the presence of
query callbacks, diagnostic callbacks, or control callbacks.

Consequently, if your application invokes query, diagnostic, or control callbacks,

and you still prefer deterministic parallel search, you can choose value 1 (one),
overriding the automatic setting and turning on deterministic parallel search. It is
then your responsibility to make sure that your callbacks do not perform
operations that could lead to opportunistic behavior and are implemented in a
thread-safe way. To meet these conditions, your application must not store and
must not update any information in the callbacks.

Determinism versus opportunism

This parameter also allows you to turn off this default setting by choosing value -1
(minus one). Cases where you might wish to turn off deterministic search include
situations where you want to take advantage of possibly faster performance of

94 CPLEX Parameters Reference

 opportunistic parallel MIP optimization in multiple threads after you have
confirmed that deterministic parallel MIP optimization produced the results you
expected.
Table 45. Values
Value Symbolic Constant Symbolic Constant Meaning

Callable Library Concert Technology

-1 CPX_PARALLEL_OPPORTUNISTIC
Opportunistic Enable opportunistic
parallel search mode
0 CPX_PARALLEL_AUTO AutoParallel Automatic: let CPLEX
decide whether to
invoke deterministic or
opportunistic search;
default
1 CPX_PARALLEL_DETERMINISTIC
Deterministic Enable deterministic
parallel search mode

See also: “global default thread count” on page 139: CPX_PARAM_THREADS, Threads

simplex perturbation switch

Decides whether to perturb problems.

Purpose

Simplex perturbation switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Perturbation_Indicator CPX_PARAM_PERIND
C++ IloCplex::Param::Simplex::Perturbation::Indicator PerInd (bool)
Java IloCplex.Param.Simplex.Perturbation.Indicator PerInd (bool)
.NET Cplex.Param.Simplex.Perturbation.Indicator PerInd (bool)
OPL perind
Python parameters.simplex.perturbation.indicator simplex.perturbation
MATLAB Cplex.Param.simplex.perturbation.indicator simplex.perturbation
Interactive simplex perturbationlimit simplex perturbationlimit
Identifier 1027 1027

Description

Decides whether to perturb problems.

Setting this parameter to 1 (one) causes all problems to be automatically perturbed

as optimization begins. A setting of 0 (zero) allows CPLEX to decide dynamically,
during solution, whether progress is slow enough to merit a perturbation. The
situations in which a setting of 1 (one) helps are rare and restricted to problems
that exhibit extreme degeneracy.

In the Interactive Optimizer, the command

set simplex perturbationlimit

accepts two arguments and actually sets two parameters simultaneously. The first
argument is a switch or indicator; its value is 1 (one) to turn on perturbation or 0

Chapter 3. List of CPLEX parameters 95

 (zero) to turn off perturbation. The second argument is a constant value to set an
amount of perturbation. See the parameter “perturbation constant” on page 54 for
more information about the value of this second argument.
Table 46. Values.
Value bool Symbol Meaning
0 false CPX_OFF Automatic: let
CPLEX choose;
default
1 true CPX_ON Turn on perturbation
from beginning

simplex perturbation limit

Sets the number of degenerate iterations before perturbation is performed.

Purpose

Simplex perturbation limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Limits_Perturbation CPX_PARAM_PERLIM
C++ IloCplex::Param::Simplex::Limits::Perturbation PerLim (int)
Java IloCplex.Param.Simplex.Limits.Perturbation PerLim (int)
.NET Cplex.Param.Simplex.Limits.Perturbation PerLim (int)
OPL perlim
Python parameters.simplex.limits.perturbation simplex.limits.perturbation
MATLAB Cplex.Param.simplex.limits.perturbation simplex.limits.perturbation
Interactive simplex limits perturbation simplex limits perturbation
Identifier 1028 1028

Description

Sets the number of degenerate iterations before perturbation is performed.

Table 47. Values
Value Meaning
0 Automatic: let CPLEX choose; default
Any positive integer Number of degenerate iterations before
perturbation

deterministic time before starting to polish a feasible solution

Sets the amount of time expressed in deterministic ticks to spend during a normal
mixed integer optimization after which CPLEX starts to polish a feasible solution

Purpose

Deterministic time before starting to polish a feasible solution

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_PolishAfter_DetTime CPX_PARAM_POLISHAFTERDETTIME
C++ IloCplex::Param::MIP::PolishAfter::DetTime PolishAfterDetTime (double)

96 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Java IloCplex.Param.MIP.PolishAfter.DetTime PolishAfterDetTime (double)
.NET Cplex.Param.MIP.PolishAfter.DetTime PolishAfterDetTime (double)
OPL polishafterdettime
Python parameters.mip.polishafter.dettime mip.polishafter.dettime
MATLAB Cplex.Param.mip.polishafter.dettime mip.polishafter.dettime
Interactive mip polishafter dettime mip polishafter dettime
Identifier 2151 2151

Description

Tells CPLEX how much time (expressed in deterministic ticks) to spend during
mixed integer optimization before CPLEX starts polishing a feasible solution. The
default value (1.0E+75 seconds) is such that CPLEX does not start solution
polishing by default.

Starting conditions

CPLEX must have a feasible solution in order to start polishing. It must also have
certain internal structures in place to support solution polishing. Consequently,
when the criterion specified by this parameter is met, CPLEX begins solution
polishing only after these starting conditions are also met. That is, there may be a
delay between the moment when the criterion specified by this parameter is met
and when solution polishing starts.

Values

Any nonnegative value in deterministic ticks; default:1.0E+75 ticks.

See also

“time before starting to polish a feasible solution” on page 101

absolute MIP gap before starting to polish a feasible solution

Sets an absolute MIP gap after which CPLEX starts to polish a feasible solution

Purpose

Absolute MIP gap before starting to polish a feasible solution

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_PolishAfter_AbsMIPGap CPX_PARAM_POLISHAFTEREPAGAP
C++ IloCplex::Param::MIP::PolishAfter::AbsMIPGap PolishAfterEpAGap (double)
Java IloCplex.Param.MIP.PolishAfter.AbsMIPGap PolishAfterEpAGap (double)
.NET Cplex.Param.MIP.PolishAfter.AbsMIPGap PolishAfterEpAGap (double)
OPL polishafterepagap
Python parameters.mip.polishafter.absmipgap mip.polishafter.absmipgap
MATLAB Cplex.Param.mip.polishafter.absmipgap mip.polishafter.absmipgap
Interactive mip polishafter absmipgap mip polishafter absmipgap
Identifier 2126 2126

Chapter 3. List of CPLEX parameters 97

 Description

Sets an absolute MIP gap (that is, the difference between the best integer objective
and the objective of the best node remaining) after which CPLEX stops
branch-and-cut and begins polishing a feasible solution. The default value (0.0) is
such that CPLEX does not invoke solution polishing by default.

Starting conditions

CPLEX must have a feasible solution in order to start polishing. It must also have
certain internal structures in place to support solution polishing. Consequently,
when the criterion specified by this parameter is met, CPLEX begins solution
polishing only after these starting conditions are also met. That is, there may be a
delay between the moment when the criterion specified by this parameter is met
and when solution polishing starts.

Values

Any nonnegative value; default: 0.0.

See also

“absolute MIP gap tolerance” on page 49

relative MIP gap before starting to polish a feasible solution

Sets a relative MIP gap after which CPLEX starts to polish a feasible solution

Purpose

Relative MIP gap before starting to polish a solution

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_PolishAfter_MIPGap CPX_PARAM_POLISHAFTEREPGAP
C++ IloCplex::Param::MIP::PolishAfter::MIPGap PolishAfterEpGap (double)
Java IloCplex.Param.MIP.PolishAfter.MIPGap PolishAfterEpGap (double)
.NET Cplex.Param.MIP.PolishAfter.MIPGap PolishAfterEpGap (double)
OPL polishafterepgap
Python parameters.mip.polishafter.mipgap mip.polishafter.mipgap
MATLAB Cplex.Param.mip.polishafter.mipgap mip.polishafter.mipgap
Interactive mip polishafter mipgap mip polishafter mipgap
Identifier 2127 2127

Description

Sets a relative MIP gap after which CPLEX will stop branch-and-cut and begin
polishing a feasible solution. The default value (0.0) is such that CPLEX does not
invoke solution polishing by default. The relative MIP gap is calculated like this:

|bestbound-bestinteger|/(1e-10+|bestinteger|)

Starting conditions

CPLEX must have a feasible solution in order to start polishing. It must also have
certain internal structures in place to support solution polishing. Consequently,

98 CPLEX Parameters Reference

 when the criterion specified by this parameter is met, CPLEX begins solution
polishing only after these starting conditions are also met. That is, there may be a
delay between the moment when the criterion specified by this parameter is met
and when solution polishing starts.

Values

Any number from 0.0 to 1.0, inclusive; default: 0.0.

See also

“relative MIP gap tolerance” on page 50

MIP integer solutions to find before starting to polish a feasible

solution
Sets the number of integer solutions to find after which CPLEX starts to polish a
feasible solution

Purpose

MIP integer solutions to find before starting to polish a feasible solution

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_PolishAfter_Solutions CPX_PARAM_POLISHAFTERINTSOL
C++ IloCplex::Param::MIP::PolishAfter::Solutions PolishAfterIntSol (long)
Java IloCplex.Param.MIP.PolishAfter.Solutions PolishAfterIntSol (long)
.NET Cplex.Param.MIP.PolishAfter.Solutions PolishAfterIntSol (long)
OPL polishafterintsol
Python parameters.mip.polishafter.solutions mip.polishafter.solutions
MATLAB Cplex.Param.mip.polishafter.solutions mip.polishafter.solutions
Interactive mip polishafter solutions mip polishafter solutions
Identifier 2129 2129

Description

Sets the number of integer solutions to find before CPLEX stops branch-and-cut
and begins to polish a feasible solution. The default value is such that CPLEX does
not invoke solution polishing by default.

Starting conditions

CPLEX must have a feasible solution in order to start polishing. It must also have
certain internal structures in place to support solution polishing. Consequently,
when the criterion specified by this parameter is met, CPLEX begins solution
polishing only after these starting conditions are also met. That is, there may be a
delay between the moment when the criterion specified by this parameter is met
and when solution polishing starts.

Values

Any positive integer strictly greater than zero; zero is not allowed; default:
9223372036800000000

See also

Chapter 3. List of CPLEX parameters 99

 “MIP integer solution-file switch and prefix” on page 66

nodes to process before starting to polish a feasible solution

Sets the number of nodes to process after which CPLEX starts to polish a feasible
solution

Purpose

Nodes to process before starting to polish a feasible solution

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_PolishAfter_Nodes CPX_PARAM_POLISHAFTERNODE
C++ IloCplex::Param::MIP::PolishAfter::Nodes PolishAfterNode (long)
Java IloCplex.Param.MIP.PolishAfter.Nodes PolishAfterNode (long)
.NET Cplex.Param.MIP.PolishAfter.Nodes PolishAfterNode (long)
OPL polishafternode
Python parameters.mip.polishafter.nodes mip.polishafter.nodes
MATLAB Cplex.Param.mip.polishafter.nodes mip.polishafter.nodes
Interactive mip polishafter nodes mip polishafter nodes
Identifier 2128 2128

Description

Sets the number of nodes processed in branch-and-cut before CPLEX starts

solution polishing, if a feasible solution is available.

When this parameter is set to 0 (zero), CPLEX completes processing at the root;
that is, it creates cuts and applies heuristics at the root.

When this parameter is set to 1 (one), it allows branching from the root; that is,
nodes are created but not solved.

When no feasible solution is available yet, CPLEX explores more nodes than the
number specified by this parameter.

Starting conditions

CPLEX must have a feasible solution in order to start polishing. It must also have
certain internal structures in place to support solution polishing. Consequently,
when the criterion specified by this parameter is met, CPLEX begins solution
polishing only after these starting conditions are also met. That is, there may be a
delay between the moment when the criterion specified by this parameter is met
and when solution polishing starts.

Values

Any nonnegative integer; default: 9223372036800000000

See also

“MIP node limit” on page 88

100 CPLEX Parameters Reference

time before starting to polish a feasible solution
Sets the amount of time in seconds to spend during a normal mixed integer
optimization after which CPLEX starts to polish a feasible solution

Purpose

Time before starting to polish a feasible solution

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_PolishAfter_Time CPX_PARAM_POLISHAFTERTIME
C++ IloCplex::Param::MIP::PolishAfter::Time PolishAfterTime (double)
Java IloCplex.Param.MIP.PolishAfter.Time PolishAfterTime (double)
.NET Cplex.Param.MIP.PolishAfter.Time PolishAfterTime (double)
OPL polishaftertime
Python parameters.mip.polishafter.time mip.polishafter.time
MATLAB Cplex.Param.mip.polishafter.time mip.polishafter.time
Interactive mip polishafter time mip polishafter time
Identifier 2130 2130

Description

Tells CPLEX how much time in seconds to spend during mixed integer
optimization before CPLEX starts polishing a feasible solution. The default value
(1.0E+75 seconds) is such that CPLEX does not start solution polishing by default.

Whether CPLEX measures CPU time or wall clock time (also known as real time)
depends on the parameter “clock type for computation time” on page 35.

Starting conditions

CPLEX must have a feasible solution in order to start polishing. It must also have
certain internal structures in place to support solution polishing. Consequently,
when the criterion specified by this parameter is met, CPLEX begins solution
polishing only after these starting conditions are also met. That is, there may be a
delay between the moment when the criterion specified by this parameter is met
and when solution polishing starts.

Values

Any nonnegative value in seconds; default:1.0E+75 seconds.

See also

“clock type for computation time” on page 35

“deterministic time before starting to polish a feasible solution” on page 96

time spent polishing a solution (deprecated)

Deprecated parameter

Chapter 3. List of CPLEX parameters 101

 Purpose

Time spent polishing a solution (deprecated)

C Name CPX_PARAM_POLISHTIME (double)
C++ Name PolishTime (double)
Java Name PolishTime (double)
.NET Name PolishTime (double)
OPL Name polishtime
Python Name deprecated and not available
MATLAB Name deprecated and not available
Interactive Optimizer mip limit polishtime
Identifier 2066

Description

This deprecated parameter told CPLEX how much time in seconds to spend after a
normal mixed integer optimization in polishing a solution. The default was zero,
no polishing time.

Instead of this deprecated parameter, use one of the following parameters to

control the effort that CPLEX spends in branch-and-cut before it begins polishing a
feasible solution:
v “absolute MIP gap before starting to polish a feasible solution” on page 97
v “relative MIP gap before starting to polish a feasible solution” on page 98
v “MIP integer solutions to find before starting to polish a feasible solution” on
page 99
v “nodes to process before starting to polish a feasible solution” on page 100
v “time before starting to polish a feasible solution” on page 101
v “optimizer time limit in seconds” on page 141

Values

Any nonnegative value in seconds; default: 0.0 (zero) seconds.

maximum number of solutions generated for solution pool by populate

Sets the maximum number of mixed integer programming (MIP) solutions
generated for the solution pool during each call to the populate procedure.

Purpose

Maximum number of solutions generated for the solution pool by populate

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_Populate CPX_PARAM_POPULATELIM
C++ IloCplex::Param::MIP::Limits::Populate PopulateLim (int)
Java IloCplex.Param.MIP.Limits.Populate PopulateLim (int)
.NET Cplex.Param.MIP.Limits.Populate PopulateLim (int)
OPL populatelim
Python parameters.mip.limits.populate mip.limits.populate
MATLAB Cplex.Param.mip.limits.populate mip.limits.populate
Interactive mip limits populate mip limits populate
Identifier 2108 2108

102 CPLEX Parameters Reference

 Description

Sets the maximum number of mixed integer programming (MIP) solutions

generated for the solution pool during each call to the populate procedure.
Populate stops when it has generated PopulateLim solutions. A solution is counted
if it is valid for all filters, consistent with the relative and absolute pool gap
parameters, and has not been rejected by the incumbent callback (if any exists),
whether or not it improves the objective of the model.

In parallel, populate may not respect this parameter exactly due to disparities
between threads. That is, it may happen that populate stops when it has generated
a number of solutions slightly more than or slightly less than this limit because of
differences in synchronization between threads.

This parameter does not apply to MIP optimization generally; it applies only to the
populate procedure.

If you are looking for a parameter to control the number of solutions stored in the
solution pool, consider instead the solution pool capacity parameter (“maximum
number of solutions kept in solution pool” on page 131: SolnPoolCapacity,
CPX_PARAM_SOLNPOOLCAPACITY).

Populate will stop before it reaches the limit set by this parameter if it reaches
another limit, such as a time limit set by the user. Additional stopping criteria can
be specified by these parameters:
v “relative gap for solution pool” on page 132: SolnPoolGap,
CPX_PARAM_SOLNPOOLGAP
v “absolute gap for solution pool” on page 130: SolnPoolAGap,
CPX_PARAM_SOLNPOOLAGAP
v “MIP node limit” on page 88: NodeLim, CPX_PARAM_NODELIM
v “optimizer time limit in seconds” on page 141: TiLim, CPX_PARAM_TILIM

Values

Any nonnegative integer; default: 20.

primal simplex pricing algorithm

Sets the primal simplex pricing algorithm.

Purpose

Primal simplex pricing algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_PGradient CPX_PARAM_PPRIIND
C++ IloCplex::Param::Simplex::PGradient PPriInd (int)
Java IloCplex.Param.Simplex.PGradient PPriInd (int)
.NET Cplex.Param.Simplex.PGradient PPriInd (int)
OPL ppriind
Python parameters.simplex.pgradient simplex.pgradient
MATLAB Cplex.Param.simplex.pgradient simplex.pgradient
Interactive simplex pgradient simplex pgradient

Chapter 3. List of CPLEX parameters 103

 API Parameter Name Name prior to V12.6.0
Identifier 1029 1029

Description
Sets the primal simplex pricing algorithm. The default pricing (0) usually provides
the fastest solution time, but many problems benefit from alternative settings.
Table 48. Values
Value Symbol Meaning
-1 CPX_PPRIIND_PARTIAL Reduced-cost pricing
0 CPX_PPRIIND_AUTO Hybrid reduced-cost & devex pricing; default
1 CPX_PPRIIND_DEVEX Devex pricing
2 CPX_PPRIIND_STEEP Steepest-edge pricing
3 CPX_PPRIIND_STEEPQSTART Steepest-edge pricing with slack initial norms
4 CPX_PPRIIND_FULL Full pricing

presolve dual setting

Decides whether CPLEX presolve should pass the primal or dual linear
programming problem to the linear programming optimization algorithm.

Purpose

Presolve dual setting

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Dual CPX_PARAM_PREDUAL
C++ IloCplex::Param::Preprocessing::Dual PreDual (int)
Java IloCplex.Param.Preprocessing.Dual PreDual (int)
.NET Cplex.Param.Preprocessing.Dual PreDual (int)
OPL predual
Python parameters.preprocessing.dual preprocessing.dual
MATLAB Cplex.Param.preprocessing.dual preprocessing.dual
Interactive preprocessing dual preprocessing dual
Identifier 1044 1044

Description

Decides whether CPLEX presolve should pass the primal or dual linear
programming problem to the linear programming optimization algorithm. By
default, CPLEX chooses automatically.

If this parameter is set to 1 (one), the CPLEX presolve algorithm is applied to the
primal problem, but the resulting dual linear program is passed to the optimizer.
This is a useful technique for problems with more constraints than variables.

When this parameter is set to 0 (zero, its default value) or 1 (one, turned on),
CPLEX disables crushing and uncrushing of the model by such routines as
CPXuncrushx. To use CPXuncrushx effectively, you must set the value of this
parameter to -1, turning off this feature.

104 CPLEX Parameters Reference

 Table 49. Values
Value Meaning
-1 Turn off this feature
0 Automatic: let CPLEX choose; default
1 Turn on this feature

presolve switch
Decides whether CPLEX applies presolve during preprocessing.

Purpose

Presolve switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Presolve CPX_PARAM_PREIND
C++ IloCplex::Param::Preprocessing::Presolve PreInd (bool)
Java IloCplex.Param.Preprocessing.Presolve PreInd (bool)
.NET Cplex.Param.Preprocessing.Presolve PreInd (bool)
OPL preind
Python parameters.preprocessing.presolve preprocessing.presolve
MATLAB Cplex.Param.preprocessing.presolve preprocessing.presolve
Interactive preprocessing presolve preprocessing presolve
Identifier 1030 1030

Description

Decides whether CPLEX applies presolve during preprocessing. When set to 1

(one), the default, this parameter invokes CPLEX presolve to simplify and reduce
problems. In other words, this parameter turns on or off presolve during
preprocessing.

To limit the number of passes that CPLEX carries out in presolve, use another
parameter: “limit on the number of presolve passes made” on page 106.
Table 50. Values
Value bool Symbol Meaning
0 false CPX_OFF Do not apply
presolve
1 true CPX_ON Apply presolve;
default

linear reduction switch

Decides whether linear or full reductions occur during preprocessing.

Purpose

Linear reduction switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Linear CPX_PARAM_PRELINEAR

Chapter 3. List of CPLEX parameters 105

 API Parameter Name Name prior to V12.6.0
C++ IloCplex::Param::Preprocessing::Linear PreLinear (int)
Java IloCplex.Param.Preprocessing.Linear PreLinear (int)
.NET Cplex.Param.Preprocessing.Linear PreLinear (int)
OPL prelinear
Python parameters.preprocessing.linear preprocessing.linear
MATLAB Cplex.Param.preprocessing.linear preprocessing.linear
Interactive preprocessing linear preprocessing linear
Identifier 1058 1058

Description

Decides whether linear or full reductions occur during preprocessing. If only linear
reductions are performed, each variable in the original model can be expressed as
a linear form of variables in the presolved model. This condition guarantees, for
example, that users can add their own custom cuts to the presolved model.

If your application uses lazy constraints (for example, you have explicitly added
lazy constraints to the model before optimization, or you have registered lazy
constraints from a callback by means of a method or routine, such as
CPXsetlazyconstraintcallbackfunc) then CPLEX turns off nonlinear reductions.
Table 51. Values.
Value Meaning
0 Perform only linear reductions
1 Perform full reductions; default

limit on the number of presolve passes made

Limits the number of presolve passes that CPLEX makes during preprocessing.

Purpose

Limit on the number of presolve passes during preprocessing

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_NumPass CPX_PARAM_PREPASS
C++ IloCplex::Param::Preprocessing::NumPass PrePass (int)
Java IloCplex.Param.Preprocessing.NumPass PrePass (int)
.NET Cplex.Param.Preprocessing.NumPass PrePass (int)
OPL prepass
Python parameters.preprocessing.numpass preprocessing.numpass
MATLAB Cplex.Param.preprocessing.numpass preprocessing.numpass
Interactive preprocessing numpass preprocessing numpass
Identifier 1052 1052

Description

Limits the number of presolve passes that CPLEX makes during preprocessing.

When this parameter is set to a positive value, presolve is applied the specified
number of times, or until no more reductions are possible.

106 CPLEX Parameters Reference

 At the default value of -1, presolve continues only if it seems to be helping.

When this parameter is set to zero, CPLEX does not enter its main presolve loop,
but other reductions may occur, depending on settings of other parameters and
characteristics of your model. In other words, setting this parameter to 0 (zero) is
not equivalent to turning off the “presolve switch” on page 105
(CPX_PARAM_PREIND, PreInd). To turn off presolve, use the “presolve switch”
on page 105 (CPX_PARAM_PREIND, PreInd) instead.
Table 52. Values
Value Meaning
-1 Automatic: let CPLEX choose; presolve
continues as long as helpful; default
0 Do not use presolve; other reductions may
still occur
Any positive integer Apply presolve specified number of times

node presolve switch

Decides whether node presolve should be performed at the nodes of a mixed
integer programming (MIP) solution.

Purpose

Node presolve switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_PresolveNode CPX_PARAM_PRESLVND
C++ IloCplex::Param::MIP::Strategy::PresolveNode PreslvNd (int)
Java IloCplex.Param.MIP.Strategy.PresolveNode PreslvNd (int)
.NET Cplex.Param.MIP.Strategy.PresolveNode PreslvNd (int)
OPL preslvnd
Python parameters.mip.strategy.presolvenode mip.strategy.presolvenode
MATLAB Cplex.Param.mip.strategy.presolvenode mip.strategy.presolvenode
Interactive mip strategy presolvenode mip strategy presolvenode
Identifier 2037 2037

Description

Decides whether node presolve should be performed at the nodes of a mixed

integer programming (MIP) solution. Node presolve can significantly reduce
solution time for some models. The default setting is generally effective at deciding
whether to apply node presolve, although runtimes can be reduced for some
models by the user turning node presolve off.

Value Meaning
-1 No node presolve
0 Automatic: let CPLEX choose; default
1 Force presolve at nodes
2 Perform probing on integer-infeasible variables
3 Perform aggressive node probing

Chapter 3. List of CPLEX parameters 107

simplex pricing candidate list size
Sets the maximum number of variables kept in the list of pricing candidates for the
simplex algorithms.

Purpose

Simplex pricing candidate list size

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Pricing CPX_PARAM_PRICELIM
C++ IloCplex::Param::Simplex::Pricing PriceLim (int)
Java IloCplex.Param.Simplex.Pricing PriceLim (int)
.NET Cplex.Param.Simplex.Pricing PriceLim (int)
OPL pricelim
Python parameters.simplex.pricing simplex.pricing
MATLAB Cplex.Param.simplex.pricing simplex.pricing
Interactive simplex pricing simplex pricing
Identifier 1010 1010

Description

Sets the maximum number of variables kept in the list of pricing candidates for the
simplex algorithms.
Table 53. Values
Value Meaning
0 Automatic: let CPLEX choose; default
Any positive integer Number of pricing candidates

MIP probing level

Sets the amount of probing on variables to be performed before MIP branching.

Purpose

MIP probing level

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_Probe CPX_PARAM_PROBE
C++ IloCplex::Param::MIP::Strategy::Probe Probe (int)
Java IloCplex.Param.MIP.Strategy.Probe Probe (int)
.NET Cplex.Param.MIP.Strategy.Probe Probe (int)
OPL probe
Python parameters.mip.strategy.probe mip.strategy.probe
MATLAB Cplex.Param.mip.strategy.probe mip.strategy.probe
Interactive mip strategy probe mip strategy probe
Identifier 2042 2042

Description

Sets the amount of probing on variables to be performed before MIP branching.

Higher settings perform more probing. Probing can be very powerful but very

108 CPLEX Parameters Reference

 time-consuming at the start. Setting the parameter to values above the default of 0
(automatic) can result in dramatic reductions or dramatic increases in solution
time, depending on the model.
Table 54. Values
Value Meaning
-1 No probing
0 Automatic: let CPLEX choose; default
1 Moderate probing level
2 Aggressive probing level
3 Very aggressive probing level

deterministic time spent probing

Limits the amount of time (expressed in deterministic ticks) spent probing.

Purpose

Time spent probing, measured deterministically

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_ProbeDetTime CPX_PARAM_PROBEDETTIME
C++ IloCplex::Param::MIP::Limits::ProbeDetTime ProbeDetTime (double)
Java IloCplex.Param.MIP.Limits.ProbeDetTime ProbeDetTime (double)
.NET Cplex.Param.MIP.Limits.ProbeDetTime ProbeDetTime (double)
OPL probedettime
Python parameters.mip.limits.probedettime mip.limits.probedettime
MATLAB Cplex.Param.mip.limits.probedettime mip.limits.probedettime
Interactive mip limits probedettime mip limits probedettime
Identifier 2150 2150

Description

Limits the amount of time (expressed in deterministic ticks) spent probing.

For a parameter limiting the amount of time spent probing in seconds, (rather than
deterministic ticks) see “time spent probing” (CPX_PARAM_PROBETIME,
ProbeTime).

Values

Any nonnegative number; default: 1e+75.

time spent probing

Limits the amount of time in seconds spent probing.

Purpose

Time spent probing

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_ProbeTime CPX_PARAM_PROBETIME

Chapter 3. List of CPLEX parameters 109

 API Parameter Name Name prior to V12.6.0
C++ IloCplex::Param::MIP::Limits::ProbeTime ProbeTime (double)
Java IloCplex.Param.MIP.Limits.ProbeTime ProbeTime (double)
.NET Cplex.Param.MIP.Limits.ProbeTime ProbeTime (double)
OPL probetime
Python parameters.mip.limits.probetime mip.limits.probetime
MATLAB Cplex.Param.mip.limits.probetime mip.limits.probetime
Interactive mip limits probetime mip limits probetime
Identifier 2065 2065

Description

Limits the amount of time in seconds spent probing.

For a parameter limiting the amount of time spent probing in deterministic ticks
(rather than seconds) see “deterministic time spent probing” on page 109
(CPX_PARAM_PROBEDETTIME, ProbeDetTime).

Values

Any nonnegative number; default: 1e+75.

indefinite MIQP switch

Decides whether CPLEX will attempt to reformulate a MIQP or MIQCP model that
contains only binary variables.

Purpose

Indefinite MIQP switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_QPMakePSD CPX_PARAM_QPMAKEPSDIND
(int)
C++ IloCplex::Param::Preprocessing::QPMakePSD QPmakePSDInd (bool)
Java IloCplex.Param.Preprocessing.QPMakePSD QPmakePSDInd (bool)
.NET Cplex.Param.Preprocessing.QPMakePSD QPmakePSDInd (bool)
OPL qpmakepsdind
Python parameters.preprocessing.qpmakepsd preprocessing.qpmakepsd
MATLAB Cplex.Param.preprocessing.qpmakepsd preprocessing.qpmakepsd
Interactive preprocessing qpmakepsd preprocessing qpmakepsd
Identifier 4010 4010

Description

Decides whether CPLEX will attempt to reformulate a MIQP or MIQCP model that
contains only binary variables. When this feature is active, adjustments will be
made to the elements of a quadratic matrix that is not nominally positive
semi-definite (PSD, as required by CPLEX for all QP and most QCP formulations),
to make it PSD, and CPLEX will also attempt to tighten an already PSD matrix for
better numerical behavior. The default setting of 1 (one) means yes, CPLEX should
attempt to reformulate, but you can turn it off if necessary; most models benefit
from the default setting.

110 CPLEX Parameters Reference

 Table 55. Values
Value bool Symbol Meaning
0 false CPX_OFF Turn off attempts to
make binary model
PSD
1 true CPX_ON On: CPLEX attempts
to make binary
model PSD; default

QP Q-matrix nonzero read limit

Specifies a limit for the number of nonzero elements to read for an allocation of
memory in a model with a quadratic matrix.

Purpose
QP Q-matrix nonzero read limit

API Parameter Name Name prior to V12.6.0

C 64-bit CPXPARAM_Read_QPNonzeros CPX_PARAM_QPNZREADLIM
(CPXLONG)
C 32-bit CPXPARAM_Read_QPNonzeros CPX_PARAM_QPNZREADLIM
(CPXINT)
C++ IloCplex::Param::Read::QPNonzeros QPNzReadLim (CPXLONG)
Java IloCplex.Param.Read.QPNonzeros QPNzReadLim (CPXLONG)
.NET Cplex.Param.Read.QPNonzeros QPNzReadLim (CPXLONG)
OPL not available
Python parameters.read.qpnonzeros read.qpnonzeros
MATLAB Cplex.Param.read.qpnonzeros read.qpnonzeros
Interactive read qpnonzeros read qpnonzeros
Identifier 4001

Description

Specifies a limit for the number of nonzero elements to read for an allocation of
memory in a model with a quadratic matrix.

This parameter does not restrict the size of a problem. Rather, it indirectly specifies
the default amount of memory that will be pre-allocated before a problem is read
from a file. If the limit is exceeded, more memory is automatically allocated.

Values

Any integer from 0 (zero) to CPX_BIGINT or CPX_BIGLONG, depending on the type of

integer; default: 5 000.

deterministic time spent in ramp up during distributed parallel

optimization
Limits the amount of time in deterministic ticks spent during ramp up of
distributed parallel optimization.

Chapter 3. List of CPLEX parameters 111

 Purpose

Time spent (measured in deterministic ticks) during ramp up of distributed parallel

optimization

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_RampupDetTimeLimit
C++ IloCplex::Param::MIP::Limits::RampupDetTimeLimit
Java IloCplex.Param.MIP.Limits.RampupDetTimeLimit
.NET Cplex.Param.MIP.Limits.RampupDetTimeLimit
OPL
Python parameters.mip.limits.rampupdettimelimit
MATLAB Cplex.Param.mip.limits.rampupdettimelimit
Interactive mip limits rampupdettimelimit
Identifier 2164

Description

This parameters specifies a limit on the amount of time measured in deterministic

ticks to spend in the ramp up phase of distributed parallel optimization. This
parameter is effective only when the “ramp up duration” on page 113 parameter
has a value of 0 (zero) or 1 (one), where 0 (zero) designates the default automatic
value that CPLEX decides the ramp up duration, and 1 (one) designates dynamic
ramp up. See “ramp up duration” on page 113 for more detail about the conditions
for time limits in ramp up.

For a parameter limiting the amount of time spent in ramp up in seconds (rather
than deterministic ticks) see “time spent in ramp up during distributed parallel
optimization.”

Values

The value 0 (zero) specifies that no time should be spent in ramp up.

Any positive number strictly greater than zero specifies a time limit in
deterministic ticks.

The default value is BIGREAL deterministic ticks; that is, (1e+75) deterministic
ticks on most platforms.

time spent in ramp up during distributed parallel optimization

Limits the amount of time in seconds spent during ramp up of distributed parallel
optimization.

Purpose

Time spent in seconds during ramp up of distributed parallel optimization

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_RampupTimeLimit
C++ IloCplex::Param::MIP::Limits::RampupTimeLimit
Java IloCplex.Param.MIP.Limits.RampupTimeLimit
.NET Cplex.Param.MIP.Limits.RampupTimeLimit
OPL

112 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Python parameters.mip.limits.rampuptimelimit
MATLAB Cplex.Param.mip.limits.rampuptimelimit
Interactive mip limits rampuptimelimit
Identifier 2165

Description

This parameters specifies a limit on the amount of time in seconds to spend in the
ramp up phase of distributed parallel optimization. This parameter is effective
only when the “ramp up duration” parameter has a value of 0 (zero) or 1 (one),
where 0 (zero) designates the default automatic value that CPLEX decides the
ramp up duration, and 1 (one) designates dynamic ramp up. See “ramp up
duration” for more detail about the conditions for time limits in ramp up.

For a parameter limiting the amount of time spent in ramp up in deterministic

ticks (rather than seconds) see “deterministic time spent in ramp up during
distributed parallel optimization” on page 111.

Values

The value 0 (zero) specifies that no time should be spent in ramp up.

Any positive number strictly greater than zero specifies a time limit in seconds.

The default value is BIGREAL seconds; that is, (1e+75) seconds on most platforms.

ramp up duration
Customizes ramp up for distributed parallel optimization.

Purpose

Ramp up duration

API Parameter Name Name prior to V12.6.0

C CPXPARAM_DistMIP_Rampup_Duration
C++ IloCplex::Param::DistMIP::Rampup::Duration
Java IloCplex.Param.DistMIP.Rampup.Duration
.NET Cplex.Param.DistMIP.Rampup.Duration
OPL
Python parameters.distmip.rampup.duration
MATLAB Cplex.Param.distmip.rampup.duration
Cplex class
compatible
Interactive distmip rampup duration
Identifier 2163

Description

During the ramp up phase of distributed parallel optimization, each worker

applies different parameter settings to the same problem as the other workers. In
other words, there is a competition among the workers to process the greatest

Chapter 3. List of CPLEX parameters 113

 number of nodes in parallel in the search tree of the distributed problem. At any
given time, each worker is a candidate to be the winner of this competition.

This parameter enables you to customize the ramp up phase for your model. Its
value has an impact on both timing parameters: “time spent in ramp up during
distributed parallel optimization” on page 112 and “deterministic time spent in
ramp up during distributed parallel optimization” on page 111.

When the value of this parameter is -1, CPLEX turns off ramp up and ignores both
of the parameters “time spent in ramp up during distributed parallel optimization”
on page 112 and “deterministic time spent in ramp up during distributed parallel
optimization” on page 111. CPLEX directly begins distributed parallel tree search.

When the value of this parameter is 2, CPLEX observes ramp up with an infinite
horizon. CPLEX ignores both of the parameters “time spent in ramp up during
distributed parallel optimization” on page 112 and “deterministic time spent in
ramp up during distributed parallel optimization” on page 111. CPLEX never
switches to distributed parallel tree search. This situation is also known as
concurrent mixed integer programming (concurrent MIP).

When the value of this parameter is 1 (one), CPLEX considers the values of both
“time spent in ramp up during distributed parallel optimization” on page 112 and
“deterministic time spent in ramp up during distributed parallel optimization” on
page 111.
v If both ramp up timing parameters are at their default value (effectively, an
infinite amount of time), then CPLEX terminates ramp up according to internal
criteria before switching to distributed parallel tree search.
v If one or both of the ramp up timing parameters is set to a non default finite
value, the CPLEX observes that time limit by executing ramp up for that given
amount of time. If the two time limits differ, CPLEX observes the smaller time
limit before terminating ramp up and switching to distributed parallel tree
search.

When the value of this parameter remains at its default, 0 (zero), CPLEX considers
the values of both timeing parameters “time spent in ramp up during distributed
parallel optimization” on page 112 and “deterministic time spent in ramp up
during distributed parallel optimization” on page 111.
v If at least one of the ramp up timing parameters is set to a finite value, then
CPLEX behaves as it does when the value of this parameter is 1 (one): first
ramping up, then switching to distributed parallel tree search.
v If both of the ramp up timing parameters are at their default value (effectively
an infinite amount of time), then CPLEX behaves as it does when the value of
this parameter is 2: concurrent MIP.

Tip: CPLEX behavior at default values is subject to change in future releases.

Values
Table 56. Values
Value Symbol Meaning
-1 CPX_RAMPUP_DISABLED CPLEX turns off ramp up.
0 CPX_RAMPUP_AUTO Automatic: let CPLEX decide.
1 CPX_RAMPUP_DYNAMIC CPLEX dynamically switches to distributed
tree search.

114 CPLEX Parameters Reference

 Table 56. Values (continued)
Value Symbol Meaning
2 CPX_RAMPUP_INFINITE CPLEX observes an infinite horizon for
ramp up; also known as concurrent MIP
optimization.

random seed
This parameter sets the random seed differently for diversity of solutions.

Purpose

Set random seed differently for diversity of solutions.

API Parameter Name Name prior to V12.6.0

C CPXPARAM_RandomSeed CPX_PARAM_RANDOMSEED
C++ IloCplex::Param::RandomSeed RandomSeed (int)
Java IloCplex.Param.RandomSeed RandomSeed (int)
.NET Cplex.Param.RandomSeed RandomSeed (int)
OPL randomseed
Python parameters.randomseed randomseed
MATLAB Cplex.Param.randomseed randomseed
Interactive randomseed randomseed
Identifier 1124 1124

Description

This parameter makes it possible for your application to manage the random seed
that CPLEX uses in some of its internal operations. Variation in the random seed
can increase diversity of results.

Values

Any nonnegative integer; that is, an integer in the interval [0, BIGINT].

The default value of this parameter changes with each release.

primal and dual reduction type

Specifies whether primal reductions, dual reductions, both, or neither are
performed during preprocessing.

Purpose

Primal and dual reduction type

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Reduce CPX_PARAM_REDUCE
C++ IloCplex::Param::Preprocessing::Reduce Reduce (int)
Java IloCplex.Param.Preprocessing.Reduce Reduce (int)
.NET Cplex.Param.Preprocessing.Reduce Reduce (int)
OPL reduce
Python parameters.preprocessing.reduce preprocessing.reduce
MATLAB Cplex.Param.preprocessing.reduce preprocessing.reduce

Chapter 3. List of CPLEX parameters 115

 API Parameter Name Name prior to V12.6.0
Interactive preprocessing reduce preprocessing reduce
Identifier 1057 1057

Description

Specifies whether primal reductions, dual reductions, both, or neither are

performed during preprocessing. These preprocessing reductions are also known as
presolve reductions.

If your application uses lazy constraints (for example, you have explicitly added
lazy constraints to the model before optimization, or you have registered lazy
constraints from a callback by means of a method or routine, such as
CPXsetlazyconstraintcallbackfunc) then CPLEX turns off dual reductions.
Table 57. Values.
Value Symbol Meaning
0 CPX_PREREDUCE_NOPRIMALORDUAL No primal or dual reductions
1 CPX_PREREDUCE_PRIMALONLY Only primal reductions
2 CPX_PREREDUCE_DUALONLY Only dual reductions
3 CPX_PREREDUCE_PRIMALANDDUAL Both primal and dual reductions;
default

simplex refactoring frequency

Sets the number of iterations between refactoring of the basis matrix.

Purpose

Simplex refactoring frequency

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Refactor CPX_PARAM_REINV
C++ IloCplex::Param::Simplex::Refactor ReInv (int)
Java IloCplex.Param.Simplex.Refactor ReInv (int)
.NET Cplex.Param.Simplex.Refactor ReInv (int)
OPL reinv
Python parameters.simplex.refactor simplex.refactor
MATLAB Cplex.Param.simplex.refactor simplex.refactor
Interactive simplex refactor simplex refactor
Identifier 1031 1031

Description

Sets the number of iterations between refactoring of the basis matrix.

Table 58. Values
Value Meaning
0 Automatic: let CPLEX choose; default
Integer from 1 to 10 000 Number of iterations between refactoring of
the basis matrix

116 CPLEX Parameters Reference

relaxed LP presolve switch
Decides whether LP presolve is applied to the root relaxation in a mixed integer
program (MIP).

Purpose

Relaxed LP presolve switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Relax CPX_PARAM_RELAXPREIND
C++ IloCplex::Param::Preprocessing::Relax RelaxPreInd (int)
Java IloCplex.Param.Preprocessing.Relax RelaxPreInd (int)
.NET Cplex.Param.Preprocessing.Relax RelaxPreInd (int)
OPL relaxpreind
Python parameters.preprocessing.relax preprocessing.relax
MATLAB Cplex.Param.preprocessing.relax preprocessing.relax
Interactive preprocessing relax preprocessing relax
Identifier 2034 2034

Description

Decides whether LP presolve is applied to the root relaxation in a mixed integer

program (MIP). Sometimes additional reductions can be made beyond any MIP
presolve reductions that were already done. By default, CPLEX applies presolve to
the initial relaxation in order to hasten time to the initial solution.

Value Symbol Meaning

-1 Automatic: let CPLEX
choose; default
0 CPX_OFF Off: do not use presolve on
initial relaxation
1 CPX_ON On: use presolve on initial
relaxation

relative objective difference cutoff

Used to update the cutoff each time a mixed integer solution is found.

Purpose

Relative objective difference cutoff

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Tolerances_RelObjDifference CPX_PARAM_RELOBJDIF
C++ IloCplex::Param::MIP::Tolerances::RelObjDifference RelObjDif (double)
Java IloCplex.Param.MIP.Tolerances.RelObjDifference RelObjDif (double)
.NET Cplex.Param.MIP.Tolerances.RelObjDifference RelObjDif (double)
OPL relobjdif
Python parameters.mip.tolerances.relobjdifference mip.tolerances.relobjdifference
MATLAB Cplex.Param.mip.tolerances.relobjdifference mip.tolerances.relobjdifference
Interactive mip tolerances relobjdifference mip tolerances relobjdifference
Identifier 2022 2022

Chapter 3. List of CPLEX parameters 117

 Description

Used to update the cutoff each time a mixed integer solution is found. The value is
multiplied by the absolute value of the integer objective and subtracted from
(added to) the newly found integer objective when minimizing (maximizing). This
computation forces the mixed integer optimization to ignore integer solutions that
are not at least this amount better than the one found so far.

The relative objective difference parameter can be adjusted to improve problem

solving efficiency by limiting the number of nodes; however, setting this parameter
at a value other than zero (the default) can cause some integer solutions, including
the true integer optimum, to be missed.

If both the relative objective difference and the “absolute objective difference
cutoff” on page 91 (CPX_PARAM_OBJDIF, ObjDif) are nonzero, the value of the
absolute objective difference is used.

Values

Any number from 0.0 to 1.0; default: 0.0.

See also

“absolute objective difference cutoff” on page 91

number of attempts to repair infeasible MIP start

Limits the attempts to repair an infeasible MIP start.

Purpose

Number of attempts to repair infeasible MIP start

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_RepairTries CPX_PARAM_REPAIRTRIES
C++ IloCplex::Param::MIP::Limits::RepairTries RepairTries (long)
Java IloCplex.Param.MIP.Limits.RepairTries RepairTries (long)
.NET Cplex.Param.MIP.Limits.RepairTries RepairTries (long)
OPL repairtries
Python parameters.mip.limits.repairtries mip.limits.repairtries
MATLAB Cplex.Param.mip.limits.repairtries mip.limits.repairtries
Interactive mip limits repairtries mip limits repairtries
Identifier 2067 2067

Description

Limits the attempts to repair an infeasible MIP start. This parameter lets you tell
CPLEX whether and how many times it should try to repair an infeasible MIP start
that you supplied. The parameter has no effect if the MIP start you supplied is
feasible. It has no effect if no MIP start was supplied.
Table 59. Values
Value Meaning
-1 None: do not try to repair

118 CPLEX Parameters Reference

 Table 59. Values (continued)
Value Meaning
0 Automatic: let CPLEX choose; default
Any positive integer Number of attempts

MIP repeat presolve switch

Specifies whether to re-apply presolve, with or without cuts, to a MIP model after
processing at the root is otherwise complete.

Purpose

Reapply presolve after processing the root node

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_RepeatPresolve CPX_PARAM_REPEATPRESOLVE
C++ IloCplex::Param::Preprocessing::RepeatPresolve RepeatPresolve (int)
Java IloCplex.Param.Preprocessing.RepeatPresolve RepeatPresolve (int)
.NET Cplex.Param.Preprocessing.RepeatPresolve RepeatPresolve (int)
OPL repeatpresolve
Python parameters.preprocessing.repeatpresolve preprocessing.repeatpresolve
MATLAB Cplex.Param.preprocessing.repeatpresolve preprocessing.repeatpresolve
Interactive preprocessing repeatpresolve preprocessing repeatpresolve
Identifier 2064 2064

Description

Specifies whether to re-apply presolve, with or without cuts, to a MIP model after
processing at the root is otherwise complete.
Table 60. Values.
Value Symbol
-1 Automatic: let CPLEX choose; default
0 Turn off represolve
1 Represolve without cuts
2 Represolve with cuts
3 Represolve with cuts and allow new root
cuts

RINS heuristic frequency

Decides how often to apply the relaxation induced neighborhood search (RINS)
heuristic.

Purpose

RINS heuristic frequency

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_RINSHeur CPX_PARAM_RINSHEUR
C++ IloCplex::Param::MIP::Strategy::RINSHeur RINSHeur (long)

Chapter 3. List of CPLEX parameters 119

 API Parameter Name Name prior to V12.6.0
Java IloCplex.Param.MIP.Strategy.RINSHeur RINSHeur (long)
.NET Cplex.Param.MIP.Strategy.RINSHeur RINSHeur (long)
OPL rinsheur
Python parameters.mip.strategy.rinsheur mip.strategy.rinsheur
MATLAB Cplex.Param.mip.strategy.rinsheur mip.strategy.rinsheur
Interactive mip strategy rinsheur mip strategy rinsheur
Identifier 2061 2061

Description

Decides how often to apply the relaxation induced neighborhood search (RINS)
heuristic. This heuristic attempts to improve upon the best solution found so far. It
will not be applied until CPLEX has found at least one incumbent solution.

Setting the value to -1 turns off the RINS heuristic. Setting the value to 0 (zero), the
default, applies the RINS heuristic at an interval chosen automatically by CPLEX.
Setting the value to a positive number applies the RINS heuristic at the requested
node interval. For example, setting RINSHeur to 20 dictates that the RINS heuristic
be called at node 0, 20, 40, 60, etc.

RINS is a powerful heuristic for finding high quality feasible solutions, but it may
be expensive.
Table 61. Values
Value Meaning
-1 None: do not apply RINS heuristic
0 Automatic: let CPLEX choose; default
Any positive integer Frequency to apply RINS heuristic

algorithm for continuous problems

Controls which algorithm is used to solve continuous models or to solve the root
relaxation of a MIP.

Purpose

Solution algorithm for continuous problems

API Parameter Name Name prior to V12.6.0

C CPXPARAM_LPMethod CPX_PARAM_LPMETHOD
C++ IloCplex::Param::RootAlgorithm RootAlg (int)
Java IloCplex.Param.RootAlgorithm RootAlg (int)
.NET Cplex.Param.RootAlgorithm RootAlg (int)
OPL rootalg
Python parameters.lpmethod lpmethod
MATLAB Cplex.Param.lpmethod lpmethod
Cplex class
API
MATLAB CPLEX Toolbox compatible lpmethod
MATLAB Optimization Toolbox compatible Simplex
Interactive lpmethod lpmethod
Identifier 1062 1062

120 CPLEX Parameters Reference

Description

Controls which algorithm CPLEX uses to solve continuous models (LPs).

In the object-oriented APIs (such as C++, Java, or .NET APIs), this parameter, as
RootAlg, also controls which algorithm CPLEX uses to solve the root relaxation of a
MIP.

In the C API and the Interactive Optimizer, there are separate parameters to
control LP, QP, and MIP optimizers, depending on the problem type. See, for
example, “algorithm for continuous quadratic optimization” on page 122 or
“algorithm for initial MIP relaxation” on page 123.

In all cases, the default setting is 0 (zero). The default setting means that CPLEX
will select the algorithm in a way that should give best overall performance.

For specific problem classes, the following details document the automatic settings.
Note that future versions of CPLEX could adopt different strategies. Therefore, if
you select any nondefault settings, you should review them periodically.

Currently, the behavior of the automatic setting is that CPLEX almost always
invokes the dual simplex algorithm when it is solving an LP model from scratch.
When it is continuing from an advanced basis, it will check whether the basis is
primal or dual feasible, and choose the primal or dual simplex algorithm
accordingly.

If multiple threads have been requested, in either deterministic or opportunistic

mode, the concurrent optimization algorithm is selected by the automatic setting
when CPLEX is solving a continuous linear programming model (LP) from scratch.

When three or more threads are available, and you select concurrent optimization
for the value of this parameter, its behavior depends on whether parallel mode is
opportunistic or deterministic (default parallel mode). Concurrent optimization in
opportunistic parallel mode runs the dual simplex optimizer on one thread, the
primal simplex optimizer on a second thread, the parallel barrier optimizer on a
third thread and on any additional available threads. In contrast, concurrent
optimization in deterministic parallel mode runs the dual optimizer on one thread
and the parallel barrier optimizer on any additional available threads.

The automatic setting may be expanded in the future so that CPLEX chooses the
algorithm based on additional problem characteristics.
Table 62. Values
Value Symbol Meaning
0 CPX_ALG_AUTOMATIC Automatic: let CPLEX
choose; default
1 CPX_ALG_PRIMAL Primal simplex
2 CPX_ALG_DUAL Dual simplex
3 CPX_ALG_NET Network simplex
4 CPX_ALG_BARRIER Barrier
5 CPX_ALG_SIFTING Sifting

Chapter 3. List of CPLEX parameters 121

 Table 62. Values (continued)
Value Symbol Meaning
6 CPX_ALG_CONCURRENT Concurrent (Dual, Barrier,
and Primal in opportunistic
parallel mode; Dual and
Barrier in deterministic
parallel mode)

algorithm for continuous quadratic optimization

Sets which algorithm to use when the C routine CPXqpopt (or the command
optimize in the Interactive Optimizer) is invoked.

Purpose

Algorithm for continuous quadratic optimization

API Parameter Name Name prior to V12.6.0

C CPXPARAM_QPMethod CPX_PARAM_QPMETHOD
C++ IloCplex::Param::RootAlgorithm RootAlg (int)
Java IloCplex.Param.RootAlgorithm RootAlg (int)
.NET Cplex.Param.RootAlgorithm RootAlg (int)
OPL rootalg
Python parameters.qpmethod qpmethod
MATLAB Cplex.Param.qpmethod qpmethod
Interactive qpmethod qpmethod
Identifier 1063 1063

Description

Sets which algorithm to use when the C routine CPXqpopt (or the command
optimize in the Interactive Optimizer) is invoked.

Currently, the behavior of the Automatic setting is that CPLEX invokes the Barrier
Optimizer for continuous QP models. The Automatic setting may be expanded in
the future so that CPLEX chooses the algorithm based on additional problem
characteristics.
Table 63. Values
Value Symbol Meaning
0 CPX_ALG_AUTOMATIC Automatic: let CPLEX
choose; default
1 CPX_ALG_PRIMAL Use the primal simplex
optimizer.
2 CPX_ALG_DUAL Use the dual simplex
optimizer.
3 CPX_ALG_NET Use the network optimizer.
4 CPX_ALG_BARRIER Use the barrier optimizer.
6 CPX_ALG_CONCURRENT Use the concurrent optimizer.

122 CPLEX Parameters Reference

algorithm for initial MIP relaxation
Sets which continuous optimizer will be used to solve the initial relaxation of a
MIP.

Purpose

MIP starting algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_StartAlgorithm CPX_PARAM_STARTALG
C++ IloCplex::Param::RootAlgorithm RootAlg
Java IloCplex.Param.RootAlgorithm RootAlg
.NET Cplex.Param.RootAlgorithm RootAlg
OPL rootalg
Python parameters.mip.strategy.startalgorithm mip.strategy.startalgorithm
MATLAB Cplex.Param.mip.strategy.startalgorithm mip.strategy.startalgorithm
Interactive mip strategy startalgorithm mip strategy startalgorithm
Identifier 2025 2025

Description

Sets which continuous optimizer will be used to solve the initial relaxation of a
MIP.

The default Automatic setting (0 zero) of this parameter currently selects the
concurrent optimizer for root relaxations of mixed integer linear programming
models (MILP) and selects the dual simplex optimizer for root relaxations of mixed
integer quadratic programming models (MIQP). The Automatic setting may be
expanded in the future so that CPLEX chooses the algorithm based on additional
characteristics of the model.

For MILP (integer constraints and otherwise continuous variables), all settings are
permitted.

For MIQP (integer constraints and positive semi-definite quadratic terms in the
objective), settings 5 (Sifting) and 6 (Concurrent) are not implemented; if you
happen to choose them, setting 5 (Sifting) reverts to 0 (zero) and setting 6
(Concurrent) reverts to 4.

For MIQCP (integer constraints and positive semi-definite quadratic terms among
the constraints), only the barrier optimizer is implemented, and therefore no
settings other than 0 (zero) and 4 are permitted.
Table 64. Values
Value Symbol Meaning
0 CPX_ALG_AUTOMATIC Automatic: let CPLEX choose;
default
1 CPX_ALG_PRIMAL Primal Simplex
2 CPX_ALG_DUAL Dual Simplex
3 CPX_ALG_NET Network Simplex
4 CPX_ALG_BARRIER Barrier
5 CPX_ALG_SIFTING Sifting

Chapter 3. List of CPLEX parameters 123

 Table 64. Values (continued)
Value Symbol Meaning
6 CPX_ALG_CONCURRENT Concurrent (Dual, Barrier, and
Primal in opportunistic mode;
Dual and Barrier in deterministic
mode)

auxiliary root threads

Partitions the number of threads to manage tasks at the root node.

Purpose

Auxiliary root threads

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_AuxRootThreads CPX_PARAM_AUXROOTTHREADS
C++ IloCplex::Param::MIP::Limits::AuxRootThreadsAuxRootThreads
Java IloCplex.Param.MIP.Limits.AuxRootThreads AuxRootThreads
.NET Cplex.Param.MIP.Limits.AuxRootThreads AuxRootThreads
OPL auxrootthreads
Python parameters.mip.limits.auxrootthreads mip.limits.auxrootthreads
MATLAB Cplex.Param.mip.limits.auxrootthreads mip.limits.auxrootthreads
Interactive mip limits auxrootthreads n mip limits auxrootthreads n
Identifier 2139 2139

Description

Partitions the number of threads for CPLEX to use for auxiliary tasks while it
solves the root node of a problem.

On a system that offers N global threads, if you set this parameter to n, where
N > n > 0

then CPLEX uses at most n threads for auxiliary tasks and at most N-n threads to
solve the root node.

See also the parameter “global default thread count” on page 139, for more general
information about parallel solving and threads.

Tip:

You cannot set n, the value of this parameter, to a value greater than or equal to N,
the number of global threads offered on your system.

Independent of the auxiliary root threads parameter, CPLEX will never use more
threads than those defined by the “global default thread count” on page 139
parameter, whether that parameter is 0 (zero), its default value, or N, a value that
you set. CPLEX also makes sure that there is at least one thread available for the
main root tasks. For example, if you set the global threads parameter to 3 and the
auxiliary root threads parameter to 4, CPLEX still uses only two threads for
auxiliary root tasks in order to keep one thread available for the main root tasks.

124 CPLEX Parameters Reference

 At its default value, 0 (zero), CPLEX automatically chooses the number of threads
to use for the primary root tasks and for auxiliary tasks. The number of threads
that CPLEX uses to solve the root node depends on these factors:
v the number of threads available to your application on your system (for
example, as a result of limited resources or competition with other applications);
v the value of the “global default thread count” on page 139 parameter
(CPX_PARAM_THREADS, Threads).
Table 65. Values
Value Meaning
-1 Off: do not use additional threads for
auxiliary tasks.
0 Automatic: let CPLEX choose the number of
threads to use; default
N>n>0 Use n threads for auxiliary root tasks

constraint (row) read limit

Specifies a limit for the number of rows (constraints) to read for an allocation of
memory.

Purpose

Constraint (row) read limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Read_Constraints CPX_PARAM_ROWREADLIM
C++ IloCplex::Param::Read::Constraints RowReadLim (int)
Java IloCplex.Param.Read.Constraints RowReadLim (int)
.NET Cplex.Param.Read.Constraints RowReadLim (int)
OPL not available
Python parameters.read.constraints read.constraints
MATLAB Cplex.Param.read.constraints read.constraints
Interactive read constraints read constraints
Identifier 1021 1021

Description

Specifies a limit for the number of rows (constraints) to read for an allocation of
memory.

This parameter does not restrict the size of a problem. Rather, it indirectly specifies
the default amount of memory that will be pre-allocated before a problem is read
from a file. If the limit is exceeded, more memory is automatically allocated.

Values

Any integer from 0 (zero) to CPX_BIGINT ; default: 30 000.

scale parameter
Decides how to scale the problem matrix.

Chapter 3. List of CPLEX parameters 125

 Purpose

Scale parameter

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Read_Scale CPX_PARAM_SCAIND
C++ IloCplex::Param::Read::Scale ScaInd (int)
Java IloCplex.Param.Read.Scale ScaInd (int)
.NET Cplex.Param.Read.Scale ScaInd (int)
OPL scaind
Python parameters.read.scale read.scale
MATLAB Cplex.Param.read.scale read.scale
Interactive read scale read scale
Identifier 1034 1034

Description

Decides how to scale the problem matrix.

Table 66. Values
Value Meaning
-1 No scaling
0 Equilibration scaling; default
1 More aggressive scaling

messages to screen switch

Decides whether or not results are displayed on screen in an application of the C
API.

Purpose

Messages to screen switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_ScreenOutput CPX_PARAM_SCRIND
MATLAB Cplex class API DisplayFunc
MATLAB Optimization Toolbox diagnostics or Diagnostics
Identifier 1035 1035

Description

Decides whether or not results are displayed on screen in an application of the

Callable Library (C API). This parameter works by adding or removing stdout to
or from the result, warning, and error channels. Consequently, good practice does
not manage stdout in those channels directly at the same time as using this
parameter; otherwise, undefined behavior can occur.

To turn off output to the screen, in a C++ application, where cplex is an instance
of the class IloCplex and env is an instance of the class IloEnv , the environment,
use cplex.setOut(env.getNullStream()) .

In a Java application, use cplex.setOut(null).

126 CPLEX Parameters Reference

 In a .NET application, use Cplex.SetOut(Null).

In a Python application, where c is an instance of the class cplex.Cplex, use

c.set_results_stream(None).
Table 67. Values
Value Symbol Meaning
0 CPX_OFF Turn off display of messages
to screen; default
1 CPX_ON Display messages on screen

sifting subproblem algorithm

Sets the algorithm to be used for solving sifting subproblems.

Purpose
Sifting subproblem algorithm

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Sifting_Algorithm CPX_PARAM_SIFTALG
C++ IloCplex::Param::Sifting::Algorithm SiftAlg (int)
Java IloCplex.Param.Sifting.Algorithm SiftAlg (int)
.NET Cplex.Param.Sifting.Algorithm SiftAlg (int)
OPL siftalg
Python parameters.sifting.algorithm sifting.algorithm
MATLAB Cplex.Param.sifting.algorithm sifting.algorithm
Interactive sifting algorithm sifting algorithm
Identifier 1077 1077

Description

Sets the algorithm to be used for solving sifting subproblems. The default
automatic setting will typically use a mix of barrier and primal simplex.
Table 68. Values
Value Symbol Meaning
0 CPX_ALG_AUTOMATIC Automatic: let CPLEX choose; default
1 CPX_ALG_PRIMAL Primal Simplex
2 CPX_ALG_DUAL Dual Simplex
3 CPX_ALG_NET Network Simplex
4 CPX_ALG_BARRIER Barrier

sifting information display

Sets the amount of information to display about the progress of sifting.

Chapter 3. List of CPLEX parameters 127

 Purpose

Sifting information display

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Sifting_Display CPX_PARAM_SIFTDISPLAY
C++ IloCplex::Param::Sifting::Display SiftDisplay (int)
Java IloCplex.Param.Sifting.Display SiftDisplay (int)
.NET Cplex.Param.Sifting.Display SiftDisplay (int)
OPL siftdisplay
Python parameters.sifting.display sifting.display
MATLAB Cplex.Param.sifting.display sifting.display
Interactive sifting display sifting display
Identifier 1076 1076

Description

Sets the amount of information to display about the progress of sifting.

Table 69. Values
Value Meaning
0 No display of sifting information
1 Display major iterations; default
2 Display LP subproblem information within
each sifting iteration

upper limit on sifting iterations

Sets the maximum number of sifting iterations that may be performed if
convergence to optimality has not been reached.

Purpose

Upper limit on sifting iterations

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Sifting_Iterations CPX_PARAM_SIFTITLIM
C++ IloCplex::Param::Sifting::Iterations SiftItLim (long)
Java IloCplex.Param.Sifting.Iterations SiftItLim (long)
.NET Cplex.Param.Sifting.Iterations SiftItLim (long)
OPL siftitlim
Python parameters.sifting.iterations sifting.iterations
MATLAB Cplex.Param.sifting.iterations sifting.iterations
Interactive sifting iterations sifting iterations
Identifier 1078 1078

Description

Sets the maximum number of sifting iterations that may be performed if

convergence to optimality has not been reached.

Values

128 CPLEX Parameters Reference

 Any nonnegative integer; default: 9223372036800000000.

simplex iteration information display

Sets how often CPLEX reports about iterations during simplex optimization.

Purpose

Simplex iteration information display

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Display CPX_PARAM_SIMDISPLAY
C++ IloCplex::Param::Simplex::Display SimDisplay (int)
Java IloCplex.Param.Simplex.Display SimDisplay (int)
.NET Cplex.Param.Simplex.Display SimDisplay (int)
OPL simdisplay
Python parameters.simplex.display simplex.display
MATLAB Cplex.Param.simplex.display simplex.display
Interactive simplex display simplex display
Identifier 1019 1019

Description

Sets how often CPLEX reports about iterations during simplex optimization.
Table 70. Values
Value Meaning
0 No iteration messages until solution
1 Iteration information after each refactoring;
default
2 Iteration information for each iteration

simplex singularity repair limit

Restricts the number of times CPLEX attempts to repair the basis when
singularities are encountered during the simplex algorithm.

Purpose

Simplex singularity repair limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Simplex_Limits_Singularity CPX_PARAM_SINGLIM
C++ IloCplex::Param::Simplex::Limits::Singularity SingLim (int)
Java IloCplex.Param.Simplex.Limits.Singularity SingLim (int)
.NET Cplex.Param.Simplex.Limits.Singularity SingLim (int)
OPL singlim
Python parameters.simplex.limits.singularity simplex.limits.singularity
MATLAB Cplex.Param.simplex.limits.singularity simplex.limits.singularity
Interactive simplex limits singularity simplex limits singularity
Identifier 1037 1037

Chapter 3. List of CPLEX parameters 129

 Description

Restricts the number of times CPLEX attempts to repair the basis when
singularities are encountered during the simplex algorithm. When this limit is
exceeded, CPLEX replaces the current basis with the best factorable basis that has
been found.

Values

Any nonnegative integer; default: 10.

absolute gap for solution pool

Sets an absolute tolerance on the objective value for the solutions in the solution
pool.

Purpose

Absolute gap for solution pool

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Pool_AbsGap CPX_PARAM_SOLNPOOLAGAP
C++ IloCplex::Param::MIP::Pool::AbsGap SolnPoolAGap (double)
Java IloCplex.Param.MIP.Pool.AbsGap SolnPoolAGap (double)
.NET Cplex.Param.MIP.Pool.AbsGap SolnPoolAGap (double)
OPL solnpoolagap
Python parameters.mip.pool.absgap mip.pool.absgap
MATLAB Cplex.Param.mip.pool.absgap mip.pool.absgap
Interactive mip pool absgap mip pool absgap
Identifier 2106 2106

Description

Sets an absolute tolerance on the objective value for the solutions in the solution
pool. Solutions that are worse (either greater in the case of a minimization, or less
in the case of a maximization) than the objective of the incumbent solution
according to this measure are not kept in the solution pool.

Values of the solution pool absolute gap (SolnPoolAGap or CPX_PARAM_SOLNPOOLAGAP)

and the solution pool relative gap (“relative gap for solution pool” on page 132:
SolnPoolGap or CPX_PARAM_SOLNPOOLGAP) may differ: For example, you may specify
that solutions must be within 15 units by means of the solution pool absolute gap
and also within 1% of the incumbent by means of the solution pool relative gap. A
solution is accepted in the pool only if it is valid for both the relative and the
absolute gaps.

The solution pool absolute gap parameter can also be used as a stopping criterion
for the populate procedure: if populate cannot enumerate any more solutions that
satisfy this objective quality, then it will stop. In the presence of both an absolute
and a relative solution pool gap parameter, populate will stop when the smaller of
the two is reached.

Values

Any nonnegative real number; default: 1.0e+75.

130 CPLEX Parameters Reference

maximum number of solutions kept in solution pool
Limits the number of solutions kept in the solution pool

Purpose

Maximum number of solutions kept in the solution pool

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Pool_Capacity CPX_PARAM_SOLNPOOLCAPACITY
C++ IloCplex::Param::MIP::Pool::Capacity SolnPoolCapacity (int)
Java IloCplex.Param.MIP.Pool.Capacity SolnPoolCapacity (int)
.NET Cplex.Param.MIP.Pool.Capacity SolnPoolCapacity (int)
OPL solnpoolcapacity
Python parameters.mip.pool.capacity mip.pool.capacity
MATLAB Cplex.Param.mip.pool.capacity mip.pool.capacity
Interactive mip pool capacity mip pool capacity
Identifier 2103 2103

Description

Sets the maximum number of solutions kept in the solution pool. At most,
SolnPoolCapacity solutions will be stored in the pool. Superfluous solutions are
managed according to the strategy set by the “solution pool replacement strategy”
on page 134 parameter (SolnPoolReplace, CPX_PARAM_SOLNPOOLREPLACE).

The optimization (whether by MIP optimization or the populate procedure) will

not stop if more than SolnPoolCapacity solutions are generated. Instead, stopping
criteria can be specified by these parameters:
v “maximum number of solutions generated for solution pool by populate” on
page 102 (PopulateLim, CPX_PARAM_POPULATELIM)
v “relative gap for solution pool” on page 132 (SolnPoolGap,
CPX_PARAM_SOLNPOOLGAP)
v “absolute gap for solution pool” on page 130 (SolnPoolAGap,
CPX_PARAM_SOLNPOOLAGAP)
v “MIP node limit” on page 88 (NodeLim, CPX_PARAM_NODELIM)
v “optimizer time limit in seconds” on page 141 (TiLim, CPX_PARAM_TILIM)

The default value for SolnPoolCapacity is 2100000000, but it may be set to any
nonnegative integer value. If set to zero, it will turn off all features related to the
solution pool.

If you are looking for a parameter to control the number of solutions generated by
the populate procedure, consider the parameter “maximum number of solutions
generated for solution pool by populate” on page 102.

Values

Any nonnegative integer; 0 (zero) turns off all features of the solution pool;
default: 2100000000.

Chapter 3. List of CPLEX parameters 131

relative gap for solution pool
Sets a relative tolerance on the objective value for the solutions in the solution
pool.

Purpose

Relative gap for the solution pool

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Pool_RelGap CPX_PARAM_SOLNPOOLGAP
C++ IloCplex::Param::MIP::Pool::RelGap SolnPoolGap (double)
Java IloCplex.Param.MIP.Pool.RelGap SolnPoolGap (double)
.NET Cplex.Param.MIP.Pool.RelGap SolnPoolGap (double)
OPL solnpoolgap
Python parameters.mip.pool.relgap mip.pool.relgap
MATLAB Cplex.Param.mip.pool.relgap mip.pool.relgap
Interactive mip pool relgap mip pool relgap
Identifier 2105 2105

Description

Sets a relative tolerance on the objective value for the solutions in the solution
pool. Solutions that are worse (either greater in the case of a minimization, or less
in the case of a maximization) than the incumbent solution by this measure are not
kept in the solution pool. For example, if you set this parameter to 0.01, then
solutions worse than the incumbent by 1% or more will be discarded.

Values of the “absolute gap for solution pool” on page 130 (SolnPoolAGap or
CPX_PARAM_SOLNPOOLAGAP) and the “relative gap for solution pool” (SolnPoolGap or
CPX_PARAM_SOLNPOOLGAP) may differ: For example, you may specify that solutions
must be within 15 units by means of the solution pool absolute gap and within 1%
of the incumbent by means of the solution pool relative gap. A solution is accepted
in the pool only if it is valid for both the relative and the absolute gaps.

The solution pool relative gap parameter can also be used as a stopping criterion
for the populate procedure: if populate cannot enumerate any more solutions that
satisfy this objective quality, then it will stop. In the presence of both an absolute
and a relative solution pool gap parameter, populate will stop when the smaller of
the two is reached.

Values

Any nonnegative real number; default: 1.0e+75.

solution pool intensity

Controls the trade-off between the number of solutions generated for the solution
pool and the amount of time or memory consumed.

132 CPLEX Parameters Reference

Purpose

Solution pool intensity

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Pool_Intensity CPX_PARAM_SOLNPOOLINTENSITY
C++ IloCplex::Param::MIP::Pool::Intensity SolnPoolIntensity (int)
Java IloCplex.Param.MIP.Pool.Intensity SolnPoolIntensity (int)
.NET Cplex.Param.MIP.Pool.Intensity SolnPoolIntensity (int)
OPL solnpoolintensity
Python parameters.mip.pool.intensity mip.pool.intensity
MATLAB Cplex.Param.mip.pool.intensity mip.pool.intensity
Interactive mip pool intensity mip pool intensity
Identifier 2107 2107

Description

Controls the trade-off between the number of solutions generated for the solution
pool and the amount of time or memory consumed. This parameter applies both to
MIP optimization and to the populate procedure.

Values from 1 (one) to 4 invoke increasing effort to find larger numbers of

solutions. Higher values are more expensive in terms of time and memory but are
likely to yield more solutions.

Effect

For MIP optimization, increasing the value of the parameter corresponds to

increasing the amount of effort spent setting up the branch and cut tree to prepare
for a subsequent call to the populate procedure.

For populate, increasing the value of this parameter corresponds, in addition, to

increasing the amount of effort spent exploring the tree to generate more solutions.
If MIP optimization is called before populate, populate will reuse the information
computed and stored during MIP optimization only if this parameter has not been
increased between calls. Similarly, if populate is called several times successively,
populate will re-use the information computed and stored during previous calls to
populate only if the solution pool intensity has not increased between calls.
Therefore, it is most efficient not to change the value of this parameter between
calls to MIP optimization and populate, nor between successive calls of populate.
Increase the value of this parameter only if too few solutions are generated.

Settings

Its default value, 0 (zero), lets CPLEX choose which intensity to apply. If MIP
optimization is called first after the model is read, CPLEX sets the intensity to
1 (one) for this call to MIP optimization and to subsequent calls of populate. In
contrast, if populate is called directly after the model is read, CPLEX sets the
intensity to 2 for this call and subsequent calls of populate.

For value 1 (one), the performance of MIP optimization is not affected. There is no
slowdown and no additional consumption of memory due to this setting.
However, populate will quickly generate only a small number of solutions.
Generating more than a few solutions with this setting will be slow. When you are
looking for a larger number of solutions, use a higher value of this parameter.

Chapter 3. List of CPLEX parameters 133

 For value 2, some information is stored in the branch and cut tree so that it is
easier to generate a larger number of solutions. This storage has an impact on
memory used but does not lead to a slowdown in the performance of MIP
optimization. With this value, calling populate is likely to yield a number of
solutions large enough for most purposes. This value is a good choice for most
models.

For value 3, the algorithm is more aggressive in computing and storing

information in order to generate a large number of solutions. Compared to values
1 (one) and 2, this value will generate a larger number of solutions, but it will slow
MIP optimization and increase memory consumption. Use this value only if setting
this parameter to 2 does not generate enough solutions.

For value 4, the algorithm generates all solutions to your model. Even for small
models, the number of possible solutions is likely to be huge; thus enumerating all
of them will take time and consume a large quantity of memory. In this case,
remember to set the “maximum number of solutions generated for solution pool
by populate” on page 102 (PopulateLim, CPX_PARAM_POPULATELIM) to a value
appropriate for your model; otherwise, the populate procedure will stop
prematurely because of this stopping criterion instead of enumerating all solutions.
In addition, a few limitations apply to this exhaustive enumeration, as explained in
Enumerating all solutions in the CPLEX User’s Manual.
Table 71. Values
Value Meaning
0 Automatic: let CPLEX choose; default
1 Mild: generate few solutions quickly
2 Moderate: generate a larger number of
solutions
3 Aggressive: generate many solutions and
expect performance penalty
4 Very aggressive: enumerate all practical
solutions

solution pool replacement strategy

Designates the strategy for replacing a solution in the solution pool when the
solution pool has reached its capacity.

Purpose

Solution pool replacement strategy

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Pool_Replace CPX_PARAM_SOLNPOOLREPLACE
C++ IloCplex::Param::MIP::Pool::Replace SolnPoolReplace (int)
Java IloCplex.Param.MIP.Pool.Replace SolnPoolReplace (int)
.NET Cplex.Param.MIP.Pool.Replace SolnPoolReplace (int)
OPL solnpoolreplace
Python parameters.mip.pool.replace mip.pool.replace
MATLAB Cplex.Param.mip.pool.replace mip.pool.replace
Interactive mip pool replace mip pool replace
Identifier 2104 2104

134 CPLEX Parameters Reference

 Description

Designates the strategy for replacing a solution in the solution pool when the
solution pool has reached its capacity.

The value 0 (CPX_SOLNPOOL_FIFO ) replaces solutions according to a first-in, first-out

policy. The value 1 (CPX_SOLNPOOL_OBJ ) keeps the solutions with the best objective
values. The value 2 (CPX_SOLNPOOL_DIV ) replaces solutions in order to build a set of
diverse solutions. When the value is 2, CPLEX considers only variables of the type
binary or integer (not continuous variables) to calculate diversity in the
replacement strategy.

If the solutions you obtain are too similar to each other, try setting
SolnPoolReplace to 2.

The replacement strategy applies only to the subset of solutions created in the
current call of MIP optimization or populate. Solutions already in the pool are not
affected by the replacement strategy. They will not be replaced, even if they satisfy
the criterion of the replacement strategy.
Table 72. Values
Value Symbol Meaning
0 CPX_SOLNPOOL_FIFO Replace the first solution
(oldest) by the most recent
solution; first in, first out;
default
1 CPX_SOLNPOOL_OBJ Replace the solution which
has the worst objective
2 CPX_SOLNPOOL_DIV Replace solutions in order to
build a set of diverse
solutions

solution target type

Specifies type of solution CPLEX targets (optimal convex or first-order satisfaction)

Purpose

Solution target type

API Parameter Name Name prior to V12.6.0

C CPXPARAM_SolutionTarget CPX_PARAM_SOLUTIONTARGET
C++ IloCplex::Param::SolutionTarget SolutionTarget (int)
Java IloCplex.Param.SolutionTarget SolutionTarget (int)
.NET Cplex.Param.SolutionTarget SolutionTarget (int)
OPL solutiontarget solutiontarget
Python parameters.solutiontarget solutiontarget
MATLAB Cplex.Param.solutiontarget solutiontarget
Interactive solutiontarget solutiontarget
Identifier 1131 1131

Chapter 3. List of CPLEX parameters 135

 Description

Specifies the type of solution CPLEX attempts to compute when CPLEX solves a
nonconvex, continuous or mixed integer quadratic model; that is, nonconvex QP or
nonconvex MIQP. In other words, the variables of the model can be continuous or
mixed integer and continuous; the objective function includes a quadratic term,
and the objective function is not positive semi-definite (non PSD).

By default, CPLEX first attempts to compute a provably optimal solution to such a

problem. If CPLEX cannot compute a provably optimal solution because the
objective function is not convex, CPLEX terminates and returns the error
CPXERR_Q_NOT_POS_DEF.

When this parameter is set to 1 (one), CPLEX searches for a globally optimal
solution to a convex model. In problems of type QP or MIQP, this setting interacts
with linearization switch for QP, MIQP.

When this parameter is set to 2, CPLEX first attempts to compute a provably

optimal solution. If CPLEX cannot compute a provably optimal solution because
the objective function is not convex, CPLEX searches for a solution that satisfies
first-order optimality conditions but is not necessarily globally optimal. In such a
case, you can query the solution status to determine the kind of solution that
CPLEX found.

When this parameter is set to 3, if the problem type is QP, CPLEX first changes the
problem type to MIQP. CPLEX then solves the problem (whether originally QP or
MIQP) to global optimality. In problems of type QP or MIQP, this setting interacts
with linearization switch for QP, MIQP.

Tip: When the value of this parameter is 3 (that is, you have instructed CPLEX to
search for a globally optimal solution to a nonconvex QP or MIQP), then
information about dual values is not available for the solution.
Table 73. Values
Value Symbol Meaning
0 CPX_SOLUTIONTARGET_AUTO Automatic: let CPLEX decide;
default
1 CPX_SOLUTIONTARGET_OPTIMALCONVEX Searches for a globally optimal
solution to a convex model.
2 CPX_SOLUTIONTARGET_FIRSTORDER Searches for a solution that
satisfies first-order optimality
conditions, but is not necessarily
globally optimal.
3 CPX_SOLUTIONTARGET_OPTIMALGLOBAL Searches for a globally optimal
solution to a nonconvex model;
changes problem type to MIQP if
necessary.

MIP strong branching candidate list limit

Controls the length of the candidate list when CPLEX uses variable selection as the
setting for strong branching.

136 CPLEX Parameters Reference

 Purpose

MIP strong branching candidate list limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_StrongCand CPX_PARAM_STRONGCANDLIM
C++ IloCplex::Param::MIP::Limits::StrongCand StrongCandLim (int)
Java IloCplex.Param.MIP.Limits.StrongCand StrongCandLim (int)
.NET Cplex.Param.MIP.Limits.StrongCand StrongCandLim (int)
OPL strongcandlim
Python parameters.mip.limits.strongcand mip.limits.strongcand
MATLAB Cplex.Param.mip.limits.strongcand mip.limits.strongcand
Interactive mip limits strongcand mip limits strongcand
Identifier 2045 2045

Description

Controls the length of the candidate list when CPLEX uses strong branching as the
way to select variables. For more detail about that parameter, see “MIP variable
selection strategy” on page 148:
v VarSel in the C++, Java, or .NET API;
v CPX_PARAM_VARSEL in the C API;
v set mip strategy variableselect 3 in the Interactive Optimizer.

Values

Any positive number; default: 10.

MIP strong branching iterations limit

Controls the number of simplex iterations performed on each variable in the
candidate list when CPLEX uses variable selection as the setting for strong
branching.

Purpose

MIP strong branching iterations limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_StrongIt CPX_PARAM_STRONGITLIM
C++ IloCplex::Param::MIP::Limits::StrongIt StrongItLim (long)
Java IloCplex.Param.MIP.Limits.StrongIt StrongItLim (long)
.NET Cplex.Param.MIP.Limits.StrongIt StrongItLim (long)
OPL strongitlim
Python parameters.mip.limits.strongit mip.limits.strongit
MATLAB Cplex.Param.mip.limits.strongit mip.limits.strongit
Interactive mip limits strongit mip limits strongit
Identifier 2046 2046

Chapter 3. List of CPLEX parameters 137

 Description

Controls the number of simplex iterations performed on each variable in the

candidate list when CPLEX uses strong branching as the way to select variables.
For more detail about that parameter, see “MIP variable selection strategy” on page
148:
v VarSel in the C++, Java, or .NET API;
v CPX_PARAM_VARSEL in the C API;
v set mip strategy variableselect 3 in the Interactive Optimizer.

The default setting 0 (zero) chooses the iteration limit automatically.

Table 74. Values
Value Meaning
0 Automatic: let CPLEX choose; default
Any positive integer Limit of the simplex iterations performed on
each candidate variable

limit on nodes explored when a subMIP is being solved

Restricts the number of nodes explored when CPLEX is solving a subMIP.

Purpose
Limit on nodes explored when a subMIP is being solved

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_SubMIPNodeLim CPX_PARAM_SUBMIPNODELIM
C++ IloCplex::Param::MIP::Limits::SubMIPNodeLim SubMIPNodeLim (long)
Java IloCplex.Param.MIP.Limits.SubMIPNodeLim SubMIPNodeLim (long)
.NET Cplex.Param.MIP.Limits.SubMIPNodeLim SubMIPNodeLim (long)
OPL submipnodelim
Python parameters.mip.limits.submipnodelim mip.limits.submipnodelim
MATLAB Cplex.Param.mip.limits.submipnodelim mip.limits.submipnodelim
Interactive mip limits submipnodelim mip limits submipnodelim
Identifier 2062 2062

Description

Restricts the number of nodes explored when CPLEX is solving a subMIP.

CPLEX solves subMIPs in these situations:

v when it builds a solution from a partial MIP start;
v when it repairs an infeasible MIP start;
v when it executes the relaxation induced neighborhood search (RINS) heuristic;
v when it branches locally;
v when it polishes a solution.

Values

Any positive integer; default: 500.

138 CPLEX Parameters Reference

symmetry breaking
Decides whether symmetry breaking reductions will be automatically executed,
during the preprocessing phase, in a MIP model.

Purpose

Symmetry breaking

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Preprocessing_Symmetry CPX_PARAM_SYMMETRY
C++ IloCplex::Param::Preprocessing::Symmetry Symmetry (int)
Java IloCplex.Param.Preprocessing.Symmetry Symmetry (int)
.NET Cplex.Param.Preprocessing.Symmetry Symmetry (int)
OPL symmetry
Python parameters.preprocessing.symmetry preprocessing.symmetry
MATLAB Cplex.Param.preprocessing.symmetry preprocessing.symmetry
Interactive preprocessing symmetry preprocessing symmetry
Identifier 2059 2059

Description

Decides whether symmetry breaking reductions will be automatically executed,

during the preprocessing phase, in a MIP model. The default level, -1, allows
CPLEX to choose the degree of symmetry breaking to apply. The value 0 (zero)
turns off symmetry breaking. Levels 1 through 5 apply increasingly aggressive
symmetry breaking.
Table 75. Values
Value Meaning
-1 Automatic: let CPLEX choose; default
0 Turn off symmetry breaking
1 Exert a moderate level of symmetry
breaking
2 Exert an aggressive level of symmetry
breaking
3 Exert a very aggressive level of symmetry
breaking
4 Exert a highly aggressive level of symmetry
breaking
5 Exert an extremely aggressive level of
symmetry breaking

global default thread count

Sets the default number of parallel threads that will be invoked by any CPLEX
parallel optimizer.

Chapter 3. List of CPLEX parameters 139

 Purpose

Global default thread count

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Threads CPX_PARAM_THREADS
C++ IloCplex::Param::Threads Threads (int)
Java IloCplex.Param.Threads Threads (int)
.NET Cplex.Param.Threads Threads (int)
OPL threads
Python parameters.threads threads
MATLAB Cplex.Param.threads threads
Interactive threads threads
Identifier 1067 1067

Description

Sets the default maximal number of parallel threads that will be invoked by any
CPLEX parallel optimizer.

For a single thread, the parallel algorithms behave deterministically, regardless of

thread parameter settings; that is, the algorithm proceeds sequentially in a single
thread.

In this context, sequential means that the algorithm proceeds step by step,
consecutively, in a predictable and repeatable order within a single thread.
Deterministic means that repeated solving of the same model with the same
parameter settings on the same computing platform will follow exactly the same
solution path, yielding the same level of performance and the same values in the
solution. Sequential execution is deterministic. In multithreaded computing, a
deterministic setting requires synchronization between threads. Opportunistic
entails less synchronization between threads and thus may offer better
performance at the sacrifice of repeatable, invariant solution paths and values in
repeated runs on multiple threads or multiple processors.

When this parameter is at its default setting 0 (zero), and your application includes
no callbacks or only an informational callback, CPLEX can use all available
threads; that is, at most 32 threads or the number of cores of the machine,
whichever is smaller. If your machine offers more than 32 threads, you can take
advantage of them by increasing the value of this parameter.

When this parameter is at its default setting 0 (zero), and your application includes
callbacks other than informational callbacks (that is, the application includes a
query, diagnostic, or control callback), then CPLEX uses one thread. In other
words, the presence of a callback turns off parallel processing when the value of
this parameter is at its default.

In order to use parallel optimization in conjunction with callbacks, you need to

set this parameter to a positive value. However, when you do so, you need to be
aware of the fact that the callbacks may be invoked concurrently.

For a description of informational, query, diagnostic, and control callbacks, see the
topic Using optimization callbacks in the CPLEX User’s Manual.

140 CPLEX Parameters Reference

 Table 76. Values
Value Meaning
0 Automatic: let CPLEX decide; default
1 Sequential; single threaded
N Uses up to N threads; N is limited by available processors and Processor
Value Units (PVU).

See also

“parallel mode switch” on page 93

optimizer time limit in seconds

Sets the maximum time, in seconds, for a call to an optimizer.

Purpose

Optimizer time limit in seconds

API Parameter Name Name prior to V12.6.0

C CPXPARAM_TimeLimit CPX_PARAM_TILIM
C++ IloCplex::Param::TimeLimit TiLim (double)
Java IloCplex.Param.TimeLimit TiLim (double)
.NET Cplex.Param.TimeLimit TiLim (double)
OPL tilim
Python parameters.timelimit timelimit
MATLAB Cplex.Param.timelimit timelimit
Cplex class
API
MATLAB Optimization Toolbox compatible MaxTime
Cplex class
API
Interactive timelimit timelimit
Identifier 1039 1039

Description

Sets the maximum time, in seconds, for a call to an optimizer. This time limit
applies also to the conflict refiner.

The time is measured in terms of either CPU time or elapsed time, according to the
setting of the “clock type for computation time” on page 35 parameter
(CPX_PARAM_CLOCKTYPE, ClockType).

The time limit for an optimizer applies to the sum of all its steps, such as
preprocessing, crossover, and internal calls to other optimizers.

In a sequence of calls to optimizers, the limit is not cumulative but applies to each
call individually. For example, if you set a time limit of 10 seconds, and you call
MIP optimization twice then there could be a total of (at most) 20 seconds of
running time if each call consumes its maximum allotment.

See also

Chapter 3. List of CPLEX parameters 141

 For a deterministic time limit on optimization, see “deterministic time limit” on
page 45 (CPX_PARAM_DETTILIM, DetTiLim).

For an introduction to time stamps measured in seconds, see the topic Timing
interface in the CPLEX User's Manual. For more detail about time stamps measured
in seconds, see the reference manual of the API that you use.
v In the Callable Library (C API), see the documentation of CPXXgettime and
CPXXgetcallbackinfo.
v In the C++ API, see the documentation of IloCplex::CallbackI::getStartTime and
IloCplex::CallbackI::getEndTime.
v In the Java API, see the documentation of IloCplex.Callback.getStartTime and
IloCplex.Callback.getEndTime.
v In the .NET API, see the documentation of Cplex.ICallback.GetStartTime and
GetEndTime.
v In the Python API, see the documentation of Callback.get_start_time and
Callback.get_end_time.
v In the MATLAB connector, see the documentation of cplex.Solution.time.

Values

Any nonnegative value in seconds; default: 1e+75.

See also

“clock type for computation time” on page 35

tree memory limit

Sets an absolute upper limit on the size (in megabytes, uncompressed) of the
branch-and-cut tree.

Purpose

Tree memory limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Limits_TreeMemory CPX_PARAM_TRELIM
C++ IloCplex::Param::MIP::Limits::TreeMemory TreLim (double)
Java IloCplex.Param.MIP.Limits.TreeMemory TreLim (double)
.NET Cplex.Param.MIP.Limits.TreeMemory TreLim (double)
OPL trelim
Python parameters.mip.limits.treememory mip.limits.treememory
MATLAB Cplex.Param.mip.limits.treememory mip.limits.treememory
Interactive mip limits treememory mip limits treememory
Identifier 2027 2027

Description

Sets an absolute upper limit on the size (in megabytes, uncompressed) of the
branch-and-cut tree. If this limit is exceeded, CPLEX terminates optimization.

Values

Any nonnegative number; default: 1e+75.

142 CPLEX Parameters Reference

deterministic tuning time limit
Sets a time limit in deterministic ticks per model and per test set (that is, suite of
models) applicable in tuning.

Purpose

Deterministic tuning time limit

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Tune_DetTimeLimit CPX_PARAM_TUNINGDETTILIM
C++ IloCplex::Param::Tune::DetTimeLimit TuningDetTiLim (double)
Java IloCplex.Param.Tune.DetTimeLimit TuningDetTiLim (double)
.NET Cplex.Param.Tune.DetTimeLimit TuningDetTiLim (double)
OPL tuningdettilim
Python parameters.tune.dettimelimit tuning.dettimelimit
MATLAB Cplex.Param.tune.dettimelimit tune.timelimit
Interactive tune dettimelimit tune dettimelimit
Identifier 1139 1139

Description

Sets a deterministic time limit per model and per test set (that is, suite of models)
applicable in tuning and measured in ticks.

When this deterministic tuning time limit is set to a finite value, then tuning finds
appropriate settings of other CPLEX parameters to minimize the deterministic time
of optimization. Furthermore, the tuning process itself is deterministic. In this
context, "a finite value" means any value strictly less than 1e+75 (such as the finite
value 1e+74).

Interaction with other parameters: nondeterministic tuning time

limit

This deterministic time limit on tuning is not compatible with the wall-clock
“tuning time limit in seconds” on page 147 (CPX_PARAM_TUNINGTILIM,
TuningTiLim). Only one of these two parameters can be set to a finite value at a
time. Any attempt to set either of these parameters to a finite value while the other
is already set to a finite value results in the error
CPXERR_PARAM_INCOMPATIBLE from the routine CPXsetdblparam or the
method setDblParam (depending on the API you are using).

Finite values of tuning time limits

If this deterministic time limit on tuning is set to a finite value, then the tuning
process itself is deterministic, and CPLEX recommends appropriate parameter
settings to minimize the deterministic optimization time.

If the wall-clock “tuning time limit in seconds” on page 147

(CPX_PARAM_TUNINGTILIM, TuningTiLim) is set to a finite value, then the
tuning process itself is nondeterministic, and it recommends appropriate
parameter settings to minimize the wall-clock optimization time.

The default value of this parameter is 1e+75 (effectively, infinite).

Chapter 3. List of CPLEX parameters 143

 Likewise, the default value of the wall-clock “tuning time limit in seconds” on
page 147 (CPX_PARAM_TUNINGTILIM, TuningTiLim) is also 1e+75 (effectively,
infinite).

If this parameter is set at its default value 1e+75, and if the “tuning time limit in
seconds” on page 147 (CPX_PARAM_TUNINGTILIM, TuningTiLim) is also set at
its default value 1e+75, then the combination is equivalent to setting the
deterministic tuning time limit to 10 000 000 ticks. Consequently, these combined
default settings make the tuning process deterministic, and CPLEX recommends
settings to minimize the deterministic optimization time.

Unlimited time per model

If you want to run a tuning session with unlimited time per model, then set one of
the tuning time limit parameters (either wall-clock “tuning time limit in seconds”
on page 147 (CPX_PARAM_TUNINGTILIM, TuningTiLim) or “deterministic tuning
time limit” on page 143 (CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim) to a
very large value that is strictly less than 1e+75 (for example, 1e+74). If you set
CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim to a finite value, then the
tuning process will be deterministic. If you set CPX_PARAM_TUNINGTILIM,
TuningTiLim to a finite value, then the tuning process will be nondeterministic.

Ticks

A tick is a unit to measure work done deterministically. The length of a

deterministic tick may vary by platform. Nevertheless, ticks are normally
consistent measures for a given platform (combination of hardware and software)
carrying the same load. In other words, the correspondence of ticks to clock time
depends on the hardware, software, and the current load of the machine. For the
same platform and same load, the ratio of ticks per second stays roughly constant,
independent of the model solved. However, for very short optimization runs, the
variation of this ratio is typically high.

Values

Any nonnegative number; default: 1e+75 ticks.

See also

“optimizer time limit in seconds” on page 141

“deterministic time limit” on page 45

“tuning time limit in seconds” on page 147

tuning information display

Specifies the level of information reported by the tuning tool as it works.

Purpose

Tuning information display

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Tune_Display CPX_PARAM_TUNINGDISPLAY
C++ IloCplex::Param::Tune::Display TuningDisplay (int)

144 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Java IloCplex.Param.Tune.Display TuningDisplay (int)
.NET Cplex.Param.Tune.Display TuningDisplay (int)
OPL tuningdisplay
Python parameters.tune.display tuning.display
MATLAB Cplex.Param.tune.display tune.display
Interactive tune display tune display
Identifier 1113 1113

Description

Specifies the level of information reported by the tuning tool as it works.

Use level 0 (zero) to turn off reporting from the tuning tool.

Use level 1 (one), the default, to display a minimal amount of information.

Use level 2 to display the minimal amount plus the parameter settings that the
tuning tool is trying.

Use level 3 to display an exhaustive report of minimal information, plus settings

that are being tried, plus logs.
Table 77. Values
Value Meaning
0 Turn off display
1 Display standard, minimal reporting; default
2 Display standard report plus parameter
settings being tried
3 Display exhaustive report and log

tuning measure
Controls the measure for evaluating progress when a suite of models is being
tuned.

Purpose

Tuning measure

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Tune_Measure CPX_PARAM_TUNINGMEASURE
C++ IloCplex::Param::Tune::Measure TuningMeasure
Java IloCplex.Param.Tune.Measure TuningMeasure
.NET Cplex.Param.Tune.Measure TuningMeasure
OPL tuningmeasure
Python parameters.tune.measure tuning.measure
MATLAB Cplex.Param.tune.measure tuning.measure
Interactive tune measure tune measure
Identifier 1110 1110

Chapter 3. List of CPLEX parameters 145

 Description

Controls the measure for evaluating progress when a suite of models is being
tuned.

Possible values are:

v CPX_TUNE_AVERAGE uses the mean average of time to compare different
parameter sets over a suite of models.
v CPX_TUNE_MINMAX uses a minmax approach to compare the time of different
parameter sets over a suite of models.
Table 78. Values
Value Meaning
CPX_TUNE_AVERAGE mean time; default
CPX_TUNE_MINMAX minmax time

tuning repeater
Specifies the number of times tuning is to be repeated on reordered versions of a
given problem.

Purpose

Tuning repeater

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Tune_Repeat CPX_PARAM_TUNINGREPEAT
C++ IloCplex::Param::Tune::Repeat TuningRepeat (int)
Java IloCplex.Param.Tune.Repeat TuningRepeat (int)
.NET Cplex.Param.Tune.Repeat TuningRepeat (int)
OPL tuningrepeat
Python parameters.tune.repeat tuning.repeat
MATLAB Cplex.Param.tune.repeat tuning.repeat
Interactive tune repeat tune repeat
Identifier 1111 1111

Description

Specifies the number of times tuning is to be repeated on reordered versions of a

given problem. The problem is reordered automatically by CPLEX permuting its
rows and columns. This repetition is helpful when only one problem is being
tuned, as repeated reordering and re-tuning may lead to more robust tuning
results.

This parameter applies to only one problem in a tuning session. That is, in the
Interactive Optimizer, this parameter is effective only when you are tuning a single
problem; in the Callable Library (C API), this parameter is effective only when you
are tuning a single problem with the routine CPXtuneparam .

Values

Any nonnegative integer; default: 1 (one)

146 CPLEX Parameters Reference

tuning time limit in seconds
Sets a nondeterministic time limit in seconds per model and per test set (that is,
suite of models) applicable in tuning.

Purpose

Nondeterministic tuning time limit (wall-clock time)

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Tune_TimeLimit CPX_PARAM_TUNINGTILIM
C++ IloCplex::Param::Tune::TimeLimit TuningTiLim (double)
Java IloCplex.Param.Tune.TimeLimit TuningTiLim (double)
.NET Cplex.Param.Tune.TimeLimit TuningTiLim (double)
OPL tuningtilim
Python parameters.tune.timelimit tuning.timelimit
MATLAB Cplex.Param.tune.timelimit tuning.timelimit
Interactive tune timelimit tune timelimit
Identifier 1112 1112

Description

Sets a nondeterministic time limit in seconds per model and per test set (that is,
suite of models) applicable in tuning. This parameter is also known as the
wall-clock time limit on tuning.

Interaction with other parameters: deterministic tuning time limit

The “deterministic tuning time limit” on page 143

(CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim) is not compatible with this
wall-clock, nondeterministic tuning time limit (CPX_PARAM_TUNINGTILIM,
TuningTiLim). Only one of these two parameters can be set to a finite value at a
time. Any attempt to set either of these parameters to a finite value while the other
is already set to a finite value results in the error
CPXERR_PARAM_INCOMPATIBLE from the routine CPXsetdblparam or the
method setDblParam (depending on your choice of API).

Finite values of tuning time limits

If this wall-clock, nondeterministic tuning time parameter

(CPX_PARAM_TUNINGTILIM, TuningTiLim) is set to a finite value, then the
tuning process itself is nondeterministic, and CPLEX recommends appropriate
parameter settings to minimize the wall-clock optimization time.

If the “deterministic tuning time limit” on page 143

(CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim) is set to a finite value, then
the tuning process itself is deterministic, and CPLEX recommends appropriate
parameter settings to minimize the deterministic optimization time.

The default value of this parameter is 1e+75 (effectively, infinite).

Likewise, the default value of the “deterministic tuning time limit” on page 143
(CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim) is also 1e+75 (effectively,
infinite).

Chapter 3. List of CPLEX parameters 147

 If this parameter is set at its default value 1e+75, and if the “deterministic tuning
time limit” on page 143 (CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim) is
also set at its default value 1e+75, then the combination is equivalent to setting the
deterministic tuning time limit to 10 000 000 ticks. Consequently, these combined
default settings make the tuning process deterministic, and CPLEX recommends
settings to minimize the deterministic optimization time.

Unlimited time per model

If you want to run a tuning session with unlimited time per model, then set one of
the tuning time limit parameters (either wall-clock “tuning time limit in seconds”
on page 147 (CPX_PARAM_TUNINGTILIM, TuningTiLim) or “deterministic tuning
time limit” on page 143 (CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim) to a
very large value that is strictly less than 1e+75 (for example, 1e+74). If you set
CPX_PARAM_TUNINGDETTILIM, TuningDetTiLim to a finite value, then the
tuning process will be deterministic. If you set CPX_PARAM_TUNINGTILIM,
TuningTiLim to a finite value, then the tuning process will be nondeterministic.

Examples

For an example of how to use general and tuning-specific time limit parameters
together, see Examples: time limits on tuning in the Interactive Optimizer in the
CPLEX User’s Manual.

Values

Any nonnegative number; default: 1e+75 seconds.

See also

“optimizer time limit in seconds” on page 141

“deterministic time limit” on page 45

“deterministic tuning time limit” on page 143

MIP variable selection strategy

Sets the rule for selecting the branching variable at the node which has been
selected for branching.

Purpose

MIP variable selection strategy

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Strategy_VariableSelect CPX_PARAM_VARSEL
C++ IloCplex::Param::MIP::Strategy::VariableSelect VarSel (int)
Java IloCplex.Param.MIP.Strategy.VariableSelect VarSel (int)
.NET Cplex.Param.MIP.Strategy.VariableSelect VarSel (int)
OPL varsel
Python parameters.mip.strategy.variableselect mip.strategy.variableselect
MATLAB Cplex.Param.mip.strategy.variableselect mip.strategy.variableselect
Cplex class
API
MATLAB Optimization Toolbox compatible BranchStrategy

148 CPLEX Parameters Reference

 API Parameter Name Name prior to V12.6.0
Interactive mip strategy variableselect mip strategy variableselect
Identifier 2028 2028

Description

Sets the rule for selecting the branching variable at the node which has been
selected for branching.

The minimum infeasibility rule chooses the variable with the value closest to an
integer but still fractional. The minimum infeasibility rule (-1) may lead more
quickly to a first integer feasible solution, but is usually slower overall to reach the
optimal integer solution.

The maximum infeasibility rule chooses the variable with the value furtherest from
an integer. The maximum infeasibility rule (1 one) forces larger changes earlier in
the tree.

Pseudo cost (2) variable selection is derived from pseudo-shadow prices.

Strong branching (3) causes variable selection based on partially solving a number
of subproblems with tentative branches to see which branch is the most promising.
This strategy can be effective on large, difficult MIP problems.

Pseudo reduced costs (4) are a computationally less-intensive form of pseudo costs.

The default value (0 zero) allows CPLEX to select the best rule based on the
problem and its progress.
Table 79. Values
Value Symbol Meaning
-1 CPX_VARSEL_MININFEAS Branch on variable with minimum
infeasibility
0 CPX_VARSEL_DEFAULT Automatic: let CPLEX choose variable to
branch on; default
1 CPX_VARSEL_MAXINFEAS Branch on variable with maximum
infeasibility
2 CPX_VARSEL_PSEUDO Branch based on pseudo costs
3 CPX_VARSEL_STRONG Strong branching
4 CPX_VARSEL_PSEUDOREDUCED Branch based on pseudo reduced costs

directory for working files

Specifies the name of an existing directory into which CPLEX may store temporary
working files.

Purpose

Directory for working files

API Parameter Name Name prior to V12.6.0

C CPXPARAM_WorkDir CPX_PARAM_WORKDIR

Chapter 3. List of CPLEX parameters 149

 API Parameter Name Name prior to V12.6.0
C++ IloCplex::Param::WorkDir WorkDir (string)
Java IloCplex.Param.WorkDir WorkDir (string)
.NET Cplex.Param.WorkDir WorkDir (string)
OPL workdir
Python parameters.workdir workdir
MATLAB Cplex.Param.workdir workdir
Interactive workdir workdir
Identifier 1064 1064

Description

Specifies the name of an existing directory into which CPLEX may store temporary
working files, such as for MIP node files or for out-of-core barrier files. The default
is the current working directory.

This parameter accepts a string as its value. If you change either the “API string
encoding switch” on page 20 or the “file encoding switch” on page 57 from their
default value to a multi-byte encoding where a NULL byte can occur within the
encoding of a character, you must take into account the issues documented in the
topic Selecting an encoding in the CPLEX User's Manual. Especially consider the
possibility that a NULL byte occurring in the encoding of a character can
inadvertently signal the termination of a string, such as a filename or directory
path, and thus provoke surprising or incorrect results.

Tip: If the string designating the path to the target directory includes one or more
spaces, be sure to include the entire string in double quotation marks.

Values

Any existing directory; default: ‘.’

memory available for working storage

Specifies an upper limit on the amount of central memory, in megabytes, that
CPLEX is permitted to use for working memory.

Purpose

Memory available for working storage

API Parameter Name Name prior to V12.6.0

C CPXPARAM_WorkMem CPX_PARAM_WORKMEM
C++ IloCplex::Param::WorkMem WorkMem (double)
Java IloCplex.Param.WorkMem WorkMem (double)
.NET Cplex.Param.WorkMem WorkMem (double)
OPL workmem
Python parameters.workmem workmem
MATLAB Cplex.Param.workmem workmem
Interactive workmem workmem
Identifier 1065 1065

150 CPLEX Parameters Reference

 Description

Specifies an upper limit on the amount of central memory, in megabytes, that

CPLEX is permitted to use for working memory before swapping to disk files,
compressing memory, or taking other actions.

Values

Any nonnegative number, in megabytes; default: 2048

See also

“directory for working files” on page 149

write level for MST, SOL files

Sets a level of detail for CPLEX to write a file in MST or SOL format.

Purpose

Write level for MST, SOL files

API Parameter Name Name prior to V12.6.0

C CPXPARAM_Output_WriteLevel CPX_PARAM_WRITELEVEL
C++ IloCplex::Param::Output::WriteLevel WriteLevel (int)
Java IloCplex.Param.Output.WriteLevel WriteLevel (int)
.NET Cplex.Param.Output.WriteLevel WriteLevel (int)
OPL not available
Python parameters.output.writelevel output.writelevel
MATLAB Cplex.Param.output.writelevel output.writelevel
Interactive output writelevel output writelevel
Identifier 1114 1114

Description

Sets the level of detail for CPLEX to write a solution to a file in SOL format or a
MIP start to a file in MST format. CPLEX writes information about a MIP start to a
formatted file of type MST with the file extension .mst. CPLEX writes information
about a solution to a formatted file of type SOL with the file extension .sol.
CPLEX records the write level at which it created a file in that file, so that the file
can be read back accurately later.

The default setting of this parameter is 0 (zero) AUTO; that is, let CPLEX decide
the level of detail. CPLEX behaves differently, depending on whether the format is
SOL or MST and on whether it is writing a solution or MIP start. For SOL files,
AUTO resembles level 1 (one): CPLEX writes all variables and their respective
values to the file. For MST files, AUTO resembles level 2: CPLEX writes discrete
variables and their respective values to the file.

When the value of this parameter is 1 (one), CPLEX writes all variables, both
discrete and continuous, with their values.

When the value of this parameter is 2, CPLEX writes values for discrete variables
only.

Chapter 3. List of CPLEX parameters 151

 When the value of this parameter is 3, CPLEX writes values of nonzero variables
only.

When the value of this parameter is 4, CPLEX writes values of nonzero discrete
variables only.

Treatment of nonzeros

With respect to levels 3 and 4, where nonzero values are significant, CPLEX
considers a value nonzero if the absolute value is strictly less than 1e-16. In the
case of SOL files, CPLEX applies this test to primal and dual variable values, that
is, both x and pi variable values. In the case of MST files, CPLEX applies this test
only to x values.

Restrictions due to reduced file size

Levels 3 and 4 reduce the size of files, of course. However, this reduced file entails
restrictions and may create surprising results when the file is re-used. Levels 3 and
4 are not equivalent to levels 1 and 2. Indeed, if a MIP start does not contain a
value for a variable expected at level 3 or 4, then this variable will be fixed to 0
(zero) when that MIP start file is processed. Specifically, at level 3, if the MIP start
does not specify a value for a variable of any type, or at level 4, if the MIP start
does not specify a value for a discrete variable, such a variable will be fixed to 0
(zero). Consequently, the same MIP start written at level 1 or 2 may produce
satisfactory solutions, but the reduced MIP start file, written at level 3 or 4,
perhaps does not lead to solutions. This surprising situation arises typically in the
case of model changes with the addition of new variables.
Table 80. Values
Value Symbol Meaning
0 AUTO Automatic: let CPLEX decide
1 CPX_WRITELEVEL_ALLVARS CPLEX writes all variables
and their values
2 CPX_WRITELEVEL_DISCRETEVARS CPLEX writes only discrete
variables and their values
3 CPX_WRITELEVEL_NONZEROVARS CPLEX writes only nonzero
variables and their values
4 CPX_WRITELEVEL_NONZERODISCRETEVARS CPLEX writes only nonzero
discrete variables and their
values

MIP zero-half cuts switch

Decides whether or not to generate zero-half cuts for the problem.

Purpose

MIP zero-half cuts switch

API Parameter Name Name prior to V12.6.0

C CPXPARAM_MIP_Cuts_ZeroHalfCut CPX_PARAM_ZEROHALFCUTS
C++ IloCplex::Param::MIP::Cuts::ZeroHalfCut ZeroHalfCuts (int)
Java IloCplex.Param.MIP.Cuts.ZeroHalfCut ZeroHalfCuts (int)
.NET Cplex.Param.MIP.Cuts.ZeroHalfCut ZeroHalfCuts (int)

152 CPLEX Parameters Reference

API Parameter Name Name prior to V12.6.0
OPL zerohalfcuts
Python parameters.mip.cuts.zerohalfcut mip.cuts.zerohalfcut
MATLAB Cplex.Param.mip.cuts.zerohalfcut mip.cuts.zerohalfcut
Interactive mip cuts zerohalfcut mip cuts zerohalfcut
Identifier 2111 2111

Description

Decides whether or not to generate zero-half cuts for the problem. The value
0 (zero), the default, specifies that the attempt to generate zero-half cuts should
continue only if it seems to be helping.

If you find that too much time is spent generating zero-half cuts for your model,
consider setting this parameter to -1 (minus one) to turn off zero-half cuts.

If the dual bound of your model does not make sufficient progress, consider
setting this parameter to 2 to generate zero-half cuts more aggressively.

For a definition of a zero-half cut, see the topic Zero-half cuts in the general topic
Cuts in the CPLEX User’s Manual. The table Parameters for controlling cutsru the
END, also in the user’s manual, includes links to the documentation of other
parameters affecting other types of cuts.
Table 81. Values
Value Meaning
-1 Do not generate zero-half cuts
0 Automatic: let CPLEX choose; default
1 Generate zero-half cuts moderately
2 Generate zero-half cuts aggressively

Chapter 3. List of CPLEX parameters 153

154 CPLEX Parameters Reference
Index
A barrier limit (continued)
growth 26
complementarity convergence
barrier (LP, QP) 25
absolute gap iterations 27 barrier (QCP) 29
solution pool 130 objective range 28 LP 25
absolute objective difference 91 barrier limits growth 26 QCP 29
accessing barrier ordering 28 QP 25
dual values of QCP 34 barrier starting algorithm 30 condition number 76
parameters 1 BarStartAlg 30 ConflictDisplay 39
sets of parameters 1 basic variable control callback 79
Advance 17 feasibility tolerance and 56 cover cut 39
advanced start 17 basis cover cut, flow 59
barrier and 17 advanced start and 17 Covers 39
basis and 17 crash ordering and 40 CPX_PARAM_ADVIND 17
node exploration limit 138 kappa computation 76 CPX_PARAM_AGGCUTLIM 18
presolve and 17 Markowitz threshold and 53 CPX_PARAM_AGGFILL 19
repair tries 118 network feasibility tolerance and 84 CPX_PARAM_AGGIND 19
root algorithm and 120 optimal and feasibility tolerance 56 CPX_PARAM_APIENCODING 20
AdvInd 17 root algorithm and 120 CPX_PARAM_AUXROOTTHREADS 124
AggCutLim 18 simplex iterations and 68 CPX_PARAM_BARALG 23
AggFill 19 simplex refactoring frequency CPX_PARAM_BARCOLNZ 23
aggregation limit 18 and 116 CPX_PARAM_BARCROSSALG 24
apiencoding 20 singularity repairs and 129 CPX_PARAM_BARDISPLAY 25
APIEncoding 20 BBInterval 30 CPX_PARAM_BAREPCOMP 25
AuxRootThreads 124 best bound interval 30 CPX_PARAM_BARGROWTH 26
best node CPX_PARAM_BARITLIM 27
absolutee mip gap and 50 CPX_PARAM_BARMAXCOR 27
B backtracking and 32 CPX_PARAM_BAROBJRNG 28
backtracking relative MIP gap and 50 CPX_PARAM_BARORDER 28
criteria for 32 target gap and 32 CPX_PARAM_BARQCPEPCOMP 29
node selection and 89 BndStrenInd 31 CPX_PARAM_BARSTARTALG 30
tolerance 32 bound strengthening 31 CPX_PARAM_BBINTERVAL 30
BarAlg 23 preprocessing 31 CPX_PARAM_BNDSTRENIND 31
BarColNz 23 bound violation CPX_PARAM_BRDIR 32
BarCrossAlg 24 feasibility (simplex) 56 CPX_PARAM_BTTOL 32
BarDisplay 25 FeasOpt 56 CPX_PARAM_CALCQCPDUALS 34
BarEpComp 25 network flow 84 CPX_PARAM_CLIQUES 35
BarGrowth 26 branching direction 32 CPX_PARAM_CLOCKTYPE 36
BarItLim 27 branching, local 68 CPX_PARAM_CLONELOG 36
BarMaxCor 27 BranchStrategy 148 CPX_PARAM_COEREDIND 37
BarObjRng 28 BrDir 32 CPX_PARAM_COLREADLIM 38
BarOrder 28 BtTol 32 CPX_PARAM_CONFLICTDISPLAY 39
BarQCPEpComp 29 CPX_PARAM_COVERS 39
barrier CPX_PARAM_CRAIND 40
advanced start and 17 C CPX_PARAM_CUTLO 41
CPX_PARAM_CUTPASS 42
detecting unbounded optimal CalcQCPDuals 34
faces 26 CPX_PARAM_CUTSFACTOR 42
callback reduced LP parameter 72
maximum absolute objective CPX_PARAM_CUTUP 43
callback, control 79
function 28 CPX_PARAM_DATACHECK 44
callbacks
barrier algorithm 23 CPX_PARAM_DEPIND 44
parallelism and 140
barrier column nonzeros 23 CPX_PARAM_DETTILIM 45
threads and 140
barrier convergence tolerance 25 CPX_PARAM_DISJCUTS 46
candidate list limit (MIP) 137
barrier crossover 24 CPX_PARAM_DIVETYPE 47
centering correction 27
barrier display 25 CPX_PARAM_DPRIIND 48
clique cut 35
barrier epsilon complementarity CPX_PARAM_EACHCUTLIM 49
Cliques 35
convergence 25 CPX_PARAM_EPAGAP 50
ClockType 36
barrier limit CPX_PARAM_EPGAP 50
CloneLog 36
absolute value of objective CPX_PARAM_EPINT 51
coefficient reduction
function 28 CPX_PARAM_EPMRK 53
preprocessing 37
centering corrections 27 CPX_PARAM_EPOPT 53
CoeRedInd 37
detecting unbounded optimal CPX_PARAM_EPPER 54
ColReadLim 38
faces 26 CPX_PARAM_EPRELAX 55

© Copyright IBM Corp. 1987, 2014 155

CPX_PARAM_EPRHS 56 CPX_PARAM_QPMAKEPSDIND 110 cut (continued)
CPX_PARAM_FEASOPTMODE 56 CPX_PARAM_QPMETHOD 122 node limit and 88
CPX_PARAM_FILEENCODING 58 CPX_PARAM_QPNZREADLIM 111 pass limit 42
CPX_PARAM_FLOWCOVERS 59 CPX_PARAM_RANDOMSEED 115 reapplying presolve and 119
CPX_PARAM_FLOWPATHS 60 CPX_PARAM_REDUCE 115 user-defined and preprocessing 105
CPX_PARAM_FPHEUR 61 CPX_PARAM_REINV 116 zero-half 152
CPX_PARAM_FRACCAND 62 CPX_PARAM_RELAXPREIND 117 CutLo 41
CPX_PARAM_FRACCUTS 62 CPX_PARAM_RELOBJDIF 117 cutoff tolerance 32
CPX_PARAM_FRACPASS 63 CPX_PARAM_REPAIRTRIES 118 CutPass 42
CPX_PARAM_GUBCOVERS 63 CPX_PARAM_REPEATPRESOLVE 119 CutsFactor 42
CPX_PARAM_HEURFREQ 64 CPX_PARAM_RINSHEUR 119 CutUp 43
CPX_PARAM_IMPLBD 65 CPX_PARAM_ROWREADLIM 125
CPX_PARAM_INTSOLFILEPREFIX 66 CPX_PARAM_SCAIND 126
CPX_PARAM_INTSOLLIM 67
CPX_PARAM_ITLIM 68
CPX_PARAM_SCRIND 126
CPX_PARAM_SIFTALG 127
D
DataCheck 44
CPX_PARAM_LANDPCUTS 69 CPX_PARAM_SIFTDISPLAY 128
dependency checking
CPX_PARAM_LBHEUR 68 CPX_PARAM_SIFTITLIM 128
preprocessing 44
CPX_PARAM_LPMETHOD 120 CPX_PARAM_SIMDISPLAY 129
DepInd 44
CPX_PARAM_MCFCUTS 70 CPX_PARAM_SINGLIM 129
deterministic
CPX_PARAM_MEMORYEMPHASIS 71 CPX_PARAM_SOLNPOOLAGAP 130
definition 93
CPX_PARAM_MIPCBREDLP 72 CPX_PARAM_SOLNPOOLCAPACITY 131
deterministic tick (definition) 45
CPX_PARAM_MIPDISPLAY 73 CPX_PARAM_SOLNPOOLGAP 132
deterministic time
CPX_PARAM_MIPEMPHASIS 74 CPX_PARAM_SOLNPOOLINTENSITY 133
and solution polishing 96
CPX_PARAM_MIPINTERVAL 75 CPX_PARAM_SOLNPOOLREPLACE 134
deterministic time limit 45
CPX_PARAM_MIPKAPPASTATS 76 CPX_PARAM_SOLUTIONTARGET 135
DetTiLim 45
CPX_PARAM_MIPORDIND 78 CPX_PARAM_STARTALG 123
Diagnostics 126
CPX_PARAM_MIPORDTYPE 78 CPX_PARAM_STRONGCANDLIM 137
DisjCuts 46
CPX_PARAM_MIPSEARCH 79 CPX_PARAM_STRONGITLIM 137
disjunctive cut 46
CPX_PARAM_MIQCPSTRAT 80 CPX_PARAM_SUBALG 86
DisplayFunc 126
CPX_PARAM_MIRCUTS 81 CPX_PARAM_SUBMIPNODELIM 138
distributed parallel optimization
CPX_PARAM_MPSLONGNUM 82 CPX_PARAM_SYMMETRY 139
ramp up 113
CPX_PARAM_NETDISPLAY 83 CPX_PARAM_THREADS 140
DiveType 47
CPX_PARAM_NETEPOPT 83 CPX_PARAM_TILIM 141
DPriInd 48
CPX_PARAM_NETEPRHS 84 CPX_PARAM_TRELIM 142
dual reduction 115
CPX_PARAM_NETFIND 84 CPX_PARAM_TUNINGDETTILIM 143
dual value
CPX_PARAM_NETITLIM 85 CPX_PARAM_TUNINGDISPLAY 144
calculating for QCP 34
CPX_PARAM_NETPPRIIND 86 CPX_PARAM_TUNINGMEASURE 145
CPX_PARAM_NODEFILEIND 87 CPX_PARAM_TUNINGREPEAT 146
CPX_PARAM_NODELIM 88 CPX_PARAM_TUNINGTILIM 147
CPX_PARAM_NODESEL 89 CPX_PARAM_VARSEL 148 E
CPX_PARAM_NUMERICALEMPHASIS 90 CPX_PARAM_WORKDIR 149 EachCutLim 49
CPX_PARAM_NZREADLIM 90 CPX_PARAM_WORKMEM 150 emphasis numerical 90
CPX_PARAM_OBJDIF 91 CPX_PARAM_WRITELEVEL 151 EpAGap 50
CPX_PARAM_OBJLLIM 92 CPX_PARAM_ZEROHALFCUTS 152 EpGap 50
CPX_PARAM_OBJULIM 92 CPXPARAM_DistMIP_Rampup_Duration 113 EpInt 51
CPX_PARAM_PARALLELMODE 93 CPXPARAM_MIP_Limits_RampupDetTimeLimit EpLin11252
CPX_PARAM_PERIND 95 CPXPARAM_MIP_Limits_RampupTimeLimit EpMrk 112 53
CPX_PARAM_PERLIM 96 CPXPARAM_Preprocessing_Aggregator 19 EpOpt 53
CPX_PARAM_POLISHAFTERDETTIME 96 CPXPARAM_Preprocessing_Fill 19 EpPer 54
CPX_PARAM_POLISHAFTEREPAGAP 97 CraInd 40 EpRelax 55
CPX_PARAM_POLISHAFTEREPGAP 98 cut EpRHS 56
CPX_PARAM_POLISHAFTERINTSOL 99 cliques (MIP) 35
CPX_PARAM_POLISHAFTERNODE 100 constraint aggregation limit and 18
CPX_PARAM_POLISHAFTERTIME 101
CPX_PARAM_POLISHTIME
covers (MIP) 39
disjunctive (MIP) 46
F
feasibility pump heuristic 61
(deprecated) 102 flow cover 59
FeasOpt
CPX_PARAM_POPULATELIM 102 flow path (MIP) 60
lower objective limit 55
CPX_PARAM_PPRIIND 103 fractional pass limit 63
mode 56
CPX_PARAM_PREDUAL 104 globally valid implied bound 65
feasopt mode 56
CPX_PARAM_PREIND 105 Gomory fractional candidate limit 62
feasopt tolerance 55
CPX_PARAM_PRELINEAR 105 Gomory fractional generation 62
FeasOptMode 56
CPX_PARAM_PREPASS 106 GUB (MIP) 63
FileEncoding 58
CPX_PARAM_PRESLVND 107 lift and project (MIP) 69
first order optimality conditions 135
CPX_PARAM_PRICELIM 108 limit by type 49
flow cover cut 59
CPX_PARAM_PROBE 108 limiting number of 42
aggregation limit 18
CPX_PARAM_PROBEDETTIME 109 MIP display and 73
flow path cut 60
CPX_PARAM_PROBETIME 109 mixed integer rounding (MIR) 81
FlowCovers 59

156 CPLEX Parameters Reference

FlowPaths 60
FPHeur 61
K mip limits populate 102
mip limits probedettime 109
FracCand 62 kappa 76 mip limits probetime 109
FracCuts 62 mip limits rampup duration 113
FracPass 63 mip limits rampupdettimelimit 112
fractional cut L mip limits rampuptimelimit 112
candidate limit 62 lazy constraint mip limits repairtries 118
generation 62 nonlinear reductions and 105 mip limits solutions 67
pass limit 63 preprocessing and 105, 115 mip limits strongcand 137
presolve reductions and 115 mip limits strongit 137
LBHeur 68 mip limits submipnodelim 138
G lift-and-project cut 69 mip limits treememory 142
mip ordertype 78
Gomory fractional cut LiftProjCuts 69
local branching heuristic 68 MIP performance
candidate limit 62
numerical difficulties 76
generation 62
mip polishafter absmipgap 97
pass limit 63
GUB cut 63 M mip polishafter dettime 96
mip polishafter mipgap 98
GUBCovers 63 Markowitz tolerance 53 mip polishafter nodes 100
maximum infeasibility rule mip polishafter solutions 99
variable selection and 148 mip polishafter time 101
H MaxIter 68 mip pool absgap 130
HeurFreq 64 MaxNodes 88 mip pool capacity 131
heuristic MaxTime 141 mip pool intensity 133
feasibility pump 61 MCFCuts 70 mip pool relgap 132
frequency 64 memory allocation 38, 90 mip pool replace 134
local branching 68 MemoryEmphasis 71 MIP start
relaxation induced neighborhood minimum infeasibility rule writing to file 151
search (RINS) 119 variable selection and 148 MIP strategy
MIP backtracking 32
bound strengthening 31 best bound interval 30
kappa computation 76
I preprocessing 31
branching direction 32
branching variable 148
ImplBd 65 solution file name 66 diving 47
implied bound cut solution file prefix 66 heuristic frequency 64
globally valid 65 writing solutions to file 66 local branching 68
incumbent MIP callback reduced LP parameter 72 node algorithm 86
backtracking and 32 mip cuts disjunctive 46 node file management 87
cutoff tolerance and 32 mip cuts flowcovers 59 node selection 89
diving and 47 mip cuts gomory 62 presolve at nodes 107
local branching heuristic and 68 mip cuts gubcovers 63 priority order 78
relaxation induced neighborhood mip cuts implied 65 probing 108
search (RINS) and 119 mip cuts mcfcut 70 quadratically constrained programs
solution pool absolute gap and 130 mip cuts mircut 81 (MIQCP) 80
solution pool relative gap and 132 mip cuts zerohalfcut 152 RINS 119
target gap and 32 mip emphasis 74 root algorithm 123
integer solution mip interval 75 strong branching and candidate
diving and 47 MIP limit limit 137
integer solution file prefix 66 aggregation for cuts 18 strong branching and iteration
integer solution limit 67 cut by type 49 limit 137
IntSolFilePrefix 66 cuts 42 mip strategy file 87
IntSolLim 67 cutting plane passes 42 mip strategy fpheur 61
iteration deterministic probing time 109 mip strategy kappastats 76
barrier centering corrections and 27 Gomory fractional cut candidates 62 mip strategy miqcpstrat 80
iteration limit nodes explored in subproblem 138 mip strategy nodeselect 89
barrier 27 passes for Gomory fractional cuts 63 mip strategy order 78
network 85 polishing time (deprecated) 102 mip strategy presolvenode 107
perturbation and (simplex) 96 probing time 109 mip strategy probe 108
refactoring of basis (simplex) ramp up time (deterministic mip strategy rinsheur 119
and 116 ticks) 112 mip strategy search 79
sifting 128 ramp up time (seconds) 112 mip strategy startalgorithm 123
simplex 68 repair tries 118 mip strategy subalgorithm 86
strong branching and (MIP) 137 size of tree 142 mip strategy variableselect 148
ItLim 68 solutions 67 mip tolerances absmipgap 50
termination criterion 88 mip tolerances lowercutoff 41
mip limits cutpasses 42 mip tolerances objdifference 91
mip limits eachcutlimit 49 mip tolerances relobjdifference 117
mip limits nodes 88

Index 157
mip tolerances upper cutoff 43 objective difference pricing (continued)
MIP tree absolute 91 network 86
advanced start and 17 relative 117 types available for dual simplex 48
MIPDisplay 73 ObjLLim 92 types available in primal
MIPEmphasis 74 ObjULim 92 simplex 103
MIPInterval 75 opportunistic primal reduction 115
MIPKappaStats 76 definition 93 priority order
MIPOrdInd 78 optimality tolerance (simplex) 53 indicator 78
MIPOrdType 78 output type to generate 78
MIPSearch 79 parallel optimization 36 Probe 108
MIQCPStrat 80 output intsolfileprefix 66 ProbeDetTime 109
MIR cut 81 output mpslong 82 ProbeTime 109
aggregation limit 18 output writelevel 151 probing
MIRCuts 81 deterministic time limit 109
mixed integer programming (MIP) MIP branching and 108
threads 140
mixed integer rounding cut 81
P time limit 109
pseudo cost
parallel optimization
MPS file format variable selection and 148
cloning log files for 36
numerical precision and 82 pseudo reduced cost
parallelism
MPSLongNum 82 variable selection and 148
callbacks and 140
multi-commodity flow cut 70 pseudo-shadow price
optimization mode 93
variable selection and 148
threads and 140
ParallelMode 93
N parameter set 1
NetDisplay 83 path cut 60 Q
NetEpOpt 83 PerInd 95 QPmakePSDInd 110
NetEpRHS 84 periodic heuristic 64 qpmethod 122
NetFind 84 PerLim 96 QPNzReadLim 111
NetItLim 85 perturbation constant (simplex) 54 quadratically constrained mixed integer
NetPPriInd 86 pivot selection 53 program (MIQCP) 80
network display 83 PolishAfterDetTime 96
network iterations 85 PolishAfterEpAGap 97
network netfind 84
network pricing 86
PolishAfterEpGap 98
PolishAfterIntSol 99
R
ramp up
network tolerances feasibility 84 PolishAfterNode 100
time limit (deterministic ticks) 112
network tolerances optimality 83 PolishAfterTime 101
time limit (seconds) 112
network with arc capacity 70 PolishTime (deprecated) 102
RampupDetTimeLimit 112
node PopulateLim 102
RampupDuration 113
best estimate 30 PPriInd 103
RampupTimeLimit 112
presolve and 107 PreDual 104
random seed 115
node file PreInd 105
RandomSeed 115
compression of 87 PreLinear 105
read columns limit 38
node relaxation in MIQCP strategy 80 PrePass 106
read constraints 125
node selection preprocessing
read nonzeros 90
backtracking and 89 bound strengthening 31
read qpnonzeros 111
best bound interval and 30 coefficient reduction 37
read scale 126
NodeAlg 86 dependency checking 44
read variables limit 38
NodeDisplayInterval 75 lazy constraints and 105, 115
Reduce 115
NodeFileInd 87 preprocessing dual 104
ReInv 116
NodeLim 88 preprocessing linear 105
relative gap
NodeSearchStrategy 89 preprocessing numpass 106
solution pool 132
NodeSel 89 preprocessing presolve 105
relative objective difference 117
nonconvex continuous QP solution preprocessing qpmakepsd 110
relaxation induced neighborhood search
type 135 preprocessing reduce 115
(RINS) 119
nonconvex MIQP solution type 135 preprocessing relax 117
RelaxPreInd 117
numerical emphasis 76 preprocessing repeatpresolve 119
RelObjDif 117
numerical precision preprocessing symmetry 139
RepairTries 118
MPS file format 82 preprocessing.aggregator 19
RepeatPresolve 119
NumericalEmphasis 90 Preprocessing.Aggregator 19
RINSHeur 119
NzReadLim 90 Preprocessing.Fill 19
root
PreslvNd 107
threads parameter 124
presolve
RootAlg 120, 122, 123
O advanced start and 17
nodes and 107
RowReadLim 125
ObjDif 91
PriceLim 108
objective
pricing
current and backtracking 32
candidate list limit 108

158 CPLEX Parameters Reference

S solution pool (continued)
intensity 133
tolerance (continued)
optimality (simplex) 53
ScaInd 126 populate limit 102 relative MIP gap 50
screen indicator 126 relative gap 132 relative MIP objective difference 117
set of parameters 1 replacement strategy 134 solution pool, absolute 130
SiftAlg 127 SolutionTarget 135 solution pool, relative 132
SiftDisplay 128 start, advanced 17 upper cutoff 43
sifting strong branching tree
iteration limit 128 candidate list and 137 memory limit (MIP) 142
node algorithm as 86 iteration limit and 137 MIP advanced start 17
root algorithm as 120 variable selection and 148 TreLim 142
sifting algorithm 127 StrongCandLim 137 tune dettimelimit 143
sifting display 128 StrongItLim 137 tune display 144
SiftItLim 128 SubMIPNodeLim 138 tune measure 145
SimDisplay 129 Symmetry 139 tune repeat 146
simplex tune timelimit 147
crash ordering 40 tuning
iterations and candidate list 137
perturbation constant 54 T deterministic time limit 143
measure 145
simplex crash 40 target gap 32
repetition of 146
simplex display 129 termination criterion
reporting level 144
simplex dual gradient 48 barrier complementarity convergence
time limit 147
simplex dual pricing 48 (LP, QP) 25
wall-clock time limit 147
simplex limit barrier complementarity convergence
TuningDetTiLim 143
iterations 68 (QCP) 29
TuningDisplay 144
lower objective function 92 barrier iterations 27
TuningMeasure 145
repairs of singularities 129 FeasOpt Phase I 55
TuningRepeat 146
upper objective function 92 MIP node limit 88
TuningTiLim 147
Simplex limit network iteration limit 85
degenerate iterations 96 simplex iteration limit 68
simplex limits iterations 68 tree size (MIP) 142
simplex limits lowerobj 92 tree size and memory 87 U
simplex limits perturbation 96 threads unbounded optimal face 26
simplex limits singularity 129 callbacks and 140
simplex limits upperobj 92 count at root 124
simplex perturbation 54 parallelism and 140
Threads 140
V
simplex perturbation indicator 95 variable selection
simplex pgradient 103 tick (definition) 45
candidate list and 137
simplex pricing 108 TiLim 141
MIP strategy 148
simplex primal pricing gradient 103 tilting 37
simplex iterations and 137
simplex refactor 116 time
variable, basic
simplex tolerances feasibility 56 as starting condition for solution
feasibility tolerance and 56
simplex tolerances markowitz 53 polishing 101
VarSel 148
simplex tolerances optimality 53 as starting condition for solution
SingLim 129 polishing, deterministic 96
singularity 129 deterministic and solution
SolnPoolAGap 130 polishing 96 W
SolnPoolCapacity 131 time limit 45 workdir 149
SolnPoolGap 132 in seconds 141 WorkDir 149
SolnPoolIntensity 133 tolerance working directory
SolnPoolReplace 134 absolute MIP gap 50 node files and 87
solution absolute MIP objective difference 91 temporary files and 149
writing to file 151 backtracking (MIP) 32 working memory
solution polishing barrier complementarity convergence limit on 150
absolute gap as starting condition (LP, QP) 25 node files and 87
for 97 basic variables and bound workmem 150
deterministic time as starting violation 56 WorkMem 150
condition for 96 complementarity convergence WriteLevel 151
integer solutions as starting condition QCP 29 writing
for 99 cutoff 32 MIP solutions to file (parameter) 66
nodes processed as starting condition cutoff and backtracking 32
for 100 feasibility (network primal) 84
relative gap as starting condition FeasOpt relaxation 55
linearization 52
Z
for 98 zero-half cuts 152
time as starting condition for 101 lower cutoff 41
ZeroHalfCuts 152
solution pool Markowitz 53
absolute gap 130 MIP integrality 51
capacity 131 optimality (network) 83

Index 159
160 CPLEX Parameters Reference



Printed in USA