Register | Sign In


Understanding through Discussion


EvC Forum active members: 65 (9162 total)
4 online now:
Newest Member: popoi
Post Volume: Total: 915,817 Year: 3,074/9,624 Month: 919/1,588 Week: 102/223 Day: 0/13 Hour: 0/0


Thread  Details

Email This Thread
Newer Topic | Older Topic
  
Author Topic:   BoxCar2D - neat little genetic algorithm demonstration
Minnemooseus
Member
Posts: 3941
From: Duluth, Minnesota, U.S. (West end of Lake Superior)
Joined: 11-11-2001
Member Rating: 10.0


Message 11 of 51 (605124)
02-17-2011 5:53 AM


50 generations
1084.6 (1:50) and 1084.3 (1.49) from two different cars.
Ran for 50 generations. No distance improvement for at least 10 generations.
Moose

Replies to this message:
 Message 13 by Theodoric, posted 02-17-2011 9:47 AM Minnemooseus has seen this message but not replied

  
Minnemooseus
Member
Posts: 3941
From: Duluth, Minnesota, U.S. (West end of Lake Superior)
Joined: 11-11-2001
Member Rating: 10.0


Message 44 of 51 (605576)
02-20-2011 11:53 PM
Reply to: Message 24 by Dr Adequate
02-18-2011 2:49 AM


Dr. A was at 1055.1
Dr. A was at 1055.1 at 2:11
I seem to remember that it took quite a while to get past that point. I took a lot of screen shots, but I seemed to have started doing such later. I need to get back to looking at the photo record, but need to get other things done now.
I got to 1135.3 at 2:02, and then stalled forever. At one generation I had 6 of 20 cars making it to that point. Any differences between them was extremely subtle.
Anyway, I cut the test off midway through generation 193 (I had it running for about 36 hours).
I saved a version of the best car and injected it into a new test. In a couple of tries it has seriously evolved to the worse in just the first few generations, when the track was still relatively short. I need to look to see if the mutations can be shut off.
Moose

This message is a reply to:
 Message 24 by Dr Adequate, posted 02-18-2011 2:49 AM Dr Adequate has not replied

Replies to this message:
 Message 45 by slevesque, posted 02-21-2011 12:44 AM Minnemooseus has replied

  
Minnemooseus
Member
Posts: 3941
From: Duluth, Minnesota, U.S. (West end of Lake Superior)
Joined: 11-11-2001
Member Rating: 10.0


Message 46 of 51 (605601)
02-21-2011 4:20 AM
Reply to: Message 45 by slevesque
02-21-2011 12:44 AM


I got to 1135.3 at 2:02 (and ran ~190 generations)
I figured out the zero mutations thing. Another thing to do is to set the population to "2", the minimum allowed, to run as few cars as possible.
I plugged my best effort (evolved from the defaults) into a zero mutations, 2 population, default track setup. There are still some tiny variables in there, you get different results in different tries. I was only able to get these tries to a ~900 best. I couldn't repeat my 1135.3 at 2:02.
I've enclosed my supposed best result parameters in a quote box, but I "hid" it to keep it from making the page way over-wide. Use "peek" to see (an try if you wish) the thing.
Moose

This message is a reply to:
 Message 45 by slevesque, posted 02-21-2011 12:44 AM slevesque has not replied

Replies to this message:
 Message 47 by Dr Jack, posted 02-21-2011 4:29 AM Minnemooseus has replied

  
Minnemooseus
Member
Posts: 3941
From: Duluth, Minnesota, U.S. (West end of Lake Superior)
Joined: 11-11-2001
Member Rating: 10.0


Message 48 of 51 (605609)
02-21-2011 5:17 AM
Reply to: Message 47 by Dr Jack
02-21-2011 4:29 AM


Re: I got to 1135.3 at 2:02 (and ran ~190 generations)
I was running the default track. I expected it to be random, but it sure looked familiar (same???) compared to the ~190 generation adventure. So I came to think it was always the same.
Moose

This message is a reply to:
 Message 47 by Dr Jack, posted 02-21-2011 4:29 AM Dr Jack has not replied

Replies to this message:
 Message 50 by slevesque, posted 02-24-2011 4:18 AM Minnemooseus has seen this message but not replied

  
Newer Topic | Older Topic
Jump to:


Copyright 2001-2023 by EvC Forum, All Rights Reserved

™ Version 4.2
Innovative software from Qwixotic © 2024