Mathematical Methods for Economics

Mathematical and Statistical Methods

Metric Spaces
The Euclidean distance between two points x and y in R is defined as |x − y |, i.e.,
as the absolute value of their difference.
For any positive integer n ∈ N, the n-fold Cartesian product of R will be denoted
Rn . We will refer to Rn as n-dimensional Euclidean space.
A point in Rn is a vector x = (x1 , . . . , xn ) where for each i = 1, . . . , n, xi is a real
number. The number xi is called the i-th coordinate of the vector x.
Vector Addition: x + y = (x1 + y1 , . . . , xn + yn )
Scalar Multiplication: αx = (αx1 , . . . , αxn ).
Given any two n-vectors x = (x1 , . . . , xn ) and y = (y1 , . . . , yn ), we write

x = y , if xi = yi , i = 1, . . . , n.
x ≥ y , if xi ≥ yi , i = 1, . . . , n.
x > y , if x ≥ y and x ̸= y .
x ≫ y , if xi > yi , i = 1, . . . , n.

Mathematical and Statistical Methods

Metric Spaces
The nonnegative and strictly positive orthants of Rn , denoted Rn+ and Rn++ ,
respectively, are defined as

Rn+ = {x ∈ Rn | x ≥ 0}

Rn++ = {x ∈ Rn | x ≫ 0}
Given x, y ∈ Rn , the Euclidean inner product of the vectors x and y , denoted x · y ,
is defined as:
n
X
x ·y = xi yi
i=1

The inner product has the following properties for any vectors x, y , z ∈ Rn and
scalars a, b ∈ R :
1. Symmetry: x · y = y · x.
2. Bilinearity: (ax + by ) · z = ax · z + by · z and
x · (ay + bz) = x · ay + x · bz.
3. Positivity: x · x ≥ 0, with equality holding if and only if x = 0.
Mathematical and Statistical Methods
Metric Spaces

Cauchy-Schwartz Inequality: For any x, y ∈ Rn we have

|x · y | ≤ (x · x)1/2 (y · y )1/2 .
The Euclidean norm (henceforth, simply the norm) of a vector x ∈ Rn , denoted
∥x∥, is defined as

n
!1/2
X
∥x∥ = xi2
i=1

The norm is related to the inner product through the identity

∥x∥ = (x · x)1/2
for all x ∈ Rn ; in particular, the Cauchy-Schwartz inequality may be written as

|x · y | ≤ ∥x∥∥y ∥.

Mathematical and Statistical Methods

Metric Spaces

The norm satisfies the following properties at all x, y ∈ Rn , and a ∈ R :

1. Positivity: ∥x∥ ≥ 0, with equality if and only if x = 0.
2. Homogeneity: ∥ax∥ = |a| · ∥x∥.
3. Triangle Inequality: ∥x + y ∥ ≤ ∥x∥ + ∥y ∥.
The distance function d is called a metric, and is related to the norm ∥ · ∥ through
the identity

d(x, y ) = ∥x − y ∥
n
for all x, y ∈ R .
The metric d satisfies the following properties at all x, y , z ∈ Rn :
1. Positivity: d(x, y ) ≥ 0 with equality if and only if x = y .
2. Symmetry: d(x, y ) = d(y , x).
3. Triangle Inequality: d(x, z) ≤ d(x, y ) + d(y , z) for all x, y , z ∈ Rn .

Mathematical and Statistical Methods

Metric Spaces

Neighborhood: An ϵ-neighborhood of a point x0 ∈ R is given by the set

Nϵ (x0 ) = {x ∈ R : d (x0 , x) < ϵ}. Simply, Nϵ (x0 ) is the set of points lying within a
distance ϵ of x0 .
A point x is a limit point of the set A if every neighborhood of x contains a point
y ̸= x such that y ∈ A.
If x ∈ A and x is not a limit point, then x is called an isolated point of A.
If x ∈ A and there exist atleast one neighborhood of x such that Nϵ x ⊂ A, then x
is said to be an interior point of A.

Mathematical and Statistical Methods

Metric Spaces

A is called a closed set if it contains all its limit points. Notice a set need not
include it’s limit points. (0, 1) - here 0 and 1 are limit points (verify) but they are
not in the set.. So (0, 1) is not closed..
All points of a set need not be limit points(0, 1) ∪ {2} − 2 is in the set but not a
limit point..
[0, 1] ∪ {2} - closed or not?
Limit point vs boundary point..
Boundary Point: A point x is called the boundary point of the set S if every Bϵ (x)
contains some point from S and S c .

Mathematical and Statistical Methods

Metric Spaces

Set A is an open set if every point of A is an interior point of A.

Alternate definition: A subset A of R is open in R if for each x ∈ A there exists a
neighborhood Nϵ (x) such that Nϵ (x) ⊆ A.
Another definition: A subset A of R is closed in R if the complement
C (A) := R\A is open in R.
To show that a set A is open, we have to find (one) ϵX > 0 for every x ∈ A such
that (x − ϵX , x + ϵX ) is contained in A.
Any union of open sets is open.
The finite intersection of open sets is open.
Any intersection of closed sets is closed.
The finite union of closed sets is closed.

Mathematical and Statistical Methods

Metric Spaces

G := {x ∈ R : 0 < x < 1}.

I := (a, b). Open.
The set I = [0, 1].
It is a closed set.
The set H = {x : 0 ≤ x < 1}.
It is neither open nor closed.
The empty set ∅, and the whole space or, universal set.
Consumption set −→ budget set- closed?

Mathematical and Statistical Methods

Metric Spaces

Bounded: A is bounded if there is a real number M and a point q ∈ A such that

d(p, q) < M for all p ∈ A.
A set X ⊂ Rn is bounded if, for every x0 ∈ X , there exists an ϵ > 0 such that
X ⊂ Nϵ (x0 ).
Easy or practicle definition: A set S ∈ Rn is compact if and only if it is both closed
and bounded.
Closedness and boundedness are not same!
Why are open sets not compact?
By an open cover of a set A in a metric space X we mean a collection {Gα } of
open subsets of X such that A ⊆ ∪α Gα .
A subset A of a metric space X is said to be compact if every open cover of A
contains a finite subcover.

Mathematical and Statistical Methods

Metric Spaces

Budget set.. consumer optimisation.

Extreme-Value Theorem- Let F be a continuous real-valued function on the closed
and bounded interval [a, b]. Then there exist points xM and xm in [a, b] such that
for all x ∈ [a, b], f (xm ) ≤ f (x) ≤ f (xM ).
Intermediate-Value Theorem- Let f a continuous real-valued function on the
closed and bounded interval [a, b]. Then for each number α (strictly) between
f (a) and f (b), there exists a point c ∈ (a, b) such that f (c) = α.
A set X is called convex if for all x&y ∈ X , we have tx + (1 − t)y ∈ X for all
t ∈ [0, 1].
If X1 and X2 are two convex set, then so is X1 ∩ X2 .

Mathematical and Statistical Methods

Real Analysis

Sequence: A sequence of real numbers is an assignment of a real number to each

natural number.
The notation {xn } means the sequence whose n-th element is xn .
Example

{1, 2, 3, 4, . . .},
 1 1
1, 2 , 3 , . . . ,
 1
1, 2 , 4, 81 , 16, . . . ,
1 − n1


Mathematical and Statistical Methods

Real Analysis

A sequence {xn } ∈ R is said to converge to x ∈ R, or x is said to be a limit of

{xn }, if for every ϵ > 0, there exists a natural number N(ϵ) such that for all
n > N(ϵ), the term xn satisfy |xn − x| < ϵ.
If a sequence has a limit, we say that the sequence is convergent; if it has no limit,
we say that the sequence is divergent.
A sequence in R can have at most one limit.

xn → x & yn → y ⇒ xn + yn → x + y .
xn → x & yn → y ⇒ xn yn → xy
xn ≤ b ∀n and xn → x, then x ≤ b
xn ≥ b ∀n and xn → x, then x ≥ b
(notice the weak inequality here.)

Mathematical and Statistical Methods

Real Analysis

Standard functions in economics- utility function, production function, cost

function, profit function..
Level sets- visualising a function in three dimension is difficult, level sets help us to
analyse it in two dimension.
A level set of the function y = f (x1 , x2 , · · · , xn ) is the set

L {(x1 , x2 , · · · , xn ) ∈ Rn : f (x1 , x2 , · · · , xn ) = c}
for some given number c ∈ R.
Better/Superior/Upper Contour set..
Worse/Inferior/Lower Contour set..
Indifference curves, isoquants..

Mathematical and Statistical Methods

Real Analysis

A function is monotonic in a given interval if it is either increasing or decreasing in

it.

x > y =⇒ f (x) ≥ f (y )
x < y =⇒ f (x) ≤ f (y )
Generally when we mention monotonic function in econ, we refer to positively
monotonic function.

Mathematical and Statistical Methods

Real Analysis

Let X be a convex set. A function f : X → R is concave for all x1 , x2 ∈ X

f (tx1 + (1 − t)x2 ) ≥ tf (x1 ) + (1 − t)f (x2 ) ∀t ∈ [0, 1].

Examples- f (x) = x α , α < 1

Weakly concave- f (x) = x.

Mathematical and Statistical Methods

Real Analysis

Let X be a convex set. A function f : X → R is convex for all x1 , x2 ∈ X

f (tx1 + (1 − t)x2 ) ≤ tf (x1 ) + (1 − t)f (x2 ) ∀t ∈ [0, 1].

α
Examples- f (x) = x , α > 1 Weakly convex- f (x) = x.
A function f is concave if and only if −f is convex.
Convexity of function is different from convexity of set!
Concave functions are quite common in economics- utility function, production
function? expenditure function..
Cost functions- convex, why?

Mathematical and Statistical Methods

Real Analysis

Properties of concave functions are quite desirable for many functions in

economics.
However the concave functions has one undesirable properties:- concavity is a
cardinal property- a monotonic transformation of a concave function need not be a
concave function!
Let f : Rn ⊇ X → R be a concave function and g : R → R an increasing and
concave function defined on an interval I containing f (x). Then the function
g [f (x)] is concave.
The concave function do have one ordinal property and this property is highly
desirable in economics- better or superior sets are convex! diminishing marginal
rate of substitution

Mathematical and Statistical Methods

Real Analysis

Let I be an interval of the real line. Then g : I → R is a (positive) monotonic

transformation of the interval / if g is a strictly increasing function on I .
Furthermore, if g is a (positive) monotonic transformation and u is a real-valued
function, then

g (u) : x → g [u(x)]
is a monotonic transformation of u.

Mathematical and Statistical Methods

Real Analysis

Preserves the ordering. ordinal Property

A property of a function is called ordinal if every monotonic transformation of a
function preserve that property.
Example- Marginal rate of substitution (MRS)- u (x1 , x2 )

∂u
∂x1
MRS = ∂u
∂x2

Elasticity of a function-

% change in f (x)
% change in x

Mathematical and Statistical Methods

Real Analysis

A function f with domain X ⊆ Rn is quasiconcave if, for every point in X , the

better set B of that point is a convex set. It is strictly quasiconcave if B is strictly
convex.
A function f : Rn → R is quasiconcave if, for all x, x ′ ∈ Rn and all λ ∈ [0, 1],

f λx + (1 − λ)x ′ ≥ min f (x), f x ′

  

Mathematical and Statistical Methods

Real Analysis

Quasiconvex Function
A function f with domain X ⊆ Rn is quasiconvex if, for every point in X , the
worse/inferior set W of that point is a convex set. It is strictly quasiconvex if W is
strictly convex.
A function f : Rn → R is quasiconvex if, for all x, x ′ ∈ Rn and all λ ∈ [0, 1],

f λx + (1 − λ)x ′ ≤ max f (x), f x ′

  

Mathematical and Statistical Methods

Real Analysis

Quasiconcave Function - Proposition: A concave function is quasiconcave. A

convex function is quasiconvex.
Any Cobb-Douglas function f (x, y ) = kx a y b is going to be quasiconcave with
k, a, b > 0.
Leontief function f (x, y ) = min{ax, by } is also quasiconcave.
Normal distribution- also quasiconcave.
Quasiconvity and quasiconvexity are ordinal properties.
Quasiconcavity is natural for many economic conceptsextremes are preferred to
average, and others..
Quasiconcavity also often serves as a sufficient condition in optimisation- to be
covered later during optimisation discussions.

Mathematical and Statistical Methods

Homogeneous and Homothetic Functions

Homogeneous Function: A real-valued function f (x1 , x2 , ..., xn ) is

homogeneous of degree k if

f (tx1 , tx2 , ..., txn ) = t k f (x1 , x2 , ..., xn ) for all t > 0

Examples: find the degree of homogeneity of the following functions.

f (x) = x α , α > 0
f (x, y ) = xy
f (x, y ) = x + y
f (x, y ) = min x, y
f (x, y ) = x 2 + y − 10
f (x, y ) = (x − 10)(x − 25)

Mathematical and Statistical Methods

Homogeneous and Homothetic Functions

Return to scale in production and homogeneity of the production

function.
If f(x) is homogeneous of degree k, f ′ (x) is going to be homogeneous
of degree k − 1.
One implication - suppose the utility function u(x, y ) is homogeneous
of degree k. (are typical utility functions used in economics
homogeneous?) Then optimal (Marshallian) demand functions take
the form:
x ∗ = k1 M and y ∗ = k2 M
For fixed prices, as income goes up, the individual spends same fixed
proportion of income on any particular good!

Mathematical and Statistical Methods

Homogeneous and Homothetic Functions

Homogeneous functions are very widely used - nice properties and

easy for derivations
However, one issue with homogeneity is its ordinal property (like
concavity)
A monotonic transformation of a homogeneous function need not be
homogeneous:

u(x, y ) = xy homogeneous of degree 2

Consider a monotonic transformation,

g (z) = z + 1, g ′ (z) > 0

Now g (u(x, y )) = xy + 1 is not a homogeneous function

Mathematical and Statistical Methods

Homogeneous and Homothetic Functions

Homothetic Function: A function is homothetic if it is a monotonic

transformation of a homogeneous function.
For a homothetic function,

(x) = f (y ) ⇒ f (tx) = f (ty ) for all t > 0

Mathematical and Statistical Methods

Implicit Function

Explicit function- y = f (x1 , x2 , · · · , xn ).

Implicit function- g (y , x1 , x2 , · · · , xn ) = k.
Example: y 5 − xy + 4x 2 = 0, difficult to have a explicit y (x) form.
It seems well defined:

y = 0 for x = 0, y = 1 for x = 1

We are interested in the behaviour of the function.

Many-a-times utility maximisation gives g (p, q, m) = 0, difficult to
separate.
We want to know how the quantity demanded changes with a change
in price and income- ∂q ∂q
∂p and ∂m .

Mathematical and Statistical Methods

Implicit Function

∂x ∂x
Similarly, in the case of firm’s behaviour, we interested in ∂p and ∂w .
. here x in input demand, p is output price, and w is input price.
Before proceeding with the derivative, we need to make sure that the
function is well-delined in the relevant neighborhood.. cont. and
differentiable.
Example: x 2 + y 2 = 1
x > 1, y =?
Consider the function g (x, y ) = k.
Suppose this function is well-defined at (x0 , y0 ).

g (x0 , y0 ) = k

Mathematical and Statistical Methods

Implicit Function

If the implicit function exist, we have

g (x, y (x)) = k
differentiating it w.r.t x at x0 , we have

∂g dx ∂g dy
(x0 , y (x0 )) + (x0 , y (x0 )) (x0 ) = 0
∂x dx ∂y dx
or,
∂g
dy ∂x (x0 , y (x0 ))
(x0 ) = − ∂g
dx
∂y (x0 , y (x0 ))

Mathematical and Statistical Methods

Implicit Function

dy
The notation y ′ (x0 ) is also used for dx (x0 ).
Implicit Function Theorem: Let G (x, y ) be a C ′ function on a ball
about (x0 , y0 ) in R2 . Suppose that G (x0 , y0 ) = c and consider the
expression

G (x, y ) = c.
If (∂G /∂y ) (x0 , y0 ) ̸= 0, then there exists a C 1 function y = y (x)
defined on an interval I about the point x0 such that:
(a) G (x, y (x)) = c for all x in I ,
(b) y (x0 ) = y6 , and
∂G
(x ,y )
(c) y ′ (x0 ) = − ∂G
∂x 0 0
(x ,y )
.
∂y 0 0

Mathematical and Statistical Methods

Implicit Function
For n variables, Implicit Function Theorem: Let G (x1 , . . . , xk , y )
be a C 1 function around the point (x1∗ , . . . , xk∗ , y ∗ ). Suppose further
that (x1∗ , . . . , xk∗ , y ∗ ) satisfies

G (x1∗ , . . . , xk∗ , y ∗ ) = c
∂G ∗
and that (x , . . . , xk∗ , y ∗ ) ̸= 0
∂y 1
Then, there is a C ′ function y = y (x1 , . . . , xn ) defined on an open
ball B about (x1∗ , . . . , xk∗ ) so that:
(a) G (x1 , . . . , xk , y (x1 , . . . , xk )) = c for all (x1 , . . . , xk ) ∈ B,
(b) y ∗ = y (x1∗ , . . . , xk∗ ), and
(c) for each index i,
∂G ∗ ∗ ∗
∂y ∗ ∂xi (x1 , . . . , xk , y )
(x1 , . . . , xk∗ ) = − ∂G ∗ ∗ ∗
∂xi ∂y (x1 , . . . , xk , y )

Mathematical and Statistical Methods

Continuous Functions

Let f : S → T , where S ⊂ Rn and T ⊂ Rl . Then, f is said to be

continuous at x ∈ S if for all ϵ > 0, there is δ > 0 such that y ∈ S
and d(x, y ) < δ implies d(f (x), f (y )) < ϵ.
d(x, y ) is the distance between x and y in Rn , while d(f (x), f (y )) is
the distance in Rl .
f is continuous at x if the value of f at any point y that is ”close” to
x is a good approximation of the value of f at x.
Example: identity function f (x) = x, ∀x ∈ R
f : R → R given by, (
0, x ≤ 0
f (x) =
1, x > 0
f(x) is continuous everywhere except at x = 0

Mathematical and Statistical Methods

Continuous Functions

A function f : S → T is said to be continuous on S if it is cont

inuous at each point in S.
Theorem: A function f : S ⊂ Rn → Rl is continuous at a point
x ∈ S if and only if for all open sets V ⊂ Rl such that f (x) ∈ V ,
there is an open set U ⊂ Rn such that x ∈ U, and f (z) ∈ V for all
z ∈ U ∩ S.
A function f : S ⊂ Rn → Rl is continuous on S if and only if for each
open set V ⊂ Rl , there is an open set U ⊂ Rn such that
f −1 (V ) = U ∩ S, where f −1 (V ) is defined by

f −1 (V ) = {x ∈ S | f (x) ∈ V }.
In particular, if S is an open set in Rn , f is continuous on S if and
only if f −1 (V ) is an open set in Rn for each open set V in Rl .

Mathematical and Statistical Methods

Continuous Functions

The continuity of a function f at a point x is a local property

It relates to the behaviour of f near x, but tells nothing about the
behaviour of f elesewhere.

Mathematical and Statistical Methods

Differentiable and continuously differentiable functions

Let S be an open set in R⋉

A function f : S → Rm is said to be differentiable at a point x ∈ S if
there exists an m × n matrix A such that for all ϵ > 0, there is δ > 0
such that y ∈ S and ∥x − y ∥ < δ implies

∥f (x) − f (y ) − A(x − y )∥ < ϵ∥x − y ∥.

Equivalently, f is differentiable at x ∈ S if
 
∥f (y ) − f (x) − A(y − x)∥
lim = 0.
y →x ∥y − x∥
Matrix A is called the derivative of f at x and is denoted as Df (x)
Df can be denoted by f ′ (x) if n = m = 1, alternatively, S ⊂ R

Mathematical and Statistical Methods

Differentiable and continuously differentiable functions

Df (x) is denoted by f ′ (x) whenever n = m = 1, i.e.. whenever S ⊂ R

and f : S → R.
If f is differentiable at all points in S, then f is said to be
differentiable on S.
When f is differentiable on S, Df itself forms a function from
S to Rmxn
If Df : S → Rmxn is ac continuous function, then f is said to be
continuously differentiable on S, and is written as C 1

Mathematical and Statistical Methods

Differentiable and continuously differentiable functions

Theorem: If f : Rn → Rm and g : Rn → Rm are both differentiable at

a point x ∈ Rn , so is (f + g ) and, in fact,

D(f + g )(x) = Df (x) + Dg (x)

Theorem: Let f : Rn → Rm and h : Rk → Rn . Let x ∈ Rk . If h is
differentiable at x, and f is differentiable at h(x), then f ◦ h is itself
differentiable at x, and its derivative may be obtained through the
”chain rule” as:

D(f ◦ h)(x) = Df (h(x))Dh(x)

Mathematical and Statistical Methods

Partial Derivatives and Differentiability

Let f : S → R, where S ⊂ Rn is an open set.

Let ej denote the vector in Rn that has a 1 in the j-th place and zeros
elsewhere (j = 1, . . . , n).
The j-th partial derivative of f is said to exist at a point x if there is
a number ∂f (x)/∂xj such that
 
f (x + tej ) − f (x) ∂f
lim = (x).
t→0 t ∂xj

Mathematical and Statistical Methods

Partial Derivatives and Differentiability

Theorem: Let f : S → R, where S ⊂ Rn is open.

1. If f is differentiable at x, then all partials ∂f (x)/∂xj exist at ẋ , and
Df (x) = [∂f (x)/∂x1 , . . . , ∂f (x)/∂xn ].
2. If all the partials exist and are continuous at x, then Df (x) exists, and
Df (x) = [∂f (x)/∂x1 , . . . , ∂f (x)/∂xn ].
3. f is C 1 on S if and only if all partial derivatives of f exist and are
continuous on S.

Mathematical and Statistical Methods

Higher Order Derivatives

Let f be a function from S ⊂ Rn to R, where S is an open set.

Assume f is differentiable on all of S.
Thus, the derivative Df = [∂f /∂x1 , . . . , ∂f /∂xn ] itself defines a function from S to
Rn .
Suppose there is x ∈ S such that the derivative Df is itself differentiable at x, i.e.,
such that for each i, the function ∂f /∂xi : S → R is differentiable at x.
Denote the partial of ∂f /∂xi in the direction ej at x by ∂ 2 f (x)/∂xj ∂xi , if i ̸= j,
and ∂ 2 f (x)/oxi2 , if i = j.
Then, f is twice-differentiable at x, with second derivative D 2 f (x), where
 2
∂ 2 f (x)

∂ f (x)
2 · · · ∂x1 ∂xn
 ∂x1 
D 2 f (x) = 
 .. .. .. 
 2. . . 

∂ f (x) ∂ 2 f (x)
∂xn ∂x1
··· ∂x 2 n

2 ′′
D f (x) is denoted as f (x) whenever n = 1 (i.e., if S ⊂ R )

Mathematical and Statistical Methods

Higher Order Derivatives

Theorem: If f : D → Rn is a C 2 function, D 2 f is a symmetric

matrix, i.e., we have

∂2f ∂2f
(x) = (x)
∂xi ∂xj ∂xj ∂xi
for all i, j = 1, . . . , n, and for all x ∈ D.

Mathematical and Statistical Methods

Quadratic Forms and Definiteness

A quadratic form on Rn is a function gA on Rn of the form

n
X
′
gA (x) = x Ax = aij xi xj
i,j=1

where A = (aij ) is any symmetric n × n matrix.

A quadratic form A is said to be

positive definite if we have x ′ Ax > 0 for all x ∈ Rn , x ̸= 0.

positive semidefinite if we have x ′ Ax ≥ 0 for all x ∈ Rn , x ̸= 0.
negative definite if we have x ′ Ax < 0 for all x ∈ Rn , x ̸= 0.
negative semidefinite if we have x ′ Ax ≤ 0 for all x ∈ Rn , x ̸= 0.

Mathematical and Statistical Methods

Quadratic Forms and Definiteness

An n × n symmetric matrix A is
negative definite if and only if (−1)k |Ak | > 0 for all k ∈ {1, . . . , n}.
positive definite if and only if |Ak | > 0 for all k ∈ {1, . . . , n}.

Moreover, a positive semidefinite quadratic form A is positive definite

if and only if |A| =
̸ 0, while a negative semidefinite quadratic form is
negative definite if and only if |A| =
̸ 0.

Mathematical and Statistical Methods

Quadratic Forms and Definiteness

A symmetric n × n matrix A is
positive semidefinite if and only if |Aπk | ≥ 0 for all k ∈ {1, . . . , n} and
for all π ∈ Π.
negative semidefinite if and only if (−1)k |Aπk | ≥ 0 for all
k ∈ {1, . . . , n} and for all π ∈ Π.

Mathematical and Statistical Methods

Quadratic Forms and Definiteness

Let f : D → R be a C 2 function, where D ⊂ Rn is open and convex.

Then,
f is concave on D if and only if D 2 f (x) is a negative semidefinite
matrix for all x ∈ D.
f is convex on D if and only if D 2 f (x) is a positive semidefinite
matrix for all x ∈ D.
If D 2 f (x) is negative definite for all x ∈ D, then f is strictly concave
on D.
If D 2 f (x) is positive definite for all x ∈ D, then f is strictly convex on
D.

Mathematical and Statistical Methods