Manual GP calculation steps

May 26, 2012 at 7:19 PM

Hi,

I need to know the manual process of GP calculation process so i can apply it to my problem.

 

for example My problem solution is like this on 6 lines;

L1 x24

L2 x53

L3 x83

L4 x105

L5 x3

L6 x42

 

Each line got 300 options so if i run first L1 and select 30 best options for L1 and then Run again

L1 (30 options)

L2 (200 options)

 

and then select 30 best options for L2 and then add L3 run like this;

L1 (30 options)

L2 (30 Options)

L3 (200 options)

 

and choose best 30 L3 options and repeat onward.

 

Is this is the way GP write lines 1 by 1?

 

Thanks

Coordinator
May 26, 2012 at 7:47 PM

Hi, 

I am afraid I didn't understand you well, so could you please be more specific.

Regarding GP calculation process it is exactly as it is specified in Koza book.

The first initialize population based on the defined initialization methods, then evaluate propulation to be able to know fitnes value of each chromosome in th epopulaton. Thes starst evolution proces performing Crossove, Mutation and Reproduction, After then it evaluate population again and performed the proces of Selection based on definde selection methods.

Then finds the best chromosome, and strats proces again.

 

If this is OK for you I am glad!

Regards

Bahrudin 

May 28, 2012 at 5:46 AM

Hi Bahrudin,
I m using Genetic Algorithm software. It work fine for small problems but when higher solution is required for a problem it require very very long time like months and years. Which is impossible. This is how GA works, I got 336 solutions available to resolve the problem. so if the problem solution is on
3 lines mean 336 x 336 x 336 = 37,933,056 posiblities which GA needed to sort out.
6 lines mean 1,438,916,737,499,140 posiblities which GA needed to sort out. This is where problem starts.
9 lines mean 54,582,509,182,892,000,000,000 posibilities which GA needed to sort out. This is where problem GA stuck up.
15 lines mean 78,539,686,037,963,600,000,000,000,000,000,000,000 posibilities which GA needed to sort out. This is where problem gets impossible for GA.
24 lines ,,,,, no idea
This 4 line code example which found the solution;
x24
x300
x39
x88
I have know some body who is using GP to resolve same issues successfully. Think of old age Optimizor. New age came I m using GA, But still getting problems as discribed above. I need a manual way to use GA to work like GP.
This idea i can think of using GA manualy to work like GP;
a) run GA for 1st line on 336 solutions
b) GA will pick lets say 10 solutions
c) then run second test like this first line (10 solutions) x second line(336 solutions) GA randomly pick solutions from both lines and do the run.
d) then lets say it find 8 good solutions for line 2 so now we got line1=10 and line2=8 possible solutions.
and it keep on doing like that and eventualy it cut the possiblities down to very few and find the solution.
Any suggestions?
Thanks
With Kind Regards,
M Faraz Babar
----- Original Message -----
From: [email removed]
To: [email removed]
Sent: Saturday, May 26, 2012 11:48 PM
Subject: Re: Manual GP calculation steps [GPdotNET:357268]

From: bhrnjica

Hi,

I am afraid I didn't understand you well, so could you please be more specific.

Regarding GP calculation process it is exactly as it is specified in Koza book.

The first initialize population based on the defined initialization methods, then evaluate propulation to be able to know fitnes value of each chromosome in th epopulaton. Thes starst evolution proces performing Crossove, Mutation and Reproduction, After then it evaluate population again and performed the proces of Selection based on definde selection methods.

Then finds the best chromosome, and strats proces again.

If this is OK for you I am glad!

Regards

Bahrudin