Subj: Soft Computing Topic: Genetic Algorithm

Soumen Paul Department of Computer Science and Informatics Haldia Institute of Technology

ICARE Complex, HIT Campus, P.O HIT,PIN 721657 Haldia, West Bengal

Soft Computing

Search and optimization

Exploration

Definition of Genetic Algorithm

The Genetic algorithm (GA) is a randomized search and optimization technique guided by the principle of natural genetic systems. Recently, there has been a great deal of interest in GAs and their application to various engineering fields.

Application of GA
The GA is being applied to a wide range of optimization and learning problems in many domains. GA is being applied to a wide range of optimization and learning problems in many domains. GA also lend themselves well to power system optimization problems, since there known to exibit robustness, require no auxulliary, and can offer significant advantages in solution methodology and optimization performance.

Genetic Algorithm

The main advantage of the GA formulation is that fairly accurate results may be obtained using a very simple algorithm. The objective of fitness function is a judge of GAs attempts. GAs do not know how to derive a problems solution, but they do know, from the objective function, how close they are to a better soln.

GA versus Conventional Optimization and search Procedure

Sometimes near optimal solutions that can be generated quickly, using GAs, are more desirable than optimal solutions which require a large amount of time.
Using GAs, problems can be modelled as optimization problems.

GA versus Conventional Optimization and search Procedure.

GAs work with a coding of the parameter set, not the parameters themselves. Therefore they can easily handle integral or discrete variables.
GAs can provide globally optimum solutions.

GA versus Conventional Optimization and search Procedure.

GA use only objective function information, not derivatives or other auxiliary knowledge. Therefore, they can deal with the non-smooth, non-continuous, and non differentiable functions. GA use probabilistic transition rules, not deterministic rules.

GA Problem Formulation

States: A population of individuals, each with an associated fitness value Operators: Transform the population into a new generation, using Darwinian principle of reproduction and survival of the fittest Goal test: Criterion for terminating the search

Formalization

Representation scheme Fitness measures Parameters for controlling the algorithm

Genetic Representation Scheme

Binary representation: Binary representation consists of binary coding, the strings are made up of 1 and 0 (i.e 11110010101011). The length of the string is determined by the precision desired for the solution.

Binary coding
2 2 2 2 11 5 2 1 Remainder 1 1 0 Binary equivalent of 11 is 1011

Procedures of GA
Suppose we wish to maximize a function of k variables, f(x1,x2,.xk): Rk->R. Suppose further that each variable xi can take values from a domain Di=[ai,bi] and f(x1,x2,.xk) > 0 for all xi in Di. We wish to optimize the function f with required precision say upto 6 decimal places for the variables. Then Di must be cut into (bi-ai)*106 equal sized ranges.

Procedures of GA
Let mi be the smallest integer such that 2mi-1<=(bi-ai)*106 <=2mi. Then mi be the size of the Chromosome.
The equivalent value for any mi bit string can be obtained as xi=xil+[(xiu-xil)/(2mi-1)]*(decoded value of string)

Fitness Measures

Numerical evaluation Fitness = f(character in S) The character representation in S is also referred as chromosomes.

Parameters for controlling the algorithm

Population size (M) Maximum number of generations in a run (G) Number of runs (N) Selection & Replacement strategy Selection function to identify the result (goal)

Genetic search program

Create some initial population (M many) randomly Iterate (G times or less)

Evaluate

the fitness of each individual

Genetic search program (contd)

Evolve

the population I.e., randomly select individuals and randomly apply operators among:
Reproduction
Crossover Mutation

Select the fittest individual as a solution Run the GA many (N) times.

Stages of GA

A GA operates through a simple cycle of stages i) Creation of a population of strings, ii) Evaluation of each string, iii) Selection of best strings and iv) Genetic manipulation to create new population of strings.

Cycle of Genetic Algorithm

Description of stages of GA

Each cycle in GA produces a new generation of possible solutions for a given problem. In the first phase, an initial population, describing representatives of the potential solution, is created to initiate the search process. The elements of the population are encoded into bitstrings, called

Description of stages of GA.

The performance of the strings, often called fitness, is then evaluated with the help of some functions, representing the constraints of the problem. Depending on the fitness of the chromosomes, they are selected for a subsequent genetic manipulation process. It should be noted that the selection process is mainly responsible for assuring survival of the best-fit individuals.

1st Step of Genetic Algorithm

After selection of the population strings is over, the genetic manipulation process consisting of two steps is carried out. In the first step, the crossover operation that recombines the bits (genes) of each two selected strings (chromosomes) is executed. Various types of crossover operators are found in the literature. The single point and two points crossover operations are illustrated in fig. 15.2 and 15.3 respectively. The crossover points of any twochromosomes are selected randomly.

2nd step of Genetic Algorithm

The second step in the genetic manipulation process is termed mutation, where the bits at one or more randomly selected positions of the chromosomes are altered. The mutation process helps to overcome trapping at local maxima. The offsprings produced by the genetic manipulation process are the next population to be evaluated.

Crossover in Genetic Algorithm

Mutation in Genetic Algorithm

Crossover Rate
In GA literature, the term cross over rate is usually denoted as PC, the probability of crossover. The probability varies from 0 to 1. This is calculated in GA by finding out the ratio of the number of pairs to be crossed to some fixed population. Typically for a population size 30 to 200, cross over rates are ranged from 0.5 to 1.

Mutation Rate Pm
Mutation rate is the probability of mutation which is used to calculate number of bits to be muted. The mutation operator preserves the diversity among the population which is also very important for the search. Typically, the simple genetic algorithm uses the population size of 30 to 200 with the mutation rates varying from 0.001 to 0.5.

Sample Problem I

Example 15.1: The GA cycle is illustrated in this example for maximizing a function f(x) = x2 in the interval 0 x 31. In this example the fitness function is f (x) itself. The larger is the functional value, the better is the fitness of the string. In this example, we start with 4 initial strings. The fitness value of the strings and the percentage fitness of the total are estimated in the 1st table

Solution by GA
Since fitness of the second string is large, we select 2 copies of the second string and one each for the first and fourth string in the mating pool.The selection of the partners in the mating pool is also done randomly. Here in 2nd table , we selected partner of string 1 to be the 2-nd string and partner of 4-th string to be the 2nd string.

Solution by GA..
The crossover points for the first-second and second-fourth strings have been selected after o-th and 2-nd bit positions respectively in 2nd table. The second generation of the population without mutation in the first generation is presented in 3rd table

Initial population and their fitness values

Table 1: Initial population and their fitness values

Mating pool strings and crossover

Table 2: Mating pool strings and crossover

Fitness value in second generation

Table 3: Fitness value in second generation

Sample problem

The function is defined as f(X) = xsin(10x)+1.0 X=[-1,2]

Solution by GA

Representation: In this case a binary vector (chromosome) has been considered to represent the real values of the variable x. Let six places after the decimal point is our required precision. The domain of the variable x has a length equal to three; the precision requirement implies the [-1,2]

Solution by GA...

It should be divided into at least 3,000,000 equal-sized intervals. This means that 22 bits are required to represent one binary vector (chromosome): 2,097,152=221<3000000<=222 =4,194,304 The mapping

Solution
The mapping from a binary string to a real number x is performed in two steps. 1. Convert the binary string from base to base10 and set it equal to x1, and 2. Find a corresponding real number x=1+[3/(222-1)]* x1, where -1.0 be left boundary and 3 is the length of the domain.

Solution .
Of course the Chromosomes (0000000000000000000000) and (1111111111111111111111) represent boundaries of domain i.e -1 and 2 respectively. For example three chromosomes V1=(1000101110110101000111) V2=(0000001110000000010000) V3=(1110000000111111000101)

Solution.
correspond to the values x1=0.637197, x2=-0.958973,x3=1.627888. consequently the evaluation function as follows eval(v1)=f(x1)=1.586345 eval(v2)=f(x2)=0.078878 eval(v3)=f(x3)=2.250650

Solution .
Genetic operators: (mutation) v31=(110000001111111000101). Correspondingly, x31=1.630818 and f(x31)=2.343555, an improvement over original value of f(x3)=2.250650. Crossover operator: v2=(00000|01110000000010000) v3=(11100|00000111111000101)

Solution .
The offspring evaluate to f(v21)=f(-0.998113)=0.940865 f(v31)=f(1.666028)=2.459245 For this particular problem, pop_size=50, Pc =0.25, Pm=0.01 The best chromosome after 200 generations was Vmax=(1111001101000100000101) which corresponds to value xmax=1.850773

Traveling Salesman Problem

In the Traveling Salesman Problem (TSP) location of all the cities is given and the salesmans task is to find the shortest route connecting them all, with each city visited only once and returns to city of origin. Since it is NP complete problem, the method of finding an optimal solution must search in a solution space that expands exponentially with the number of cities.

Steps of Genetic Algorithm

Step 1: Generate random population of n Chromosome. Step 2: Evaluate the fitness f(x) of each Chromosome x in the population. Step 3: Create a new population by repeating the following. Steps until the new population by repeating the following.

Steps of Genetic Algorithm

a): [Selection] select two parent Chromosomes from a population according to their fitness (the better fitness, the bigger chance to be selected.) b): [crossover] with a crossover probability Crossover the parents to form new offspring (children). If no Crossover was performed, offspring is the exact copy of parents.

Steps of Genetic Algorithm..

c): [Mutation] with a mutation probability mutate new offspring at each locus (position in Chromosome). d) [Accepting] place new offspring in the new population. Step 4: Use new generated population for a further run of the algorithm. Step 5: If the end condition is satisfied, then stop and return the best. Solution in the current population. Go to step 2.

Fitness Measure
Fitness function fit(tk) (k=1,2,pop_size) is defined as fit(fx) = 1/[(ci,ci+1)+ d(cn,c1)] where d(ci,ci+1): travelling distance from city ci to ci+1. If city ci is represented by pixel at point P(x,y) and city ci+1 is represented by pixel at point q(s,t) and d(ci,ci+1) is Euclidean distance between two points P(x,y) and q(s,t) which is defined as d(ci,ci+1)=|(x-s)|+ |(y-t)| = (x-s)2+(y-t)2

Greedy Crossover Operation

Algortthm: Step 1: Select randomlytwo parent tours P1 and P2 to create first offspring Step 2: Start with a random city C. Step 3: If the edge <C,Cright> or <C,Cleft> is used in both P1 and P2 then choose common edge.

Greedy Crossover Operation.)

Step 4: Otherwise compare Cs right side edge in each of P1 and P2. Step 5: Choose shorter one if it will not introduce a cycle. Step 7: If longer one created introduces a cycle, then extend the tour by a carefully selected edge. To generate second offspring, left edges are compared instead of right edges.

Sample Problem of TSP

1 2

Distance Table
1 2 3 4 5 6 0 10 17 20 17 10
1

0 10 17 20 17
2

0 10 17 20
3

0 10 17
4

0 10
5

0
6

Solution of TSP by GA
(Parent 1)P1: 1 3 4 2 5 6 (Parent 2)P2: 1 2 3 6 5 4 Offspring1: (if randomly starting from city 6) 6 5 4 2 3 1 -> 1 3 2 4 5 6 Offspring2: (if randomly starting from city 3) 3 2 1 6 5 4 -> 1 6 5 4 3 2

Application of Greedy Crossover operation through GA

Assume that we have two parents : P1 and P2 . To generate the first offspring, we randomly start from citry 6. The two right sides are <6,1> in P1 and <6,5> in P2. The distance table shows that <6,1> and <6,5> have exactly the same length. In this situation , the edge in P1 would normally be selected. However, in our example <6,5> is a common edge between the two parents, so <6,5> is choosen.

Application of Greedy Crossover operation through GA.

Next, from city 5, we have two right side edges <5,6> and <5,4>. This time, we choose <5,4> because the common edge <5,6> would create a cycle. From city 4, the edges are <4,2> and <4,1>: the shorter one <4,2> is choosen. Then from city 2 , there are <2,5> and <2,3> : the shorter one <2,3> is chosen,

Application of Greedy Crossover operation through GA.

Finally, we have only city 1 left, which has to be the last one in the tour. The second offspring is generated starting from a different random city 3. This time, the left edges are compared. Similarly, all of the best edges in P1 and P2 have been inherited by offspring 2.

Thank you!