2011-06-11T00:04:21 *** antimatroid has quit IRC (Ping timeout: 250 seconds) 2011-06-11T00:17:50 *** antimatroid has joined #aichallenge 2011-06-11T00:20:08 aichallenge: janzert epsilon * ra65ad39 / (worker/jailguard.py worker/sandbox.py): A few more jail fixes - http://bit.ly/lWroOc 2011-06-11T00:36:20 *** xQuasar has joined #aichallenge 2011-06-11T00:47:43 *** deltOr has joined #aichallenge 2011-06-11T00:49:00 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-06-11T01:34:53 *** Mathnerd314 has quit IRC (Read error: Connection reset by peer) 2011-06-11T01:36:52 *** antimatroid has left #aichallenge 2011-06-11T01:40:33 *** Zaphus has joined #aichallenge 2011-06-11T01:46:32 *** antimatroid has joined #aichallenge 2011-06-11T01:50:59 Hey Antimatroid 2011-06-11T01:52:40 antimatroid: are you there ? got a question about scoring for you 2011-06-11T01:56:31 *** Pyronaut has joined #aichallenge 2011-06-11T02:07:12 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Read error: Connection reset by peer) 2011-06-11T02:29:19 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-11T02:50:22 Zaphus: i am now 2011-06-11T02:50:29 ahh, but you aren't :p 2011-06-11T02:57:42 Hi guys, I know I asked yesterday about people building a C# starter package, Just wondering if we still need one since I am part way through finishing up one for my own uses (Simply parsing the game data and storing it in various lists, very basic), or is there already someone assigned for the job. 2011-06-11T03:00:05 pretty sure someone's already filled that hole 2011-06-11T03:00:51 wait, maybe they haven't 2011-06-11T03:00:55 https://github.com/aichallenge/aichallenge/tree/epsilon/ants/dist/starter_bots 2011-06-11T03:00:59 that's where all the current ones are 2011-06-11T03:01:24 so far as I know were still in the state of several having been started but no finished ones have appeared yet 2011-06-11T03:01:27 pyronaut: are you familiar with the "strategy" for starter bots? 2011-06-11T03:02:06 There isn't any current C# starterbots, but on the forums and on git people have "said" that they will do it. But so far no one has. 2011-06-11T03:02:18 I do know someone is on the ladder with a C# bot that I thought possibly could be doing it. 2011-06-11T03:02:21 meh, do it and we'll use yours 2011-06-11T03:02:27 Just asking so I don't step on anyones toes. 2011-06-11T03:02:39 s'all good i think 2011-06-11T03:02:45 if you do, i'm sure they'll get over it ;) 2011-06-11T03:03:05 janzert: did you see the conversation about blocking :\ 2011-06-11T03:03:08 antimatroid, Somewhat, can pull the strategies from the other starter packages. 2011-06-11T03:03:15 damage option allows blocking 2011-06-11T03:03:34 yes, we already knew that though 2011-06-11T03:03:38 in my biased opinion i think the c++ bot is best to base it off :P 2011-06-11T03:03:43 i changed it recently though btw 2011-06-11T03:03:44 I didn't read the conversation 2011-06-11T03:04:00 it allows blocking of corridors of >2 width 2011-06-11T03:04:05 i didn't realise that 2011-06-11T03:04:15 hmm, I thought we know that too 2011-06-11T03:04:25 it's limited to 3 though isn't it? 2011-06-11T03:04:29 no 2011-06-11T03:04:32 at least with the current radius 2011-06-11T03:04:34 no 2011-06-11T03:04:47 :P. Not too familiar with C++ syntax. Will use something that I can copy and paste over and just change the syntax :). Anyway thank you! 2011-06-11T03:04:57 oh, it can block any width? 2011-06-11T03:05:20 that could be bad enough to change 2011-06-11T03:05:22 not sure on the actual water configurations needed, but yeah, there were examples of larger corridors that were blocked 2011-06-11T03:06:25 http://pastebin.com/CjKYcN5J 2011-06-11T03:06:57 note the middle a's are within the battle radius of the top/bottom b's 2011-06-11T03:07:35 imo it's be harder to block with attack2 = 5 again, but even then it was still blockable i think 2011-06-11T03:07:43 but i haven't properly tested that either 2011-06-11T03:08:10 hmm :( 2011-06-11T03:08:23 yeah, that's what i thought 2011-06-11T03:08:36 we basically hope it doesn't stagnate the top games or change the battle option :\ 2011-06-11T03:09:27 mcleopold's who i really want to talk to about it 2011-06-11T03:09:31 and maybe sigh 2011-06-11T03:09:35 i missed sigh earlier :\ 2011-06-11T03:10:07 yep, I'm still trying to stay out of game mechanic decisions if possible :} 2011-06-11T03:10:18 hehe yep, fair enough 2011-06-11T03:12:44 *** qacek has quit IRC (Quit: WeeChat 0.3.5) 2011-06-11T03:14:33 Not to be a pain again, but just FYI, I could not for the life of me get the tools.zip file download to work. It has hardcoded file paths that will only work with the repo, not with how the file structure is in the zip. 2011-06-11T03:19:21 *** xQuasar has quit IRC (Ping timeout: 240 seconds) 2011-06-11T03:19:43 *** Palmik has joined #aichallenge 2011-06-11T03:23:55 *** xQuasar has joined #aichallenge 2011-06-11T03:41:41 *** Quasar has joined #aichallenge 2011-06-11T03:42:07 *** Quasar is now known as Guest54815 2011-06-11T03:44:47 *** xQuasar has quit IRC (Ping timeout: 240 seconds) 2011-06-11T04:33:31 *** UncleVasya has joined #aichallenge 2011-06-11T04:38:28 *** Zaphus has joined #aichallenge 2011-06-11T04:38:32 *** berak has joined #aichallenge 2011-06-11T04:45:47 *** stocha has joined #aichallenge 2011-06-11T04:46:52 if blocking becomes to much of a burden, rather than going back 2 month and endlessly arguing on battle resolution again, refine the water evaporate idea. because it affect map dynamic which isn't settled anyway. And it solves all blocking problems. 2011-06-11T04:47:15 people will still have to block if that is efficient. But can't count on that for winning against all opponents. 2011-06-11T04:47:46 evaporating water is a nice garantee that games will finish also. 2011-06-11T04:48:07 it's certainly harder to have stalemate on a no water map. 2011-06-11T04:49:03 of course you would have to change the sending protocol, wich rely on water never changing. 2011-06-11T04:49:42 but that's not something people will argue endlessly with, like fighting resolution. 2011-06-11T04:49:49 well, my 2 cents in :p 2011-06-11T04:50:26 if only fixed sized (<6) twisted corridor can be blocked, needing a lot of ant, we need to test to see if it is really a problem anyway. 2011-06-11T04:50:41 most block are unidirectionnal also. 2011-06-11T04:51:26 i found a bug in stocha bot. Where it would behave badly with certain map sizes. 2011-06-11T04:52:08 too bad i can't see easilly the map size on the test server :p to find games where the bug hit hard. 2011-06-11T04:52:20 *** stocha has quit IRC (Quit: Page closed) 2011-06-11T04:52:24 *** carlop has joined #aichallenge 2011-06-11T04:59:25 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-11T05:03:11 *** antimatroid1 has joined #aichallenge 2011-06-11T05:03:39 *** antimatroid has quit IRC (Ping timeout: 255 seconds) 2011-06-11T05:05:04 *** Racko has joined #aichallenge 2011-06-11T05:24:25 *** amstan has quit IRC (Remote host closed the connection) 2011-06-11T05:31:41 *** smiley1983 has quit IRC (Ping timeout: 240 seconds) 2011-06-11T05:33:59 *** smiley1983 has joined #aichallenge 2011-06-11T05:39:37 *** mcstar has joined #aichallenge 2011-06-11T05:41:39 *** Accoun has quit IRC () 2011-06-11T05:53:37 *** antimatroid1 has quit IRC (Ping timeout: 260 seconds) 2011-06-11T05:55:16 *** Zaphus has joined #aichallenge 2011-06-11T06:12:25 *** antimatroid has joined #aichallenge 2011-06-11T06:21:58 *** StijnD has joined #aichallenge 2011-06-11T06:23:31 i havent really followed the combat rules discussion so apologies if it's been said before ... but if there were also some chance in the battles ('dice roll' with weighted probabilities depending on the configuration) then that would mean defences could be broken with enough ants attacking, albeit at a high cost ... 2011-06-11T06:24:37 george1914: you would be fine with probabilistic fight fules? 2011-06-11T06:24:46 i couldn't think of much worse personally 2011-06-11T06:24:55 that would be horrible - I'd be blaming the engine for every battle I lost 2011-06-11T06:24:57 hard to debug your bot, hard to tell if someone fairly won etc. etc. 2011-06-11T06:25:33 i'm not fond of disappearing water either 2011-06-11T06:25:41 I see no issue with the defenses. it probably will be the case that someone will work out good defenses - but people will also work out good attack patterns (they exist too) where you can hunt down and kill ants 2011-06-11T06:25:52 well, re won fairly - i rely on the server to get that right - not an issue for contestants 2011-06-11T06:26:03 yes, but what we don't want is everyone expanding then setting up shop with defenses 2011-06-11T06:26:11 that would be really boring and make for long drawn out games 2011-06-11T06:26:34 george1914: yes, but "fairness" is messy without bringing chance into it 2011-06-11T06:26:48 and for debugging - not so hard, is it - i get the map from the game server each go, so it is not any harder - i just have to read the map as usual 2011-06-11T06:27:08 yeah, but what about on small test cases when you are testing the behaviour of your bot 2011-06-11T06:27:16 it'll be much harder to work out what's happened each turn 2011-06-11T06:28:26 otherwise you could have the death rule be something like, if you have enemies in range, your probability of dying is 1-1/noEnemies 2011-06-11T06:28:30 stay away from randomness, keep the battle with set rules. throw in some maps that are mostly open area, the defensive bots will get there a$$es kicked by the attacking ones on those 2011-06-11T06:28:47 Zaphus: yeah i agree, more open maps 2011-06-11T06:29:02 i'll be working on map stuff after the 23rd 2011-06-11T06:29:47 any form of randomness sucks, because you can feel like the dice were rolled against you - it's bad enough that the matchmaker is unpredictable and your ranking is at it's mercy, plus that some starting positions are worse than others (from what I have seen) 2011-06-11T06:30:04 starting positions worse? 2011-06-11T06:30:12 only based on your relative neighbours? 2011-06-11T06:30:19 well, if you play games like risk etc, the randomness is a fine part of the game 2011-06-11T06:30:26 the ranking algorithm is meant to somewhat account for your matchings 2011-06-11T06:30:43 so are asymmetric games in computer games 2011-06-11T06:30:43 rthere will be 1000s of battles per game, so good averaging. 2011-06-11T06:30:59 but taking it to ai and ranking these bots, i prefer more fairness enforced 2011-06-11T06:32:19 the main source of 'unfair' randomness would/will be who you play against and where you start compared to opponents ... whereas unfairness over a few 1000 battles is small 2011-06-11T06:32:48 yeah, true, i still don't like random fighting :P 2011-06-11T06:33:03 anyway, it was just a suggestion, fair enough if yous prefer other options 2011-06-11T06:33:13 I suggest that someone tries to make a defensive bot that can win, to show that there is a problem. the one I've created (Zaphus_blocktest) does well, but that is mostly because it plays the same as 'Zaphus' for the first 200 turns. it then steals some second places due to the blocking because nobody else is doing it yet 2011-06-11T06:33:15 it's something people have suggested a lot 2011-06-11T06:33:22 i was curious at your interest given your placing in pw :) 2011-06-11T06:33:59 i don't mind defense, in fact i think it'll be awesome, provided it isn't the only strategy people use 2011-06-11T06:34:13 i want to see defense structures broken down, people going around them and coming in from the back, etc. 2011-06-11T06:34:42 i see two main pros with chance battle: 2011-06-11T06:34:43 -block are no more possible (defence still have a bonus, cause they have position, but 1000v5 stalemate should not be possible) 2011-06-11T06:34:45 -micro management is way more interesting, cause now i think that almost all 2v1 battle between good bot end with a death for side (that's almost the same of 1v1 battle, so go together is better) 2011-06-11T06:34:54 btw, the 'defense' set up by Zaphus_blocktest is only occasionally real defense, most of the lines it makes are breakable - it was just easier to write that way. it beats dumb attacks every time though 2011-06-11T06:35:24 i'm working on my blocking bot, but i don't have a lot of time, i hope that tomorrow can be ready, but don't know 2011-06-11T06:36:18 I'm going to modify this one sometime this weekend, but it's actually quite hard to find really useful blocks. finding a place that will keep a couple of ants alive to the end is relatively easy, but won't get you a win 2011-06-11T06:36:22 but I see that changing thing now is not a good thing 2011-06-11T06:36:31 *** olexs has joined #aichallenge 2011-06-11T06:37:26 sorry, I want to write that go together sucks now, cause 2v1 end almost the same way of 1v1 2011-06-11T06:37:38 1 death for side 2011-06-11T06:38:39 yeah, you have to move very cleverly to get a 2v1 guaranteed win 2011-06-11T06:39:12 http://aichallengebeta.hypertriangle.com/visualizer.php?game=20224 - this is where blocking will cause problems... my bot wins the map because the leading bot couldnt cope with how many ants I let it have, so it timed out and gave me the bonus :-) 2011-06-11T06:39:53 i haven't investigated yet, but i think that a clever alone ant can alway kill one of the attacker ant, or keep his position 2011-06-11T06:41:07 clever is good - clever with obstacles (water) is very good 2011-06-11T06:42:37 well, i guess if we choose maps that always have big open routes (in addition to blockable routes) then it will be fine - we just dont want 'mazes' that everyone locks up on 2011-06-11T06:43:56 if you had both open and 'closed' maps, then i would make my bot play differently for each one - 'proper' play for the open one and corridor blocking/fighting play for the closed one 2011-06-11T06:45:41 *** carlop has quit IRC (Ping timeout: 240 seconds) 2011-06-11T06:48:04 *** Pyronaut has quit IRC (Ping timeout: 252 seconds) 2011-06-11T06:50:03 *** olexs has quit IRC (Quit: Leaving.) 2011-06-11T06:50:54 *** carlop has joined #aichallenge 2011-06-11T07:10:40 *** Guest54815 is now known as xQuasar 2011-06-11T07:10:50 I think something's wrong with the contest server 2011-06-11T07:10:51 14 2011-06-11 15:10:13 Compile Error: error while compiling submission - Java 2011-06-11T07:10:51 Output file /home/contest/compiled/tmp6w7SJq/bot/MyBot.jar was not created. 2011-06-11T07:11:20 Did it run out of disk space or something? 2011-06-11T07:14:20 that's another problem with defense structures 2011-06-11T07:14:24 the games will always go to max turns 2011-06-11T07:15:03 antimatroid, any idea what my error is about? 2011-06-11T07:15:10 nope sorry 2011-06-11T07:15:35 Who should I ask about it 2011-06-11T07:15:36 :o 2011-06-11T07:16:09 amstan, mcleopold maybe? 2011-06-11T07:16:14 probably amstan 2011-06-11T07:16:16 not really sure 2011-06-11T07:16:38 ok. thanks 2011-06-11T07:17:45 *** carlop has quit IRC (Ping timeout: 255 seconds) 2011-06-11T07:20:08 *** Quasar has joined #aichallenge 2011-06-11T07:20:34 *** Quasar is now known as Guest91645 2011-06-11T07:23:47 antimatroid: did you see githup issue #181? about fixing the c++ starterbot, now that we do not get the 'r' message when food goes away? (iirc you were the person to ask about this?) 2011-06-11T07:24:05 *** xQuasar has quit IRC (Ping timeout: 276 seconds) 2011-06-11T07:25:55 *** antimatroid has quit IRC (Ping timeout: 250 seconds) 2011-06-11T07:29:42 Is anyone else getting this kind of error (java) ? 2011-06-11T07:29:43 Output file /home/contest/compiled/tmpmlea3E/bot/MyBot.jar was not created. 2011-06-11T07:30:14 *** george1914 has quit IRC (Ping timeout: 252 seconds) 2011-06-11T07:32:24 *** locutus2 has joined #aichallenge 2011-06-11T07:34:44 *** FireFly has joined #aichallenge 2011-06-11T07:41:00 *** antimatroid has joined #aichallenge 2011-06-11T07:43:47 *** Eruonen has joined #aichallenge 2011-06-11T07:56:12 *** Guest91645 is now known as xQuasar 2011-06-11T07:59:58 *** xQuasar has quit IRC () 2011-06-11T08:03:42 *** dlila has joined #aichallenge 2011-06-11T08:05:26 *** locutus2 has quit IRC (Ping timeout: 252 seconds) 2011-06-11T08:05:29 *** george1914 has joined #aichallenge 2011-06-11T08:06:03 antimatroid: Let's suppose you want quickly evaluate your bot's new version strength. You have played 30 games. What is more important for you to see: average place, average scores, total scores, times you was eliminated, something else? 2011-06-11T08:06:03 Anyone's else answers are welcome. 2011-06-11T08:06:35 i test my bots behaviour with small localised examples 2011-06-11T08:06:53 then i might consider my strategy after watching it play games 2011-06-11T08:06:59 and see where it could be done better 2011-06-11T08:07:06 then try to alter my strategy based off that 2011-06-11T08:07:42 an easyish way to run your bot locally is to just feed "made up input" to it and see what moves it makes 2011-06-11T08:11:49 *** StijnD has quit IRC (Read error: Connection reset by peer) 2011-06-11T08:15:54 *** stocha has joined #aichallenge 2011-06-11T08:24:36 *** locutus2 has joined #aichallenge 2011-06-11T08:24:45 we need to activate the java bot, to judge blocking. Most strong bot are concerned with the java loading problem 2011-06-11T08:25:07 interesting, I havent done that (feed made-up situations). wish we could dump one frame of the replay into a set of input files ready to go :-) 2011-06-11T08:25:11 from the given link, i doubt that zaphus blocker will fare well. 2011-06-11T08:25:24 i think he would loose against himself non blocking :p 2011-06-11T08:26:04 indeed, it's not so much blocking as 'awakening the idea of blocking for other people' 2011-06-11T08:26:27 have you tryed blocker zaphus against non blocker zaphus locally ? (or even on the server after all, but that will be slow) 2011-06-11T08:26:28 it doesnt even try to do proper blocks, just roughly close to blocking - I wrote it at 1am last night, close enough 2011-06-11T08:26:48 fair enough 2011-06-11T08:27:11 the blocker version will always lose, since the non-blocker will beat it's score. the only win it gets against Zaphus is when my main bot times out because it has so many ants :-) 2011-06-11T08:27:49 e.g. this one: http://aichallengebeta.hypertriangle.com/visualizer.php?game=20074 (replay has the wrong final order of bots, the blocker wins that one) 2011-06-11T08:28:01 well, at some point score was supposed to meseare one ability go fight with good ration. Blocker is supposed to get a good ratio. 2011-06-11T08:29:41 yeah, lots of stuff that you 'should' do - and if people do it then good on them I reckon. the odds of being able to get far enough in front to block and win, but you would have lost otherwise, seem slim. more likely blocking is how the other bots fill out the lower places :-) 2011-06-11T08:30:26 i don't know 2011-06-11T08:30:32 i think in some situation blocking will be good 2011-06-11T08:30:35 if used well 2011-06-11T08:30:42 but the question is : how to use it well ? 2011-06-11T08:30:55 and then you need to be able to do it in the first place. 2011-06-11T08:31:34 and to find the situation - I think the combination will be hard to do . right now suicide (1-1) attacks are a perfectly effective strategy once you have a lead 2011-06-11T08:34:05 stocha: you helped convince me of the spawn points! :P 2011-06-11T08:34:45 Zaphus: I wrote the c++ starter bot without an engine, just feeding fake data to it :P 2011-06-11T08:35:08 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-11T08:35:11 antimatroid: Nicely done ! 2011-06-11T08:40:23 *** Naktibalda has joined #aichallenge 2011-06-11T08:41:13 *** stocha has joined #aichallenge 2011-06-11T08:43:45 oh my god. there is no C start bot ?? 2011-06-11T08:44:19 anyone with some skill in c to write a base for me ? :p 2011-06-11T08:44:32 i wanted to rewrite the stocha bot in c. 2011-06-11T08:44:47 (so i can run in the messy test environnement :p ) 2011-06-11T08:45:20 i'll have a look at the cpp i guess 2011-06-11T08:47:04 i guess i should try to write a c starter. Then people can point out what i did wrong. hopefully :p 2011-06-11T08:47:34 mmm, where is the starter speck ? 2011-06-11T08:48:10 *** delt0r_ has joined #aichallenge 2011-06-11T08:48:15 *** olexs has joined #aichallenge 2011-06-11T08:49:02 stocha: use the c++ bot :P 2011-06-11T08:49:21 c++ can be wonderful if you don't need to use some features 2011-06-11T08:50:00 *** deltOr has quit IRC (Ping timeout: 268 seconds) 2011-06-11T08:50:15 is there a starter bot testcase ? 2011-06-11T08:50:31 stocha: https://github.com/aichallenge/aichallenge/wiki/Ants-Starter-Pack-Guide 2011-06-11T08:51:08 https://github.com/aichallenge/aichallenge/wiki/Ants-Strategy-Guide 2011-06-11T08:51:19 there is that for anyone to look at/make suggestions for at some point too 2011-06-11T08:52:03 i guess some people like myself will want a pure C stuff. 2011-06-11T08:52:45 beside, if you do it in c, you don't want it to show as "c++" on the langage info :p 2011-06-11T08:52:52 i thought someone did a c bot 2011-06-11T08:52:58 but maybe they never finished/uploaded it 2011-06-11T08:53:02 i certainly can't seems to find it 2011-06-11T08:54:01 wow starter bot is a pretty evolved thing. 2011-06-11T08:54:10 i guess i'll do something simpler to start with. 2011-06-11T08:54:41 like : pick a random direction for each ant, with no care. 2011-06-11T08:57:58 evolved how? 2011-06-11T08:58:02 antimatroid: sorry if i missed your reply, but did you see github issue #181? i was wanting to ask you about it, if i could ... 2011-06-11T08:58:10 ? 2011-06-11T08:58:56 george1914: i think i've updated this 2011-06-11T08:58:57 well, it seems that the 'r' is not sent any more 2011-06-11T08:59:02 ah, right 2011-06-11T08:59:09 yeah, the c++ bot got updated earlier this week 2011-06-11T08:59:22 yes 2011-06-11T08:59:54 but the current c++ download still has the 'r' line in it 2011-06-11T09:00:14 and (as of yesterday) i thought it was getting the food locations wrong 2011-06-11T09:00:22 *** StijnD has joined #aichallenge 2011-06-11T09:00:29 from spawning (both in and out of vision) 2011-06-11T09:00:39 what do you mean it still has the 'r' line? 2011-06-11T09:00:41 evolved like you have to remember water :p 2011-06-11T09:00:50 in State.cc 2011-06-11T09:00:54 https://github.com/aichallenge/aichallenge/blob/epsilon/ants/dist/starter_bots/cpp/State.cc 2011-06-11T09:00:55 and then test if a direction go to water of food. 2011-06-11T09:00:56 it's not there 2011-06-11T09:01:12 stocha: you have to remember the water as you're only told it once 2011-06-11T09:01:13 1 minute ... (checking again) 2011-06-11T09:01:36 george1914: i updated it only a few days ago 2011-06-11T09:01:42 i had been slack, sorry :) 2011-06-11T09:01:58 i also fixed the timer so it will work on all windows compilers (hopefully) now 2011-06-11T09:02:22 but if someone uses the windows timer on linux, they'll get clock time rather than wall time 2011-06-11T09:02:32 hopefully people listen to my comment warning 2011-06-11T09:02:44 i just downloaded again ... i see else if(inputType == "r") //removal of food square 2011-06-11T09:02:59 is it the case that your new c++ has not been uploaded? 2011-06-11T09:03:11 are you donwloading from the beta site? 2011-06-11T09:03:16 that probably hasn't been updated 2011-06-11T09:03:16 yes 2011-06-11T09:03:20 donwload from github 2011-06-11T09:03:22 ah, my error ... 2011-06-11T09:03:31 right, ok 2011-06-11T09:03:33 not really your error, just confusion ;) 2011-06-11T09:03:36 i will try that, thanks 2011-06-11T09:03:55 certainly i have been confused ... :) 2011-06-11T09:04:19 how do people emit debug information from a bot ? 2011-06-11T09:04:34 like i want to see what the engine send to the bot :p 2011-06-11T09:04:41 i submitted an issue 'that's not my bot' because it was clearly not moving as mine did in games - turn out it was chasing phantom food ... :) 2011-06-11T09:05:09 does anyone have visual c++ installed on windows? 2011-06-11T09:05:22 george1914: yeah, sorry about that :P 2011-06-11T09:05:30 i've known about the problem since 'r' lines got removed and been slack 2011-06-11T09:05:44 i was waiting to do it when i fixed the timer, and was slack about that to be precise 2011-06-11T09:05:54 no problem, thanks for working on the contest, and for replying to my questions ... :) 2011-06-11T09:06:08 np 2011-06-11T09:06:49 i dont know this stuff so well, but i have seen preprocessor stuff to distinguish windows/linux specific code in c++ - might be how you neatly beat the 2 timers problem ... 2011-06-11T09:07:06 did i see something about there being an issue from you about it? 2011-06-11T09:07:11 is that still open? I might close it 2011-06-11T09:07:59 yeah that's what i'm doing with the timer now, see here https://github.com/aichallenge/aichallenge/blob/epsilon/ants/dist/starter_bots/cpp/Timer.h 2011-06-11T09:08:02 me? issue #181 is the only one open related to me 2011-06-11T09:08:42 how do one store a variable size list in c ? 2011-06-11T09:08:59 the test server is current very slow, their are played very few games in the last hour 2011-06-11T09:09:05 (that's not promising in now :p ) 2011-06-11T09:09:07 "ps: george1914 posed this question on #aichallenge 10 june" that was the comment i saw, all good 2011-06-11T09:09:18 ok, good 2011-06-11T09:09:27 anyone now c in there by any chances ? 2011-06-11T09:09:37 is their a problem or make the admins changes on the server 2011-06-11T09:09:42 i only know c++ 2011-06-11T09:09:50 i can flounder around in python 2011-06-11T09:10:45 the first python program i wrote was the map gen for pw 2011-06-11T09:11:34 stocha: why you using c? is this for learning? if not then i would just use c++ if i were you - should be almost as fast etc 2011-06-11T09:11:40 every parameter in the starting phase is given as plain integer ? 2011-06-11T09:11:50 stocha: i can make an c starter bot based on the c++ starterbot 2011-06-11T09:11:52 *** Frontier has joined #aichallenge 2011-06-11T09:11:59 stocha: I am happy to field any questions when you don't understand c++ stuff :P 2011-06-11T09:12:13 george1914: well, c is more straight forward :p And yes it's also for "learning" 2011-06-11T09:12:21 (i do not know c btw - but am reliably informed that c++ will be about as fast in most cases) 2011-06-11T09:12:32 i find c++ easier than c 2011-06-11T09:12:36 locutus2: how much time would it take you ? 2011-06-11T09:12:43 very rarely need to use pointers or dynamic memory 2011-06-11T09:12:50 in fact the only time i do is for game tree's 2011-06-11T09:13:05 otherwise I just pass references around etc. 2011-06-11T09:13:14 george1914: c++ allow you to write plain c. but does not always run on plain c compilation 2011-06-11T09:13:28 stocha: code::blocks with mingw compiler! 2011-06-11T09:13:38 antimatroid: uh ? 2011-06-11T09:13:45 c++ <3 :P 2011-06-11T09:14:02 C++ is just a way to make C langage messy :p 2011-06-11T09:14:05 even more messy 2011-06-11T09:14:14 only if you're doing it wrong 2011-06-11T09:14:41 mm, i'm not convinced myself. it's just that c is older, and still used in a lot of tasks 2011-06-11T09:15:04 and c is a subset of c++ afterall 2011-06-11T09:15:28 beside, there were c bots in previous contests. 2011-06-11T09:15:32 i think 3-4 hours (must implement some stuff like queue which in c++ is standard) 2011-06-11T09:15:48 george1914: have you squared your euclidean metric locally? 2011-06-11T09:15:56 i try the c starter bot 2011-06-11T09:15:58 stocha: as you wish :) i think people have all kinds of opinions on this, go with what you find best - me, i like the 'fancier' stuff in c++, and dislike what little i know of the ugly/raw stuff in c - but go with what you like. 2011-06-11T09:16:00 that way all your distance calculations are integers and no square root calculations 2011-06-11T09:16:01 locutus2: maybe i should try to do it, for learning, then you can review ? 2011-06-11T09:16:13 *** Vio has joined #aichallenge 2011-06-11T09:16:27 antimatroid: ? how do you mean? elaborate please ... 2011-06-11T09:16:46 ah, for speed? 2011-06-11T09:16:49 yeah 2011-06-11T09:16:57 i dopnt recall 2011-06-11T09:17:09 how do you scanf the input ? 2011-06-11T09:17:14 stocha: thats a good idea and i can review your code. In past i am used alot c but later i switch to c++ for oop 2011-06-11T09:17:24 stocha: https://github.com/aichallenge/aichallenge/blob/f444f48a75fe2b866602c4de27dea643b5aeefd8/ants/dist/starter_bots/cpp/State.cc 2011-06-11T09:17:28 for planetwars, i wrote some ok code, then later optimised using profiler 2011-06-11T09:17:29 that's the state file for c++ 2011-06-11T09:17:48 something about 'premature optimisation is the root of all evil' ... 2011-06-11T09:18:29 *** StijnD has quit IRC (Ping timeout: 276 seconds) 2011-06-11T09:18:37 yeah optimization is a nasty beast. It will come to haunt you, if you don't plan for it. And even more so if you do plan for it :p 2011-06-11T09:18:54 meh, i plan my strategies out well in advanced to progamming 2011-06-11T09:18:57 Maybe optimization is a Zen concept. 2011-06-11T09:19:02 ... most code i wrote was not critical for speed - the 10% or so that was, i improved later 2011-06-11T09:19:06 i plan my optimisations into my strategy 2011-06-11T09:19:48 antimatroid: trouble is, i can't read others code. So i'm starting from the spec. 2011-06-11T09:20:02 yeah i can't read other people's code either 2011-06-11T09:20:24 can someone add to the spec, the fact that the values are plain integer ? 2011-06-11T09:20:39 antimatroid: in most things i have written, it has been the ideas/plan and correct implementaion that are critical 2011-06-11T09:20:41 "viewradius2" # view radius squared etc 2011-06-11T09:20:43 erm silly me, i have "std::"'s on that page, when i don't need them 2011-06-11T09:20:55 spped came low down in importance 2011-06-11T09:21:09 george1914: you obviously didn't participate in tron? :P 2011-06-11T09:21:19 and is -usually- easyish to improve if needed 2011-06-11T09:21:27 i was in tron 2011-06-11T09:21:31 31st i think 2011-06-11T09:21:34 tron was all about who could squeeze the most out of a game tree 2011-06-11T09:21:40 *** _0x47 <_0x47!~slash@p5B12F1F8.dip.t-dialin.net> has quit IRC (Read error: Connection reset by peer) 2011-06-11T09:21:45 but i didnt do alpha-beta :( 2011-06-11T09:21:46 in a lot of case, i was haunted back for not caring for optimization in the architecture. i would have to rebuild the architecture from scratch. 2011-06-11T09:21:54 i used nash equilibrium 2011-06-11T09:21:59 and yes, you are right there 2011-06-11T09:22:02 *** _0x47 <_0x47!~slash@p5B12F1F8.dip.t-dialin.net> has joined #aichallenge 2011-06-11T09:22:16 could you explain your strategy? 2011-06-11T09:22:24 and in a lot of case, i was haunted back for caring to much with optimization in the architecture. I would have to rebuild the archicture from scratch. 2011-06-11T09:22:29 maybe i'm a bad architect :p 2011-06-11T09:22:37 george1914: http://csclub.uwaterloo.ca/contest/profile_games.php?user_id=2788 we got disqualified for timing out :( 2011-06-11T09:22:46 i generrally have managed to have modular stuff ... so i can change individual functions/classes at later points, so easier to optimise specific parts ... 2011-06-11T09:23:00 we were ~10th at the time 2011-06-11T09:23:14 (did that one with a friend 2011-06-11T09:23:35 ah, you were one of the unfortunates? i am sorry ... very bad luck 2011-06-11T09:23:46 is this friend participating now? 2011-06-11T09:23:55 i tried to only use <0.5s because i was so scared of timing out ... 2011-06-11T09:24:01 UncleVasya: they didn't in tron, i doubt they'll find time for ants 2011-06-11T09:24:10 i didn't realise it was disqualificaiton 2011-06-11T09:24:18 we didn't time out on the official server until that happened 2011-06-11T09:24:24 it was incredibly devestating 2011-06-11T09:24:29 i bet 2011-06-11T09:24:54 so, are you now an organiser rather than participant? 2011-06-11T09:25:04 i am a contributer 2011-06-11T09:25:06 i can compete still 2011-06-11T09:25:11 aha, good :) 2011-06-11T09:25:14 locutus2: what about i use a chained list for storing stuff issued from the engine ? 2011-06-11T09:25:33 only people who can't compete are those with access to the main servers 2011-06-11T09:25:40 what i can do is all open source 2011-06-11T09:25:48 ok, i see 2011-06-11T09:25:49 rather than trying my hand at resizing arrays 2011-06-11T09:26:01 stocha: just learn c++ :P 2011-06-11T09:26:04 stl is your friend 2011-06-11T09:26:30 maybe realloc is good enougth 2011-06-11T09:26:54 mm, i think i will use chained list 2011-06-11T09:26:57 i agree with antimatroid, stocha :) 2011-06-11T09:27:00 c++ is almost my perfect language 2011-06-11T09:27:32 http://advice.cio.com/esther_schindler/linus_torvalds_why_c_sucks 2011-06-11T09:27:35 if i could use templates in .cc files, overload anything as an operator, overload operator precedence, have a better type system and treat functions like variables, then i would be in heaven 2011-06-11T09:27:49 i love the syntax 2011-06-11T09:29:15 also, there's some weird thing that if you don't use templates then you have to do <> functions in .cc instead of .h otherwise it wont compile :\ 2011-06-11T09:29:27 i still don't understand why that happens, george1914 any idea? 2011-06-11T09:30:08 oh, and another gripe is the fact that vectors of bools have some weird type that wont let you return a reference to an element of one 2011-06-11T09:30:11 sorry, way beyond my level ... i hardly used templates 2011-06-11T09:30:27 which means you can't reference a matrix struct of bools with a location, it doesn't work! 2011-06-11T09:30:47 i am/was physicist, i write fortran (with data encapsulation) in c++ :/ 2011-06-11T09:30:59 how old are you? 2011-06-11T09:31:06 antimatroid: re timeouts, being disqualified for them really sucks. is that the plan for this one or people just lose the one match they timeout on ? it's almost impossible to code for timeouts against the wall-clock perfectly 2011-06-11T09:31:06 ancient 2011-06-11T09:31:11 1914 old? 2011-06-11T09:31:16 :) 2011-06-11T09:31:24 no 1977 2011-06-11T09:31:34 Zaphus: I didn't shut up at the start of planet wars until that unecessarily harsh rule was changed ;) 2011-06-11T09:31:52 1977 ? wow, now I feel really old.. I wrote my first code in '79 2011-06-11T09:32:06 * antimatroid was born in 88 :P 2011-06-11T09:32:28 i used to work with a guy on uk c++ standards committe - he got very excited about discussions like your there - but it was mostly greek to me 2011-06-11T09:32:31 george1914: do you know much game theory? 2011-06-11T09:32:33 owner is always an int ? 2011-06-11T09:32:36 you make me feel old ... :( 2011-06-11T09:32:43 https://github.com/antimatroid/GTL 2011-06-11T09:32:48 i have that which i'm slowly working on 2011-06-11T09:32:50 me? i know nash equilibrium stuff 2011-06-11T09:33:05 used it with tree for tron 2011-06-11T09:33:08 i have extensive form game stuff too, but it's not up on github 2011-06-11T09:33:17 i'm old too. in 1979 i was already born ! 2011-06-11T09:33:27 instead of usual minimax+alpha-beat 2011-06-11T09:33:33 sorry, alpha-beta 2011-06-11T09:33:46 was this on a game tree? 2011-06-11T09:33:48 or game matrix? 2011-06-11T09:34:07 we didn't use alpha-beta with our bot 2011-06-11T09:34:19 we had a pretty sweet payoff function though 2011-06-11T09:34:24 *** carlop has joined #aichallenge 2011-06-11T09:34:38 why not alpha-beat? gave you 2times depth, right? 2011-06-11T09:35:18 generating my tree is the costly part, evaluating it isn't that expensive, it's not alright to just prune with alpha beta while generating 2011-06-11T09:35:33 i would probably use it in the future when evaluating the tree, but i was a complete noob back then 2011-06-11T09:35:55 *** Vio has quit IRC (Ping timeout: 250 seconds) 2011-06-11T09:35:57 i would try to use a* search too 2011-06-11T09:36:08 the nash? well i searched eg 10moves deep on the game tree, but moves were simultaneous, so each move had a payoff matrix, with maybe mixed strategy as the nash equilibrium ... 2011-06-11T09:36:23 mixed strategy? 2011-06-11T09:36:33 that would have been expensive to computer, you need to solve a linear program 2011-06-11T09:36:36 hm,, maybe i am getting my terms wrong 2011-06-11T09:36:42 i mean 2011-06-11T09:36:49 eg 2 players are close to each other 2011-06-11T09:37:08 so on next move, we might eg crash, or 1 traps the other, etc 2011-06-11T09:37:28 best strategy can be to make some move with certain probability, right? 2011-06-11T09:37:42 there is not a unique best move for me to make 2011-06-11T09:38:17 yes, it took a little code to do it, and some cpu cost 2011-06-11T09:38:29 but i did that at each node in the game tree 2011-06-11T09:38:36 we broke our payoff function into something like the following lost < separating with less squares < together with them being able to access more squares first < draw < separate with the same number of squares < together with the same number of squares < together with more squares accessible first < win 2011-06-11T09:38:51 i am unsure whether that is what i do here, i suspect i should do ... 2011-06-11T09:39:28 hm 2011-06-11T09:39:59 and we'd also recycle part of our tree each turn 2011-06-11T09:40:03 omg there are replays with 4000 ants now Oo 2011-06-11T09:40:11 are there default values for struct field in c ? 2011-06-11T09:40:16 ah, now that recycling sounds clever 2011-06-11T09:40:18 no just kiddin 2011-06-11T09:40:18 *** StijnD has joined #aichallenge 2011-06-11T09:40:27 at the end of the turn, we'd pick out move, then make that the tree and delete the rest, then at the start of the turn, work out their move and do the same thing, then keep generating the tree from there 2011-06-11T09:40:37 like if a struct contains a int x, can you assign a default value to x ? 2011-06-11T09:40:53 probably not. .... mm c sucks :p 2011-06-11T09:41:12 stocha: c++ is your friend 2011-06-11T09:41:21 no, linus said it was bad. 2011-06-11T09:41:26 linus is a fool 2011-06-11T09:41:29 i'll take his word on that. 2011-06-11T09:41:42 no i like him. since he said that c++ was bad. 2011-06-11T09:41:43 * antimatroid ducks 2011-06-11T09:42:29 is a single file starter c bot a good thing ? 2011-06-11T09:42:40 terrible idea if you want it to be useful 2011-06-11T09:42:45 my tron bot was single file :P 2011-06-11T09:42:47 it was awful 2011-06-11T09:43:01 well, it's not a bot. it's the basis of a starter bot. 2011-06-11T09:43:47 george1914: templates really are wonderful 2011-06-11T09:43:52 stocha: is this a cunning plan of yours, to sabotage the other contestants by writing questionable starter bots? :) 2011-06-11T09:44:35 what do you use them for? 2011-06-11T09:44:46 eg in this contest, how would they help? 2011-06-11T09:45:06 i mean, i use them in eg std::vector<> etc, but to write them? 2011-06-11T09:45:55 well matrices for example? 2011-06-11T09:45:58 antimatroid: ah, that reminds me - i got error in g++ from your previous c++ starterbot - it didnt like NULL - changing to '0' instead seemed to be ok ... 2011-06-11T09:46:12 NULL where? 2011-06-11T09:46:24 1 moment ... 2011-06-11T09:46:55 what's the best way of making a program break out of 32bits vs 64bits ? 2011-06-11T09:46:57 in c. 2011-06-11T09:47:23 i mean, using int and long in structure will not do that, if you use sizeof right ? 2011-06-11T09:48:15 * antimatroid wishes there were more game theory people hanging around, hmph :( 2011-06-11T09:48:27 in timer.h 2011-06-11T09:48:29 NULL 2011-06-11T09:48:44 oh, hmmmm 2011-06-11T09:48:51 what in game theory are you wondering about? 2011-06-11T09:48:55 that's always worked for my, but i've always used mingw 2011-06-11T09:49:08 nah, it's just one of my main interests :P 2011-06-11T09:49:12 *** StijnD has quit IRC (Ping timeout: 252 seconds) 2011-06-11T09:49:20 i'm doing my honours thesis on game theory, and have that library 2011-06-11T09:49:30 ah, right 2011-06-11T09:50:48 maybe i dont know enough to know that there are important things i dont know ... but making a payoff matrix and deciding what to do seemed to be about it. mixed strategies seemed to be as far as i needed to go ... 2011-06-11T09:51:15 for tron? 2011-06-11T09:51:28 yes, for tron 2011-06-11T09:51:36 but i think also for most other problems ... 2011-06-11T09:51:56 stocha: have you looked on github? There is a c starterbot (ok, dump). 2011-06-11T09:52:28 minimax on a zero sum game tree gives you the best worst outcome when they are playing optimally 2011-06-11T09:52:54 if they play less than optimally, you do at least as well, probably better, this condition doesn't hold for non-zero sum games 2011-06-11T09:52:58 which ants is not 2011-06-11T09:53:23 aha? now that is something i could have done with knowing for tron ... 2011-06-11T09:53:45 it's a bit iffy when you're treating the game as sequential rather than simultaneous 2011-06-11T09:53:51 but imo it's fairly sound in most instances 2011-06-11T09:54:03 but, can i not improve on minimax for simultaneous? my guess was/is that you could 2011-06-11T09:54:10 i believe people came up with some examples where you might be able to game the fact that the bot was using minimax 2011-06-11T09:54:17 ah, so you say it is 'almost the best' ? 2011-06-11T09:54:21 i think a main part of this contest is resolve rock paper scissor situation modeling other player behavior, not solving the game against an optimal opponent 2011-06-11T09:54:34 well define best? 2011-06-11T09:54:39 look at chess, you can't solve that 2011-06-11T09:54:40 in this game, i am very unsure what to do 2011-06-11T09:54:48 not with our current constraints computationally 2011-06-11T09:54:54 yeah, i have no idea for this 2011-06-11T09:55:01 given a payoff function, what is best move? 2011-06-11T09:55:15 i considered partitioning the ants up into sets that can affect each other then game treeing it, but toooooooooooooo^99999 many states 2011-06-11T09:55:33 yes, i was wondering about that 2011-06-11T09:55:40 i would make the move that maximised my worst outcome for each combination of moves from the other ants 2011-06-11T09:55:41 hard to partition though ... 2011-06-11T09:56:08 not too hard, do a bfs from each ant collecting and continuing searches as you find other ants 2011-06-11T09:56:29 the hardest part of not adding ants that can't affect each other because of water on either end of your search 2011-06-11T09:56:35 i still haven't worked out a neat way to do that 2011-06-11T09:56:55 hm - i meant: if i have 2 ants far apart, then kind of ... 2011-06-11T09:57:04 it is 2 cases of 1-ant decisions 2011-06-11T09:57:12 not a 2-ant decision 2011-06-11T09:57:30 so 2*5 possible moves 2011-06-11T09:57:36 instead of 5*5 possible moves 2011-06-11T09:58:14 if you have 5 ants with 4 moves each, that's 4^5 states to look at 2011-06-11T09:58:32 5^5 (can stand still) 2011-06-11T09:58:43 but yes 2011-06-11T09:58:52 yeah, i'm assuming some might be next to water or something :P 2011-06-11T09:58:59 ok, fine 2011-06-11T09:59:04 that's not that bad 2011-06-11T09:59:07 1024 states 2011-06-11T09:59:17 and 5 independent ants is 5*4 2011-06-11T09:59:19 you have to do battle resolution on the terminal n odes though 2011-06-11T09:59:45 you could game tree some battles then heuristic the rest 2011-06-11T09:59:53 and limit the max size of a partition 2011-06-11T10:00:01 yes 2011-06-11T10:00:14 so for me, i think that good partitioning is the hard part 2011-06-11T10:00:26 dont know if/how well i can do it 2011-06-11T10:00:42 yeah, do you go for ants close to the battles or far out first? 2011-06-11T10:01:00 what split do you have between your ants and others ants in a partition 2011-06-11T10:01:10 and in an area that is all battles, what to do? 2011-06-11T10:01:22 also, you can do backward induction with this 2011-06-11T10:01:38 but i think it is a good avenue to explore ... 2011-06-11T10:01:44 backward induction? 2011-06-11T10:02:00 just at each node, bring up say your worst outcome at each of an opponents nodes, then your best outcome at each of your nodes, then go back through and select the optimal moves at each of your nodes afterwards 2011-06-11T10:02:28 so minimax? 2011-06-11T10:02:32 not quite 2011-06-11T10:02:49 minimax would have you bring up the worst outcome to you from the optimal moves of them at each node 2011-06-11T10:03:01 you are now bringing up the worst regardless of what they are doing 2011-06-11T10:03:12 so dropping the assumption that they are rational/optimal opponents 2011-06-11T10:03:19 ah 2011-06-11T10:03:31 btw, do you know much about the contest details? 2011-06-11T10:03:33 if you wanted to get really complex, you could tit for tat who you wanted to cooperate with/who plays optimmally 2011-06-11T10:03:38 depends what 2011-06-11T10:03:49 ah, tit for tat :) 2011-06-11T10:03:57 that would/will be interesting :) 2011-06-11T10:04:02 wont* :P 2011-06-11T10:04:12 it would be awesome, but no way trying to cooperate will payoff in this 2011-06-11T10:04:14 re contest ... 2011-06-11T10:04:30 well, it might do ... 2011-06-11T10:04:42 do you know eg the 2-dimensional prisoners dilemma 2011-06-11T10:04:49 Is it worth updating the starter bot to watch the timer and avoid timeouts, so people can see how to do it (like the fact that you need to set it to ~half the supplied turn value if prev. contests are anything to go by 2011-06-11T10:05:08 i already show them how to check how long they've taken 2011-06-11T10:05:11 or even the round-robin PD, that tit for tat won? 2011-06-11T10:05:12 they should be able to work it out from there 2011-06-11T10:05:57 ah, yes you do. 2011-06-11T10:06:04 you only get cooperation being optimal for repeated prisoners dilemma if it's infinitely repeated or the end is probabilitically determined (pseudo infinite) 2011-06-11T10:06:24 probabilistically* 2011-06-11T10:06:38 ie if the end time is not known ... 2011-06-11T10:07:04 who wrote that win32 timer function ? 2011-06-11T10:07:06 yeah, so the game ends with some probabilty, not sure what the requirements (if there are any) on the probability required for cooperation 2011-06-11T10:07:10 but here we are kind of repeated - 100 moves in a game 2011-06-11T10:07:15 Zaphus: I went googling :P 2011-06-11T10:07:41 antimatroid: I thought these comps used wall clock time (much to my hatred) ? that code gives CPU time 2011-06-11T10:08:00 Zaphus: it gives wall clock on windows 2011-06-11T10:08:15 oh, ok 2011-06-11T10:08:15 it's to everyone's hatred that we use wall clock, blame windows 2011-06-11T10:08:24 why blame windows ? 2011-06-11T10:08:26 it's too hard to use clock time with cross platform 2011-06-11T10:08:37 we've looked at doing it numerous times and never found a way 2011-06-11T10:08:49 so you run servers on different platforms ? 2011-06-11T10:08:53 no 2011-06-11T10:08:56 but people test on windows 2011-06-11T10:09:05 *** paratrechina has joined #aichallenge 2011-06-11T10:09:10 it's a cross platform contest, the servers are all ubuntu i think 2011-06-11T10:09:17 that should be irrelevant, they also test on slower and faster machines than the server 2011-06-11T10:09:48 if you used CPU time, it would at least be consistent between one run and the next on the server, I found it was inconsistent in PW 2011-06-11T10:10:15 yeah, i don't like how the servers were used for planet wars games 2011-06-11T10:10:20 it will hopefully be better with this 2011-06-11T10:10:24 george1914: if the end is guaranteed after a fixed number of turn, cooperation is not an equilibrium (don't matter if there are 10 or 100K move left), but for ants I don't know how many intelligent bot there are in final contest, that should understand if their opponent are collaborating or not 2011-06-11T10:10:45 we are now "pausing" the processes of bots while they aren't picking their moves, and only one bot is picking moves on a processor at a time 2011-06-11T10:11:02 carlop: why would I collaborate with another bot ? 2011-06-11T10:11:05 previously you had 1 second shared with a bot, so worst case 0.5 seconds - other stuff 2011-06-11T10:11:34 antimatroid: thats a big change... didnt realise, so I can't do any processing between turns (never did, but it was theoretically possible in PW) 2011-06-11T10:11:47 yep, hopefully we've fixed that ;) 2011-06-11T10:12:03 carlop: if you, me, and antimatroid write 1 each, then we have a nucleus :) 2011-06-11T10:12:28 Zaphus: try to collaborate with me.... you don't need to defend border agaisnt me, trust me :p 2011-06-11T10:12:39 it's not about cooperating with certain other bots, it's about cooperating when it's mutually beneficial in that game for you to 2011-06-11T10:12:47 yes 2011-06-11T10:13:05 I guess so... again, good luck to anyone who manages that! 2011-06-11T10:13:15 yep, that's what i've concluded 2011-06-11T10:13:19 and decided not to cooperate 2011-06-11T10:13:31 i don't see it paying off 2011-06-11T10:13:35 seriously, sometimes is better cooperate for ending 1 and 2 than kill your enemy (and kill your ant too) on the first few turn 2011-06-11T10:13:52 i'm more referring to cooperating with a battle to kill an enemy 2011-06-11T10:13:58 rather than perhaps a macro cooperation 2011-06-11T10:14:01 i'm talking micro 2011-06-11T10:14:22 antimatroid: since you like game theory - do you know axelrod's prisoners' dilemma tournaments? 2011-06-11T10:14:26 where you need to assume they can play reasonably optimally, which i'm rather concerned about my own ability to do :P 2011-06-11T10:14:34 george1914: no? 2011-06-11T10:14:55 well, they were where i first came across g.t. 2011-06-11T10:15:01 maybe have a read on it 2011-06-11T10:15:09 I think that both macro and micro cooperation are almost impossible, but we will see 2011-06-11T10:15:23 the tit-for-tat won, and won the rematch 2011-06-11T10:15:26 i tend to agree 2011-06-11T10:15:31 of course it did :P 2011-06-11T10:15:34 even against 'complex' opponents 2011-06-11T10:15:43 it was a shock at the time 2011-06-11T10:15:49 why? 2011-06-11T10:15:56 why a shock? 2011-06-11T10:16:07 it can also depend on who the opponents are in the contest 2011-06-11T10:16:13 i think the question is why are you not surprised? 2011-06-11T10:16:17 that's one of the things that makes games so complicated 2011-06-11T10:16:30 all this feels like analysis-paralysis, trying to imagine what people might do and whether to stop it (blocking, cooperation, etc) when it's all theoretical :-) 2011-06-11T10:16:40 because tit for tat is a great strategy for generating cooperation in such situations 2011-06-11T10:16:51 well anyway, tit-for-tat 'enforced cooperation' 2011-06-11T10:16:52 yes 2011-06-11T10:16:55 Zaphus: i'm studying pure maths :P 2011-06-11T10:17:03 and possibly this can be done in other cases 2011-06-11T10:17:09 ants i am not sure about 2011-06-11T10:17:21 antimatroid: yeah, did that decades ago (pure maths + comp sci) 2011-06-11T10:17:21 but maybe 2011-06-11T10:17:42 i did maths, comp sci and economics majors for undergrad :P 2011-06-11T10:17:51 soooo many overloads 2011-06-11T10:17:58 ouch! 2011-06-11T10:18:26 antimatroid: to youngsters like you it seems obvious hey? i guess it has since become part of 'standard knowledge' 2011-06-11T10:18:38 Who is imagine? 2011-06-11T10:18:52 yeah, that's the deal with heindsight :P 2011-06-11T10:19:02 :) 2011-06-11T10:19:14 it's like minimax with games like ches 2011-06-11T10:19:20 or nash equilibria 2011-06-11T10:19:32 yes 2011-06-11T10:19:38 nash coming up with and proving existence of a nash equilibria doesn't seem all that amazing to me 2011-06-11T10:19:47 it's a fairly straight forward application of fixed points 2011-06-11T10:19:57 anyway, should we agree a standard 'ant dance' now, that we dont attack each other? 2011-06-11T10:20:02 it is an awesome application of them though, and not a trivial feat 2011-06-11T10:20:31 I'll arrange my ants to spell out the words "PLAY NICE" in 3x5 font 2011-06-11T10:20:41 :) 2011-06-11T10:20:44 i'll draw a penis 2011-06-11T10:21:02 that will never stop being funny to me 2011-06-11T10:21:10 or "KILL HIM ---->" 2011-06-11T10:21:15 finish him! 2011-06-11T10:21:18 ftfy 2011-06-11T10:21:19 i want to signal - 'i am willing to be a peace with you, BUT if you attack me, then i will irrationally attack you' 2011-06-11T10:21:37 cya all, need sleep 2011-06-11T10:21:37 *** Accoun has joined #aichallenge 2011-06-11T10:21:38 george1914: have you played multiplayer flash galcon (pw)? 2011-06-11T10:21:46 no 2011-06-11T10:21:48 that's my strategy on that 2011-06-11T10:21:53 aha 2011-06-11T10:21:54 touch me and you lose 2011-06-11T10:21:55 any idea when this will be going live ? this month ? 2011-06-11T10:22:05 haha, alright, everyone keen for some galcon? 2011-06-11T10:22:15 http://www.galcon.com/flash/ 2011-06-11T10:22:18 go to multiplayer 2011-06-11T10:22:19 i play risk-like game 'warlight' and try that strategy sometimes 2011-06-11T10:22:34 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-11T10:22:39 mutually assured destruction can often work, even without chat communication 2011-06-11T10:22:46 i'm in Upsilon-1 2011-06-11T10:23:06 Zaphus: play one game 2011-06-11T10:23:13 carlop, UncleVasya ^^^ 2011-06-11T10:23:14 hm, i need a few days to write my bot for multiplayer galcon ... 2011-06-11T10:23:21 no, this is a human game 2011-06-11T10:23:41 :/ i think that is not for me ... 2011-06-11T10:23:46 nawww 2011-06-11T10:23:51 sorry 2011-06-11T10:24:06 UncleVasya: its me 2011-06-11T10:24:07 it's like playing human multiplayer planet wars with imperfect information 2011-06-11T10:24:26 so, back to this contest's details ... 2011-06-11T10:24:37 start time? roughly? 2011-06-11T10:24:46 when beta finishes :P 2011-06-11T10:24:49 how are matches picked? 2011-06-11T10:24:54 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-11T10:24:55 we were really hoping to have started ages ago 2011-06-11T10:25:03 eg similar ranked players, or random player 2011-06-11T10:25:12 i'm not sure about the matching script, ask janzert maybe? (sorry janzert :P) 2011-06-11T10:25:21 similar 2011-06-11T10:25:28 *** StijnD has joined #aichallenge 2011-06-11T10:25:35 random was awful in tron 2011-06-11T10:25:47 if random - then i need to eg identify and target weak/stupid bots ... 2011-06-11T10:25:59 i skip this PW match.. 2011-06-11T10:26:06 is similar, then i can assume intelligence ... 2011-06-11T10:26:23 meh, i prefer to write my bots more generally than that :P 2011-06-11T10:26:30 any idea if vision distance will change (much) ? 2011-06-11T10:26:37 i don't think that'll change 2011-06-11T10:26:41 there's no need for it to 2011-06-11T10:26:49 you should always play with both better and worst bot 2011-06-11T10:27:00 what is the food spawn rate? 2011-06-11T10:27:04 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-11T10:27:06 the start of the contest has everyone start from scratch 2011-06-11T10:27:11 they play bad bots then 2011-06-11T10:27:17 food spawn rate isn't decided 2011-06-11T10:27:23 sure, butafter a few games ... 2011-06-11T10:27:24 i'd like to see it lowered personally 2011-06-11T10:27:47 i don't like flooding the maps with ants 2011-06-11T10:27:51 is food rate fixed across games 2011-06-11T10:27:52 i'd rather more strategic battles 2011-06-11T10:28:06 ie does it change depending on map area? 2011-06-11T10:28:06 don't think that's decided 2011-06-11T10:28:17 part of beta is us testing what works best 2011-06-11T10:28:19 hm 2011-06-11T10:28:22 yes 2011-06-11T10:28:26 we haven't been able to test stuff like that up until now 2011-06-11T10:28:54 i would like to write bot soon, but would prefer not to have to change much when/if rules/settings change ... 2011-06-11T10:29:09 yeah, i would wait till at least the official launch 2011-06-11T10:29:19 how does a bot know the spawnrate? 2011-06-11T10:29:26 it doesn't 2011-06-11T10:29:32 i read it on the website, and code it in? 2011-06-11T10:29:35 right 2011-06-11T10:29:54 you might be able to do it if it's ever "locked in", otherwise don't assume stuff :P 2011-06-11T10:29:59 i hate people who do that :P 2011-06-11T10:30:07 (poke at map stuff from pw) 2011-06-11T10:30:23 ? 2011-06-11T10:30:35 remember when the maps changed towards the end of planet wars? 2011-06-11T10:30:42 i dont mean try to hard-code specific maps 2011-06-11T10:30:54 i think you can approximate spawnrate after a while into the game, but you don't know it 2011-06-11T10:31:01 people assumed that the other map generator was meant to be used in the final contest, that wasn't actually the case, they just assumed it 2011-06-11T10:31:36 i just mean - knowing the rate at which food arrives at a square is useful for deciding how often to check areas - what is Prob(food now there) 2011-06-11T10:31:44 yeah i know 2011-06-11T10:31:51 yes - but in PW it was known that the maps would change ... 2011-06-11T10:31:55 atm, the official answer is you shouldn't assume you know the rate 2011-06-11T10:32:02 so they had little reason to complain 2011-06-11T10:32:07 in reality, after a while you can probably assume it'll stay where it's at 2011-06-11T10:32:15 i guess so 2011-06-11T10:32:18 they complained anyway :P 2011-06-11T10:32:26 and i agree the change was too late 2011-06-11T10:32:33 but i'd been waiting for the change and no one did it 2011-06-11T10:32:38 so i did when i finished exams 2011-06-11T10:32:38 i just wondered if a big map got less food-per-unit-area than a small one? 2011-06-11T10:32:55 dunno, we haven't even worked out what size maps to use yet 2011-06-11T10:32:55 hm 2011-06-11T10:33:04 i think the games on the beta server are too long? 2011-06-11T10:33:12 do we really need them to take that long to work out which bot is better? 2011-06-11T10:33:18 coudl we decrease the size of the maps? 2011-06-11T10:33:20 in other word 2011-06-11T10:33:21 s 2011-06-11T10:34:30 i had slight dissappointment in tron ... 2011-06-11T10:34:43 maps finished small, in order to spped up games 2011-06-11T10:34:49 'speed' 2011-06-11T10:34:49 yeah i hated that 2011-06-11T10:34:53 and they weren't random enough 2011-06-11T10:35:04 but i had tried a clever 'long range win spotting' method 2011-06-11T10:35:06 we built out bot with tests on 50x50 maps 2011-06-11T10:35:20 that would have done well against minimax bots 2011-06-11T10:35:31 but hey, nothing is perfect 2011-06-11T10:35:41 that's one of my biggest requests, if we say something needs to be general enough for "whatever" then the final contest needs to meet that level of generality 2011-06-11T10:35:48 otherwise it comes down to who made the best assumptions 2011-06-11T10:35:59 yes 2011-06-11T10:36:27 but, i've been pretty happy with both contests 2011-06-11T10:36:36 a1k0n and bocsi were clear winners 2011-06-11T10:36:39 they were brilliant 2011-06-11T10:36:47 (the contest, and the players :) 2011-06-11T10:36:52 yep 2011-06-11T10:36:55 tron was my favourite 2011-06-11T10:37:06 planet wars was a great problem, but almost impenetrably from an ai point of view 2011-06-11T10:37:19 they were both good 2011-06-11T10:37:32 but in both cases i tried 'clever' ai tricks 2011-06-11T10:37:47 and though otehrs must be doing the same (because they were beating me) 2011-06-11T10:37:57 what clever ai tricks did you use in pw? 2011-06-11T10:38:08 i used a very horribly buggy game tree 2011-06-11T10:38:31 and then, i was surprised and slightly dissappointed when i found others had beaten me by more efficient 'standard' methods 2011-06-11T10:38:41 i had hoped that clever methods would win 2011-06-11T10:38:52 i'm not sure if my methods count as clever 2011-06-11T10:39:02 i try to throw everything i can at the problems from a game theory perspective 2011-06-11T10:39:07 *** stocha has joined #aichallenge 2011-06-11T10:39:13 (ok the winners on both cases did some smart stuff - but not that different from others) 2011-06-11T10:39:34 well, i tried neural net for position evaluation 2011-06-11T10:39:55 and minimisation methods for picking my parameters 2011-06-11T10:40:19 i had big plans for move selection 2011-06-11T10:40:20 (play bots, and combine the parameters from the best bots, to produce offspring) etc 2011-06-11T10:40:39 aha? 2011-06-11T10:40:40 trying to set it up as some kind of linear program, so you could use simplex to iterate towards the optimal solution 2011-06-11T10:40:46 but never got there 2011-06-11T10:40:50 this is for PW? 2011-06-11T10:40:55 i tried using dynamic programming for move selection too 2011-06-11T10:40:56 yeah 2011-06-11T10:41:01 right 2011-06-11T10:41:02 really it needed to be an integer program too 2011-06-11T10:42:04 so what stopped you? 2011-06-11T10:42:20 technical issues, or theory, or exams? 2011-06-11T10:42:22 :) 2011-06-11T10:42:23 time 2011-06-11T10:42:26 contrainsts 2011-06-11T10:42:27 aha 2011-06-11T10:42:35 1 second doesn't give you much 2011-06-11T10:42:46 oh, right, machine time 2011-06-11T10:42:50 i was also doing 5 units and attending measure theory classes :P 2011-06-11T10:43:09 i would not have expected that to be such a problem ... 2011-06-11T10:43:34 that is - if you couldnt do it in 1s, i doubt you could do it in 10s either? 2011-06-11T10:43:44 probably not 2011-06-11T10:43:48 ok 2011-06-11T10:43:51 but i ended up not doing it :P 2011-06-11T10:44:08 i tried doing game trees or matrices 2011-06-11T10:44:13 in that case it is nearer to 'theoretical' imo 2011-06-11T10:44:17 aha? 2011-06-11T10:44:18 ended up finding it easier to iteratively generate tree's 2011-06-11T10:44:33 just keep a queue of extendable non-terminal nodes 2011-06-11T10:44:42 and extend nodes as long as you want 2011-06-11T10:44:58 right, i think i should go do some stuff ... 2011-06-11T10:45:02 turn it into a priority queue if you're into that kind of thing 2011-06-11T10:45:03 nice chatting with you 2011-06-11T10:45:13 yep, see ya 2011-06-11T10:45:16 thanks for the info on the game specs 2011-06-11T10:45:20 np 2011-06-11T10:45:43 btw - are there any minor tasks a semi-competent c++ person like me can help with? 2011-06-11T10:46:04 if there were, i could volunteer ... 2011-06-11T10:46:12 also eg website edits etc 2011-06-11T10:46:16 ? 2011-06-11T10:46:27 website edits etc. talk to amstan about 2011-06-11T10:46:34 there isn't really anything to do c++ wise :( 2011-06-11T10:46:38 aha 2011-06-11T10:46:42 i have had to learn python to do stuff 2011-06-11T10:46:56 yes, i feared that mightbe the case 2011-06-11T10:47:26 well, thanks again. see you around. 2011-06-11T10:47:29 you could work on the strategy guide :P 2011-06-11T10:47:35 and good luck in the cempetition when it starts 2011-06-11T10:47:43 ah, strategy guide? 2011-06-11T10:47:48 that i could do 2011-06-11T10:47:56 https://github.com/aichallenge/aichallenge/wiki/Ants-Strategy-Guide 2011-06-11T10:47:58 similar to the ones from the last 2 contests? 2011-06-11T10:48:05 yeah, that's what i started a while ago 2011-06-11T10:48:09 it's very rough 2011-06-11T10:48:52 what is covered looks good 2011-06-11T10:50:02 i spilled blood: http://users.hszk.bme.hu/~pf650/screenie.mp4 2011-06-11T10:50:13 ? 2011-06-11T10:51:18 fullfood map ? 2011-06-11T10:51:41 yep 2011-06-11T10:52:01 nice isnt it? 2011-06-11T10:52:17 godamnit! 2011-06-11T10:52:21 dunnow. i already tryed it. 2011-06-11T10:52:22 i can't view the tron games 2011-06-11T10:52:24 i don't know why 2011-06-11T10:52:26 it makes me sad 2011-06-11T10:52:48 in fact, i made 80% food map also. 256x256. to stress test my simplest bot. 2011-06-11T10:52:57 *** UncleVasya has joined #aichallenge 2011-06-11T10:53:12 its a stress-test for my engine, and it works now, can handle food 2011-06-11T10:53:23 there was a bug before, but now its ok 2011-06-11T10:53:37 you have your own engine ? 2011-06-11T10:53:41 yep 2011-06-11T10:53:54 UncleVasya: you asked who is imagine, its me, why? 2011-06-11T10:53:55 what map generator will use in there ? 2011-06-11T10:54:09 i just load the official maps 2011-06-11T10:54:47 *** _0x47 <_0x47!~slash@p5B12F1F8.dip.t-dialin.net> has quit IRC (Quit: Leaving.) 2011-06-11T10:55:04 why use your own engine ? 2011-06-11T10:55:22 mcstar: Hungary, Lisp... I thought that it's possibly Him. 2011-06-11T10:55:28 gain programming experience, and for faster testing 2011-06-11T10:55:36 no 2011-06-11T10:55:41 what testing feature will you add ? 2011-06-11T10:55:44 btw theres a Hungry bot lol 2011-06-11T10:57:19 stocha: i was considering GP but i changed my mind, i wont do that, the engine waS developed to facilitate that 2011-06-11T10:58:49 mgngn ? 2011-06-11T11:00:31 theres awful lot of consonants in that word 2011-06-11T11:00:49 what does it mean? 2011-06-11T11:02:31 where is the standard implementation of hashmap in c standard libraries ? 2011-06-11T11:03:01 mcstar: its basically a grunt, with a "?" 2011-06-11T11:03:20 oh 2011-06-11T11:03:31 so you can resume it to "?" i suppose. 2011-06-11T11:03:45 ok 2011-06-11T11:03:56 or . Mm, interesting, i wonder what more there is to it. 2011-06-11T11:04:05 that's closer i think. 2011-06-11T11:04:08 i get it 2011-06-11T11:04:24 do you want me to answer that "grunt"? 2011-06-11T11:04:24 acually, you can also write it ngngm if you want. 2011-06-11T11:04:31 sure :) 2011-06-11T11:04:32 or do they at all? 2011-06-11T11:04:47 uh ? 2011-06-11T11:04:58 do whom at all what ? 2011-06-11T11:05:02 i dont know if its customary to answer grunts at all 2011-06-11T11:05:04 is it? 2011-06-11T11:05:12 what's customary ? 2011-06-11T11:05:26 do they usually do that 2011-06-11T11:05:32 = customary 2011-06-11T11:05:43 grunt are great. my less than childs use it everytime with great sucess :) 2011-06-11T11:05:50 so did my dog. 2011-06-11T11:05:59 less than 2 years childs 2011-06-11T11:06:04 thats new, didnt know dogs grunt 2011-06-11T11:06:17 well, they make sounds. 2011-06-11T11:06:55 back to track: is there a specific question i need to answer? 2011-06-11T11:06:58 my dog used to says ; arrwwghwaaaahhmm. 2011-06-11T11:07:19 wich means, it's not that i'm bored, but i'll accept any suggestion at doing something. Maybe going out ? 2011-06-11T11:07:36 very educational :D 2011-06-11T11:07:49 mcstar: what are the debugging utilities of your engine ? :) 2011-06-11T11:08:02 print 2011-06-11T11:08:05 :D 2011-06-11T11:08:23 basic prints ? doesn't the official engine allow that ? 2011-06-11T11:08:40 i dont know, i didnt look at it 2011-06-11T11:08:47 ah okay. 2011-06-11T11:09:04 its python, and its too complex for me 2011-06-11T11:09:06 i build a graphical debugging engine. 2011-06-11T11:09:13 i'm a bit disapointed at the result. 2011-06-11T11:09:26 but it enabled me not to look inside the official stuff :p 2011-06-11T11:09:28 what do you mean exactly by debugginh? 2011-06-11T11:09:47 you known, finding unwanted behaviors and features. 2011-06-11T11:09:48 debugging as a compiler or interpreter does? 2011-06-11T11:10:12 it seems that i missed an important undesired feature though. like nothing work with map size = 65 2011-06-11T11:10:33 i never tested map size = 65 for some reason. Or if i did, i took the wrong conclusion out of it. 2011-06-11T11:10:34 cause of course, i can debug my bot in my engine 2011-06-11T11:11:15 im on the way of being that cooperative, that i only need to load a file with class definitions from my bot into the engine and thats all 2011-06-11T11:11:26 the point was : if people add some features to their own engine, maybe we should tell it to the official team. so they can add it to the official one. 2011-06-11T11:11:45 im not going into that direction 2011-06-11T11:11:59 i dont want to extend its functionality 2011-06-11T11:12:33 i think some basic debugging features are needed into the official engine. 2011-06-11T11:12:46 but i didn't look to hard at it. so can't make suggestion :p 2011-06-11T11:13:21 well, since i wrote my code,i pretty much know what happens for what reason, so i dont need "debugging features" 2011-06-11T11:14:02 now im going to develop path finding 2011-06-11T11:14:07 gonna be interesting 2011-06-11T11:14:28 oh, and the ant can see the whole map for the time being 2011-06-11T11:14:49 the fog of war is not implemented yet 2011-06-11T11:14:51 anyone knows where the standard hashmap in c standard lib hides ? 2011-06-11T11:15:13 i can develop pathfinding and even battle strategies before that 2011-06-11T11:15:36 what is that? 2011-06-11T11:15:51 hashmap? 2011-06-11T11:16:06 function-names vs. code? 2011-06-11T11:16:36 i should have written address of code 2011-06-11T11:17:02 hashmap like binding a String value with a value. 2011-06-11T11:17:19 ok 2011-06-11T11:17:28 static string associations 2011-06-11T11:17:56 why is that important? 2011-06-11T11:18:35 btw you can use strings on object files, it gives the readable characters back 2011-06-11T11:18:58 ImportError: No module named Queue 2011-06-11T11:19:20 when trying to run play_one_game.cmd 2011-06-11T11:19:26 stocha: you can also read the man page of nm 2011-06-11T11:20:03 anyone knows how to solve this? 2011-06-11T11:20:13 StijnD: you need at least python 2.6 2011-06-11T11:21:04 installed 3.2 2011-06-11T11:21:16 that might be too new 2011-06-11T11:21:23 the current stable is 2.7.x 2011-06-11T11:22:04 there must be a lot of python guys here, ask them 2011-06-11T11:22:39 alright, im gonna simulate a couple of wildfires on fullfood maps, then code some 2011-06-11T11:22:48 bb 2011-06-11T11:22:53 *** mcstar has quit IRC (Quit: WeeChat 0.3.5) 2011-06-11T11:23:14 I think it might be teh same problem as this guy had http://ai-contest.com/forum/viewtopic.php?f=25&t=1401&sid=ffc30ca4e08bc0345a6b2b9e67e5b6df#p9046 2011-06-11T11:28:32 stocha: in c you have no standard container classes. In c++ you have as standard the STL for various cointainers 2011-06-11T11:29:02 must implement it yourself 2011-06-11T11:29:54 stocha: i see you have already begin with the c-starter 2011-06-11T11:30:16 locutus2: yeah. goes on slowly. but i'm confident i can do it :p Testing is a nut case though 2011-06-11T11:30:41 stocha: have you read my comment about the c-starter-bot in the github repositry? 2011-06-11T11:30:46 locutus2: what if my scanf buffer overflow my int because i didn't get the thing right ? how can i know i blundered ? 2011-06-11T11:30:57 nop i read nothing yet :p 2011-06-11T11:31:24 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-11T11:31:54 locutus2: where can i read it ? :p 2011-06-11T11:32:16 stocha: on the gitbug repositry you can find an c-starter-bot under ant/dist 2011-06-11T11:32:44 StijnD: You need python < 3 2011-06-11T11:32:45 aahhh ! i see it 2011-06-11T11:32:47 i have him tested short locally and its runs, but plays only random moves 2011-06-11T11:32:54 but but .. i was trying to write one :p 2011-06-11T11:33:17 yeah that was my plan too. playing only random moves. i mean it's enough for a simple starter bot. 2011-06-11T11:34:21 perhaps you can lock their a little bit by problems 2011-06-11T11:34:24 what's map in gameinfo ? 2011-06-11T11:34:49 ? 2011-06-11T11:35:25 you track individual ants ? how do you do that ? what is the "myant.id" for ? 2011-06-11T11:35:51 to admins: on the test server aren't any games played the last hours 2011-06-11T11:36:12 locutus2: basic_ant i called that other_ant 2011-06-11T11:36:49 locutus2: in the spec isn't player an integer rather than a *char ? 2011-06-11T11:37:07 basic_ant.player 2011-06-11T11:38:31 stocha: yes, the players are int 2011-06-11T11:38:52 it's a bit misleading then 2011-06-11T11:39:39 i don't know how anyone prefers c to c++ :\ 2011-06-11T11:40:17 antimatroid: sorry to come back to this again, but in the code on githun, where do you remove food squares? (ie we see are not told a square we can see has food, so we should remove food that used to be there from State) ? 2011-06-11T11:40:18 while 42 ! Is it common to put 42. or a private joke ? 2011-06-11T11:41:10 stocha: in the maps the various ant are marked by a, b, ... 2011-06-11T11:41:25 george1914: in Square.h/cc in reset() it makes isFood = 0, and when resetting state i think it clears the food vector/set 2011-06-11T11:41:29 but during a game you get an int 2011-06-11T11:41:34 locutus2: you only process first char ? :) 2011-06-11T11:42:18 antimatroid I installed python 2011-06-11T11:43:15 stocha: you don't must process any chars because you get an int, only the playgame.py on the server reads from the map files, to the bots ints are send 2011-06-11T11:43:15 george1914: it's also worth getting rid of "isVisible" and adding an int "lastSeen" 2011-06-11T11:43:18 antimatroid/george1914: a possible problem is that you lost information on food that you can't seen (eg *....a....*, if you take a food you don't remember the other one), but i think that this is not so basic, and everyone should do it for himself 2011-06-11T11:43:20 starting everything at 0 2011-06-11T11:43:32 locutus2 : i wanted to put a function pointer into a hashmap. but i had no hashmap ready. and it's to much effort for me :p 2011-06-11T11:43:36 carlop: yeah you have to do that yourself now 2011-06-11T11:43:41 i'll pastebin what i am doing about that 2011-06-11T11:43:42 antimatroid: so reset should be called every move? 2011-06-11T11:43:55 george: it is called in endTurn i think? 2011-06-11T11:44:19 locutus2: i mean you only processed first char for keywords, and then another discriminating char if needed. 2011-06-11T11:44:38 locutus2: i was goint to build a sort of switch statement on the whole string. 2011-06-11T11:45:01 carlop: http://pastebin.com/XwrDEBui i call that after updateVisionInformation 2011-06-11T11:45:15 and oldFood is made = food just before I clear it in State::reset 2011-06-11T11:45:28 food etc. is also sets in my bot 2011-06-11T11:46:12 ok, i see what you both say 2011-06-11T11:47:03 stocha: i think its easer to read the whole word and compare 2011-06-11T11:47:16 I have a grid struct too, with an operator to index it with a location 2011-06-11T11:47:24 thanks. i guess it would be nice for the starterbot to include that feature for remembering out-of-vision food, plus removing it from State once we see it has gone? - would be nice for new competitiors to have? 2011-06-11T11:47:31 grid[loc] kicks the shit out of grid[loc.row][loc.col] 2011-06-11T11:47:41 locutus2: i was doing just that. but you used only discrimating char, so you can build a switch statement. i would have though of it :p 2011-06-11T11:47:48 george1914: meh, let the competitors write it themselves 2011-06-11T11:47:53 locutus2: have you fully test the bot ? 2011-06-11T11:47:54 stocha: i have find a hashmap-implementation for c on http://elliottback.com/wp/hashmap-implementation-in-c/ 2011-06-11T11:47:55 it's a good way of getting to know the starter bot 2011-06-11T11:48:03 perhaps you can try this 2011-06-11T11:48:07 yes, maybe 2011-06-11T11:49:04 locutus2: well, one would have to test it. and it adds a dependency. i would maybe have tryed if it was standard. but i don't need it really do i. i can by a if( ) map also. 2011-06-11T11:49:05 i even have state++ to process collions/battles etc. :D 2011-06-11T11:49:22 i love operator overloading <3 2011-06-11T11:49:34 locutus2: i'll try to replace my stocha bot wich don't really work on the test server, by your random bot, and see how it goes. 2011-06-11T11:49:45 antimatroid: your function seem right, and i think that both grid[location] and lastSeen are must do 2011-06-11T11:50:10 you can't do grid[loc] when it's just a vector of vecotrs 2011-06-11T11:50:17 and i want the starter bot to be reasonably simple 2011-06-11T11:50:21 so i left that out 2011-06-11T11:50:27 it's already got a bunch of files :P 2011-06-11T11:50:57 you can't do template matrix and index A[loc] 2011-06-11T11:51:08 the way vectors of bools are done, it just doesn't work, it's really really annoying! 2011-06-11T11:51:13 I just downloaded the source from github 2011-06-11T11:51:31 should I be able to run play_one_game fromt the map ants? 2011-06-11T11:51:43 StijnD: yeah 2011-06-11T11:51:49 from a console/command window? 2011-06-11T11:51:51 yes, starting bot should be simple, the current one is seem good 2011-06-11T11:52:44 ImportError: no module named StringIO :( 2011-06-11T11:52:54 mm apparently author is flagcapper-google. form 14 may. why didn't i see it when i downloaded the full repository ? 2011-06-11T11:53:01 I added the python map to my path in windows 2011-06-11T11:53:29 just wondering, why in the site they tell "don't upload your best bot" ? 2011-06-11T11:53:33 StijnD: not sure sorry, i'm not very knowledgeable on that kind of stuff 2011-06-11T11:53:41 np :p 2011-06-11T11:53:52 carlop: because the beta server is hosted on a contestants server 2011-06-11T11:54:03 while we don't expect there to be any "cheating" there are potential security concerns 2011-06-11T11:54:11 so if you want to be anal, don't upload your best bot 2011-06-11T11:54:27 https://github.com/flagcapper/GoogleAI-Ants-Starter-Pack#readme 2011-06-11T11:54:44 that awfully like an old c starter bot. 2011-06-11T11:54:57 stocha: its not in the official package, but in a submodule. in the INSTALL.md its says you must do 'git sunmodule init' and 'git submodule update'. then you get it 2011-06-11T11:54:57 i'm confused. 2011-06-11T11:55:14 ah. i don't have git. 2011-06-11T11:55:23 why isn't it in the main branch ? 2011-06-11T11:55:32 you made a commentary on it ? 2011-06-11T11:57:02 i don't know, i'am not on of the maintainers 2011-06-11T11:57:12 http://www.reddit.com/r/AskReddit/comments/hx53w/which_of_your_favorite_movies_seems_to_be_hated/c1z35mg that must be a troll 2011-06-11T12:00:21 *** StijnD has quit IRC (Ping timeout: 240 seconds) 2011-06-11T12:02:31 the thing doesn't compile well at home. 2011-06-11T12:02:41 there is a non declared fun. 2011-06-11T12:03:19 *** Accoun has quit IRC () 2011-06-11T12:03:23 i'll try to upload it on the server 2011-06-11T12:05:14 *** Racko has quit IRC (Ping timeout: 252 seconds) 2011-06-11T12:05:22 *** Accoun has joined #aichallenge 2011-06-11T12:05:34 locutus2: http://aichallengebeta.hypertriangle.com/visualizer.php?game=20336 2011-06-11T12:06:04 blocker bot doesn't seem to block well. or is it not the blocker ? 2011-06-11T12:06:40 *** StijnD has joined #aichallenge 2011-06-11T12:06:48 alright got it to work 2011-06-11T12:07:20 I guess mcstar was right and installed older version of python 2011-06-11T12:07:30 python 2.7.1 2011-06-11T12:08:43 :) 2011-06-11T12:09:04 stocha: i still think you should try your hand at c++ 2011-06-11T12:09:18 StijnD: you will be using c++ yes? 2011-06-11T12:09:25 i am herding everyone to c++ 2011-06-11T12:09:26 it didn't recognized the MyBot.c O_o 2011-06-11T12:09:35 what version of pyton is needed ? 2.7.1 or 3.2 ? 2011-06-11T12:09:50 2.7.1 it would seem 2011-06-11T12:09:53 it's definitely < 3 2011-06-11T12:10:10 <3 and >=3 have incompatible things from what i understand 2011-06-11T12:10:11 hm, something seems wrong on the beta server 2011-06-11T12:10:14 but i understand little about python 2011-06-11T12:10:23 game results on my user page and in game viewers are different 2011-06-11T12:10:39 this current server doesn't support c ?? 2011-06-11T12:10:57 I'll be using java :p 2011-06-11T12:11:08 StijnD: inconceivable! 2011-06-11T12:11:15 don't have much c++ expercience 2011-06-11T12:11:22 perfect opportunity! 2011-06-11T12:11:44 the transition wouldn't be to difficuld I heard 2011-06-11T12:12:33 StijnD: it isn't. I switched from java to c++ for this contest, only had very little c++ exp before, but it works out fine 2011-06-11T12:12:59 ok nice 2011-06-11T12:13:16 StijnD: just harrass me when you don't understand something :) 2011-06-11T12:13:30 i think i wrapped it insied a directory and the worker didn't like it 2011-06-11T12:13:39 let's see how it goes now 2011-06-11T12:13:40 but I might start with java for now until my exams are over and then when I have more time I can start learning c++ 2011-06-11T12:13:51 you're in Australia? 2011-06-11T12:13:59 StijnD: go with c. c is the next big thing ! 2011-06-11T12:14:08 stocha: :D 2011-06-11T12:14:11 http://aichallengebeta.hypertriangle.com/visualizer.php?game=20337 "player 5" should be me (leviathan), and my player page says I actually won that... wtf? 2011-06-11T12:14:17 it's so much ... worst than pascal ! yeah ! 2011-06-11T12:14:43 StijnD: where do you live? 2011-06-11T12:14:58 belgium 2011-06-11T12:15:10 stocha: wrong bot, Zaphus_BlockTest is the blocker, locutus2_bot is identical to my locutus2 bot. Its for testing of stability of the trueskill rating system 2011-06-11T12:17:39 no, brainfuck rules! 2011-06-11T12:17:43 mm this time i was able to compile the c starter bot. but .. i wonder if my mac files are possible to compile, seeing the error i got. 2011-06-11T12:17:49 locutus2: ? 2011-06-11T12:17:50 *** Kingpin13 has joined #aichallenge 2011-06-11T12:19:14 *** StijnD has quit IRC (Read error: No route to host) 2011-06-11T12:19:25 *** StijnD has joined #aichallenge 2011-06-11T12:20:55 locutus2: ah brainfuck would be a computer language okay. 2011-06-11T12:21:03 stocha: as i see version 9 has compiled and get through the tests. what differs to version 10 2011-06-11T12:21:07 i though it was an ant bot 2011-06-11T12:21:10 antimatroid, did u say something after StijnD: where do you live? 2011-06-11T12:21:11 belgium 2011-06-11T12:21:23 cause my connction failed 2011-06-11T12:21:26 locutus2: version 10 inlcludes a c file i generated myself :) 2011-06-11T12:22:13 i think i put in a hidden rep in v 8, i wanted to check that c files generated on my laptop would be okay 2011-06-11T12:22:55 nup sorry :P 2011-06-11T12:23:06 i have exams atm too, so i was randomly guessing you're in australia 2011-06-11T12:24:17 stocha: seems you have many illegall characters in the file 2011-06-11T12:24:57 carlop: btw, i've does quite a few tests on that processOldFoodInformation function 2011-06-11T12:25:03 pretty sure it handles food information correctly 2011-06-11T12:25:34 but blind its not easy to say what currently really wrong 2011-06-11T12:26:30 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-11T12:38:16 *** amriedle has joined #aichallenge 2011-06-11T12:38:51 *** locutus2 has quit IRC (Ping timeout: 252 seconds) 2011-06-11T12:40:07 *** UncleVasya has joined #aichallenge 2011-06-11T12:47:58 *** Vio has joined #aichallenge 2011-06-11T12:49:51 *** StijnD has quit IRC (Ping timeout: 252 seconds) 2011-06-11T12:50:33 *** locutus2 has joined #aichallenge 2011-06-11T13:01:07 *** dlila has quit IRC (Quit: Leaving) 2011-06-11T13:06:21 replay on map with rotational symmetry: http://pastebin.com/PzcWkefX 2011-06-11T13:06:36 and the map file: http://pastebin.com/4waYMJVY 2011-06-11T13:12:36 for people who like mazes: http://pastebin.com/E0bymR15 2011-06-11T13:12:52 and the map file: http://pastebin.com/WUcNdm4V 2011-06-11T13:13:20 *** amriedle has quit IRC (Quit: Lost terminal) 2011-06-11T13:14:23 I'm not actually suggesting that mazes would be good to play in (I know they would be hopeless) but it might be possible to make some good maps with a modified maze-digging algo which added more loops and interconnections (and used a large granularity like the first of those two maps I posted) 2011-06-11T13:18:31 the rotation and tiling code can work with other map varieties too, but the mazes just posed an interesting challenge :) 2011-06-11T13:20:14 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-06-11T13:25:58 *** olexs has quit IRC (Quit: Leaving.) 2011-06-11T13:29:30 *** stocha has joined #aichallenge 2011-06-11T13:29:53 <_flag> stocha: I'm the one that wrote the C starter bot if you have any questions :P 2011-06-11T13:30:12 i agree that cpu time on ubuntu server should be used. The max value should be configurable easilly so people can locally extend it (for windows user for example) 2011-06-11T13:30:52 _flag: okay :) I puted it on the test server under the name stocha. while i finish porting one of my bots. 2011-06-11T13:31:38 <_flag> stocha: okay, let me know if there's any issues 2011-06-11T13:35:04 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-11T13:38:36 *** locutus2 has quit IRC () 2011-06-11T13:42:25 *** carlop has quit IRC (Remote host closed the connection) 2011-06-11T13:54:50 *** Diversus has joined #aichallenge 2011-06-11T13:55:23 Tell me please why not registred 2011-06-11T13:55:47 *** Kingpin13 is now known as Kingpin13|food 2011-06-11T13:57:27 *** Naktibalda has quit IRC (Remote host closed the connection) 2011-06-11T14:06:02 *** _0x47 <_0x47!~slash@p57B3E5ED.dip.t-dialin.net> has joined #aichallenge 2011-06-11T14:08:32 *** locutus2 has joined #aichallenge 2011-06-11T14:10:13 Diversus: if you're looking for a confirmation email for registration, they haven't been activated yet, but there's an activation link after the signup page. 2011-06-11T14:14:21 What about McLeopold's ant map pastebin? 2011-06-11T14:14:58 It would be nice to have a real preview of the maze map rather than ascii art 2011-06-11T14:15:45 *** carlop has joined #aichallenge 2011-06-11T14:16:53 *** amstan has joined #aichallenge 2011-06-11T14:16:53 *** ChanServ sets mode: +o amstan 2011-06-11T14:18:43 *** Kingpin13|food is now known as Kingpin13 2011-06-11T14:20:56 aichallenge: janzert epsilon * r2c25489 / worker/jailguard.py : Refactor out one jailguard thread - http://bit.ly/kkcM5S 2011-06-11T14:21:56 @map pastebin 2011-06-11T14:21:56 Frontier: map pastebin could be http://ants.zeroviz.us/. 2011-06-11T14:22:00 *** Diversus has quit IRC (Ping timeout: 252 seconds) 2011-06-11T14:23:33 Frontier: http://imgur.com/XOkas 2011-06-11T14:23:50 smiley1983: look at this, much better, isn't it? http://ants.zeroviz.us/I1hCP/ 2011-06-11T14:26:22 Ah, thankyou for showing me that :) 2011-06-11T14:26:29 It is much better indeed 2011-06-11T14:29:46 http://ants.zeroviz.us/8ZU9b/ 2011-06-11T14:30:47 *** smiley1983 has quit IRC (Quit: leaving) 2011-06-11T14:30:59 *** Redgis has joined #aichallenge 2011-06-11T14:32:17 smiley1983: ... and the generator for that rot sym map ? 2011-06-11T14:32:40 nice i want more of those 2011-06-11T14:32:49 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-11T14:35:15 btw, i liked this world map. did some one keep it ? 2011-06-11T14:39:11 berak: I can download it from the beta server 2011-06-11T14:39:32 link will do 2011-06-11T14:40:21 http://ants.zeroviz.us/A1n5K/ 2011-06-11T14:40:27 ty 2011-06-11T14:43:45 anyone have problem running the test tools ? 2011-06-11T14:46:00 *** UncleVasya has joined #aichallenge 2011-06-11T14:48:58 when running test_bot.sh or play_one_game.sh I have a lot of python error that i can't understand, someone can take a look ? these is the report: http://pastebin.com/9UvrXDKx 2011-06-11T15:01:18 *** akranis has joined #aichallenge 2011-06-11T15:05:08 carlop, can't reproduce that (since i'm on win) but does it get better, if you put the whole ants thing into a folder, that does NOT have blanks in the name ? 2011-06-11T15:09:00 *** Vio has quit IRC (Ping timeout: 240 seconds) 2011-06-11T15:13:52 *** StijnD has joined #aichallenge 2011-06-11T15:14:04 i bet, that this def get_cmd_wd(cmd) function in playgame.py fails on spaces in the path 2011-06-11T15:16:14 i've renamed the folder, but show the same error.. 2011-06-11T15:16:42 don't know where can be the problem, maybe later i reboot under win 2011-06-11T15:17:54 *** _0x47 <_0x47!~slash@p57B3E5ED.dip.t-dialin.net> has quit IRC (Ping timeout: 255 seconds) 2011-06-11T15:19:08 getting paranoid here, ignore me 2011-06-11T15:20:41 *** StijnD has quit IRC (Ping timeout: 240 seconds) 2011-06-11T15:21:10 aichallenge: janzert epsilon * r35eb895 / setup/worker_server_info.py.template : Up memory_limit in worker config to match default - http://bit.ly/inxUtk 2011-06-11T15:21:17 *** StijnD has joined #aichallenge 2011-06-11T15:23:15 aichallenge: janzert epsilon * rc2c21f5 / setup/worker_setup.py : Up jail resource limits to allow current beta server game limits - http://bit.ly/jO4kC1 2011-06-11T15:26:15 contestbot: later tell McLeopold looks like jskills crashes even with 1000 iterations on game 20458 2011-06-11T15:26:15 janzert: Ready to serve, my lord. 2011-06-11T15:28:22 *** Naktibalda has joined #aichallenge 2011-06-11T15:29:02 aichallenge: Marco Leise epsilon * re860754 / (2 files in 2 dirs): visualizer: fixed incorrect player order when final ranks are changed by food bonus - http://bit.ly/izOo5g 2011-06-11T15:29:02 aichallenge: Marco Leise epsilon * rb986352 / (9 files in 4 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://bit.ly/jRZry7 2011-06-11T15:31:22 *** _0x47 <_0x47!~slash@p5B12DCE1.dip.t-dialin.net> has joined #aichallenge 2011-06-11T15:34:30 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-06-11T15:38:40 *** Accoun has quit IRC () 2011-06-11T15:43:24 *** paratrechina has quit IRC (Ping timeout: 252 seconds) 2011-06-11T15:43:59 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-11T15:47:15 *** Redgis has quit IRC (Quit: ... mains libres) 2011-06-11T15:54:57 *** Accoun has joined #aichallenge 2011-06-11T15:55:35 *** MuTa-ofd00m has joined #aichallenge 2011-06-11T15:56:18 *** BrykBot has joined #aichallenge 2011-06-11T15:58:06 *** stocha has joined #aichallenge 2011-06-11T16:00:38 *** BrykBot has quit IRC (Ping timeout: 252 seconds) 2011-06-11T16:01:00 *** stocha has quit IRC (Client Quit) 2011-06-11T16:12:13 *** EQRF has joined #aichallenge 2011-06-11T16:24:19 *** Eruonen has quit IRC (Read error: Connection reset by peer) 2011-06-11T16:36:09 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 4.0.1/20110518052719]) 2011-06-11T16:43:25 *** dlila has joined #aichallenge 2011-06-11T16:47:06 *** Eruonen has joined #aichallenge 2011-06-11T16:48:38 *** olexs has joined #aichallenge 2011-06-11T16:48:43 *** deltOr has joined #aichallenge 2011-06-11T16:50:53 *** delt0r_ has quit IRC (Ping timeout: 276 seconds) 2011-06-11T16:53:48 *** olexs has quit IRC (Quit: Leaving.) 2011-06-11T16:56:23 *** Palmik has quit IRC (Remote host closed the connection) 2011-06-11T17:03:06 *** berak has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2011-06-11T17:06:58 *** Frontier has quit IRC (Quit: Leaving.) 2011-06-11T17:09:27 *** Frontier has joined #aichallenge 2011-06-11T17:11:11 *** locutus2 has left #aichallenge 2011-06-11T17:28:43 *** qacek has joined #aichallenge 2011-06-11T17:44:05 *** sigh has joined #aichallenge 2011-06-11T18:04:50 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-11T18:18:24 *** Zaphus has joined #aichallenge 2011-06-11T18:18:39 *** sigh has joined #aichallenge 2011-06-11T18:21:03 *** StijnD has quit IRC (Ping timeout: 255 seconds) 2011-06-11T18:21:11 *** sigh has quit IRC (Client Quit) 2011-06-11T18:35:43 *** xQuasar has joined #aichallenge 2011-06-11T18:38:14 *** EQRF has left #aichallenge ("http://www.last.fm/music/earthquake+retrofit") 2011-06-11T18:51:02 *** akranis has quit IRC (Quit: Lämnar) 2011-06-11T19:12:08 *** olexs has joined #aichallenge 2011-06-11T19:19:58 *** paratrechina has joined #aichallenge 2011-06-11T19:28:51 *** paratrechina has quit IRC (Quit: Page closed) 2011-06-11T19:49:11 *** olexs has quit IRC (Quit: Leaving.) 2011-06-11T19:51:13 *** carlop has quit IRC (Ping timeout: 260 seconds) 2011-06-11T19:55:06 *** MuTa-ofd00m has quit IRC (Ping timeout: 255 seconds) 2011-06-11T20:05:52 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-11T20:19:11 *** Kingpin13 has quit IRC (Quit: this used to be the life, but I don't need another one) 2011-06-11T20:46:07 aichallenge: Marco Leise epsilon * ra50f0f8 / (2 files): 2011-06-11T20:46:07 aichallenge: visualizer: GET parameter 'user' will give the user with that id the 2011-06-11T20:46:07 aichallenge: first color; useful to identify the own bot when clicking on games in 2011-06-11T20:46:07 aichallenge: your profile - http://bit.ly/iy6dbD 2011-06-11T20:53:33 @later tell McLeopold: I didn't want to mess with the php files, especially since I don't have a working setup with database to test my changes. Could you add the current user id to game links on the profile page? The issue is here: https://github.com/aichallenge/aichallenge/issues/176 2011-06-11T20:53:33 Frontier: Job's done. 2011-06-11T21:21:19 *** FireFly has quit IRC (Quit: swatted to death) 2011-06-11T21:26:02 *** xQuasar has quit IRC (Ping timeout: 268 seconds) 2011-06-11T21:31:47 *** xQuasar has joined #aichallenge 2011-06-11T21:35:29 *** Quasar has joined #aichallenge 2011-06-11T21:35:55 *** Quasar is now known as Guest87571 2011-06-11T21:38:21 *** xQuasar has quit IRC (Ping timeout: 252 seconds) 2011-06-11T21:48:06 *** Cyndre_ has joined #aichallenge 2011-06-11T21:52:04 *** Cyndre has quit IRC (Ping timeout: 260 seconds) 2011-06-11T22:07:59 *** Antimatter__ has joined #aichallenge 2011-06-11T22:10:02 *** Antimatter__ has quit IRC (Client Quit) 2011-06-11T22:50:05 *** Eruonen has quit IRC () 2011-06-11T22:51:04 *** amstan has joined #aichallenge 2011-06-11T22:51:04 *** ChanServ sets mode: +o amstan 2011-06-11T23:00:56 *** Accoun has quit IRC () 2011-06-11T23:13:03 *** dlila has quit IRC (Quit: Leaving) 2011-06-11T23:18:00 *** smiley1983 has joined #aichallenge 2011-06-11T23:18:22 *** Accoun has joined #aichallenge 2011-06-11T23:54:17 smiley1983: those maze map with "thicker" than one ant corridors is awesome! 2011-06-11T23:54:23 http://ants.zeroviz.us/8ZU9b/ that 2011-06-11T23:55:39 i would suggest making the ends lines more circular or square 2011-06-11T23:57:13 who was i talking to about using markov chains about tracking the probabilities of where ants are? 2011-06-11T23:57:17 i just became a fan of that idea again