SRI KRISHNA COLLEGE OF ENGINEERING AND TECHNOLOGY

Kuniamuthur, Coimbatore, Tamilnadu, India

An Autonomous Institution, Affiliated to Anna University,
Accredited by NAAC with “A” Grade & Accredited by NBA (CSE, ECE, IT, MECH, EEE, CIVIL & MCT)

DEPARTMENT OF M.TECH-CSE

Course : Design and Analysis of Algorithms

Module : Fundamentals of Algorithm Analysis –
Empirical Analysis of Algorithm
Faculty :S.SREERAJ/AP-M.TECH/CSE

04/07/2023 1
 EMPIRICAL ANALYSIS OF ALGORITHM

• How algorithms, both non-recursive and

recursive, can be analyzed mathematically.
• Though these techniques can be applied
success-fully to many simple algorithms, the
power of mathematics, even when enhanced
with more advanced.

04/07/2023 2
 CONT…

• In fact, even some seemingly simple algorithms

have proved to be very difficult to analyze with
mathematical precision and certainty.
• The principal alternative to the mathematical
analysis of an algorithm’s efficiency is its
empirical analysis.
• This approach implies steps spelled out in the
following plan.

04/07/2023 3
 GENERAL PLAN FOR THE EMPIRICAL
ANALYSIS OF ALGORITHM TIME EFFICIENCY
• Understand the experiment’s purpose.
• Decide on the efficiency metric M to be
measured and the measurement unit (an
operation count vs. a time unit).
• Decide on characteristics of the input sample
(its range, size, and so on).
• Prepare a program implementing the algorithm
(or algorithms) for the experimentation.
• Generate a sample of inputs.

04/07/2023 4
 CONT…

• Run the algorithm (or algorithms) on the

sample’s inputs and record the data observed.
• Analyze the data obtained.
• Alternatively, you can implement one of several
known algorithms for generating (pseudo)
random numbers.
• The most widely used and thoroughly studied of
such algorithms is the linear congruential
method.

04/07/2023 5
 ALGORITHM

ALGORITHM Random (n, m, seed, a, b)

//Generates a sequence of n pseudorandom
numbers according to the linear congruential
method
//Input: A positive integer n and positive integer
parameters m, seed, a, b
//Output: A sequence r1,...,rn of n pseudorandom
integers uniformly distributed among integer
values between 0 and m − 1

04/07/2023 6
 CONT…

//Note: Pseudorandom numbers between 0 and 1

can be obtained by treating the integers
generated as digits after the decimal point
r0 ← seed
for i ← 1 to n do
 ri ← (a ∗ ri−1 + b) mod m

04/07/2023 7
 TYPICAL SCATTER PLOTS

04/07/2023 8
 CONT…

04/07/2023 9
 IS EMPIRICAL ANALYSIS EASY OR
HARD?
Five aspects to consider in the process of
analysing an algorithm
• correctness,
• work done,
• space used,
• simplicity or clarity,
• optimality.

04/07/2023 10
 DOING EMPIRICAL ANALYSIS

In order to do empirical analysis to verify and

expand on the theoretical analysis of an algorithm
we need to
• understand the theoretical analysis
• decide on what should be measured
• decide on appropriate hardware
• decide on an appropriate implementation
language
• decide on appropriate data structures

04/07/2023 11
• implement the algorithms
• implement some form of timing device
• create the input data sets necessary to produce
the measurements we need
• measure the performance of the algorithm on
the different input data sets created to meet our
aim
• interpret the results
• relate the results to the theoretical analysis

04/07/2023 12
 ISSUES TO CONSIDER

• Empirical analysis introduces many more factors

that need to be controlled for in some way.
• It is much less obvious how to perform a
rigorous analysis in the presence of so many
factors.
• Practical considerations prevent complete
testing.

04/07/2023 13
 MEASURES OF PERFORMANCE

• For the time being, we focus on sequential

algorithms.
• What is an appropriate measure of
performance?
• What is the goal?
• Possible measures

04/07/2023 14
 MEASURING TIME

• User time
• System time
• Wallclock time

04/07/2023 15
REPRESENTATIVE OPERATION COUNTS

• In some cases, we may want to count

operations, rather than time
• What operations should we count?

04/07/2023 16
 TEST SETS

• Example: Sorting

04/07/2023 17
 COMPARING ALGORITHMS

• Given a performance measure and a test set,

the question still arises how to decide which
algorithm is “better.”
• We can do the comparison using some sort of
summary statistic.
• These statistics hide information useful for
comparison.

04/07/2023 18
 ACCOUNTING FOR STOCHASTICITY

• Empirical Analysis
• Wallclock Time
• Parallel Processing
• Multiple Identical Runs
• Statistical Analysis

04/07/2023 19
 PERFORMANCE PROFILES

• Comparison of Algorithms
• Visual Summary

04/07/2023 20
 EXAMPLE PERFORMANCE PROFILE

04/07/2023 21
 EMPIRICAL VERSUS THEORETICAL
ANALYSIS
• Sequential Algorithms
• Tuning of Implementational Details
• Parallel Algorithms
• Model of Computation
• Scalability

04/07/2023 22
04/07/2023 23
04/07/2023 24
04/07/2023 25