2011-01-14T00:00:29 hmm not as well as I had hoped, but my fix does save ~35-40 secs per 1000 turns 2011-01-14T00:05:32 lol 2011-01-14T00:07:01 my original prediction was about 1-2mins but it did not scale as planned. 2011-01-14T00:09:15 *** amriedle has quit IRC (Quit: Lost terminal) 2011-01-14T00:24:36 *** Meatkat has quit IRC (Ping timeout: 255 seconds) 2011-01-14T00:27:25 *** Meatkat has joined #aichallenge 2011-01-14T00:56:54 in what game you guys are on? 2011-01-14T01:17:34 http://queue.acm.org/icpc/ 2011-01-14T01:17:40 this link is better for that snowman game 2011-01-14T01:34:25 *** sigh has left #aichallenge ("Leaving") 2011-01-14T01:53:00 *** nann has joined #aichallenge 2011-01-14T01:56:29 *** Apophis_ has quit IRC (Ping timeout: 240 seconds) 2011-01-14T02:45:40 *** sigh has joined #aichallenge 2011-01-14T02:56:10 *** mega1 has joined #aichallenge 2011-01-14T03:00:54 *** tapwater has quit IRC (Quit: tapwater) 2011-01-14T03:49:20 *** aerique has joined #aichallenge 2011-01-14T03:50:50 *** Naktibalda has joined #aichallenge 2011-01-14T04:30:48 *** Palmik has joined #aichallenge 2011-01-14T04:58:03 *** boegel has joined #aichallenge 2011-01-14T05:08:55 Zombies.coolness >> Ants.coolness http://ai-contest.com/forum/viewtopic.php?f=21&t=1243&p=8396#p8396 2011-01-14T05:09:31 Meatkat: i was directed to a new contest today i'm interested enough in to have started making a bot 2011-01-14T05:09:49 http://queue.acm.org/icpc/ 2011-01-14T05:09:53 that's the best link 2011-01-14T05:14:44 antimatroid: that looks pretty fun, nice and short as well 2011-01-14T05:15:24 It'll be over just as this one starts 2011-01-14T05:16:04 yep 2011-01-14T05:16:15 i'm not going to put a huge amount of effort in 2011-01-14T05:16:27 just making some kind of tree strategy and will see how it does 2011-01-14T05:16:38 throwing in a few other algorithms to pick out moves 2011-01-14T05:18:16 Cool, I'm tempted but I'm currently doing BattleCode 2011-01-14T05:18:39 do you need a team member? someone messaged me on reddit about that the other day 2011-01-14T05:18:45 i haven't got around to responding to them yet 2011-01-14T05:19:13 Hmmm... 2 questions. Are they smart? and do they get stuff done? 2011-01-14T05:19:26 We're looking for 1 more 2011-01-14T05:22:27 no idea 2011-01-14T05:22:34 i'll check their username, hold up 2011-01-14T05:23:02 http://www.ai-contest.com/profile.php?user_id=6026 2011-01-14T05:23:53 haha... when you said that you'll check their username, I thought you were going to stalk their reddit account :P 2011-01-14T05:24:30 :P 2011-01-14T05:24:59 i don't think i'm that amazing that i could judge someone from comments on a website 2011-01-14T05:25:18 not that that would stop me from doing it 2011-01-14T05:26:04 Loading... what do you guys think of the zombie theme instead of the ant theme? 2011-01-14T05:26:54 From Oct 9-Oct 30th their entry didn't compile. Is that a noob thing or a C# thing? I heard they had problems 2011-01-14T05:28:28 amstan: what's the go with that simplified ants game? 2011-01-14T05:28:48 you have food popping up during the game, that's iffy with respect to staying symmetric 2011-01-14T05:29:05 Meatkat: no idea sorry 2011-01-14T05:37:01 *** mceier has joined #aichallenge 2011-01-14T05:37:27 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-01-14T05:38:38 *** janzert has joined #aichallenge 2011-01-14T05:39:44 Cool, well I'll keep him in mind. I want to what the limiting factor on my current team is before I throw someone else into the mix. Thanks! 2011-01-14T05:42:49 I want to know* what 2011-01-14T06:11:43 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-01-14T07:02:21 *** mega1 has quit IRC (Read error: Operation timed out) 2011-01-14T07:16:03 *** delt0r_ has joined #aichallenge 2011-01-14T07:17:49 *** delt0r___ has quit IRC (Ping timeout: 260 seconds) 2011-01-14T07:34:34 sigh: another reason why it makes more sense to index grids by row col 2011-01-14T07:35:01 if you read a grid by x y, then it is going to be in the file in a different format to how it's really being represented 2011-01-14T07:43:20 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-01-14T07:44:31 *** sigh has quit IRC (Remote host closed the connection) 2011-01-14T07:54:29 *** chl_ has quit IRC (Ping timeout: 260 seconds) 2011-01-14T08:01:08 *** chl has joined #aichallenge 2011-01-14T08:17:59 *** amstan has joined #aichallenge 2011-01-14T08:17:59 *** ChanServ sets mode: +o amstan 2011-01-14T08:21:46 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-01-14T08:23:07 *** janzert has joined #aichallenge 2011-01-14T08:26:20 *** amstan has quit IRC (Ping timeout: 246 seconds) 2011-01-14T09:50:09 *** Vogons has joined #aichallenge 2011-01-14T09:53:42 antimatroid: too bad if it's no symmetric 2011-01-14T09:53:51 you can't expect everything to be fair 2011-01-14T09:54:00 it'll be random 2011-01-14T09:59:39 *** chl has quit IRC (Ping timeout: 240 seconds) 2011-01-14T09:59:59 *** chl has joined #aichallenge 2011-01-14T10:27:35 *** mega1 has joined #aichallenge 2011-01-14T10:50:26 *** Mekanik has joined #aichallenge 2011-01-14T10:55:43 *** aerique has quit IRC (Quit: ...) 2011-01-14T11:09:26 *** Palmik has quit IRC (Ping timeout: 260 seconds) 2011-01-14T11:36:51 *** boegel has quit IRC (Ping timeout: 260 seconds) 2011-01-14T12:03:12 *** Mathnerd314 has quit IRC (Ping timeout: 250 seconds) 2011-01-14T12:14:53 antimatroid: my habits are still to go with (x, y), but i agree with you that (row, col) is superior 2011-01-14T12:15:25 antimatroid: i've paid the price for going with (x, y) *again* just yesterday in my tron engine, and soon i will have to go through my code and fix it 2011-01-14T12:16:28 antimatroid: another reason for (row, col) is that it's more cache coherent when you traverse it in left-to-right, top-to-bottom order. to make (x, y) cache coherent you have to traverse it top-to-bottom, left-to-right 2011-01-14T12:24:20 lol http://www.youtube.com/watch?v=3RAjfNdZ0hw 2011-01-14T12:48:37 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2011-01-14T13:11:13 *** dr- has joined #aichallenge 2011-01-14T13:11:20 *** Cyndre has quit IRC (Quit: Leaving) 2011-01-14T13:13:58 *** dr- has quit IRC (Quit: WeeChat 0.2.6.3) 2011-01-14T13:17:46 *** Frontier has joined #aichallenge 2011-01-14T13:25:51 *** McLeopold has joined #aichallenge 2011-01-14T13:34:59 *** dr- has joined #aichallenge 2011-01-14T13:42:05 jmcarthur: :) 2011-01-14T13:42:30 amstan_irssi: I was a fan of things being symmetric 2011-01-14T13:52:42 *** Naktibalda has joined #aichallenge 2011-01-14T13:56:28 randomness is important for some games, but i think not for this one 2011-01-14T13:57:42 i also just don't like spawning food in this game i think. when the food doesn't spawn, the game has distinct phases, but when food does spawn it makes the strategy basically the same throughout the whole game 2011-01-14T14:03:31 i want randomness with map generators, but i also want symmetric games 2011-01-14T14:03:51 and one can replenish symmetrically from the start of hte game, but that's iffy 2011-01-14T14:08:22 have the rate of food spawning slowly decrease to 0 by say 3/4 of the game time is completed 2011-01-14T14:09:27 i still think it's fine to just have all food placed at the start 2011-01-14T14:09:44 with each ant having a view radius you don't know the entire map at the start anyway 2011-01-14T14:10:02 having it spawn does mean there is a reason to keep ants back in an area you already cleared though 2011-01-14T14:10:33 meh, that's not that interesting strategy wise 2011-01-14T14:10:48 i don't like where the ants game is going 2011-01-14T14:10:50 otherwise you should just keep pushing all your ants outward 2011-01-14T14:11:02 I think it's better than without though 2011-01-14T14:11:22 the objective of the game is completely different to the example game i had 2011-01-14T14:11:58 janzert: i have started making a bot for a new contest though.. 2011-01-14T14:12:07 http://queue.acm.org/icpc/ 2011-01-14T14:12:08 the snowball one? 2011-01-14T14:12:09 that's the contest 2011-01-14T14:12:11 yeah 2011-01-14T14:12:12 * janzert nods 2011-01-14T14:12:32 didn't read all the details but looked interesting 2011-01-14T14:13:22 there setup is certainly less compute intensive, e.g. you only get a maximum of 6 games per day 2011-01-14T14:13:52 i think if we went with McLeopold's hills idea then that would be plenty of incentive to hold your ants back in some areas 2011-01-14T14:13:56 and from what I can tell only the players get the results (and only from 3 of them?) so you can't go watch a bunch of matches 2011-01-14T14:14:50 janzert: i'll be surprised if something like tcp doesn't pop up surely 2011-01-14T14:15:06 i think they made moving more complicated than was necessary 2011-01-14T14:15:19 and still haven't fully workout out what's happening with the throwing stuff 2011-01-14T14:15:35 :} 2011-01-14T14:16:01 is there a forum with discussion? 2011-01-14T14:17:48 *** tapwater has joined #aichallenge 2011-01-14T14:18:26 haven't found one yet :P 2011-01-14T14:18:42 or an irc, i'm just playing around with code for an ai game 2011-01-14T14:18:45 it's been far too long 2011-01-14T14:19:00 heh 2011-01-14T14:42:38 *** dr- has quit IRC (Remote host closed the connection) 2011-01-14T14:48:23 the game could have been more awesome 2011-01-14T14:48:39 you can only walk if the change in height is less than 2 2011-01-14T14:48:44 leq 2011-01-14T14:49:07 *** Cold-Phoenix has joined #aichallenge 2011-01-14T14:53:32 Whoever is playing minecraft, please watch this: http://www.youtube.com/watch?v=b3TUskvZ2Qc&NR=1 2011-01-14T14:53:35 jmcarthur: with my hills idea, it really needs spawning to work well 2011-01-14T14:54:02 A bot needs 2 conflicting goals, IMO, to have a fun game. 2011-01-14T14:54:30 Without spawning, there is no incentive to leaving the hills. 2011-01-14T15:04:58 antimatroid: having food that respawns is interesting 2011-01-14T15:05:21 antimatroid: since it makes it advantageous for players that are expanding their ants, and making sure they cover a larger area 2011-01-14T15:10:13 *** boegel has joined #aichallenge 2011-01-14T15:26:36 *** boegel has quit IRC (Ping timeout: 240 seconds) 2011-01-14T15:39:13 *** boegel has joined #aichallenge 2011-01-14T15:56:35 *** boegel has quit IRC (Quit: *poof!*) 2011-01-14T16:09:00 *** Meatkat has quit IRC (Ping timeout: 240 seconds) 2011-01-14T16:15:07 *** Naktibalda has quit IRC (Remote host closed the connection) 2011-01-14T16:54:42 *** Mathnerd314 has joined #aichallenge 2011-01-14T17:01:25 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2011-01-14T17:06:13 *** Migi32 has joined #aichallenge 2011-01-14T17:17:49 *** sigh has joined #aichallenge 2011-01-14T17:41:17 *** Mathnerd314 has joined #aichallenge 2011-01-14T17:44:56 *** Frontier has quit IRC (Ping timeout: 276 seconds) 2011-01-14T18:32:58 *** Migi32 has quit IRC (Quit: BOINC) 2011-01-14T18:37:09 *** Cyndre has joined #aichallenge 2011-01-14T18:54:25 *** mega1 has quit IRC (Ping timeout: 260 seconds) 2011-01-14T19:24:45 *** Cold-Phoenix has quit IRC () 2011-01-14T19:44:19 *** Mathnerd314 has quit IRC (Ping timeout: 264 seconds) 2011-01-14T19:54:05 *** Meatkat has joined #aichallenge 2011-01-14T19:58:03 *** McLeopold has left #aichallenge 2011-01-14T20:45:44 *** amstan has joined #aichallenge 2011-01-14T20:45:44 *** ChanServ sets mode: +o amstan 2011-01-14T20:46:45 *** mceier has quit IRC (Quit: leaving) 2011-01-14T20:56:42 i want to come up with an awesome but completely stupid-sounding game 2011-01-14T20:57:39 like mixing a classic game like go with something completely unrelated like a shooter 2011-01-14T20:58:10 i have no idea how it would work 2011-01-14T21:06:00 jmcarthur: so, I reran the trueskill stuff, and looked at the middle of the pack. it is a mess compared to the official results. 2011-01-14T21:06:00 on one hand, I think it is fine, because considering the sigmas, people are pretty close to where they were. 2011-01-14T21:06:26 but i haven't checked to close, and am not sure what a good measure would be to check it. 2011-01-14T21:06:56 Levenshtein distance? 2011-01-14T21:07:22 but then we'd just have to decide on a threshold anyway, and i don't know how we'd go about doing that 2011-01-14T21:07:40 oh, nevermind anyway 2011-01-14T21:07:51 yeah, with the sigmas, it is hard to tell. 2011-01-14T21:07:58 Levenshtein measures insertions, deletions, and substitutions, but not swaps 2011-01-14T21:08:08 swaps would be the interesting one i think 2011-01-14T21:08:18 hmm 2011-01-14T21:08:23 well, not strictly 2011-01-14T21:09:41 i tried tweaking the true skill parameters to get a the actual ratings the same, but that didn't really work out. 2011-01-14T21:10:10 bayeselo uses logistic, and trueskill uses normal, so a I am not sure if that is a factor. 2011-01-14T21:12:28 i doubt it would make a huge difference. 2011-01-14T21:14:35 Damerau–Levenshtein distance includes transposition apparently 2011-01-14T21:15:21 if we wanted to take the time i bet we could try to tweak the trueskill parameters using hill climbing with Damerau–Levenshtein distance. i'm not sure how close we would get 2011-01-14T21:15:40 really, it's just hard to make this comparison since trueskill is order-dependent in the matches 2011-01-14T21:16:22 yes, there is that issue too. 2011-01-14T21:18:34 *** janzert has quit IRC (Ping timeout: 272 seconds) 2011-01-14T21:37:42 for me the most important thing is the final rankings being as accurate as possible 2011-01-14T21:37:50 i don't care at all how accurate they are before that 2011-01-14T21:38:32 well, very very barely care, particularly compared to final ranking accuracy 2011-01-14T21:39:34 the places where I see differences, though, appear to be because of the sigmas in the ratings. I *believe* that the diffs are inside the confidence interval. 2011-01-14T21:40:01 well, not because of the sigmas, but within, or something like that. 2011-01-14T21:40:35 accurate estimates, that is what an old boss used to say... 2011-01-14T21:41:02 dmj111: did you play with snowballs at all? 2011-01-14T21:41:27 i have mostly finished setting up my own code, i'll start writing some basic strategies (like a ragebot) soon 2011-01-14T21:41:48 i just watched a live game for the first time :) 2011-01-14T21:42:40 their starter bot isn't necessarily as stupid as i first thought 2011-01-14T21:43:02 once my guys move over there, they're in a good position to throw balls from directly behind the walls at you 2011-01-14T21:43:28 still stupid, just not as stupid as i thought 2011-01-14T21:43:30 ha. is there a way to run it without the gui? 2011-01-14T21:43:54 not sure 2011-01-14T21:43:57 I don't see it in the directions. not useful at the moment, but for later testing, it might be good. 2011-01-14T21:44:18 worst comes to worst, surely someone will write another engine 2011-01-14T21:45:01 that may be my saving excuse not to take it seriously. 2011-01-14T21:45:27 it's fun :P 2011-01-14T21:46:15 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-01-14T21:46:40 i was describing it to a coworker today, and his office mate didn't realize we were talking about a game. all he heard was me saying something about "if you keep hitting the kid with snowballs, they won't be able to get back up." 2011-01-14T21:52:00 haha 2011-01-14T22:03:49 *** janzert has joined #aichallenge 2011-01-14T22:08:06 interesting, the kids can walk through each other diagonally 2011-01-14T22:18:48 *** Accoun has quit IRC (Ping timeout: 265 seconds) 2011-01-14T22:18:52 *** Accoun has joined #aichallenge 2011-01-14T22:41:43 the collision detection for this game is just messy 2011-01-14T22:41:49 they can pass through each other diagonally or swap squares, but god forbid they try to pass through to same square to get through to another one 2011-01-14T22:46:27 *** amstan_ has joined #aichallenge 2011-01-14T22:46:27 *** ChanServ sets mode: +o amstan_ 2011-01-14T22:48:40 *** sigh has quit IRC (Remote host closed the connection) 2011-01-14T23:00:07 *** chris__0076 has joined #aichallenge 2011-01-14T23:02:54 *** Chris_0076 has quit IRC (Ping timeout: 240 seconds) 2011-01-14T23:06:07 *** chris___0076 has joined #aichallenge 2011-01-14T23:09:33 *** chris__0076 has quit IRC (Ping timeout: 264 seconds) 2011-01-14T23:27:33 *** chris___0076 is now known as chris_0076