4.

Logarithmic Loops

Learning objective:
to introduce algorithms that perform logarithmic number
of iterations

Lecture Plan:
4.1 Exponents and Logarithms
4.2 Typical algorithms
4.3 Example
4.1 Exponents and Logarithms

Starting with definitions (next slide)

 Exponent: 2x = 2  2  …  2

x 2 x
(x times)
1
-2 4
1 10 x log2x
-1 1
2
-2
0 1 4
8
1
1 2 -1
2
2 4 6 1 0
3 8 2 1
4
4 16 3 4 2
2 8 3
1 16 4
-2 1 2 3 4 6 8 10

Logarithm:
Logarithm to the base 2 of a number x is the power
to which 2 must be raised in order to produce x.

log2x is the unique real number y such that 2y = x

 Exponent: 2x = 2  2  …  2

x 2 x
(x times)
1
-2 4
1 10 x log2x
-1 1
2
-2
0 1 4
8
1
1 2 -1
2
2 4 6 1 0
3 8 2 1
4
4 16 3 4 2
y = log2x
2 8 3
1 16 4
-2 0 1 2 3 4 6 8 10
-2
Logarithm:
Logarithm to the base 2 of a number x is the power
to which 2 must be raised in order to produce x.

log2x is the unique real number y such that 2y = x

Summary Table
 Further Reading

Textbooks for Fundamental Mathematical Concepts

(available online)
- S. Lipschutz, M.L. Lipson: Schaum's outline of discrete
mathematics. McGraw-Hill, 2009 (Chapter 3, pp. 49-50)
- K.H. Rosen: Discrete Mathematics and its Applications. ISE
EBook Online Access for Discrete Mathematics and Its
Applications, 8th edition, McGraw-Hill, 2018
(Appendix A.2.2, pp. 1805-1807)
- R. Johnsonbaugh: Discrete Mathematics. Prentice-Hall,
2008 (Appendix B, pp. 636-638)
 4.2 Typical Algorithms

Loops:
- control variable is doubled
or
- control variable is halved
 i  1 i  n
while i < n do while i > 1 do
application code application code
i  i*2 i  i/2

Assume n=2k. This implies k=log2n

Cond. to i at the i at the Cond. to i at the i at the
Iter. continue beginning end of Iter. continue beginning end of
the loop of iteration iteration the loop of iter. iteration
i<n i>1
1 T 1 2 1 T n n/2
Thus the number of iterations is k=log2n.
2 T 2 4 2 T n/2 n/4
3 T 4 8 3 T n/4 n/8
4 T 8 16 4 T n/8 n/16
     
k T 2k-1 2k = n k T n/2k-1 n/2k=1
k+1 F k+1 F

The algorithm stops when condition i<n The algorithm stops when condition i>1
is no longer satisfied. This happens when is no longer satisfied. This happens when
at iteration k variable i reaches n=2k. at iteration k variable i reaches n/2k =1.
 i  1 i  n
while i < n do while i > 1 do
application code application code
i  i*2 i  i/2

Assume 2k-1< n ≤ 2k. This implies k=log2n.

Cond. to i at the i at the Cond. to i at the i at the

Iter. continue beginning end of Iter. continue beginning end of
the loop of iteration iteration the loop of iter. iteration
i<n i>1
1 T 1 2 1 T n n/2
Thus the number of iterations is k=log2n.
2 T 2 4 2 T n/2 n/4
3 T 4 8 3 T n/4 n/8
4 T 8 16 4 T n/8 n/16
     
k T 2k-1 2k ≥ n k T n/2k-1 n/2k≤1
k+1 F k+1 F

The algorithm stops when condition i<n The algorithm stops when condition i>1
is no longer satisfied. This happens after is no longer satisfied. This happens after
iteration k when i reaches 2k, 2k ≥ n. iteration k when i reaches n/2k, n/2k ≤1.
 4.3 Example

Find the number of binary digits in the binary representation

of a positive decimal integer n

n = ak2k + ak-12k-1+ … + a12+ a0 = (𝑎𝑘 … 𝑎1 𝑎0 )2 where ak=1

Given binary representation

(a positive
decimal integer)

Output: 𝑘 + 1
 4.3 Example

Find the number of binary digits in the binary representation

of a positive decimal integer n

(13)10 = 123 + 122+021+ 120 = (1101)2

Given binary representation

Output: 𝑘 + 1 = 4
 n = ak2k + ak-12k-1+ … + a12+ a0, where ak=1

ALGORITHM number-of-binary-digits(n)
//Input: A positive decimal integer n
//Output: The number of binary digits in the
// binary representation of n
count  1
while n > 1 do
count  count+1
n  n/2
return count
n n
Iteration (at the start of count (at the end of
iteration) iteration)
1
Tracing for n=13: 1 13 2 6
2 6 3 3
3 3 4 1
4 1
 n = ak2k + ak-12k-1+ … + a12+ a0, where ak=1

ALGORITHM number-of-binary-digits(n)
//Input: A positive decimal integer n
//Output: The number of binary digits in the
// binary representation of n
count  1
while n > 1 do
count  count+1
n  n/2
/
return count
As we have seen, if the control variable is halved, then the number of iterations is
k=log2n.
In the current example, the control variable is halved and rounded down: n  n/2 .
This implies that the number of iterations is bounded by k=log2n.

(a) What is the algorithm’s main (most frequent) operation? division

(b) What is an upper bound on the number of times it is repeated? log2n
(c) Determine the class O(?) the algorithm belongs to. O(log2n)
 n = ak2k + ak-12k-1+ … + a12+ a0, where ak=1

ALGORITHM number-of-binary-digits(n)
//Input: A positive decimal integer n
//Output: The number of binary digits in the
// binary representation of n
count  1
while n > 1 do
count  count+1
+
n  n/2
/
return count
As we have seen, if the control variable is halved, then the number of iterations is
k=log2n.
In the current example, the control variable is halved and rounded down: n  n/2 .
This implies that the number of iterations is bounded by k=log2n.

(a) What are the algorithm’s basic operations? division, summation

(b) What is an upper bound on the number of times they are repeated? 2log2n
(c) Determine the class O(?) the algorithm belongs to. O(log2n)
 n = ak2k + ak-12k-1+ … + a12+ a0, where ak=1

ALGORITHM number-of-binary-digits(n)
count  1
while n > 1 do
count  count+1
n  n/2
return count

Algorithm correctness: The algorithm counts how many times operation

n/2 is performed before reaching n=1.
With the initial value count=1, it returns the number of divisions plus 1.
Consider n = ak2k + ak-12k-1+ … + a222 + a12+ a0
The first division results in n = ak2k-1 + ak-12k-2+ … + a22 + a1
The second division results in n = ak2k-2 + ak-12k-3+ … + a2
Thus after k divisions n=ak=1 (the first digit of the binary representation),
and count=k+1 (the number of binary digits).
 Conclusions

Make sure you understand

- the definition of logarithm
- the analysis of the two basic pseudocodes for n=2k
(the material for 2k-1< n ≤ 2k is not compulsory)
- the last example and its analysis

Further examples will be provided in tutorials and in lectures

on searching and sorting