2011-11-23T00:02:32 *** delt0r_ has joined #aichallenge 2011-11-23T00:04:54 *** smarsh has quit IRC (Quit: Page closed) 2011-11-23T00:10:11 *** Antimony has joined #aichallenge 2011-11-23T00:10:35 http://ants.fluxid.pl/replay.11570 meh 2011-11-23T00:10:43 all the other hills die early so my bots just go into a stalemate 2011-11-23T00:10:48 half my games are like this now >.< 2011-11-23T00:12:36 i'm not sure, but i think red has enough ants to take purple 2011-11-23T00:12:40 *** praveen has joined #aichallenge 2011-11-23T00:13:19 :P 2011-11-23T00:27:21 *** Areks has joined #aichallenge 2011-11-23T00:43:56 *** Palmik has joined #aichallenge 2011-11-23T00:52:17 @later tell amstan the planetwars site should be completely setup on the archive server, tron is broken as it apparently needs Zend 2011-11-23T00:52:17 janzert1: I think that worked... 2011-11-23T00:55:58 @later i think is not acceptable, give me a straight answer. 2011-11-23T00:55:59 roflmao: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2011-11-23T00:56:31 *** janzert1 is now known as janzert 2011-11-23T01:00:13 *** besh has joined #aichallenge 2011-11-23T01:03:06 *** aarossig has quit IRC (Read error: Connection reset by peer) 2011-11-23T01:07:01 bwark 2011-11-23T01:09:31 *** xar0l has joined #aichallenge 2011-11-23T01:11:36 *** aarossig has joined #aichallenge 2011-11-23T01:12:28 *** kiv has joined #aichallenge 2011-11-23T01:14:50 *** roflmao has quit IRC (Quit: Leaving.) 2011-11-23T01:22:57 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-11-23T01:34:25 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-23T01:44:14 *** xar0l has quit IRC (Quit: Page closed) 2011-11-23T01:49:06 *** praveen has quit IRC (Ping timeout: 265 seconds) 2011-11-23T01:49:28 *** ThatSnail has quit IRC (Ping timeout: 240 seconds) 2011-11-23T01:52:39 *** Euphony has joined #aichallenge 2011-11-23T01:56:08 *** HaraKiri has joined #aichallenge 2011-11-23T01:56:31 *** Fandekasp has joined #aichallenge 2011-11-23T01:59:48 *** analyst74 has quit IRC (Quit: HydraIRC -> http://www.hydrairc.com <- Nine out of ten l33t h4x0rz prefer it) 2011-11-23T02:00:02 *** Fandekasp has quit IRC (Client Quit) 2011-11-23T02:02:11 *** peyton has quit IRC (Quit: peyton) 2011-11-23T02:02:40 *** choas has joined #aichallenge 2011-11-23T02:03:10 *** Garf has quit IRC (Quit: Make a new plan, Stan!) 2011-11-23T02:04:48 *** Garf has joined #aichallenge 2011-11-23T02:07:13 *** Antimony has quit IRC (Ping timeout: 258 seconds) 2011-11-23T02:09:41 *** flag has quit IRC (Ping timeout: 252 seconds) 2011-11-23T02:10:57 *** kara has joined #aichallenge 2011-11-23T02:11:32 *** b0rder has quit IRC (Quit: 离开) 2011-11-23T02:12:20 *** HaraKiri has quit IRC (Ping timeout: 260 seconds) 2011-11-23T02:12:50 *** peyton has joined #aichallenge 2011-11-23T02:22:32 *** Fandekasp has joined #aichallenge 2011-11-23T02:24:37 *** choas has quit IRC (Ping timeout: 240 seconds) 2011-11-23T02:28:07 *** Fandekasp has joined #aichallenge 2011-11-23T02:28:42 *** Fandekasp has joined #aichallenge 2011-11-23T02:41:34 *** Fandekasp has quit IRC (Quit: Lost terminal) 2011-11-23T02:44:28 *** peyton has quit IRC (Quit: peyton) 2011-11-23T02:51:09 I think I finally fixed foraging 2011-11-23T02:51:11 http://paste.aichallenge.org/JN7u7/ 2011-11-23T02:51:26 going to bed now xD 2011-11-23T03:04:44 *** bmh has joined #aichallenge 2011-11-23T03:04:44 *** bmh has joined #aichallenge 2011-11-23T03:09:35 *** treeform has quit IRC (Read error: Connection reset by peer) 2011-11-23T03:09:43 *** treeform has joined #aichallenge 2011-11-23T03:11:14 *** bmh has quit IRC (Quit: bmh) 2011-11-23T03:11:40 *** epicmonkey has joined #aichallenge 2011-11-23T03:12:14 *** Rav3nsW0rd has quit IRC (Quit: Leaving) 2011-11-23T03:13:00 *** magiik has quit IRC (Ping timeout: 260 seconds) 2011-11-23T03:13:16 *** g0llum has joined #aichallenge 2011-11-23T03:14:59 *** Conorach has joined #aichallenge 2011-11-23T03:15:51 *** speakeasy has joined #aichallenge 2011-11-23T03:17:05 *** epicmonkey has quit IRC (Ping timeout: 244 seconds) 2011-11-23T03:17:55 *** aerique has joined #aichallenge 2011-11-23T03:18:06 *** mj41 has joined #aichallenge 2011-11-23T03:21:37 *** Euphony has quit IRC (Quit: Page closed) 2011-11-23T03:22:35 *** ikaros has joined #aichallenge 2011-11-23T03:22:53 *** Cyndre has quit IRC (Quit: Leaving) 2011-11-23T03:23:15 hi there, can you give me a hand at finding the lowest value in collum 3 of a python list? 2011-11-23T03:24:04 queue* 2011-11-23T03:24:56 *** Rav3nsW0rd has joined #aichallenge 2011-11-23T03:26:55 *** Rav3nsW0rd has quit IRC (Read error: Connection reset by peer) 2011-11-23T03:27:26 *** Rav3nsW0rd has joined #aichallenge 2011-11-23T03:27:40 *** Blkt has joined #aichallenge 2011-11-23T03:29:11 *** grwip has joined #aichallenge 2011-11-23T03:34:37 *** mj41 has quit IRC (Ping timeout: 258 seconds) 2011-11-23T03:37:28 *** thestinger has quit IRC (Ping timeout: 240 seconds) 2011-11-23T03:37:40 *** Rav3nsW0rd has quit IRC (Quit: Leaving) 2011-11-23T03:37:52 *** thestinger has joined #aichallenge 2011-11-23T03:38:06 *** Rav3nsW0rd has joined #aichallenge 2011-11-23T03:43:27 *** Rav3nsW0rd has quit IRC (Quit: Leaving) 2011-11-23T03:47:53 *** mj41 has joined #aichallenge 2011-11-23T03:49:41 *** muhkuh has joined #aichallenge 2011-11-23T03:51:57 *** muhkuh has quit IRC (Client Quit) 2011-11-23T03:53:49 *** treeform has quit IRC (Remote host closed the connection) 2011-11-23T03:54:01 *** Rav3nsW0rd has joined #aichallenge 2011-11-23T04:00:48 *** magiik has joined #aichallenge 2011-11-23T04:04:53 *** Jak_o_Shadows has joined #aichallenge 2011-11-23T04:19:53 *** gcflymoto has joined #aichallenge 2011-11-23T04:27:35 http://ants.fluxid.pl/replay.11225 I wonder what makes A's ants suddenly spread, and then group again, at turn 544 or 584 for example 2011-11-23T04:28:39 *** Surya has joined #aichallenge 2011-11-23T04:28:40 *** ALplus has joined #aichallenge 2011-11-23T04:31:44 Hey there. I read some new servers have been added. o/ 2011-11-23T04:32:57 oh, eight servers 2011-11-23T04:33:13 prices must be low today 2011-11-23T04:34:30 *** NotABug has joined #aichallenge 2011-11-23T04:35:31 *** Fandekasp has joined #aichallenge 2011-11-23T04:44:24 *** antimatroid has quit IRC (Ping timeout: 244 seconds) 2011-11-23T04:52:05 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-23T04:58:11 *** sigh has joined #aichallenge 2011-11-23T05:11:45 *** mcstar has joined #aichallenge 2011-11-23T05:13:46 *** ALplus has quit IRC () 2011-11-23T05:16:55 *** antimatroid has joined #aichallenge 2011-11-23T05:34:49 *** gcflymoto has quit IRC (Ping timeout: 265 seconds) 2011-11-23T05:40:27 *** peyton has joined #aichallenge 2011-11-23T06:31:52 *** TheLinker has joined #aichallenge 2011-11-23T06:32:49 *** NotABug has quit IRC (Ping timeout: 265 seconds) 2011-11-23T06:34:48 *** praveen has joined #aichallenge 2011-11-23T06:54:07 good day everyone 2011-11-23T07:04:29 *** flag has joined #aichallenge 2011-11-23T07:08:51 *** pairofdice has joined #aichallenge 2011-11-23T07:10:52 *** mj41 has quit IRC (Ping timeout: 245 seconds) 2011-11-23T07:11:43 *** b0rder has joined #aichallenge 2011-11-23T07:25:27 *** mj41 has joined #aichallenge 2011-11-23T07:41:10 *** Conorach_ has joined #aichallenge 2011-11-23T07:43:47 *** Conorach has quit IRC (Ping timeout: 245 seconds) 2011-11-23T07:46:06 *** besh has quit IRC (Quit: Page closed) 2011-11-23T07:46:13 *** flag has quit IRC (Ping timeout: 240 seconds) 2011-11-23T07:50:19 *** u_ has joined #aichallenge 2011-11-23T07:50:24 *** delt0r_ has quit IRC (Ping timeout: 244 seconds) 2011-11-23T07:54:40 *** speakeasy has quit IRC (Ping timeout: 240 seconds) 2011-11-23T07:57:39 *** Fandekasp has quit IRC (Quit: Lost terminal) 2011-11-23T08:01:12 *** amstan_ has joined #aichallenge 2011-11-23T08:01:12 *** ChanServ sets mode: +o amstan_ 2011-11-23T08:03:17 *** delt0r_ has joined #aichallenge 2011-11-23T08:09:52 *** Fandekasp has joined #aichallenge 2011-11-23T08:15:27 *** ikaros has joined #aichallenge 2011-11-23T08:41:47 *** u_ has quit IRC (Quit: u_) 2011-11-23T08:44:53 Fluxid: I see that the behavior of A_laptop was enemy dependent (enable coverage view and you see its behavior changing every time it discovers a new enemy) 2011-11-23T08:57:20 *** jotauve has joined #aichallenge 2011-11-23T08:58:23 hi guys 2011-11-23T08:58:31 I need help with C++ 2011-11-23T08:58:37 about the tutorial.sh script 2011-11-23T08:58:42 can anybody help me please? 2011-11-23T09:01:49 ...Go ahead 2011-11-23T09:04:40 whats that? 2011-11-23T09:05:16 *** rajanaresh has quit IRC (Ping timeout: 252 seconds) 2011-11-23T09:05:48 I think I already got that 2011-11-23T09:06:14 it was how to open MyBot with the play_one_game script 2011-11-23T09:06:29 but after looking for a while in the forum I already fot that 2011-11-23T09:06:41 the forum search engine doesn't work really well hehe 2011-11-23T09:06:45 thank you anyway! 2011-11-23T09:07:51 run, not open, right? 2011-11-23T09:07:54 *** antonh has quit IRC (Ping timeout: 244 seconds) 2011-11-23T09:08:48 yes run 2011-11-23T09:08:59 I already got to run the play_one_game.sh 2011-11-23T09:09:03 with my bot 2011-11-23T09:09:08 but the tutorial.sh the documentation says 2011-11-23T09:09:12 I'm not able to run it 2011-11-23T09:09:12 *** sigh has quit IRC (Remote host closed the connection) 2011-11-23T09:09:19 *** ThatSnail has joined #aichallenge 2011-11-23T09:09:20 it returns this 2011-11-23T09:09:27 *** antonh has joined #aichallenge 2011-11-23T09:09:37 raceback (most recent call last): File "/home/javic3/Desktop/aichallenge/tools/engine.py", line 93, in run_game sandbox.start(bot_cmd) File "/home/javic3/Desktop/aichallenge/tools/sandbox.py", line 385, in start raise SandboxError('Failed to start {0}'.format(shell_command)) SandboxError: Failed to start ['./uc3m_bot/MyBot'] 2011-11-23T09:10:48 jotauve: use the full path to your bot 2011-11-23T09:11:39 oh jesus 2011-11-23T09:11:42 I already got it 2011-11-23T09:11:45 haha 2011-11-23T09:11:52 *** Harpyon has joined #aichallenge 2011-11-23T09:11:57 thank you very much mcstar 2011-11-23T09:13:56 *** AntDroid has joined #aichallenge 2011-11-23T09:21:16 *** jotauve has quit IRC (Quit: Page closed) 2011-11-23T09:23:55 *** Surya has quit IRC (Ping timeout: 265 seconds) 2011-11-23T09:27:25 *** Areks has quit IRC (Ping timeout: 240 seconds) 2011-11-23T09:29:52 *** AlliedEnvy has quit IRC (Ping timeout: 240 seconds) 2011-11-23T09:35:38 it's funny how hard some problems can be 2011-11-23T09:41:27 :) 2011-11-23T09:41:45 i suck at writing the correct evaluation function 2011-11-23T09:42:06 i cant make a general one 2011-11-23T09:42:12 well, generally good one 2011-11-23T09:45:40 *** praveen has quit IRC (Ping timeout: 265 seconds) 2011-11-23T09:47:30 *** cbad has quit IRC (Read error: Connection reset by peer) 2011-11-23T09:52:26 *** cbad has joined #aichallenge 2011-11-23T09:52:26 *** cbad has joined #aichallenge 2011-11-23T09:59:47 *** mrspandex has joined #aichallenge 2011-11-23T10:00:17 *** peyton has quit IRC (Quit: peyton) 2011-11-23T10:00:49 Hey All, I was thinking of multithreading my algorithm to improve perfromance - does anyone know if this is worthwhile on the real challenge server? 2011-11-23T10:01:25 *** Palmik has quit IRC (Ping timeout: 240 seconds) 2011-11-23T10:02:04 mrspandex, unless it's green threads, forget that 2011-11-23T10:04:24 multithreading isn't allowed 2011-11-23T10:10:34 *** Palmik has joined #aichallenge 2011-11-23T10:14:28 *** foRei has joined #aichallenge 2011-11-23T10:16:57 ah, well thanks for saving me the time 2011-11-23T10:17:56 yea, we all had that idea .. ;( 2011-11-23T10:19:48 that would be mobile phone against supercomputer :p 2011-11-23T10:26:02 *** userjjb has joined #aichallenge 2011-11-23T10:27:14 *** Palmik has quit IRC (Ping timeout: 252 seconds) 2011-11-23T10:29:26 i'm still waiting for someone hacking the university network into a hive mind playing on tcp.. 2011-11-23T10:31:26 most of us have access to clusters, so why hack? 2011-11-23T10:32:01 i have a cluster of 2 and a half. 2011-11-23T10:34:12 i have access to ~10 fermi gpu's and say 80 xenon core 2011-11-23T10:34:19 What's the max allowed memory usage for a bot? 2011-11-23T10:34:34 on a supermicro workstation ~ 48 gb 2011-11-23T10:34:39 mcstar , ;) go for it ! 2011-11-23T10:35:05 userjjb, like 2 gb 2011-11-23T10:35:10 ok 2011-11-23T10:35:17 efficiently using such a system, would require way too much work 2011-11-23T10:35:38 and probably wouldnt let me to run a "bot" anyway 2011-11-23T10:36:05 2? not 1 gb? 2011-11-23T10:36:10 k, ok, my thoughts just running wild there .. 2011-11-23T10:36:42 *** tobym has joined #aichallenge 2011-11-23T10:47:34 *** mrspandex has quit IRC (Quit: Page closed) 2011-11-23T10:50:28 *** Areks has joined #aichallenge 2011-11-23T10:52:28 *** praveen has joined #aichallenge 2011-11-23T10:53:09 *** UncleVasya has joined #aichallenge 2011-11-23T10:53:21 lol: http://aichallenge.org/visualizer.php?game=119352&user=53 2011-11-23T10:53:59 *** b0rder has quit IRC (Ping timeout: 240 seconds) 2011-11-23T10:55:36 i *LOVE* it 2011-11-23T10:56:11 wow 2011-11-23T10:57:02 ;) and both of you won ! 2011-11-23T10:58:59 both stinger won - http://ants.fluxid.pl/replay.12309 - see the formation in the end 2011-11-23T11:00:35 UncleVasya: very nice play 2011-11-23T11:01:03 ty 2011-11-23T11:04:26 bugs 2011-11-23T11:04:27 i hate them 2011-11-23T11:05:11 * avdg wants to kill division by zero for modulus 2011-11-23T11:06:05 antimatroid: what were you referring to earlier? 2011-11-23T11:06:06 x = value != 0 ? value % max : 0; <- I don't know why this fails 2011-11-23T11:06:19 when? 2011-11-23T11:06:22 earlier 2011-11-23T11:06:24 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-23T11:06:52 *** g0llum has joined #aichallenge 2011-11-23T11:06:55 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-23T11:07:09 avdg: what are you trying o achieve? 2011-11-23T11:07:12 t* 2011-11-23T11:07:29 0%max? 2011-11-23T11:07:31 value should be in range max, but its allowed to be negative as positive 2011-11-23T11:07:44 advg, you never imagine, that all your hives aer gone ? 2011-11-23T11:07:50 (and I don't want to keep it negative or positive) 2011-11-23T11:08:08 meh, I don't care about bot performance atm 2011-11-23T11:08:15 I'm more worried about my c++ skills 2011-11-23T11:09:00 about the expression: looks like I still have that division by zero 2011-11-23T11:09:33 i think 0%anything is 0 2011-11-23T11:10:02 yeah 2011-11-23T11:10:02 Yea 2011-11-23T11:10:15 shouldnt you be more worried about max not being 0? 2011-11-23T11:10:19 * avdg checks his variables 2011-11-23T11:10:27 if anythinf is 0, that's dividing by zero 2011-11-23T11:10:41 yeah, I got it... 2011-11-23T11:10:51 g0llum: ok, i meant something, not anything :D 2011-11-23T11:10:57 hehe 2011-11-23T11:11:01 using it before its declared 2011-11-23T11:11:16 and no errors 2011-11-23T11:11:27 I'm probably relying too much on them atm 2011-11-23T11:11:40 declared? 2011-11-23T11:11:48 that throws a compile time error 2011-11-23T11:11:55 bleh declared/defined 2011-11-23T11:11:59 not the same 2011-11-23T11:12:15 so x is not initilized before you use it? 2011-11-23T11:12:37 max was not set 2011-11-23T11:12:44 it doesn't matter for x 2011-11-23T11:13:34 antimatroid: "it's funny how hard some problems can be" 2011-11-23T11:14:15 oh, :P 2011-11-23T11:17:57 btw, whois simple8DFMMFDGR on tcpants ? 2011-11-23T11:21:42 *** Antimony has joined #aichallenge 2011-11-23T11:22:03 *** d0nss1 has joined #aichallenge 2011-11-23T11:22:04 come out of your hive, we all know, you're here ;) 2011-11-23T11:28:02 that didn't work, nvm. 2011-11-23T11:28:13 *** Areks has quit IRC (Ping timeout: 240 seconds) 2011-11-23T11:35:32 awesome: http://aichallenge.org/visualizer.php?game=117892&user=432 2011-11-23T11:36:09 pwned 2 turns before turn limit 2011-11-23T11:38:08 *** rajanaresh has joined #aichallenge 2011-11-23T11:39:28 n1 a1k0n, pretty aggressive fights too. 2011-11-23T11:40:26 yeah it's definitely a little too gung-ho 2011-11-23T11:43:01 *** JorgeB has quit IRC (Ping timeout: 258 seconds) 2011-11-23T11:46:51 a1k0n: for sure that orange has map prediction 2011-11-23T11:49:25 my bot gets so lost on that map unless someone initiates an attack and it can follow the ants back to the source 2011-11-23T11:50:21 *** d0nss1 has quit IRC (Quit: Page closed) 2011-11-23T11:50:26 a1k0n: hi. I try to upload a zip with greedy_bot but got 'General upload error: 3' few times. What's this problem about? 2011-11-23T11:50:37 sry, I mean amstan_: 2011-11-23T11:52:10 *** amstan_ has quit IRC (Ping timeout: 260 seconds) 2011-11-23T11:52:36 weird. https://github.com/aichallenge/aichallenge/blob/epsilon/website/check_submit.php#L14 2011-11-23T11:52:37 nvm, success. 2011-11-23T11:52:51 what was it? 2011-11-23T11:54:34 Greedy bot is buggy 2011-11-23T11:55:04 UncleVasya, yea, anyone, who ends up below, di not try hard enough. 2011-11-23T11:57:59 $errors[] = "Nuh-uh. The contest is over. No more submissions."; 2011-11-23T11:59:09 *** raemde_ has joined #aichallenge 2011-11-23T11:59:21 *** raemde_ is now known as raemde 2011-11-23T11:59:57 *** JorgeB has joined #aichallenge 2011-11-23T12:01:27 *** IchBinDasWir has quit IRC (Read error: Operation timed out) 2011-11-23T12:04:56 *** smarsh has joined #aichallenge 2011-11-23T12:05:17 . 2011-11-23T12:06:08 *** McLeopold has left #aichallenge 2011-11-23T12:11:02 all i ever get on fluxid's i s a bloody nose.. 2011-11-23T12:12:36 Are bots on TCP much stronger than top20 on the official server? 2011-11-23T12:12:38 every day my bot goes down 10 in the tcp server rankings and I have to improve it to climb back up 2011-11-23T12:13:21 UncleVasya: well, it's where a lot of them test/run their new versions 2011-11-23T12:13:31 the bots on the site are mostly ancient 2011-11-23T12:13:45 and you get a huge amount of time for your turn on the tcp server 2011-11-23T12:14:12 so if people have some sort of game tree/payoff matrix thing they are going to be much stronger 2011-11-23T12:14:45 knapsack01 ? 2011-11-23T12:18:29 luckily most improvements to my bot just involve running a bunch of versions to find the best constant/equation for something 2011-11-23T12:19:24 *** ikaros has joined #aichallenge 2011-11-23T12:20:50 again, it's fun, nonetheless. 2011-11-23T12:21:58 yeah 2011-11-23T12:24:33 thestinger, you really mutiplied, like agent smith 2011-11-23T12:25:31 nvm, .. 2011-11-23T12:26:43 *** seletskiy has quit IRC (Ping timeout: 244 seconds) 2011-11-23T12:28:24 *** seletskiy has joined #aichallenge 2011-11-23T12:29:29 since, *tomorrow*, i'll have an .idea, to take rhem all down! hehe 2011-11-23T12:31:03 *** Blkt has quit IRC (Remote host closed the connection) 2011-11-23T12:32:02 eliminated, once again, now make that a week ;) 2011-11-23T12:33:04 *** bmh has joined #aichallenge 2011-11-23T12:34:06 strcat_nplayers better win a bunch of games now... 2011-11-23T12:34:21 *** mj41 has quit IRC (Read error: Operation timed out) 2011-11-23T12:35:16 *** epicmonkey has joined #aichallenge 2011-11-23T12:35:20 http://ants.fluxid.pl/replay.12425 that's the first game after fixing a horrible bug 2011-11-23T12:37:39 uh oh 2011-11-23T12:37:53 nope, another loss :( 2011-11-23T12:38:15 http://ants.fluxid.pl/replay.12427 on a terrible map though 2011-11-23T12:38:47 *** Antimony has quit IRC (Ping timeout: 245 seconds) 2011-11-23T12:39:15 *** Antimony has joined #aichallenge 2011-11-23T12:39:49 can't win em all 2011-11-23T12:45:34 *** Harpyon has quit IRC (Quit: Computer has gone to sleep.) 2011-11-23T12:46:28 *** Harpyon has joined #aichallenge 2011-11-23T12:52:39 *** Hexren has joined #aichallenge 2011-11-23T12:52:41 *** Hexren has left #aichallenge 2011-11-23T12:54:53 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-23T12:57:14 *** bhasker has joined #aichallenge 2011-11-23T12:57:54 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-11-23T13:01:09 *** Antvolution has joined #aichallenge 2011-11-23T13:12:22 *** Cyndre has joined #aichallenge 2011-11-23T13:17:45 *** Redgis has joined #aichallenge 2011-11-23T13:20:43 *** Rav3nsW0rd has quit IRC (Excess Flood) 2011-11-23T13:21:31 *** Rav3nsW0rd has joined #aichallenge 2011-11-23T13:24:24 *** Antimony_ has joined #aichallenge 2011-11-23T13:25:52 *** Antimony has quit IRC (Ping timeout: 245 seconds) 2011-11-23T13:26:04 *** Antimony_ is now known as Antimony 2011-11-23T13:28:03 *** epicmonkey has quit IRC (Ping timeout: 258 seconds) 2011-11-23T13:28:21 *** treeform has joined #aichallenge 2011-11-23T13:28:47 *** treeform has joined #aichallenge 2011-11-23T13:31:52 *** smarsh has quit IRC (Ping timeout: 265 seconds) 2011-11-23T13:39:34 *** Palmik has joined #aichallenge 2011-11-23T13:41:47 *** xathis has joined #aichallenge 2011-11-23T13:49:42 *** amstan has joined #aichallenge 2011-11-23T13:49:42 *** ChanServ sets mode: +o amstan 2011-11-23T13:50:30 thestinger: Do you use game tree/minimax ? 2011-11-23T13:52:22 not atm 2011-11-23T13:52:29 rewriting combat right now though 2011-11-23T13:52:56 the current bots playing on the tcp server just assume the enemy doesn't move 2011-11-23T13:53:05 my combat branch knows they might move 2011-11-23T13:53:54 Ok strcat_dense is really doing well, so was wondering if its using gametree 2011-11-23T13:54:12 nope, it's just a bugfix for movement 2011-11-23T13:54:12 that really answers what proper foraging and little defense could do. 2011-11-23T13:54:37 well, my combat basically just rips apart the static defenses - if they are exposed 2011-11-23T13:54:58 in a small area where it can't surround them it will just result in a stalemate 2011-11-23T13:54:59 what is the name of the bot, combat branch one 2011-11-23T13:55:06 not running on the server yet 2011-11-23T13:55:10 ok 2011-11-23T13:55:12 all of them can deal with unmoving enemies tho 2011-11-23T13:55:39 strcat_combatzones was my old combat branch but it never worked 2011-11-23T13:55:47 ok got to implement them, still in early stages 2011-11-23T13:55:51 it just took a lot longer to do the same thing as strcat_rewrite 2011-11-23T13:56:06 and I mean 300ms instead of 4ms to do the same thing xD 2011-11-23T13:56:42 from 300ms to 4ms is impressive 2011-11-23T13:57:00 from 4 ms to 300 ms is not 2011-11-23T13:57:11 ^ yeah, that 2011-11-23T13:57:16 oh sorry :) 2011-11-23T13:57:39 we can still call that impressive 2011-11-23T14:05:07 *** clapautius has joined #aichallenge 2011-11-23T14:09:11 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-11-23T14:11:01 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-23T14:13:18 *** LouisMartin has joined #aichallenge 2011-11-23T14:14:14 About rules, is there any mention of maximum bot (source code) size? 2011-11-23T14:14:35 well, I think the max size for the zip is 2MB 2011-11-23T14:14:51 and there's probably a compilation time limit 2011-11-23T14:15:54 ok, do you remember where you've read that or heard that from? 2011-11-23T14:16:05 On this channel from the admins 2011-11-23T14:17:06 https://github.com/aichallenge/aichallenge/blob/epsilon/worker/compiler.py 2011-11-23T14:17:14 looks like the default timelimit is 600 secs (10 mins) 2011-11-23T14:17:47 so, don't include the linux kernel in your bot :P 2011-11-23T14:17:48 2MB compressed text is something like 8MB? 2011-11-23T14:17:50 *** Anilm3 has joined #aichallenge 2011-11-23T14:17:59 Hi 2011-11-23T14:18:24 I bet funrolled loops compress quite well 2011-11-23T14:19:00 damn, guess my Linux kernel bot is out :~( lol 2011-11-23T14:19:17 thanks for the info thestinger 2011-11-23T14:19:57 iirc, you can get ask for more than 2MB if you need it (I think I read that on the forum) 2011-11-23T14:20:22 pairofdice: depends on many factors, but could be a lot. 8 mb sounds normal, yep 2011-11-23T14:28:41 *** Euphony has joined #aichallenge 2011-11-23T14:38:59 *** andar has quit IRC (Ping timeout: 244 seconds) 2011-11-23T14:43:46 * avdg wonders about the advantages of a linux bot 2011-11-23T14:46:18 *** clapautius has quit IRC (Ping timeout: 265 seconds) 2011-11-23T14:51:41 *** andar has joined #aichallenge 2011-11-23T14:52:06 They're all linx botten 2011-11-23T14:54:02 *** LouisMartin has quit IRC (Ping timeout: 265 seconds) 2011-11-23T14:56:18 *** Accoun has quit IRC () 2011-11-23T15:00:20 *** thestinger has quit IRC (Quit: be back later) 2011-11-23T15:02:36 yippie my bot kills a random bot 2011-11-23T15:02:44 in front of water 2011-11-23T15:04:25 i can hear you all say: oh, thats soo lame 2011-11-23T15:04:33 i dont even commont on it 2011-11-23T15:04:34 pfff 2011-11-23T15:04:40 e* 2011-11-23T15:06:26 *** moises_ has joined #aichallenge 2011-11-23T15:09:03 whoa 2011-11-23T15:09:06 that is soo nice 2011-11-23T15:09:13 it kills through water 2011-11-23T15:10:33 *** raemde has quit IRC (Read error: Connection reset by peer) 2011-11-23T15:11:40 *** flag has joined #aichallenge 2011-11-23T15:13:01 *** raemde has joined #aichallenge 2011-11-23T15:14:18 *** Fandekasp has quit IRC (Read error: Operation timed out) 2011-11-23T15:20:08 *** mj41 has joined #aichallenge 2011-11-23T15:20:47 *** rajanaresh has quit IRC (Ping timeout: 240 seconds) 2011-11-23T15:21:23 *** Accoun has joined #aichallenge 2011-11-23T15:24:27 mcstar: hmm, water only restricts movements right? I thought you know that already 2011-11-23T15:24:42 ? 2011-11-23T15:24:52 what makes you think i didnt know? 2011-11-23T15:25:03 just wondering 2011-11-23T15:25:32 it positioned itself right on water, and killed the enemies spawning from the hill 2011-11-23T15:25:43 (right on->right beside) 2011-11-23T15:35:31 which map? 2011-11-23T15:36:55 maze 04 01 2011-11-23T15:37:29 the hill is mroe protected, but stupid ants went for the water... 2011-11-23T15:40:27 *** mj41 has quit IRC (Ping timeout: 245 seconds) 2011-11-23T15:48:58 *** praveen has quit IRC (Quit: Page closed) 2011-11-23T15:50:47 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-23T15:51:36 *** Palmik has quit IRC (Remote host closed the connection) 2011-11-23T15:55:44 *** Rinum has joined #aichallenge 2011-11-23T15:56:02 what's the difference between Collaborative Diffusion and Influence Maps? 2011-11-23T15:56:17 i guess nothings 2011-11-23T15:56:23 Not much 2011-11-23T15:56:39 i think its an umbrella for quite similar methods 2011-11-23T15:57:28 seems like that collaborative diffusion article is total bs 2011-11-23T15:57:35 just a rip off of influence maps 2011-11-23T15:57:38 which one? 2011-11-23T15:57:51 the only one that comes up on google 2011-11-23T15:58:09 http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&ved=0CDAQFjAB&url=http%3A%2F%2Fwww.cs.colorado.edu%2F~ralex%2Fpapers%2FPDF%2FOOPSLA06antiobjects.pdf&ei=WV7NTuWIC-fU2AWl8JSlDw&usg=AFQjCNEH_hMVBLfFWPZqB2HMxpdYRTrQMA 2011-11-23T15:58:18 www.cs.colorado.edu/~ralex/papers/PDF/OOPSLA06antiobjects.pdf 2011-11-23T15:58:32 ive seen that after someone linked it in here 2011-11-23T15:58:39 seemed ok 2011-11-23T15:58:48 it's exactly the same as influence maps though 2011-11-23T15:58:55 ah yes 2011-11-23T15:59:19 ive never heard collaborative diffusion before someone started saying it here 2011-11-23T15:59:27 *** epicmonkey has joined #aichallenge 2011-11-23T15:59:31 right same here 2011-11-23T15:59:59 i think they've got the wrong idea if they think the dude that wrote the paper created the idea 2011-11-23T16:00:04 but i came across some pretty old stuff on influence maps 2011-11-23T16:00:08 some decades old 2011-11-23T16:00:20 http://www.gamedev.net/page/resources/_/technical/artificial-intelligence/the-core-mechanics-of-influence-mapping-r2799 2011-11-23T16:00:21 right 2011-11-23T16:00:36 that one is fairly recent i think? 2011-11-23T16:00:39 and similar stuff appears in papers from 70's 2011-11-23T16:00:48 but i think they dont call it that 2011-11-23T16:01:24 yeah 2011-11-23T16:01:28 i linked that page in ~2 weeks ago? 2011-11-23T16:02:02 btw, the area graphs are more interesting in that article than the influence stuff 2011-11-23T16:02:10 no that it tells much about anything 2011-11-23T16:02:27 well, ok, it has a code snippet 2011-11-23T16:02:32 i forgot about that 2011-11-23T16:02:59 i stripped my bot of any kind of influence stuff weeks ago 2011-11-23T16:03:05 lol 2011-11-23T16:03:14 didn't work out? 2011-11-23T16:03:22 *** delt0r_ has joined #aichallenge 2011-11-23T16:03:23 its black magic 2011-11-23T16:03:28 and dumb 2011-11-23T16:03:51 i mean if you have much time to tune things idk, it might work out 2011-11-23T16:03:58 lol 2011-11-23T16:04:05 but for the precision stuff im doing now its unacceptable 2011-11-23T16:04:17 but i can imagine using it for some global stuff 2011-11-23T16:04:22 so I'm guessing you're sticking to good old fashioned path-finding? 2011-11-23T16:04:25 but i dont need it right now 2011-11-23T16:04:46 i never wanted to replace pathfinding with influence thingy 2011-11-23T16:04:51 i always had pathfinding 2011-11-23T16:05:03 are you using minimax? 2011-11-23T16:05:08 sort of 2011-11-23T16:05:24 the tree is too shallow to call it anything like that 2011-11-23T16:05:25 i think 2011-11-23T16:05:36 but its outcome-enumeration, yes 2011-11-23T16:05:39 well right, it'd have to be some variation of it 2011-11-23T16:06:23 how's that working out? 2011-11-23T16:06:29 its good 2011-11-23T16:06:45 and i dont have speed problems yet 2011-11-23T16:06:55 but my code is a mess right now 2011-11-23T16:07:11 and i have many functinalities that need to be cleaned up 2011-11-23T16:07:23 i need a better coordination framework 2011-11-23T16:07:52 haha better than influence maps 2011-11-23T16:08:06 i hope so 2011-11-23T16:08:22 i take a shower 2011-11-23T16:29:33 *** NezumiTejina has joined #aichallenge 2011-11-23T16:31:08 I have to parallelize some algorithm for school... it has to be an algorithm which hasn't been parallelized... I'm all out of ideas :( 2011-11-23T16:31:29 what sort? 2011-11-23T16:31:37 any kind 2011-11-23T16:31:49 ah, so you can choose 2011-11-23T16:32:00 right 2011-11-23T16:32:14 Parallellize this! 2011-11-23T16:32:26 I've tried the basics.. graph algos, sorting, optimization... all have parallel versions :( 2011-11-23T16:32:48 pairofdice: does something like that exist? lol 2011-11-23T16:32:53 *** foRei has quit IRC (Quit: Bye) 2011-11-23T16:34:03 pairofdice: I guess not.... lol I was expecting some site with list of algos to be parallelized x.x 2011-11-23T16:34:09 ... 2011-11-23T16:34:32 i think meant Ants 2011-11-23T16:34:37 oh lol 2011-11-23T16:34:40 Joke 2011-11-23T16:35:07 *** ltriant has joined #aichallenge 2011-11-23T16:35:11 Rinum: that criteria is pretty tight 2011-11-23T16:35:17 rium* 2011-11-23T16:35:30 yeah... 2011-11-23T16:35:44 i guess you dont need to follow it a 100% 2011-11-23T16:36:02 I actually thought about making a parallel version of an influence mapping algo lol 2011-11-23T16:36:10 I'm not sure if that exists 2011-11-23T16:36:30 *** Euphony has quit IRC (Ping timeout: 265 seconds) 2011-11-23T16:37:11 lol 2011-11-23T16:37:16 what part? 2011-11-23T16:37:39 the spreading the influence part? 2011-11-23T16:37:42 spreading the 'scent'? 2011-11-23T16:37:57 have you heard of partial differential equations? 2011-11-23T16:38:01 i don't know what else could be made concurrent 2011-11-23T16:38:04 yes I have 2011-11-23T16:38:09 so? 2011-11-23T16:38:17 you see the connection 2011-11-23T16:38:28 hmmm 2011-11-23T16:38:49 basically you have a lattice and apply a finite differencemethod on it 2011-11-23T16:39:04 so it has been parallleized to pieces 2011-11-23T16:39:08 *** McLeopold has joined #aichallenge 2011-11-23T16:39:14 i did too 2011-11-23T16:39:14 you're right 2011-11-23T16:39:16 damn 2011-11-23T16:39:20 i did adaptive refinement too 2011-11-23T16:39:23 bad starting position? http://paste.aichallenge.org/hROdr/ 2011-11-23T16:39:44 McLeopold: beautiful 2011-11-23T16:39:52 looks like I'm stuck then :/ 2011-11-23T16:40:19 McLeopold: looks like the doors no wraith ships 2011-11-23T16:40:25 on* 2011-11-23T16:40:25 how about a map where every square is a hill? 2011-11-23T16:40:41 *** thestinger has joined #aichallenge 2011-11-23T16:40:44 uh oh 2011-11-23T16:40:48 aham 2011-11-23T16:40:50 i got it 2011-11-23T16:41:01 it has much diagonal lines 2011-11-23T16:41:03 That looks like the lower intestine 2011-11-23T16:41:09 thats why i like that map(the water) 2011-11-23T16:41:52 McLeopold: youre not kidding? 2011-11-23T16:42:06 what would be the purpose of that? 2011-11-23T16:42:24 its like before hills, when ants could spawn from anywhere 2011-11-23T16:42:26 I am kidding, but it would be interesting 2011-11-23T16:42:40 McLeopold: the map would initially be filled with ants then, and most of them would die instantly 2011-11-23T16:42:51 no 2011-11-23T16:42:53 not most of them 2011-11-23T16:42:55 all would 2011-11-23T16:42:57 only on boundaries 2011-11-23T16:43:04 why? 2011-11-23T16:43:07 depends how the hills are distributed 2011-11-23T16:43:21 on every single square 2011-11-23T16:43:28 no heaans by hives 2011-11-23T16:43:31 i like 2011-11-23T16:43:32 he means* 2011-11-23T16:43:34 I guess you'd have to hope all ants around you collieded 2011-11-23T16:43:48 yes but teams together or randomly placed? 2011-11-23T16:44:14 and that, if you had a chuck of your own hills together 2011-11-23T16:44:43 *** Rinum has quit IRC (Ping timeout: 265 seconds) 2011-11-23T16:45:07 *** moises_ has quit IRC (Remote host closed the connection) 2011-11-23T16:45:09 McLeopold: thats a great map btw, cause ants can run long wo hitting water 2011-11-23T16:45:12 Fluxid: are you around? 2011-11-23T16:45:21 cant* 2011-11-23T16:45:23 cant* 2011-11-23T16:45:48 McLeopold: will you surprise everyone with those cell mazes after submissions are closed or will we see them before? 2011-11-23T16:46:13 we should put them up on fluxid's server, with suitable starting points anyway 2011-11-23T16:46:24 McLeopold: does it support more than just 4-fold mirror symmetry? 2011-11-23T16:46:41 a1k0n: he had translation the other day 2011-11-23T16:47:31 xathis: I'm working hard to get them up a soon as possible 2011-11-23T16:47:36 so i guess i shoudl save it 2011-11-23T16:47:48 McLeopold: you know what we should do? 2011-11-23T16:47:49 McLeopold: that's good 2011-11-23T16:48:04 amstan: shut down the servers? 2011-11-23T16:48:10 McLeopold: a map that's looks like it's symmetric one way, but it's actually symmetric in other ways 2011-11-23T16:48:28 like.. water blocks in a grid, 4x4 repeating grid 2011-11-23T16:48:53 them we can fool the 2 bots that try to detect symmetry... 2011-11-23T16:49:09 i'm sure quite a few detect symmetry 2011-11-23T16:49:37 except they wouldn't be fooled as soon as they find a single inconsistency 2011-11-23T16:49:48 *** se__ has joined #aichallenge 2011-11-23T16:49:56 a1k0n: yeah, they might just segfault :) 2011-11-23T16:50:17 a1k0n: depends on the implementation 2011-11-23T16:50:17 I'm not oppossed, but I don't think it will gain us much 2011-11-23T16:50:32 McLeopold: one map can't hurt 2011-11-23T16:51:04 I think I have this combat crap half-working 2011-11-23T16:52:17 http://paste.aichallenge.org/ugoOB/ 2011-11-23T16:52:42 McLeopold: lol 2011-11-23T16:52:46 what is that, a symmetry test? 2011-11-23T16:52:49 if there is a tcp guy around, please put that very stupid map up :D 2011-11-23T16:52:54 also.. how many hills is that? lol 2011-11-23T16:52:59 it looks like a lot in the corner 2011-11-23T16:53:27 :< 2011-11-23T16:54:08 yeah i think there are extra hills on the edges 2011-11-23T16:54:22 it's nothing but hills 2011-11-23T16:54:40 yeah, the hills on the edges are displayed oddly 2011-11-23T16:55:16 I think it is hill picture overlap 2011-11-23T16:55:45 yeah, the raw map looks fine 2011-11-23T16:57:25 http://paste.aichallenge.org/Dbk8a/ 2011-11-23T16:57:33 *** Redgis has quit IRC (Ping timeout: 244 seconds) 2011-11-23T17:01:55 I wonder if that map was designed to crash bots 2011-11-23T17:03:16 http://ants.fluxid.pl/replay.12716 this almost makes me think my combat is working 2011-11-23T17:04:34 there are definitely lots of weird bugs 2011-11-23T17:08:51 *** LouisMartin has joined #aichallenge 2011-11-23T17:12:14 Rinum: That's the topic for a thesis 2011-11-23T17:12:24 I hope you're not getting that assignment in some undergrad class 2011-11-23T17:15:39 *** AlliedEnvy has joined #aichallenge 2011-11-23T17:16:08 *** se__ has quit IRC (Ping timeout: 265 seconds) 2011-11-23T17:17:20 *** iglo has joined #aichallenge 2011-11-23T17:18:05 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-23T17:19:34 i want jimmy's face out of mine 2011-11-23T17:20:56 i just love adblock <3 2011-11-23T17:25:53 mcstar: you can just click x in the corner anyway without adblock 2011-11-23T17:26:12 every time i load wiki? 2011-11-23T17:26:14 noo 2011-11-23T17:27:07 I think it makes an anti-panhandling cookie and doesn't show it again 2011-11-23T17:27:29 in the past it showed up again and again for me 2011-11-23T17:27:34 i didnt take the change now 2011-11-23T17:27:44 chance* 2011-11-23T17:28:21 *** ikaros has joined #aichallenge 2011-11-23T17:28:37 im readind a topic unknown to me previously 2011-11-23T17:28:40 g 2011-11-23T17:28:53 and i was reading some research papers 2011-11-23T17:29:13 i have to say wikipedia gives me a much better understanding on this random topic 2011-11-23T17:29:19 than the author gave 2011-11-23T17:31:09 *** Garf has quit IRC (Quit: Make a new plan, Stan!) 2011-11-23T17:31:36 *** NezumiTejina has quit IRC (Quit: Page closed) 2011-11-23T17:36:25 Test Error: compiled, but failed test cases 2011-11-23T17:36:36 how to fix that 2011-11-23T17:36:38 ? 2011-11-23T17:37:38 turn 1 bot 0 invalid actions: 2011-11-23T17:37:38 o 1 18 # invalid direction 2011-11-23T17:37:38 o 3 18 # invalid direction 2011-11-23T17:38:16 you aren't giving directions 2011-11-23T17:38:29 'o row col direction' is the format 2011-11-23T17:39:06 i think he is 2011-11-23T17:39:10 hmm 2011-11-23T17:39:14 isnt that a space there? 2011-11-23T17:39:14 *** dom7b5 has quit IRC (Quit: Ex-Chat) 2011-11-23T17:39:34 (kidding) 2011-11-23T17:40:01 it had no problems before 2011-11-23T17:40:10 im not sure weather the invalid directions cause your failed test though 2011-11-23T17:40:32 http://aichallenge.org/specification.php#bot_output 2011-11-23T17:41:04 *** Jak_o_Shadows has joined #aichallenge 2011-11-23T17:44:15 void State::makeMove(const Location &loc, int direction) 2011-11-23T17:44:15 { 2011-11-23T17:44:15 cout << "o " << loc.row << " " << loc.col << " " << CDIRECTIONS[direction] << endl; 2011-11-23T17:44:24 its the same as in the starterpack 2011-11-23T17:45:28 run your bot with valgrind 2011-11-23T17:46:04 direction needs to be 0..3 inclusive or you're accessing an array out of bounds 2011-11-23T17:46:05 you probably called it with an invalid "direction" 2011-11-23T17:46:56 hmm 2011-11-23T17:47:17 ok I check the startvalues 2011-11-23T17:47:33 http://ants.fluxid.pl/replay.12812 I need to teach my ants to run away now 2011-11-23T17:50:23 *** Rinum has joined #aichallenge 2011-11-23T17:50:57 ok it was some startvalue 2011-11-23T17:50:59 hmm... seems like these days ppl are interested in parallel algo implementations on GPUs... not CPUs 2011-11-23T17:51:11 thanks 2011-11-23T17:51:42 *** Conorach_ has quit IRC (Ping timeout: 244 seconds) 2011-11-23T17:51:49 thestinger: I should update my fluxid bot 2011-11-23T17:52:02 my current bot would have probably eaten your entire attacking army 2011-11-23T17:52:04 GPUs are massively parallel 2011-11-23T17:52:26 CPUs are considerably less 2011-11-23T17:52:44 BenJackson: yeah, they are too dumb to run away atm 2011-11-23T17:52:52 they just stand there if the odds are bad 2011-11-23T17:52:58 (my ants) 2011-11-23T17:53:33 cpu's are good in tasks gpu's in scaled data 2011-11-23T17:53:46 the bind I'm in right now is that the ignorant aggression of my V7 bot seems to be a factor in its success 2011-11-23T17:54:01 improved combat actually shifts the "trade ants" balance too far 2011-11-23T17:54:13 pairofdice: that's probably why 2011-11-23T17:54:34 ...probably :) 2011-11-23T17:55:42 modern GPU's have hundreds of 'cores' 2011-11-23T17:55:50 lol 2011-11-23T17:56:10 *** iglo has quit IRC (Remote host closed the connection) 2011-11-23T17:56:20 Probably up to thousands by now 2011-11-23T17:56:23 I guess I know what I'll do then 2011-11-23T17:56:40 graphics fall into the category of "embarassingly parallel" 2011-11-23T17:56:59 I'll find a parallel algo... and implement it with GPU 2011-11-23T17:57:00 meaning that there is such a high degree of independence between the operations (each pixel in this case) that parallelizing is trivial 2011-11-23T17:57:33 it's probably been done though 2011-11-23T17:57:43 whereas something like A* would be very hard to parallelize because of the interdependence of each step on the next 2011-11-23T17:58:00 *** kara has quit IRC () 2011-11-23T17:58:05 Rinum: Like I said before, doing truly new work is thesis level 2011-11-23T17:58:30 bfs is a little bot more scalable than a*, though its not perfect 2011-11-23T17:58:30 if you're new to this stuff you should take any problem and try to parallelize it regardless of how many have come before you 2011-11-23T17:58:43 which I'm basically doing... a miniature thesis 2011-11-23T17:59:02 avdg: depends on what the "visit" is and what its ordering dependencies are 2011-11-23T17:59:10 I'm not new to parallelizing stuff, I just need something to work on x.x 2011-11-23T17:59:24 improving bot? 2011-11-23T17:59:33 bots aren't allowed to be parallel :) 2011-11-23T17:59:33 write it in an other language? 2011-11-23T17:59:35 nah, school work 2011-11-23T17:59:57 ah, in that context 2011-11-23T17:59:59 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-11-23T18:00:06 it's for school... we have to do something like an undergraduate level thesis 2011-11-23T18:00:23 you have to take efficiency into consideration 2011-11-23T18:00:35 modern gpus support every bit of c++ 2011-11-23T18:00:44 but that doesnt mean you can actually use that power 2011-11-23T18:00:59 you still ahve to write in a particular fashion to get fast code 2011-11-23T18:01:06 even trivially paralizable algorithms might suck on the gpu for the first time 2011-11-23T18:01:17 that's true 2011-11-23T18:01:22 I'm more probably more interested in learning the hardware atm :p 2011-11-23T18:02:53 *** Anilm3 has quit IRC (Quit: leaving) 2011-11-23T18:03:02 the key is to find a way to use max processing power while maximizing memory throughput 2011-11-23T18:03:57 this sounds too general... 2011-11-23T18:04:12 *** treeform has quit IRC (Remote host closed the connection) 2011-11-23T18:04:25 but really, you must consider memory efficiency very seriously 2011-11-23T18:04:41 you must do enough computations while memory is being fetched 2011-11-23T18:04:46 pre-optimalization atm for me 2011-11-23T18:05:23 I'll remember that, thanks 2011-11-23T18:06:00 I worked on early NUMA machines. Those lead to some interesting algorithms due to memory cost 2011-11-23T18:06:02 it would probably more efficient to make the code breakable into units 2011-11-23T18:06:15 and each unit care about their own variables 2011-11-23T18:06:24 *** treeform has joined #aichallenge 2011-11-23T18:06:41 i always forget that acronym 2011-11-23T18:06:53 nn uniform... 2011-11-23T18:06:58 mamemory access 2011-11-23T18:07:00 f 2011-11-23T18:07:05 non uniform memory access 2011-11-23T18:07:57 modern multi-chip intel stuff is actually all NUMA now (because the memory controller is in the CPU now) but the difference between remote and local memory access is small enough that most people don't care 2011-11-23T18:08:43 *** treeform has quit IRC (Remote host closed the connection) 2011-11-23T18:08:47 In the early days when the remote memory was truly remote (accessible across some kind of bus or ring) it could be 100x or even 1000x worse to access remote memory 2011-11-23T18:09:33 so imagine algorithms where the answer you want is already in memory but it's actually cheaper to recompute it 2011-11-23T18:09:45 interesting 2011-11-23T18:10:18 gpu global memory has some hundred times latency of the on-chip mem 2011-11-23T18:10:58 BenJackson: i guess there was some kiind of cache still? 2011-11-23T18:11:05 oh yeah 2011-11-23T18:11:19 "cc-NUMA" is "cache coherent NUMA" 2011-11-23T18:11:20 most people drop the cc 2011-11-23T18:11:53 gpu texture memory isnt coherent within kernel calls 2011-11-23T18:12:06 coherency is expensive 2011-11-23T18:12:23 for one thing if someone has your memory cached remotely now your invalidate incurs the remote access penalty 2011-11-23T18:12:56 BenJackson: are you familiar with clojure's concurrency solution? 2011-11-23T18:13:08 no, not familiar with clojure 2011-11-23T18:13:31 it has a multi-version software concurrency control mechanism or whatnot 2011-11-23T18:13:49 ah 2011-11-23T18:13:50 it manages to keep a consistent state of data among threads 2011-11-23T18:14:00 I thought about doing something like that for another language 2011-11-23T18:14:11 it might not be "current" but its consistent 2011-11-23T18:14:13 basically I wanted to make a bunch of threads think they were still alone 2011-11-23T18:14:30 by determining when they did dependent things and then unwinding the lower 2011-11-23T18:14:32 *loser 2011-11-23T18:14:49 and if that happened a few times by actual locking 2011-11-23T18:14:50 BenJackson: just like in time warp os? 2011-11-23T18:15:19 not familiar with that 2011-11-23T18:15:47 ah, yes, exactly like that 2011-11-23T18:16:08 http://fmg-www.cs.ucla.edu/reiher/Time_Warp.html 2011-11-23T18:16:21 i saw an interview with beckman 2011-11-23T18:16:56 i forgot his full name 2011-11-23T18:17:12 the cleverest application I've heard about of that principle is the apple lisa 2011-11-23T18:17:18 brian beckman 2011-11-23T18:17:22 it had a MC68000 which does not support the 68881 MMU 2011-11-23T18:17:33 so if you fault it can't restart an instructino 2011-11-23T18:17:39 (like a 68020+ could) 2011-11-23T18:17:50 so they put in two and ran the same thing on both 2011-11-23T18:17:53 with one 1 clock behind 2011-11-23T18:17:59 o if you fault it can't rest 2011-11-23T18:18:16 when the frontrunner faulted they'd take the fault and resume execution on the one that was behidn 2011-11-23T18:18:19 ah ok 2011-11-23T18:18:19 and then trade places 2011-11-23T18:18:32 was 1 intruction enough? 2011-11-23T18:19:23 the problem was just that it couldn't restart the faulting instruction 2011-11-23T18:19:37 so eg "load from 0x1000" causing a fault would resume at the next instruction not having loaded 2011-11-23T18:20:05 ah it was th '010 that added vm support 2011-11-23T18:20:23 oops the mmu was the 68851 2011-11-23T18:22:47 lol 2011-11-23T18:22:51 i made up a word 2011-11-23T18:22:57 "somewhy" 2011-11-23T18:23:19 =for some reason 2011-11-23T18:23:30 or am i too late? 2011-11-23T18:24:49 looks good, its not taken 2011-11-23T18:25:25 somewhy.com? 2011-11-23T18:26:16 ah, no, i meant i couldnt find it in the dictionary 2011-11-23T18:26:30 huh, somehow I went up to 18th on aichallenge. I think some top bots must be resubmitting 2011-11-23T18:26:58 yeah, back to coding... 2011-11-23T18:27:02 oh, just a ton of us packed in right around 82.0 2011-11-23T18:27:04 thanks for the reminder 2011-11-23T18:30:39 http://aichallenge.org/visualizer.php?game=119352&user=1115 2011-11-23T18:31:56 yep, he linked that on 2011-11-23T18:31:58 in 2011-11-23T18:32:07 lol 2011-11-23T18:34:17 rofl 2011-11-23T18:34:17 oh man.. 2011-11-23T18:34:24 reading people's code is entertaining 2011-11-23T18:34:40 and educational too 2011-11-23T18:34:52 amstan: People Magazin's? 2011-11-23T18:35:00 the submissions 2011-11-23T18:35:07 noway 2011-11-23T18:35:12 you do that? 2011-11-23T18:35:17 there's all these cool observations about it.. 2011-11-23T18:35:25 like some use tabs 2011-11-23T18:35:50 and on empty lines, they do it like i do, they don't just leave the line empty, but keep the tabs in 2011-11-23T18:35:53 mcstar: why not? 2011-11-23T18:35:54 *** tdubellz_ has joined #aichallenge 2011-11-23T18:36:13 *** shamik has joined #aichallenge 2011-11-23T18:36:35 nothing, i just hope the pupper master wont brainhack you 2011-11-23T18:36:41 puppet* 2011-11-23T18:37:13 i'm very catious, like i don't move stuff off the server(unless people want their code back) 2011-11-23T18:37:14 *** tdubellz has quit IRC (Disconnected by services) 2011-11-23T18:38:04 have you looked at top bots' code? 2011-11-23T18:38:13 yep, interesting stuff 2011-11-23T18:39:59 oh god, i don't even get this one 2011-11-23T18:40:04 *** tdubellz_ is now known as tdubellz 2011-11-23T18:40:40 so many function arguments! 2011-11-23T18:40:43 *** shamik has quit IRC (Ping timeout: 265 seconds) 2011-11-23T18:44:05 *** treeform has joined #aichallenge 2011-11-23T18:45:59 *** ThatSnail has quit IRC (Ping timeout: 240 seconds) 2011-11-23T18:52:17 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-23T18:52:57 *** treeform has quit IRC (Read error: Connection reset by peer) 2011-11-23T18:53:25 *** bmh has joined #aichallenge 2011-11-23T18:55:01 *** bmh has quit IRC (Client Quit) 2011-11-23T18:56:10 *** Rinum has quit IRC (Quit: Page closed) 2011-11-23T18:57:32 *** treeform has joined #aichallenge 2011-11-23T19:04:33 *** fpcfan has quit IRC (Ping timeout: 244 seconds) 2011-11-23T19:05:35 *** xathis has quit IRC (Ping timeout: 248 seconds) 2011-11-23T19:06:08 *** fpcfan has joined #aichallenge 2011-11-23T19:06:22 *** tobym has quit IRC (Remote host closed the connection) 2011-11-23T19:07:39 *** grwip has quit IRC (Quit: Leaving) 2011-11-23T19:10:52 *** userjjb has quit IRC () 2011-11-23T19:16:02 *** mcstar has quit IRC (Quit: WeeChat 0.3.6) 2011-11-23T19:18:50 *** replore has joined #aichallenge 2011-11-23T19:19:09 *** Harpyon_ has joined #aichallenge 2011-11-23T19:19:31 aichallenge: Jack Palevich epsilon * r80118d7 / (3 files in 2 dirs): 2011-11-23T19:19:31 aichallenge: Fix errors detected by DartEditor 2011-11-23T19:19:31 aichallenge: Also changed Config.set to Config.update to work around 2011-11-23T19:19:31 aichallenge: http://code.google.com/p/dart/issues/detail?id=561 2011-11-23T19:19:31 aichallenge: Now people can edit their Dart Ant Bots using the DartEditor. - http://git.io/gZkHxA 2011-11-23T19:19:31 aichallenge: Janzert epsilon * re19c074 / (3 files in 2 dirs): 2011-11-23T19:19:32 aichallenge: Merge pull request #378 from jackpal/dart-starter-pack-update 2011-11-23T19:19:32 aichallenge: Fix errors detected by DartEditor - http://git.io/8HibqQ 2011-11-23T19:19:35 *** BoAnd has quit IRC (Ping timeout: 260 seconds) 2011-11-23T19:20:09 *** Harpyon_ has quit IRC (Client Quit) 2011-11-23T19:20:51 *** tdubellz has quit IRC (Quit: restarting) 2011-11-23T19:21:14 *** tdubellz has joined #aichallenge 2011-11-23T19:29:00 *** Antvolution has left #aichallenge 2011-11-23T19:32:38 weird, I can't track down this bug 2011-11-23T19:40:04 *** ThatSnail has joined #aichallenge 2011-11-23T19:40:56 *** Harpyon has quit IRC (Quit: Computer has gone to sleep.) 2011-11-23T19:42:02 *** jstrong has joined #aichallenge 2011-11-23T19:42:44 *** jstrong is now known as roflmao 2011-11-23T19:44:55 *** BoAnd has joined #aichallenge 2011-11-23T19:47:28 *** rajanaresh has joined #aichallenge 2011-11-23T19:52:27 *** ltriant has quit IRC (Read error: Connection reset by peer) 2011-11-23T19:52:40 *** ltriant has joined #aichallenge 2011-11-23T19:54:39 *** Extrarius has quit IRC (Ping timeout: 248 seconds) 2011-11-23T19:55:10 aichallenge: janzert epsilon * rb78d414 / website/sql.php : Average worker game rate over 30 minutes - http://git.io/6oRjVg 2011-11-23T19:56:19 *** fpcfan has quit IRC (Ping timeout: 244 seconds) 2011-11-23T19:56:37 *** rajanaresh has quit IRC (Ping timeout: 240 seconds) 2011-11-23T19:58:38 *** Anilm3 has joined #aichallenge 2011-11-23T19:59:11 hey guys 2011-11-23T20:05:31 *** Anilm3 has quit IRC (Quit: Page closed) 2011-11-23T20:05:51 *** flag has quit IRC (Ping timeout: 248 seconds) 2011-11-23T20:06:17 *** janzert1 has joined #aichallenge 2011-11-23T20:08:14 *** janzert has quit IRC (Ping timeout: 252 seconds) 2011-11-23T20:13:22 *** bmh has joined #aichallenge 2011-11-23T20:21:57 meh, I'm about to add my first goto in years 2011-11-23T20:22:19 *** Extrarius has joined #aichallenge 2011-11-23T20:22:34 *** fpcfan has joined #aichallenge 2011-11-23T20:22:41 * avdg tries to fix it 2011-11-23T20:22:42 *** pairofdice has quit IRC (Quit: leaving) 2011-11-23T20:23:08 avdg: repeat after me, 'goto' does not exist 2011-11-23T20:23:11 *** fpcfan has quit IRC (Ping timeout: 240 seconds) 2011-11-23T20:23:28 goto is usefull :-) 2011-11-23T20:23:43 oh well, within boundaries of course 2011-11-23T20:25:40 that whole goto hate is entirely misplaced 2011-11-23T20:25:56 it was relevant when people didn't use any other loop constructs 2011-11-23T20:27:13 now people know that for and while exist, so they understand that goto has limited usefulness 2011-11-23T20:29:11 *** b0rder has joined #aichallenge 2011-11-23T20:29:12 *** bmh has quit IRC (Quit: Page closed) 2011-11-23T20:29:25 ^ yay 2011-11-23T20:30:59 well, my goto is gone now 2011-11-23T20:31:42 well that was an epic melee: http://ants.fluxid.pl/replay.12955 2011-11-23T20:32:46 I wonder if pavlog is also securing the surrounding area 2011-11-23T20:33:43 a1kon also had an efficient attack in that game 2011-11-23T20:34:15 * avdg likes frame 164, which is the opposite of the desired defense strategy 2011-11-23T20:35:31 where in frame 164? 2011-11-23T20:35:41 oh, how i just go right through the ants surrounding me 2011-11-23T20:36:42 *** McLeopold has left #aichallenge 2011-11-23T20:39:26 I think there's a bit of RPS here 2011-11-23T20:39:52 which rps? 2011-11-23T20:39:56 "semi-ignorant aggression" (which is how I'd characterize my current aichallenge submission) is pretty effective 2011-11-23T20:40:14 if you gather well it's going to do well against any low-end bots or bots which make mistakes or lack strong defense 2011-11-23T20:40:27 but it does poorly against near-perfect-combat bots which can absorb the attacks 2011-11-23T20:40:39 but near-perfect-combat implies some timidity 2011-11-23T20:40:54 and the timidity does poorly against everything else but other near-perfect-combat bots 2011-11-23T20:41:13 I've been trying to do the near-perfect-combat thing 2011-11-23T20:41:15 can't get it nailed down 2011-11-23T20:41:19 too difficult 2011-11-23T20:41:41 so near-perfect-combat beats semi-ignorant-aggression in 1v1, but if you are on the same map together the aggressive bot will probably take other hills and score more points 2011-11-23T20:42:15 I think the combat rules for this challenge are kind of poor 2011-11-23T20:42:27 it would have been hard to see it when they were chosen 2011-11-23T20:42:45 but I think the problem is that the problem isn't "hard" enough to prevent everyone from "solving" it 2011-11-23T20:43:06 and two bots with "solved" combat just stalemate mostly 2011-11-23T20:43:21 or they just have to suicide each other 2011-11-23T20:46:04 *** fpcfan has joined #aichallenge 2011-11-23T20:48:51 *** TheLinker has quit IRC (Quit: Bye) 2011-11-23T20:55:58 how the hell does A do it 2011-11-23T20:56:00 he can predict your moves 2011-11-23T20:56:21 learning algorithm? 2011-11-23T20:56:33 roflmao: example game? 2011-11-23T20:57:10 if it can predict, I wonder how good the predications are 2011-11-23T20:57:16 anyone know who A is on official yet? 2011-11-23T20:57:26 FlagCapper 2011-11-23T20:57:36 #5 on aichallenge 2011-11-23T20:57:36 ahh 2011-11-23T20:57:39 *** janzert1 is now known as janzert 2011-11-23T20:57:39 http://ants.fluxid.pl/replay.4828 2011-11-23T20:57:55 so he's 16 years old 2011-11-23T20:57:58 and he kicks all of our asses 2011-11-23T20:58:00 yes 2011-11-23T20:58:04 oops now #4 2011-11-23T20:58:08 slightly depressing :P 2011-11-23T20:58:14 yeah, and heading for 3 2011-11-23T20:58:30 * avdg doesn't care about age 2011-11-23T20:58:33 #rankings 2011-11-23T20:58:35 *** TheLinker has joined #aichallenge 2011-11-23T20:58:36 @rankings 2011-11-23T20:58:37 amstan: Top 10 players: xathis(91.5), pguillory(88.5), delineate(86.8), FlagCapper(86.3), RVeerdonk(86.2), MomoBot(84.9), Komaki(84.8), ThisIsNotABug(84.7), lazarant(84.3), GarySWest(84.3) 2011-11-23T20:58:52 i should probably get that going sometime 2011-11-23T20:59:01 then auto@rankings every 15 min 2011-11-23T20:59:04 anyone opposed? 2011-11-23T20:59:10 A's play in that game isn't that great that I see 2011-11-23T20:59:12 16 year olds are smart enough to be starcraft pros and (often) have plenty of time to throw away on fun stuff like this 2011-11-23T20:59:16 was there a particular spot? 2011-11-23T20:59:20 once an hour maybe 2011-11-23T20:59:26 janzert: ok 2011-11-23T21:00:25 contestbot: later tell mcleopold I would like to look at your mapgen code, integrate it in mine, can you post the code in a branch sometime? 2011-11-23T21:00:25 amstan: Yes master! 2011-11-23T21:00:29 amstan: 15 minutes too spammy, imo 2011-11-23T21:01:41 janzert: weren't the workers supposed to have screen sessions with stuff in them? 2011-11-23T21:05:34 aichallenge: Alexandru Stan epsilon * rffa5c72 / manager/worker_ssh.py : added worker count to worker_ssh - http://git.io/C8_beA 2011-11-23T21:17:34 janzert: are we going to be resetting the rankings at the start of the final tourny? 2011-11-23T21:17:35 *** thestinger has quit IRC (Ping timeout: 240 seconds) 2011-11-23T21:17:52 some people are complaining that xanthis has a low sigma, so it's impossible to catch up to him 2011-11-23T21:18:02 *** thestinger has joined #aichallenge 2011-11-23T21:23:55 HOLY COW REV BREAKTHROUGH 2011-11-23T21:24:29 *** Jak_o_Shadows1 has joined #aichallenge 2011-11-23T21:25:55 his sigma is only like 0.06 less than mine 2011-11-23T21:26:06 which is only worth 0.18 score 2011-11-23T21:26:20 wouldn't move xathis out of 1st 2011-11-23T21:26:38 *** Jak_o_Shadows has quit IRC (Ping timeout: 244 seconds) 2011-11-23T21:26:50 @amstan: what's much more likely (imo) is that he'd settle somewhere else if he reset to the bottom 2011-11-23T21:26:51 BenJackson: I'm sorry Dave, err BenJackson; I cannot 'amstan:'. 2011-11-23T21:27:05 @openthepodbaydoors 2011-11-23T21:27:06 BenJackson: I have no idea what you mean. 2011-11-23T21:28:01 *** JorgeB has quit IRC (Quit: Computer has gone to sleep.) 2011-11-23T21:50:38 *** treeform has quit IRC (Remote host closed the connection) 2011-11-23T22:04:24 @later tell amstan workers only get a screen session if I need to restart it manually for some reason, they normally just run straight 2011-11-23T22:04:24 janzert: I come to serve. 2011-11-23T22:05:15 @later tell amstan yes rankings are completely reset at the beginning of finals but xathis has no inherit advantage with a low sigma right now that wouldn't be overcome by a better bot within a few days of games 2011-11-23T22:05:15 janzert: Aye, aye, sir 2011-11-23T22:09:41 for those curious here are the current top 20 mu http://pastebin.com/F1KMqZdK 2011-11-23T22:14:11 and maybe slightly more interesting here it is with sigma and rank shown as well http://pastebin.com/fJWVtiW4 2011-11-23T22:14:32 paratrechina is the rising start at the moment 2011-11-23T22:17:54 that looks pretty right to me 2011-11-23T22:18:05 given the thousands of games played on tcp 2011-11-23T22:18:18 though i'm not sure whether pguillory is above flagcapper 2011-11-23T22:18:41 need moar games 2011-11-23T22:18:53 well flagcapper is still quite new so the mu will move around more for him of course still 2011-11-23T22:18:56 *** treeform has joined #aichallenge 2011-11-23T22:19:17 yeah he and i haven't played too many games yet 2011-11-23T22:19:19 but I don't think he'll surpass xathis with this submission 2011-11-23T22:19:26 no, def not 2011-11-23T22:20:16 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-11-23T22:20:36 *** deltree_ has joined #aichallenge 2011-11-23T22:34:27 *** Antimony has joined #aichallenge 2011-11-23T23:01:26 *** Euphony has joined #aichallenge 2011-11-23T23:03:54 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-23T23:16:05 what amazes me is that xathis NAILED it one his very first submission 2011-11-23T23:16:05 *** avdg has quit IRC (Quit: Leaving.) 2011-11-23T23:16:09 he hasn't changed it a bit 2011-11-23T23:16:21 he just knew what the hell he needed to do and did it right off the bat 2011-11-23T23:16:35 on his very first submission** 2011-11-23T23:18:10 *** avdg has joined #aichallenge 2011-11-23T23:22:23 *** avdg has quit IRC (Ping timeout: 252 seconds) 2011-11-23T23:23:43 *** Antimony has joined #aichallenge 2011-11-23T23:26:19 *** bmh has joined #aichallenge 2011-11-23T23:26:39 hi 2011-11-23T23:27:49 *** Euphony has quit IRC (Ping timeout: 265 seconds) 2011-11-23T23:27:54 heh, made a bot that grows to 100% map coverage at least 2011-11-23T23:28:13 otoh it waited until it had 195 ants to attack a RandomBot.py... 2011-11-23T23:28:21 BenJackson: awesome! 2011-11-23T23:31:45 *** rajanaresh has joined #aichallenge 2011-11-23T23:33:07 roflmao: rae you joking? 2011-11-23T23:33:50 there was a several-month-long beta period before the contest officially launched, during which xathis (and all the other guys on the leaderboard on day 1) developed their bots 2011-11-23T23:34:44 there was a very small difference just beforehand, but not much 2011-11-23T23:34:54 a1k0n: Can you copypasta context? I don't want to read the logs 2011-11-23T23:37:23 ooops 2011-11-23T23:37:24 *** Cyndre has quit IRC (Quit: Leaving) 2011-11-23T23:37:28 used a function in a template I hadn't used before 2011-11-23T23:37:33 turned out I implemented assignment backwards 2011-11-23T23:38:01 bmh: roflmao was just amazed xathis's first version is winning 2011-11-23T23:38:34 aah I see a1k0n 2011-11-23T23:38:36 that explains it then :P 2011-11-23T23:39:46 I'll win yet, just you wait and see! :) 2011-11-23T23:42:12 I believe in you! 2011-11-23T23:45:55 hey guys, I want to know what kind of algorithms are people using to explore. I mean I see the top player's bots and their ants seem to know what they are doing and where they are going right from the time they spawn. My bot is so naive I used brownian motion to explore unless you see a food or enemy ant hill in which case there is pick and attack strategy respectively. How do those top bot's ants know exactly where they are goin 2011-11-23T23:46:15 How are those bots keeping track of what is explored and what is to be explored in the map? 2011-11-23T23:46:55 rajanaresh: i just have a "lastSeen" int in each square 2011-11-23T23:47:00 that's easy rajanaresh 2011-11-23T23:47:07 keeping track of what is explored and what is to be explored 2011-11-23T23:47:13 the algorithms the top bots use are more complex 2011-11-23T23:47:25 they use some sort of flood fill from the edges to keep the distribution even 2011-11-23T23:47:37 if you just hunt unexplored tiles you tend to get a kind of "directional" pattern 2011-11-23T23:47:43 like a snake that goes across the maze 2011-11-23T23:47:53 if you instead use some sort of edge flood fill system 2011-11-23T23:47:57 i have a function to meet enemy ants as soon as possible on one of the enemies shortest paths to my hills 2011-11-23T23:48:01 you get a "dissipation" thing going on 2011-11-23T23:48:04 i like that :P 2011-11-23T23:48:15 I tried telling my bots to explore the closest unexplored square to the hill to get more even coverage, but it didn't work very well 2011-11-23T23:48:37 what I do is I first explore never seen tiles 2011-11-23T23:48:38 do you reuse ants that have been used for exploring food collecting? 2011-11-23T23:48:39 i just try to minimize the distance between all squares on the map and the nearest ants 2011-11-23T23:48:57 it's a very.. calculus approach 2011-11-23T23:49:07 so you basically just go to the nearest unseen tile? 2011-11-23T23:49:15 but it's not a continuous function 2011-11-23T23:49:27 basically but it's more subtle 2011-11-23T23:49:29 I don't have a strategy, but I will have one. 2011-11-23T23:49:38 i assume the probability food is present is proportional to lastSeen-turn 2011-11-23T23:49:44 er, turn-lastSeen 2011-11-23T23:50:07 aaah 2011-11-23T23:50:08 interesting 2011-11-23T23:50:12 and i try to maximize the likelihood of finding new food 2011-11-23T23:50:13 so if turn last seen is greater 2011-11-23T23:50:21 then that tile has preference 2011-11-23T23:50:21 anyway, in my current bot I just made my ants try to stay at least at least 3 away from each other, so they naturally disperse 2011-11-23T23:50:23 that's pretty neat 2011-11-23T23:50:27 it works well except on dense mazes 2011-11-23T23:51:11 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-23T23:51:41 does anyone try to keep track of stored food? 2011-11-23T23:51:50 not hard :P 2011-11-23T23:51:54 if you get some symmetrical detection stuff going on 2011-11-23T23:52:04 that can really improve your exploring/foraginig 2011-11-23T23:52:06 i haven't done it yet but i plan to 2011-11-23T23:52:13 same 2011-11-23T23:52:22 has anyone does symmetry detection? 2011-11-23T23:52:24 antimatroid: wait, how do you know what tiles have been seen. Do you update your lastseen in tiles for every ants sqrt(viewradius)? 2011-11-23T23:52:25 what do you mean by stored food? 2011-11-23T23:52:28 until I get at least rudimentary combat, everything else is secondary 2011-11-23T23:52:29 animatroid: bugnuts has 2011-11-23T23:52:30 i did my best to make it hard to exploit 2011-11-23T23:52:30 BenJackson has, I think 2011-11-23T23:52:33 you know, i keep talking about doing it but i haven't 2011-11-23T23:52:39 i'm just going to implement it 2011-11-23T23:52:43 before I try symmetry, I'm going to add basic triangulation 2011-11-23T23:52:48 it sounds very difficult though 2011-11-23T23:52:55 the triangulation idea is super clever 2011-11-23T23:52:59 translational symmetry is easy 2011-11-23T23:53:07 triangulation? 2011-11-23T23:53:07 a1k0n: I'm a super clever guy ;) 2011-11-23T23:53:09 rajanaresh: in turn 0 i calculate the translation values to get locatins within the view radius of a square then just stamp vision at the start of each turn 2011-11-23T23:53:18 it'd help more than symmetry detection in the early game 2011-11-23T23:53:25 a1k0n: :D 2011-11-23T23:53:28 what's triangulation? 2011-11-23T23:53:30 roflmao: If you see an ant on turn N, you know that the enemy hill is no more than N steps away from where that ant is 2011-11-23T23:53:39 nice 2011-11-23T23:53:47 could be useful on multihill maps 2011-11-23T23:53:52 hahaha that's extremely clever 2011-11-23T23:53:55 and if you see three ants from player b on turn N at totally different spots... 2011-11-23T23:54:03 lol 2011-11-23T23:54:10 Antimony: and you know how many hills there are because you can count your own hills! 2011-11-23T23:54:28 you don't know the number of players :P 2011-11-23T23:54:42 bmh: NO WAY! 2011-11-23T23:54:46 antimatroid: whatever. model it. 2011-11-23T23:54:48 It sounds nice in theory, but I think it's too much effort ot be worth it 2011-11-23T23:55:10 that level of triangulation sounds ridicueasy 2011-11-23T23:55:20 Antimony: exploiting symmetry is VERY useful for pathfinding 2011-11-23T23:55:24 which is a large component of the game 2011-11-23T23:55:26 although I don't see much of the triangle in it 2011-11-23T23:55:32 I'm talking about triangulation, not symmetry 2011-11-23T23:55:35 and knowing enemy hill locations is useful 2011-11-23T23:55:37 oh 2011-11-23T23:55:53 antimatroid: and you do that for every ant. Don't you think it's computationally expensive? I mean does it scale with more ants? Consider radius approximately 7 so you have approx. 49 squares for each ant to update. Is that what's happening? 2011-11-23T23:55:54 I do plan on doing symmetry once I get rudimentary combat working, if ever 2011-11-23T23:55:55 roflmao: like a1k0n said -- if you see a couple of ants early enough in the game and far enough apart, you can pin-point their hill (approximately) 2011-11-23T23:56:03 ah 2011-11-23T23:56:10 rajanaresh: computers are fast :) 2011-11-23T23:56:20 I think that the cases in which triangulation would be useful are very limted 2011-11-23T23:56:24 is that really going to give a boost though? at least in my personal experience 2011-11-23T23:56:24 yeah 2011-11-23T23:56:33 in my personal experience I usually find hills BEFORE I'm ready to attack them 2011-11-23T23:56:47 finding hills isn't that much of an issue 2011-11-23T23:56:54 it might be with better bots later 2011-11-23T23:57:06 symmetry is much more effective at finding hills 2011-11-23T23:57:13 i haven't implemented it yet but i plan to add into my combat to avoid an enemy seeing a particular hill for the first time 2011-11-23T23:57:32 lol 2011-11-23T23:57:44 maybe I should keep track of which players have seen my hills 2011-11-23T23:57:51 not hard to do 2011-11-23T23:58:08 how do you know they didn't sneak an ant in in a place you can't see? 2011-11-23T23:58:12 just add a set of locations to each square and insert an ant player when it's in range at the start of each turn :) 2011-11-23T23:58:19 shit :P 2011-11-23T23:58:20 unless you keep the entire radius of your hill visible every turn 2011-11-23T23:58:35 i plan to keep the radius visible when possible, but obviously not always the case 2011-11-23T23:58:40 i guess you just go on what you do know 2011-11-23T23:58:44 Antimony: I'm going to politely ask my ants to sacrifice themselves to stop someone from seeing my hill 2011-11-23T23:59:01 bmh: his point was an enemy might see a hill without you knowing 2011-11-23T23:59:03 arrrg, there are so many features I need to implement 2011-11-23T23:59:06 but you just assume it hasn't seen it then 2011-11-23T23:59:12 antimatroid: yeah... 2011-11-23T23:59:14 but I'm held back because I still have no combat 2011-11-23T23:59:35 my source files are like 2500 lines worth now :|