You are on page 1of 1

Assignment-1 Report

Submitted by: Hritu Raj (B18ME023) && Deepak Arjariya (B18ME021)

Choosing the starting state:


We will choose an improved state instead of a random state.
For getting improved state we have arranged the papers of greater similarity in one session and
the paper having the most dissimilarity with papers already arranged, we put them in parallel
sessions if available.
Similarly, we have arranged the papers for other time slots in the decreasing order of similarity.

Applying genetic algorithm

At first, we arrange all papers in a row. Then divide them in t (no. of time slots) parts. Within the
time slots we divide them in p(no. of parallel sessions) such that each smallest unit of list has k
papers.
Assume we have k=2 ;p=2; t=3
Let our initial state be the following:

0 1 2 3 4 5 6 7 8 9 10 11

Above, papers under a time slots are in one colour, while papers in parallel sessions are divided
by blue lines within the same colour.

We calculate the fitness score of each t (within the same colour) part.
The fitness score is the goodness score if we have a time slot.

Now, we will follow a genetic algorithm in which the group which has less fitness score will
mutate as they have more probability of improving their score while the groups having more
fitness score will mutate only if their fellow groups reach his fitness score.

Assume the fitness score of yellow is 3, fitness score of greens is 4 and goodness score of red
Is 5. Then we will mutate green and yellow to improve their fitness score.

If the fitness score comes more than their previous one, we will store that else we will ignore
that.

Until we reach the fitness score as equal to red we will not mutate the papers in red with
another group as it is already contributing more to the overall goodness score.
We will continue the process for a reasonable time and then we will accept the value which
comes at the end of our operation.

You might also like