Lecture # 02 - New

Design and Analysis of Algorithm
Tanveer Ahmed Siddiqui
Department of Computer Science

COMSATS University, Islamabad
Recap
Lecture No 1
Department of Computer Science 2

Recap
Algorithm It is sequence of unambiguous instructions for solving a

(Definition) particular problem in a finite amount of time
Properties Input, Output, Precision, Determinism, Finiteness,

Definiteness, Correctness, Generality
Building Blocks Sequence, Decision Structure, Repetition Structure
Representation Structured Description, Flow Chart, Pseudo Code
Important Point There may exist several algorithms for solving the same
problem.

YOUR TURN

TASK 1
 Design an algorithm to calculate which known

locations are closest to current known
location.

Lecture No 2
Framework for designing and

analyzing algorithms for new problems
(Fundamentals of Algorithmic Problem Solving)

Today Covered
After completing this lecture, you will be able to

understand:
 Framework for designing and analyzing
algorithms for new problems
 A sequence of steps one typically goes through in
designing an algorithm
 Problem-solving strategy with a computer
 Fundamental design concepts and principles
 Design algorithms for solving simple
problems.

Two main issues related to algorithms
 Solving problem by a computer is challenging and

exciting with possibilities for creative satisfaction(why)
 when we are solving a problem with a computer, we are
essentially teaching a dumb servant how to do a job?
 Solving a problem (or getting job done) by a computer
can be quite a demanding task(why)
 Because it requires:
 Considerable thoughts
 Careful planning, and
 Attention to details

TASK 2
 Outline a sequence of steps one typically goes

through in designing an algorithm(You may give your
description in English)

Steps of development of an algorithm
 Understand the problem /Writing

the statement of the problem
 Develop a model/choose
appropriate data structure
 Select appropriate designing
technique.
 Specify the algorithm
 Prove correctness of the algorithm
 Analyze the complexity of the
algorithm
 Implement the algorithm/code

Designing an algorithm is easy if you
Understand
the
problem

Understanding the problem
“problem definition” phase.

 Decide what must be done(understand clearly the requirements
in a particular problem).
 From the problem statement which may be vague or imprecise, we have to
find out a set of precisely define tasks(how).
PROBLEM:
Some friends of yours are working on techniques for coordinating
groups of mobile robots. Each robot has a radio transmitter that it
uses to communicate with a base station, and your friends find that
if the robots get too close to one another, then there are problems
with interference among the transmitters. So, a natural problem
arises how to plan the motion of the robots in such a way that each
robot gets to its intended destination, but in the process the robots
don’t come close enough together to cause interference problems.

 From the problem statement which may be vague or

imprecise, we have to find out a set of precisely
define tasks(how).
 Ask originator of the problem various questions and answer
them, until the problem is clear.
 Do a few small examples by hand, think about special cases,
etc.
 Some relevant questions that can be asked are:
 Do we understand the vocabulary used in
the raw statement of the problem?
 The problem may be in a particular field having its own
vocabulary or technical words. Do we understand them clearly,
at least clear enough to develop the algorithm?
 E.g., what does interference- free schedule means in above
Department ofproblem?
Computer Science
 What information is given?

 Precisely what are the given data? In other words, what is the
input to algorithm?
 What are the given conditions?
 What is to be found out?
 What should be the output from the algorithm and the
corresponding program?
 Trivial as it may sound answering this question is not that easy
when it comes to quite a few real-life problems.
 How to recognize a valid solution?
 We should understand how precisely a correct solution is
defined; so that we can include in the algorithm means to
detect them and reject invalid or unacceptable solutions.

 What information required to define the

algorithm, is missing?
 The raw problem statement may not have given all the
necessary information, or some information may be vague or
conflicting.

 What assumptions have been made?

 Most of the times, assumptions are unstated, even the originator
of problem may not be aware of them.
 They are to identified and taken care of while developing the
algorithm.
 For example, “write a function which, given an integer
number input, outputs its factorial”.
 Here, the fact that the integer is non-negative is unspecified, but
assumed.
 Also, the assumption that function is supposed to work only up to
some relatively small input (like 20 or so) is unstated.
 Many more questions are possible depending upon
the field and nature of the problem.

Develop a
Model

Developing a model
 A model is an abstraction of reality.
 It will help people to represent a reality in an
understandable form.
 We have to develop a model where calculations can
be done.
 A model may be in the form of a
 Graph
 Matrix
 Equation(s)
 sequence of logic
 Mathematical problem with an objective function and
a set of constraints etc.
There are no specific rules for this step, mainly experience
countsDepartment
and ofthis is where the art of programming comes in.
Computer Science
Develop a Model
PROBLEM:
Some friends of yours are working on techniques for
coordinating groups of mobile robots. Each robot has a
radio transmitter that it uses to communicate with a base
station, and your friends find that if the robots get too
close to one another, then there are problems with
interference among the transmitters. So a natural
problem arises: how to plan the motion of the robots in
such a way that each robot gets to its intended
destination, but in the process the robots don’t come
close enough together to cause interference problems.
How can we model this problem?
We can model this problem with an undirected graph.


know
Design
Techniques

Design Techniques
 An algorithm design techniques (or “strategy” or

“paradigm”) is a general approach for solving
problems algorithmically that is applicable to a
variety of problems from different area of
computing.
 The design approach depends mainly on the
model chosen.

Algorithm design techniques/strategies

steps of development of an algorithm

know
Methods of
Specifying an
Algorithm

Methods of Specifying an Algorithm
 Commonly used methods as discussed in first

lecture are:
 Flow chart
 Natural language
 Pseudo code
 A mixture of a natural language and programming
language-like structures
 Precise and succinct.
 Pseudo-code in this course
 omits declarations of variables
 use indentation to show the scope of such statements as
for, if, and while.
 use  for assignment

Pseudo code Details
 Control flow
if … then … [else …]

 while … do …  Expressions
 repeat … until …
 Assignment
 for … do … (like  in Java)
 Indentation replaces braces  Equality testing
 Method declaration (like  in Java)
Algorithm method (arg [, arg…]) n 2
Superscripts and other
mathematical
Input …
formatting allowed
Output …

Prove the
Correctness of
the Algorithm

Checking the correctness of the algorithm
 Once an algorithm has been specified, you
have to prove its correctness which is one of
the difficult, but essential task in algorithm
development.
 The correctness of algorithm means the
correctness of the logic/steps of the
algorithm.
 How can be sure that algorithm always
produces the correct output?

 One possible method: tracing the algorithm’s

performance for few specific inputs can be very
worthwhile activity.(Dry run)
 Make a Trace Table: usually takes the form of a multi-
column, multi-row table; With each column showing a
variable, and each row showing each number input into
the algorithm and the subsequent values of the
variables.
 This approach cannot prove the algorithm’s
correctness conclusively(why)
 It is impossible to test the algorithm for all set of inputs
values.
 It is not sufficient for programs for applications where
reliability
Department is of paramount importance.
of Computer Science
 Other way: ?
 rely on some mathematical proof to ensure that the
algorithm will always yield the correct output.
 Correctness of algorithm can be confirmed by the
uses of some:
 Inference Rules
 Proving Techniques.

 A common technique for proving correctness of
an algorithm is to use the
 Mathematical Induction(Why)
 because an algorithm’s iteration provides a natural
sequence of steps needed for such proofs.
 Try to justify each step. This may involve some condition
which existed before and after the step is executed.
 Try to provide some proof that algorithm will terminate
and while doing so, it would have read all appropriate
input data and would have generated all the expected
output.

Analyzing the
Complexity of
the algorithm

Analyzing the complexity of the algorithm
 After correctness, by far the most important is

efficiency.
 There are two kinds of efficiency:
 Time efficiency: indicates how fast the algorithm runs
 Space efficiency. indicates how much extra memory
the algorithm needs.

Examples

Example-1
 Problem: Design an algorithm that check whether all
the elements in a given array of n elements are
distinct.
 Solution:
 What is input
 An array A of n numbers e.g. 4 1 8 9 3 7 10 2 3 1
 What should be the output

 Returns “true” if all the elements in A are distinct
 Which designing technique?
 Brute-Force
 Logic/Idea
Take the first element and check whether it is in the rest of the array,
take the second element and check if it is in the rest of the array, and
so on…
Example-1
 Problem: Design an algorithm that check whether all
the elements in a given array of n elements are
distinct.
 Logic/Idea
Take the first element and check whether it is in the rest of the array,
take the second element and check if it is in the rest of the array, and
so on…

Example-1
 Problem: Design an algorithm that check
whether all the elements in a given array of n
elements are distinct.
Representation

Example-2
 Problem: Design an algorithm that find the minimum

difference between any pair in the given array.
 Solution:
 Understand the problem
 An array of numbers.

 Given a list of distinct unsorted integers.

 The smallest absolute difference between any pair in
the given array

Example-2

 Solution:
 Input: {1, 5, 3, 19, 18, 25}
 Output: 1
 Explanation: Minimum difference is between 18 and 19
 Input: {30, 5, 20, 9}
 Output: 4
 Explanation: Minimum difference is between 5 and 9
 Input: {1, 19, -4, 31, 38, 25, 100}
 Output: 5
 Explanation: Minimum difference is between 1 and -4
Example-2

 Solution:
 Brute-force
 Logic/Idea?
Take first element and find its difference from rest of the
numbers in array, take the second element and find its
difference from rest of the numbers in array, and so on…

Example-2
 Problem: Design an algorithm that find the
minimum difference between any pair in the
given array.
Representation

Example-2
 Can you make as many improvements as you
can in this algorithmic solution to the
problem?
 You can: decrease the number of times the
innermost loop is executed, make that loop run
faster (at least for some inputs), or, more
significantly, design a faster algorithm from scratch.

Example-3: Your Turn
 Problem: Design an algorithm that find the two closest

elements in an array of numbers.
 Solution:
 An array of numbers.
 What does two closest elements in an array of

numbers means?
 The pair of elements that have the smallest absolute
difference between them.

 Find the pair of elements that have the smallest
absolute difference between them? If there are multiple

pairs, find them all.
Example-3

 Solution:
 Given a list of distinct unsorted integers.

 Input: arr[] = {10, 50, 12, 100}
 Output: (10, 12)
 The closest elements are 10 and 12
 Input : arr[] = {5, 3, 4, 2}
 Output : (2, 3), (3, 4), (4, 5)

Example-3

 Solution:
 Logic/Idea?
Take first element and find its difference from rest of the
numbers in array, take the second element and find its
difference from rest of the numbers in array, and so on.
Return the pair whose minimum distance is minimum.

Example-3
Problem: Design an algorithm that find the two

closest elements in an array of numbers.
Representation

TASK 1
 Problem: Design an algorithm to calculate which known

locations are closest to current known location.
 Solution:

TASK 1

 Solution:
 A list P of n (n ≥ 2) points p1(x1, y1), . . . , pn(xn, yn) and current
known location p0(x0, y0),
 For simplicity, we consider the two-dimensional case of the closest-
pair problem. We assume that the points in question are specified in
a standard fashion by their (x, y) Cartesian coordinates

TASK 1

 Solution:
 Compute the distance between each pair of distinct points and
find a pair with the smallest distance. Of course, we do not
want to compute the distance between the same pair of points
twice. To avoid doing so, we consider only the pairs of
points (pi, pj ) for which i < j.
 How to compute the distance between pair of points?
 This distance is typically measured using the Haversine
formula, which calculates the great-circle distance between
two points on the Earth’s surface.

Develop the formula
 The Haversine Formula:

 Calculate geographic distance on earth
 The haversine formula is a very accurate way of
computing distances between two points on the
surface of a sphere using the latitude and longitude
of the two points.

Develop the Model
 The Haversine Formula:

TASK 1

 Solution:
 What is input
 GPS coordinate of current known location say P(x, y)
 A list L of n GPS coordinate of known locations.
P1 P2 P3 P4 P5 P6
 What should be the output
 Return the index of known location whose distance
from P is minimum
 Brute-force
 Logic/Idea?
Take first element and find its distance from P, take the second
element and find its distance from P, and so on… . Store these
distances in an array A and find minimum from A.
Design Technique
 Brute Force
 Algorithm Steps: Now, let’s outline the
algorithm to calculate the nearest location
based on longitude and latitude:
1. Define the reference point with its longitude and
latitude coordinates.
2. Iterate over the set of locations and calculate the
distance between each location and the reference
point using the Haversine formula.
3. Keep track of the minimum distance encountered so
far and the corresponding location.
4. After iterating over all the locations, the location
with the minimum distance is the nearest location.
Design Technique

Haversine formula
 For your convenience, I have written the

function for Haversine formula in Python
syntax.

What we learn?
CONCLUSION

Conclusion
 Understanding the problem
 Asking questions, do a few examples by hand, think about
special cases, etc.
 Deciding on
 Exact vs. approximate problem solving
 Appropriate data structure
 Choose appropriate Designing technique
 Specify using e.g., Pseudo code.
 Prove correctness
 Analyzing an algorithm
 Time efficiency : how fast the algorithm runs
 Space efficiency: how much extra memory the algorithm needs.
 Simplicity and generality
 Implement/Coding an algorithm.

Conclusion
 As a rule, a good algorithm is

a result of repeated effort and
rework.
 The Joy of Algorithms
Even if you have been fortunate
enough to get an algorithmic idea
that seems perfect, you should still
try to see whether can be
improved.

Lecture # 02 - New

Uploaded by

Copyright:

Available Formats

Lecture # 02 - New

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture # 02 - New

Uploaded by

Copyright:

Available Formats

Design and Analysis of Algorithm

Tanveer Ahmed Siddiqui

Department of Computer Science

Department of Computer Science 2

Algorithm It is sequence of unambiguous instructions for solving a

Properties Input, Output, Precision, Determinism, Finiteness,

Representation Structured Description, Flow Chart, Pseudo Code

Department of Computer Science

Department of Computer Science

 Design an algorithm to calculate which known

Department of Computer Science

Framework for designing and

(Fundamentals of Algorithmic Problem Solving)

Department of Computer Science 6

After completing this lecture, you will be able to

Department of Computer Science

 Solving problem by a computer is challenging and

Department of Computer Science

 Outline a sequence of steps one typically goes

Department of Computer Science

 Understand the problem /Writing

Department of Computer Science

Designing an algorithm is easy if you

Department of Computer Science

“problem definition” phase.

Department of Computer Science

 From the problem statement which may be vague or

 What information is given?

Department of Computer Science

 What information required to define the

Department of Computer Science

 What assumptions have been made?

Department of Computer Science

Designing an algorithm is easy if you

Department of Computer Science

We can model this problem with an undirected graph.

Department of Computer Science

Designing an algorithm is easy if you

Department of Computer Science

 An algorithm design techniques (or “strategy” or

Department of Computer Science

Department of Computer Science

Designing an algorithm is easy if you

Department of Computer Science

 Commonly used methods as discussed in first

Department of Computer Science

Department of Computer Science 25

Designing an algorithm is easy if you

Department of Computer Science

Department of Computer Science

 One possible method: tracing the algorithm’s

Department of Computer Science

Department of Computer Science

Department of Computer Science

 After correctness, by far the most important is

Department of Computer Science

Department of Computer Science

 What should be the output

Department of Computer Science