Author |
Topic: Floating Tuple Elimination pairing algorithm (Read 3469 times) |
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: Floating Tuple Elimination pairing algorithm
« Reply #15 on: Dec 7th, 2009, 6:52am » |
Quote Modify
|
Quote:With fte.cpp, ... the last argument takes the name of a file containing earlier games that were part of the elimination tournament (which must exist as an empty file if there hasn't been any games yet). |
| Should the order of the games in this file be such that games from the most recent round are at the top; or should those games be at the bottom? Same question for the past games file for colors.cpp. Did some tests and it looks like the more recent games should be at the bottom.
|
« Last Edit: Dec 7th, 2009, 7:15am by omar » |
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: Floating Tuple Elimination pairing algorithm
« Reply #16 on: Dec 7th, 2009, 10:16am » |
Quote Modify
|
Yes, in both applications all game files are interpreted as being in chronological order, which makes it congruous with the order in which the files are taken as arguments. Hopefully you didn't take the code before I made that correction shortly after the upload.
|
|
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: Floating Tuple Elimination pairing algorithm
« Reply #17 on: Dec 7th, 2009, 5:36pm » |
Quote Modify
|
I downloaded the code this morning. For the file that is the last argument to Fte.cpp how do you specify a bye.
|
|
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: Floating Tuple Elimination pairing algorithm
« Reply #18 on: Dec 7th, 2009, 6:40pm » |
Quote Modify
|
No mention of byes in earlier rounds is necessary; the algorithm looks at the number of times each player has played so far.
|
|
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: Floating Tuple Elimination pairing algorithm
« Reply #19 on: Dec 8th, 2009, 8:28am » |
Quote Modify
|
Thanks. I think I've got it working now with the tournament management tool. I simulated last years WC tournament. Here is the output: http://arimaa.com/arimaa/events/showGames.cgi?e=aaaa1 In round 6 which is the first round when your program is invoked, it seems to be doing a fold pairing where as the pairing program we actually used did a slide pairing. It did a good job of preserving the color assignment from the swiss. Looks pretty good to me. Let me know if you notice any problems. If you want to experiment with it let me know and I'll set you up with an account.
|
« Last Edit: Dec 8th, 2009, 9:04am by omar » |
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: Floating Tuple Elimination pairing algorithm
« Reply #20 on: Dec 8th, 2009, 9:04am » |
Quote Modify
|
Here is a simulation of the 2009 computer championship. http://arimaa.com/arimaa/events/showGames.cgi?e=aaaa2 The pairing is the same (not considering color) up to round 3 and then it differs.
|
« Last Edit: Dec 8th, 2009, 9:10am by omar » |
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: Floating Tuple Elimination pairing algorithm
« Reply #21 on: Dec 8th, 2009, 7:35pm » |
Quote Modify
|
on Dec 8th, 2009, 8:28am, omar wrote:In round 6 which is the first round when your program is invoked, it seems to be doing a fold pairing where as the pairing program we actually used did a slide pairing. |
| The already existing preference to maximize the sum of the squares of differences in rank in the current floating tuple elimination algorithm invariably implies a preference for a folding scheduling (where not overruled by higher priorities).
|
|
IP Logged |
|
|
|
|