Computer Chess Club Archives

Search

Messages

Subject: Early BWTC results

Author: Steven Edwards

Date: 12:19:46 05/12/05

Go up one level in this thread

I'm now trying the GA on the 1,001 BWTC suite with a habitat of 64 organisms.

1.  It took 826.58 seconds (400 MHz PPC) to load the test suite and apply all
possible precalculations.

2. The precalculation count is roughly 1,001 positons times (approximately) 32
moves per position times 20 microfeature recognitions per resulting position;
about 640,000 recognition attempts.

3. It took 5097.69 seconds to calculate the merits for the 64 initial organisms.

4. It's taking about 80 seconds per habitat cycle to breed and test a new
organism.

5. The highest ranking initial organism matched a best move in 383 of the 1,001
problems.

6. I'm going to let this thing run overnight.

• New champion organism at generation 48 Steven Edwards 13:43:48 05/12/05
  • New champion organism at generation 103; solves 475/1001 Steven Edwards 14:39:15 05/12/05
    • New champion organism at generation 111; solves 598/1001 Steven Edwards 14:49:14 05/12/05
      • New champion organism at generation 297; solves 699/1001 Steven Edwards 20:21:21 05/12/05
        • New champion organism at generation 358; solves 702/1001 Steven Edwards 20:33:07 05/12/05
      • Re: New champion organism at generation 111; solves 598/1001 Matthew Hull 15:14:44 05/12/05
        • Mutator code snippet Steven Edwards 20:27:33 05/12/05

This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.