2011-12-17T00:00:02 *** u_ has quit IRC (Quit: u_) 2011-12-17T00:13:35 *** treeform has quit IRC (Remote host closed the connection) 2011-12-17T00:29:23 *** u_ has joined #aichallenge 2011-12-17T00:31:50 *** u_ has quit IRC (Client Quit) 2011-12-17T00:34:46 255926: could have been a bit more aggressive toward the end, one advantage of people stopping-up their holes is that you don't know what resources you need to raze that hole 2011-12-17T00:35:13 Should have really thrown a bunch more in, or perhaps made some special formation 2011-12-17T00:35:34 Special formation generally means heaps of code (and slow) 2011-12-17T00:36:32 BenJackson: http://tcpants.com/replay.2466 I should not be doing this well against your bot 2011-12-17T00:39:50 *** Gnome has joined #aichallenge 2011-12-17T00:42:26 A few early timeouts in a field of 8 are like low-hanging fruit, if you can grab them quick then you generally will win on score even with a blunder or two 2011-12-17T01:02:37 *** bgs100 has quit IRC (Quit: night) 2011-12-17T01:11:43 *** treeform has joined #aichallenge 2011-12-17T01:19:20 rwest_: helps to take your neighbor early like you did 2011-12-17T01:19:29 at least as far as I can see on this laptop, where that game is tiny and laggy 2011-12-17T01:23:09 how can I tell if my bot has built? 2011-12-17T01:23:12 (on the server) 2011-12-17T01:23:45 bottom of your profile page 2011-12-17T01:23:57 if your rank is "no skillz" or 7700+ it probably built 2011-12-17T01:24:48 Could not find module `Data.Vector': doh 2011-12-17T01:25:00 hooray -- they didn't bother installing the haskell libraries. 2011-12-17T01:31:01 sweet. My bot won't run on the server because they're using an old version of GHC. 2011-12-17T01:31:22 nice. 2011-12-17T01:32:10 *** yoden-cloud has quit IRC (Ping timeout: 258 seconds) 2011-12-17T01:32:30 If I wasn't so tired, I'd be grumpy. 2011-12-17T01:33:12 *** bmh has quit IRC (Quit: FTS.) 2011-12-17T01:34:28 wtf 2011-12-17T01:34:46 my ant is walking off a location, then standing still to collect food at its old location when there's no food there :\ 2011-12-17T01:34:59 i don't think food even ever existed there 2011-12-17T01:38:30 does play_one_game_live thing ever screw up showing food? 2011-12-17T01:39:04 *** TL_ has quit IRC (Ping timeout: 258 seconds) 2011-12-17T01:39:17 never used live 2011-12-17T01:39:30 but it just pipes the input to the regular visualizer running in a java canvas 2011-12-17T01:40:27 what the hell is it with me and Speedy_Consoles 2011-12-17T01:40:32 I've played him 5 times today 2011-12-17T01:40:38 3 times in consecutive games 2011-12-17T01:45:04 interesting 2011-12-17T01:45:10 the engine is definitely telling me food is turning up 2011-12-17T01:45:15 but the visualiser isn't showing it 2011-12-17T01:45:37 live is overrated :) 2011-12-17T01:49:00 *** analyst74 has joined #aichallenge 2011-12-17T01:55:51 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-17T01:57:25 *** Palmik has joined #aichallenge 2011-12-17T01:58:11 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-17T02:07:25 got back to #1 on tcpants 2011-12-17T02:07:36 . o O ( all hail! king of the looosers! ) 2011-12-17T02:08:47 *** Kurnevsky has joined #aichallenge 2011-12-17T02:14:19 http://tcpants.com/replay.2521 2011-12-17T02:14:26 kind of too bad that ended at 750 2011-12-17T02:22:51 *** itzkow has joined #aichallenge 2011-12-17T02:25:17 http://tcpants.com/replay.2401 2011-12-17T02:25:20 that game makes me sort of dizzy 2011-12-17T02:25:22 all the tiling 2011-12-17T02:31:08 *** snuglepuss has joined #aichallenge 2011-12-17T02:36:39 *** bhasker has joined #aichallenge 2011-12-17T02:36:40 wassup 2011-12-17T02:36:47 watching my tcpants.com wins 2011-12-17T02:37:07 :) 2011-12-17T02:37:14 btw is there a live game visualizer? 2011-12-17T02:37:19 that would be nice 2011-12-17T02:37:24 if one could watch the game live 2011-12-17T02:37:34 same visualizer but there's a way to stream it 2011-12-17T02:37:39 play_one_game_live .sh or .cmd 2011-12-17T02:37:47 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-17T02:37:47 ah i meant the games on tcp 2011-12-17T02:38:01 as they are being played from my perspective 2011-12-17T02:38:07 oh, no 2011-12-17T02:38:13 I think thestinger added some better output 2011-12-17T02:38:17 my bot isn't getting any better 2011-12-17T02:38:18 stats from his bot 2011-12-17T02:38:26 I guess the C++ one logs the map by default 2011-12-17T02:38:33 yea that i have 2011-12-17T02:38:48 i guess i could write something on top 2011-12-17T02:38:51 of my logs 2011-12-17T02:38:53 to grab the state 2011-12-17T02:38:55 and visualize it 2011-12-17T02:39:22 i implemented minmax for battle but my eval function just blows 2011-12-17T02:39:33 thinking of just implementing the CombatAnt battle strategy 2011-12-17T02:39:50 no matter how i tweak my eval it just keeps doing stupid things 2011-12-17T02:40:16 and also the fact that i have to limit my move generation severly for it to run in a reasonable time 2011-12-17T02:41:34 I never did see how to apply minmax to this 2011-12-17T02:41:36 bhasker: how are you doing minimax/ 2011-12-17T02:41:42 my way is quite simple and works reasonably well 2011-12-17T02:42:03 i am creating islands 2011-12-17T02:42:08 and doing minimax on the island 2011-12-17T02:42:14 which results in some idiotic things sometimes 2011-12-17T02:42:16 full minimax? 2011-12-17T02:42:22 no just a 2 ply 2011-12-17T02:42:28 what moves? 2011-12-17T02:42:28 one move per side 2011-12-17T02:42:41 usually i make the move i want to make for my ant 2011-12-17T02:42:50 and for enemy i pick a few random moves + 1 move towards my ant 2011-12-17T02:43:03 and i add a stay at same place move as well 2011-12-17T02:43:14 like a dummy move for every ant 2011-12-17T02:43:33 then i make all my moves, make all enemy moves then evaluate 2011-12-17T02:44:00 and any ant involved in the minmax in my favour is not moved after that 2011-12-17T02:44:28 the problem usually results in the fact that my path finding in the island uses a wrapped world like the main state which causes issues 2011-12-17T02:44:37 because it makes wrong assumptions about which ants are closeby 2011-12-17T02:44:43 and within battle range 2011-12-17T02:45:04 something on the top end of island will think the one at bottom is in battle range which in reality will never happen 2011-12-17T02:45:41 i just pass my battle functions locations/moves 2011-12-17T02:45:58 it can look at stuff in state if it wants, but everything it does to simulate battle etc. it does with sets 2011-12-17T02:46:26 for my ants, a "move" is a move for each ant 2011-12-17T02:46:57 yes same here 2011-12-17T02:47:03 hmm 2011-12-17T02:47:36 *** sigh has quit IRC (Remote host closed the connection) 2011-12-17T02:50:14 *** snuglepuss has quit IRC (Remote host closed the connection) 2011-12-17T02:50:16 http://tcpants.com/replay.2562 2011-12-17T02:50:19 stupid stupid bot 2011-12-17T02:51:03 tough map though 2011-12-17T02:51:32 so is everyone playing on tcpants.com now? 2011-12-17T02:51:35 and not on fluxid? 2011-12-17T02:51:45 I am anyway 2011-12-17T02:51:51 soooooo much faster 2011-12-17T02:51:52 I might be on fluxid too 2011-12-17T02:51:54 oh 2011-12-17T02:52:15 why is it faster on tcpants? 2011-12-17T02:52:18 better server? 2011-12-17T02:52:36 we suspect some fluxid players don't respect the move time limit 2011-12-17T02:52:53 btw what exactly is the limit 2011-12-17T02:52:58 my bot goes over it for sure 2011-12-17T02:53:04 500ms or 1second? 2011-12-17T02:53:20 500ms 2011-12-17T02:53:31 but the tcp server actually waits 5s to avoid network timeouts 2011-12-17T02:53:34 heh my bot will probably timeout every game 2011-12-17T02:55:54 my bot usually takes under 100ms i think 2011-12-17T02:56:12 *** Fandekasp has joined #aichallenge 2011-12-17T02:56:19 nice 2011-12-17T02:56:30 btw why the nice long line of ants antimatroid 2011-12-17T02:57:08 bhasker: i'm trying to improve my movement out from hills 2011-12-17T02:57:13 i'm not sure it's going very well :P 2011-12-17T02:57:15 oh, I always thought the time limit is 1sec 2011-12-17T02:57:22 :/ 2011-12-17T02:57:36 http://ants.fluxid.pl/replay.36798 2011-12-17T02:57:41 there's the problem with fluxid 2011-12-17T02:57:45 though its good in a way you are not wasting time moving around edges etc 2011-12-17T02:57:49 I pick a game to see why I didn't do well 2011-12-17T02:57:52 oh, I timed out 2011-12-17T02:58:03 in fact I was winning before that 2011-12-17T02:58:14 heh 2011-12-17T02:58:22 it took 900 turns to take my hill after I QUIT MOVING 2011-12-17T02:58:25 so I feel good about that 2011-12-17T02:58:29 btw do you guys just set an alarm and stop processing when you hit timelimit or something? 2011-12-17T02:58:45 yeah 2011-12-17T02:58:51 haha look at the huge dip in green AFTER EVERYONE ELSE IS OUT OF THE GAME 2011-12-17T02:58:56 nearly kills emself 2011-12-17T02:59:12 bhasker: I tell each thing what proportion of the time limit it can use up to 2011-12-17T02:59:24 then if it's over that it stops doing whatever i told it to do 2011-12-17T02:59:42 bhasker: I do whatever I want to do and hope it's fast enough 2011-12-17T03:00:00 heh same here ben 2011-12-17T03:00:02 but I did do profiling and optimization and I've got a big record of games where I don't go over 500ms with my cpus which are slower 2011-12-17T03:00:04 * antimatroid would like to see BenJacksons tron bot :P 2011-12-17T03:00:04 except my code is really crappy lol 2011-12-17T03:00:13 hypothetical bot obviously :P 2011-12-17T03:00:22 yeah I'm sorry I missed tron 2011-12-17T03:00:32 I didn't really get PW when I saw it so it didn't capture my imagination 2011-12-17T03:00:54 what's not to get? 2011-12-17T03:00:58 bhasker: secondary backup is my bot's willingness to sacrifice goes up over 200 and then again over 250 ants 2011-12-17T03:01:15 antimatroid: I just had no intuitive sense of what the game was about 2011-12-17T03:01:22 *** Gnome has quit IRC (Quit: Gnome) 2011-12-17T03:03:57 ah k 2011-12-17T03:04:23 btw do you guys also have something in the battle logic as to whom to attack 2011-12-17T03:04:26 *** sigh has joined #aichallenge 2011-12-17T03:04:30 if you see ants from multiple players 2011-12-17T03:04:35 do you normally pick which ones to attack 2011-12-17T03:04:49 or do you fight any battles you can win 2011-12-17T03:04:55 ah sigh! 2011-12-17T03:05:02 haven't seen a sighbot this time around 2011-12-17T03:05:18 yeah, I sat this one out 2011-12-17T03:06:02 *** Kurnevsky has left #aichallenge 2011-12-17T03:06:23 I moved to a new city and started a new job earlier in the year, so didn't have as much time 2011-12-17T03:06:49 how's the competition going? 2011-12-17T03:07:06 i have a crappy bot 2011-12-17T03:07:15 :) 2011-12-17T03:07:17 not much time this time and i frankly don't have a good idea of what to improve 2011-12-17T03:07:33 so probably going to land up somewhere in the 700-800's for all i know 2011-12-17T03:08:01 ah yeah, that happens 2011-12-17T03:08:16 http://bhickey.net:2080/replay.17221 2011-12-17T03:08:23 I accidentally watched a bhickey replay 2011-12-17T03:08:29 where if you look at orange's first hill loss 2011-12-17T03:08:36 it just lets blue go right in! 2011-12-17T03:08:43 and I panicked because I thought that was fixed 2011-12-17T03:08:48 but of course that's much older code 2011-12-17T03:10:17 heh 2011-12-17T03:10:27 my bot does that too though i think i fixed some stuff around it 2011-12-17T03:10:37 earlier it would just move away from the hill and have no visibility on it 2011-12-17T03:10:41 now i try and leave some ants behind 2011-12-17T03:10:43 around hill 2011-12-17T03:10:49 once i have more than a handful 2011-12-17T03:11:00 old versions of my bot did that 2011-12-17T03:11:16 http://tcpants.com/replay.2568 2011-12-17T03:11:22 here it did reasonably okay 2011-12-17T03:11:29 current version is all about defending my base... at their bse 2011-12-17T03:11:32 but got jacked by both the players 2011-12-17T03:11:46 BenJackson: http://tcpants.com/replay.2567 my bot sucks at actually getting hills :P 2011-12-17T03:11:52 which is you? 2011-12-17T03:11:59 bhasker: 2011-12-17T03:12:27 malazanAnt 2011-12-17T03:12:31 the yellow 2011-12-17T03:12:39 combat code is a bit better now 2011-12-17T03:12:55 the #1 thing I saw about that earlier is how much you wobble back and forth 2011-12-17T03:13:14 if you consider how many extra moves you'd get if you didn't waste so many 2011-12-17T03:13:25 yea its mostly because ants don't remember where they were headed 2011-12-17T03:13:38 so every turn they can get new priorities 2011-12-17T03:13:45 so sometimes they end up wobbling 2011-12-17T03:13:46 :-\ 2011-12-17T03:13:56 i need to add some logic around remembering where an ant was headed 2011-12-17T03:14:01 and only switch for higher priorites 2011-12-17T03:14:05 I like to think of some goals as "stable" in the sorting sense 2011-12-17T03:14:13 so if you move toward them you only become more attracted to them 2011-12-17T03:14:44 well sometimes the wobbling is also because my ants if you notice are still sometimes heading towards the same food 2011-12-17T03:14:51 thats a bug in my food collection 2011-12-17T03:14:55 more than one ant goes for same food 2011-12-17T03:14:59 and if you pick bots individually for tasks you need to make sure your selection and movement of an ant makes him better suited to be re-selected next time 2011-12-17T03:15:00 so that causes some useless wobbling too 2011-12-17T03:15:15 yeah individually scheduling food is important 2011-12-17T03:15:22 in fact individually scheduling everything is probably important 2011-12-17T03:15:28 but I resisted that for too long to change it all now 2011-12-17T03:16:32 heh mine is harder to fix given the way i am doing it 2011-12-17T03:16:42 i do a multisource BFS and move the ants to closest move next to them 2011-12-17T03:16:51 closest target 2011-12-17T03:16:59 so if two ants have the same food as closest target 2011-12-17T03:17:01 they will move to it 2011-12-17T03:17:20 not sure how to fix that :-\ 2011-12-17T03:17:53 yeah my original code was all about multi-source bfs 2011-12-17T03:17:55 since the ant has no idea what goal its moving towards 2011-12-17T03:18:13 but despite the fact that you will get obvious unwanted clumping from that I resisted changing for a long time 2011-12-17T03:18:35 if I was starting over I'd simplify my combat and spend all my time thinking about how to maneuver ants 2011-12-17T03:18:41 heh 2011-12-17T03:19:51 BenJackson: great forum post 2011-12-17T03:20:12 if only academic people cared about such things :( 2011-12-17T03:20:51 lol i actually won this one 2011-12-17T03:20:52 http://tcpants.com/replay.2579 2011-12-17T03:21:01 even though i timedout 2011-12-17T03:21:03 :=\ 2011-12-17T03:21:14 antimatroid: thanks 2011-12-17T03:21:44 I'm getting really excited while my bot is playing, anyone else gets that? 2011-12-17T03:22:49 itzkow: I get nervous for mine 2011-12-17T03:23:05 i try not to watch 2011-12-17T03:23:14 then hope it's a screen of "o"s when i look 2011-12-17T03:23:25 bhasker: you have a problem I share to some extent: you are reliant on outside influences (like food and movement of other ants) to jostle you into action sometimse 2011-12-17T03:24:00 *** Palmik has quit IRC (Remote host closed the connection) 2011-12-17T03:26:26 can you ever see "good comebacks" in this? 2011-12-17T03:26:34 there were some pretty good comebacks in planet wars 2011-12-17T03:26:35 oh yeah 2011-12-17T03:26:44 BenJackson: i think your example before was one? 2011-12-17T03:26:47 http://aichallenge.org/visualizer.php?game=252102&user=3473 2011-12-17T03:26:53 yes, my fave game :) 2011-12-17T03:26:55 but check xathis 2011-12-17T03:27:00 he's the king of that sort of thing 2011-12-17T03:27:32 "get out of my way, I have a game to win!" 2011-12-17T03:28:52 derp, played a game, my ants were braindead 2011-12-17T03:29:13 sigh 2011-12-17T03:29:17 stupid 2 hour waits 2011-12-17T03:29:37 oh I hate that 2011-12-17T03:29:59 I wonder how much value will the 1on1 games take in the finals 2011-12-17T03:30:22 I think those weird walls must have caused me grief in that game 2011-12-17T03:30:43 I find mazes very random 2011-12-17T03:30:45 http://tcpants.com/replay.2590 my bots not currently doing very well 2011-12-17T03:31:13 well shit I wanted to quit working on this and not be tempted by a last minute submission 2011-12-17T03:31:17 but that must be a bug I should fix 2011-12-17T03:31:44 i might need a "finish them" function 2011-12-17T03:34:42 frag stealing is funny 2011-12-17T03:35:20 by the way BenJackson. what do you think of my new bot? 2011-12-17T03:41:19 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-12-17T03:46:02 Hmmm. With BFS, if you don't enqueue a node, other processed nodes could "go around" the skipped one and do its siblings, right? 2011-12-17T03:49:13 wow that map is a perfect example of how testing against yourself doesn't work 2011-12-17T03:49:22 my bots all do terribly on that 2011-12-17T03:50:54 bhasker: beat you :P 2011-12-17T03:57:53 interesting 2011-12-17T03:58:28 say you have a game with scores at 5,1,1 and a player with score 1 times out, the other player is allowed to continue on to decreasing their ranking 2011-12-17T03:58:31 because they could increase it 2011-12-17T04:15:14 *** kire has joined #aichallenge 2011-12-17T04:24:35 You know how it is when you're optimizing neural networks 2011-12-17T04:24:47 It gets to be a bit of a weighting game 2011-12-17T04:27:53 *** rmmh has quit IRC (Quit: T-10h (IL -> CO)) 2011-12-17T04:39:26 when exactly is the deadline? 2011-12-17T04:39:28 timezone? 2011-12-17T04:40:10 *** scribble has joined #aichallenge 2011-12-17T04:40:16 11:59 EST 2011-12-17T04:41:10 yep 2011-12-17T04:41:15 we need a timer 2011-12-17T04:43:10 EST where though? 2011-12-17T04:43:23 EST in Australia pretty much means +10 2011-12-17T04:44:08 US east coast 2011-12-17T04:44:35 I'm still thorn which bot to submit 2011-12-17T04:44:59 I have a bot thats about 1.0 mu better on fluxid ant tcpats, but ended up about 3 mu below the previous one on aichallenge 2011-12-17T04:45:05 10:59am monday morning for me 2011-12-17T04:45:28 Garf: I imagine aichallenge rankings are all over the show atm 2011-12-17T04:45:39 *** analyst74 has quit IRC (Quit: HydraIRC -> http://www.hydrairc.com <- The alternative IRC client) 2011-12-17T04:46:14 I submitted my best alternate, gonna see how that ends up tonight 2011-12-17T04:46:21 if it's higher, I'll just leave it in place 2011-12-17T04:47:05 I think on previous contents it was also shown TCP doesn't always accurately reflect the real server well 2011-12-17T04:47:35 the difference between the two is that one tries to see as much terrain as possible, while the other tries to explore as fast as possible 2011-12-17T04:47:51 1 seems better on fluxid, 2 on aichallenge 2011-12-17T04:49:13 *** kire has quit IRC (Quit: Leaving) 2011-12-17T04:49:28 *** kire has joined #aichallenge 2011-12-17T04:51:57 The servers seem to have gone in and out of CPU availability, so the timeout ratio swings back and forth 2011-12-17T04:53:52 technically, is it possible to play more than one game per (bot) account on Fluxid.pl and such? 2011-12-17T04:54:25 maybe if you log in twice with the same name? dunno if it allows that 2011-12-17T04:57:49 http://aichallenge.org/visualizer.php?game=254975&user=757 FML TURN LIMIT REACHED 2011-12-17T04:57:57 Vaenom: no 2011-12-17T04:58:47 Fluxid: nawww, don't think you could have won anyway 2011-12-17T04:59:03 i'd be still 2nd 2011-12-17T04:59:08 but no tie 2011-12-17T05:00:07 yeah, just the come back was admirable 2011-12-17T05:00:16 would be even more so could one bring it back and win 2011-12-17T05:01:38 i think for some reason xmagnum went suicidal 2011-12-17T05:06:23 *** TL_ has joined #aichallenge 2011-12-17T05:08:43 heh, xathis *is* really pushing everyone elses rating down 2011-12-17T05:09:13 *** huisho has quit IRC (Quit: Page closed) 2011-12-17T05:10:06 *** Jacob_Strauss has joined #aichallenge 2011-12-17T05:11:23 *** kire has quit IRC (Remote host closed the connection) 2011-12-17T05:17:22 *** ZenWraithBot has joined #aichallenge 2011-12-17T05:17:30 !timeleft 2011-12-17T05:17:30 1 day, 18:41:29.698715 until cutoff 2011-12-17T05:17:35 There 2011-12-17T05:18:53 *** Kurnevsky has joined #aichallenge 2011-12-17T05:30:11 *** dvladim has joined #aichallenge 2011-12-17T05:30:52 *** Boscop has joined #aichallenge 2011-12-17T05:39:16 *** treeform has quit IRC (Remote host closed the connection) 2011-12-17T05:39:32 *** ikaros has joined #aichallenge 2011-12-17T05:40:23 *** SankaD has joined #aichallenge 2011-12-17T05:43:33 *** SankaD has quit IRC (Client Quit) 2011-12-17T05:44:56 *** Darhuuk_ has quit IRC (Read error: Operation timed out) 2011-12-17T05:49:24 *** Darhuuk has joined #aichallenge 2011-12-17T05:56:42 *** Murashka has joined #aichallenge 2011-12-17T05:57:53 *** Whitecoma has joined #aichallenge 2011-12-17T05:58:50 *** Whitecoma has quit IRC (Client Quit) 2011-12-17T06:06:06 http://aichallenge.org/visualizer.php?game=254611&user=1740 2011-12-17T06:06:10 that doesn't look very normal? 2011-12-17T06:07:04 *** replore has joined #aichallenge 2011-12-17T06:09:13 *** Boscop has quit IRC (Changing host) 2011-12-17T06:09:13 *** Boscop has joined #aichallenge 2011-12-17T06:13:25 *** Redgis has joined #aichallenge 2011-12-17T06:13:48 *** mcstar has joined #aichallenge 2011-12-17T06:16:39 *** Palmik has joined #aichallenge 2011-12-17T06:19:11 a1k0n: congratulations on second place :P 2011-12-17T06:24:45 *** jab_bott has joined #aichallenge 2011-12-17T06:25:32 *** hkraal^afk is now known as hkraal_ 2011-12-17T06:26:53 jab_bott: been using you visualizer extension, thanks, it was very useful 2011-12-17T06:26:57 r 2011-12-17T06:27:41 mcstar: glad to hear :) 2011-12-17T06:27:46 i would love to know how many people expect to finish in the top x 2011-12-17T06:28:12 something like 3*x I imagine :P 2011-12-17T06:28:21 antimatroid: hum - I have no idea - I haven't been following the rankings much until yesterday when I finally implemented battle (I've been busy) 2011-12-17T06:28:59 antimatroid: i expect to finish in the top x definitely 2011-12-17T06:29:19 mcstar: what about the end of the universe 2011-12-17T06:29:19 my last bot without battle was in place 542 - so I expect this one to get into the top 500, maybe higher 2011-12-17T06:29:28 *** zyberkiddy has joined #aichallenge 2011-12-17T06:29:30 according to janzert it's only most probably likely or something 2011-12-17T06:29:36 :p 2011-12-17T06:29:49 have you been perusing my source code? 2011-12-17T06:30:02 i had a variable called endOfTheUniverse 2011-12-17T06:30:07 or something 2011-12-17T06:30:08 haha 2011-12-17T06:30:17 top 200 would be nice I guess - but looking at the battle intelligence of some of the better bots, I highly doubt mine can compete with that within the short space of time we have left 2011-12-17T06:30:29 i want top 10 2011-12-17T06:30:37 i'll be happy with top 50 and first in australia 2011-12-17T06:30:46 >=100 for me 2011-12-17T06:30:49 (both of those) 2011-12-17T06:31:13 what's your bot called? 2011-12-17T06:31:24 antimatroid :) 2011-12-17T06:31:27 i haven't uploaded yet 2011-12-17T06:31:37 you've never uploaded? 2011-12-17T06:31:40 nope 2011-12-17T06:31:50 how do you know how good it is then? 2011-12-17T06:31:51 i did in beta, but it wasn't very good 2011-12-17T06:31:59 i played a bit on tcp in the last few day 2011-12-17T06:32:01 s 2011-12-17T06:32:01 tcp server testing probably :) 2011-12-17T06:32:06 just ironing out a few kinks 2011-12-17T06:32:37 problem is.. i noticed tcp and official behave very differently 2011-12-17T06:33:00 yeah that's fine 2011-12-17T06:33:01 http://tcpants.com/replay.1461 2011-12-17T06:33:08 there was a game the other day against some good bots which i won 2011-12-17T06:33:09 I haven't even had time to look into tcp server stuff - just implemented the basics and started to look at combat this week - wrote this entire complicated thing that didn't work at all, then last night I stripped most of it out and wrote a much simpler thing that works pretty well 2011-12-17T06:34:09 sounds familiar 2011-12-17T06:34:26 yea hehe 2011-12-17T06:34:29 well.. I say pretty well, actually it's a bit thick, but does the job in about 50% of cases - trouble is I know the good bots will avoid those and kill me on the other 50% 2011-12-17T06:35:56 antimatroid: which bot is yours in that replay? 2011-12-17T06:36:04 SkyCaptain 2011-12-17T06:37:18 ahh good job 2011-12-17T06:37:29 precint assault superfusion district 2011-12-17T06:37:43 lapd 2011-12-17T06:38:20 mcstar: you know the visualizer works on http://paste.aichallenge.org/ - some interesting pastes there 2011-12-17T06:38:23 *** zyberkiddy has quit IRC (Ping timeout: 245 seconds) 2011-12-17T06:38:36 *** UncleVasya has joined #aichallenge 2011-12-17T06:38:45 mcstar: :) 2011-12-17T06:38:48 best game ever 2011-12-17T06:38:53 nobody ever knows of it 2011-12-17T06:39:06 it just has a difficult to remember name 2011-12-17T06:39:17 antimatroid: how are you doing time-wise? 2011-12-17T06:39:22 sure, but once you hear it you should know of the awesomenes 2011-12-17T06:39:26 mcstar: laughing 2011-12-17T06:39:33 usually under 100ms 2011-12-17T06:39:38 thats nice 2011-12-17T06:39:46 my battle is oh so cheap 2011-12-17T06:40:06 mine too, i have a bfs that takes most of the time, ill take a look at it 2011-12-17T06:40:19 i dont know why is that so slow 2011-12-17T06:40:38 in every 5 turns, i go from 50ms to 250 say 2011-12-17T06:41:12 i have a max search path for all bfs's except my last one 2011-12-17T06:41:12 its a multisource bfs, with a zillion sources though 2011-12-17T06:41:24 at which point i bfs from any remaining targets and move ants as i find them 2011-12-17T06:41:52 i update closest hill info etc. only when i need to from vision info or change of hill info 2011-12-17T06:42:03 and only generate that up to vision depth etc. 2011-12-17T06:42:12 antimatroid: do you still have the battle code that you first wrote? or does it work on a dissimilar principle? 2011-12-17T06:42:15 and i store a lot of stuff about battle locations for each location 2011-12-17T06:42:19 mostly set up in turn 0 2011-12-17T06:42:33 depends on which battle code you think was my first 2011-12-17T06:42:47 my first attempt was a full minimax, but that was obviously never going to work 2011-12-17T06:42:56 you had an enumeration with agressive pruning 2011-12-17T06:43:08 thats what i remember 2011-12-17T06:43:09 i never did any pruning with it, i gave up and did something else 2011-12-17T06:43:19 ok, thats what i was asking 2011-12-17T06:43:46 a battle partition has a set of my ants and a set of enemy ants, for my ants I pick out like a set of offensive moves, defensive, etc., similarly for the enemy, then i run minimax on that 2011-12-17T06:44:02 so there's like 4-5 moves total for my set of ants and about 4 moves for the enemies 2011-12-17T06:44:12 jab_bott: can you link an example? 2011-12-17T06:44:25 i do partitions too 2011-12-17T06:44:33 i have a cluster search algorithm 2011-12-17T06:44:52 but i dont enumerate at all any more 2011-12-17T06:45:33 mine just depends on me being able to heuristically pick out decent sample moves 2011-12-17T06:45:55 my entire bot is based on multi-source BFS - it's not really that good tbh, a little bit too brute-force, but som eclever tweaks in there 2011-12-17T06:47:08 lol, guy pasted a replay with the bot name: MyBotHuman 2011-12-17T06:47:08 it runs super-fast though since I re-wrote it in C - I should probably time it 2011-12-17T06:47:21 MyBotHuman.php to be precise 2011-12-17T06:47:43 jab_bott: pls give a link to an interesting paste 2011-12-17T06:47:47 there are too many 2011-12-17T06:48:03 mcstar: looks like all the good ones have gone off the bottom - It must clip old ones 2011-12-17T06:48:46 I'll see if I can make one with my bot now, you'll see some of the inner workings of my bot 2011-12-17T06:48:49 my debug vis looks interesting, it has nice hue-maps, lines and stars 2011-12-17T06:49:10 jab_bott: thats probably not a good idea 2011-12-17T06:49:12 post a paste or some images :) 2011-12-17T06:49:31 im against too much sharing 2011-12-17T06:49:47 it is still not late to incorporate things 2011-12-17T06:49:53 mcstar: it's fine - I really don't mind, I'm not in this to be aggressively competitive - I'd rather learn and help others learn... I'm pretty sure I've got more to learn than I have to give, heh! 2011-12-17T06:50:39 well, had my new alg. worked out better, i would expect to cut my ranking in half 2011-12-17T06:50:39 *** GeorgeSebastian has joined #aichallenge 2011-12-17T06:52:12 mcstar: yeah, same here - had my bot turned out to be a no. 1 winner, I'd expect to be in no. 1 by now 2011-12-17T06:52:24 *** GeorgeSebastian has quit IRC (Client Quit) 2011-12-17T06:52:25 "They didn't list it because they didn't think it "counts". Nobody "counts", we only care what you did and can do!" 2011-12-17T06:52:47 jab_bott: i didnt mean it like that :D 2011-12-17T06:52:55 mcstar: :D 2011-12-17T06:53:20 i mean that particular alg. has potential, im just too stupid to make the best of it in this time limit 2011-12-17T06:56:02 i think i wont change anything further.. too few time and it sorta works so far 2011-12-17T06:56:13 few->little 2011-12-17T06:56:16 ill just relax and watch 2011-12-17T06:56:17 yea 2011-12-17T06:56:29 lies 2011-12-17T06:56:38 gotta do something stupid at the last minute and submit :) 2011-12-17T06:56:40 hehe 2011-12-17T06:56:41 *** nha has joined #aichallenge 2011-12-17T06:56:52 well on the first submission the chances are high :)) 2011-12-17T06:56:58 hrm... the replay file for this game I just played is 118M - perhaps not so good for uploading 2011-12-17T06:57:11 yes, but it'll be funnny if i pull it off :p 2011-12-17T06:57:11 i couldnt break the 80 skill barrier (yet) 2011-12-17T06:57:22 mcstar: could? :P 2011-12-17T06:57:44 obviously not, maybe this new submission will 2011-12-17T06:57:56 it was marginally worse 2011-12-17T06:57:59 like 79.5 2011-12-17T06:58:38 ikaros: what was wrong with lue? 2011-12-17T06:58:40 lua 2011-12-17T06:58:56 nothing.. i like to experiment :) 2011-12-17T06:59:02 i should have stayed with lua :P 2011-12-17T06:59:12 would have saved time and such :) 2011-12-17T06:59:13 was it slow or somethign? 2011-12-17T06:59:21 no its really fast with luajit 2011-12-17T06:59:28 i had a prototype 2011-12-17T06:59:43 74 skill isnt particularly bad :D 2011-12-17T06:59:48 and wanted to rewrite everything when i knew what to do with the bot 2011-12-17T07:00:03 it had no real combat back then 2011-12-17T07:00:44 code was a mess :) .. but well 2011-12-17T07:00:57 mine still is :) 2011-12-17T07:01:07 i tried some languages.. wrote the bot in lua->go->c++11 2011-12-17T07:01:12 and for fun in ruby 2011-12-17T07:01:13 lol 2011-12-17T07:01:17 thats a lot 2011-12-17T07:01:18 which was too slow 2011-12-17T07:01:20 c++11 ftw 2011-12-17T07:01:31 c++11 is nice BUT 2011-12-17T07:01:36 i wrote it in common lisp, and then in c++ and instantly switched to c++11 2011-12-17T07:01:44 i had to change alot when i did my first submission in c++11 2011-12-17T07:01:49 because the use gcc4.5 2011-12-17T07:01:55 and i used 4.6 2011-12-17T07:01:57 ... 2011-12-17T07:02:17 ikaros: don't say things like this to me 2011-12-17T07:02:30 better for you :) 2011-12-17T07:02:35 to check it before deadline hehe 2011-12-17T07:02:47 ikaros: uoset and uomap doest have yet have emplace methods even in 4.6 2011-12-17T07:02:49 foreach loops and nullptr type wont work 2011-12-17T07:03:00 -have 2011-12-17T07:03:14 pretty sure all i'm using is unordered_set and auto 2011-12-17T07:03:29 yea i just didnt think about the servers having an older version of gcc.. (stupid me) 2011-12-17T07:03:31 mcstar: <3 auto 2011-12-17T07:03:53 90% im doing my next aichallenge in lua only 2011-12-17T07:03:54 antimatroid: but lambdas arent polymorphic 2011-12-17T07:04:09 it's just the most fun and still so fast 2011-12-17T07:04:58 does lua have a repl? 2011-12-17T07:05:11 (i guess so) 2011-12-17T07:05:32 yea 2011-12-17T07:05:35 stupid question really 2011-12-17T07:05:42 hehe 2011-12-17T07:07:02 i hope by the time of the next challenge i will be able to think in a functional way 2011-12-17T07:07:05 lua itself is really nice but luajit is just a masterpiece.. never seen a jit for a dynamic language that fast.. and it is a one man project.. awesome 2011-12-17T07:07:33 ikaros: clojure? 2011-12-17T07:07:48 ok yea clojure is pretty fast too 2011-12-17T07:07:51 best of lisp + java hotspot 2011-12-17T07:07:53 yea 2011-12-17T07:08:03 i wanted to try that one day.. but never did hehe 2011-12-17T07:08:22 i only went as far as a couple tutorials 2011-12-17T07:08:24 i want to see someone win with haskell 2011-12-17T07:08:26 did some scheme years ago.. but i liked ocaml more than the lisps in terms of syntax 2011-12-17T07:08:36 oh 2011-12-17T07:08:43 damn, ok, i can take this... 2011-12-17T07:08:47 hehe 2011-12-17T07:09:03 i know most think ocaml has the ugly syntax lol 2011-12-17T07:09:18 brb 2011-12-17T07:09:20 i decided couple days ago i will learn haskell and not ocaml 2011-12-17T07:09:49 *** yahim91 has joined #aichallenge 2011-12-17T07:09:58 antimatroid: remember, there were some guys here who wanted to be the first in haskell cat. but they didnt think they will win overall 2011-12-17T07:10:11 bmh? 2011-12-17T07:10:17 cant remember 2011-12-17T07:10:23 might have been him i think 2011-12-17T07:10:25 one guy, but he was talking in plural 2011-12-17T07:10:29 yeah 2011-12-17T07:10:49 i will definitely look at their code 2011-12-17T07:11:02 i think it's been mostly him in the end 2011-12-17T07:11:16 a1k0n always does interesting stuff 2011-12-17T07:12:12 pretty excited to see xathis solution 2011-12-17T07:14:19 mcstar: why Haskell instead of OCaml? 2011-12-17T07:18:12 mcstar: you still there? http://paste.aichallenge.org/EwMd9/ 2011-12-17T07:18:37 Just made my bot fight against some older ones on a small map - the replay file is only 13M for that one! 2011-12-17T07:20:37 we can see rhe solutions after the finals? 2011-12-17T07:21:25 yahim91: a lot of people share their code once submissions close 2011-12-17T07:21:42 we're yet to not have first place release their code, and i'm pretty sure xathis has said he intents to release his 2011-12-17T07:21:51 jab_bott, i guess the blue boxes are for exploration 2011-12-17T07:22:08 awsome 2011-12-17T07:22:37 how do you decide the specific one "sliding in" again 2011-12-17T07:22:42 if you know what i mean :) 2011-12-17T07:23:30 you know something thaat would be really awesome would be a thing in the profile where a competitor could tick a "release my code" button 2011-12-17T07:23:41 happening the first time on turn 12 with : 32,77 and 40,82 2011-12-17T07:23:42 and if ticked the code would be available for download 2011-12-17T07:23:57 was afk 2011-12-17T07:25:10 bugnuts: if you write the ncessary code, they might just merge it 2011-12-17T07:25:32 yeah maybe after the deadline.... 2011-12-17T07:26:49 ikaros: yeah - blue are start nodes for exploration BFS (not shown) red are combat 2011-12-17T07:27:13 yea 2011-12-17T07:27:25 i had something similar too in the beginning of the contest 2011-12-17T07:27:29 for exploration 2011-12-17T07:27:45 but didnt work out a good way to make previously seen tiles unseen again 2011-12-17T07:28:08 *** kire has joined #aichallenge 2011-12-17T07:28:11 so minimizing "never seen" tiles worked.. but not minimizing "invisible" tiles 2011-12-17T07:28:25 *** stilkin has joined #aichallenge 2011-12-17T07:28:27 ikaros: I track both visible and unseen 2011-12-17T07:28:30 hello 2011-12-17T07:29:04 *** Zaphus has joined #aichallenge 2011-12-17T07:29:04 ikaros: you notice the grid of blue squares marching around - if you turn on fog you'll notice that they line up with the edges fo visible squares - to ensure my ants keep re-exploring seen tiles that are no longer visible 2011-12-17T07:29:51 I wanted to ask when submission closes, it says on the 18th, but at what time ? 2011-12-17T07:30:01 midnight EST 2011-12-17T07:30:11 -5 utc 2011-12-17T07:30:17 EST will do fine 2011-12-17T07:30:18 ;-) 2011-12-17T07:30:27 (I live in GMT+1) 2011-12-17T07:30:39 +1 2011-12-17T07:31:00 ok jab_bott so your ants have some kind of memory? so they dont jitter between invisible areas? 2011-12-17T07:31:17 wait that didn make any sense >_< 2011-12-17T07:31:20 *** olexs has joined #aichallenge 2011-12-17T07:31:55 ikaros: not really memory as such - they have a "preferred direction" and if all else is equal they will go in that direction, it's reset every time they move so by default they will keep going the same way 2011-12-17T07:31:58 5AM I guess 2011-12-17T07:32:07 yes 2011-12-17T07:32:09 monday 2011-12-17T07:32:12 no 2011-12-17T07:32:14 thanks! 2011-12-17T07:32:14 5 am 2011-12-17T07:32:16 6 2011-12-17T07:32:18 6 2011-12-17T07:32:20 6 2011-12-17T07:32:22 6 2011-12-17T07:32:22 true :) 2011-12-17T07:32:24 i hate when i mistype somethign 2011-12-17T07:32:37 something** 2011-12-17T07:32:40 ;-) 2011-12-17T07:32:40 ok thats what i meant with memory.. you keep track of the ants 2011-12-17T07:32:45 dont aggrevate me bro 2011-12-17T07:32:49 and right now its ~7:30am EST (17th) right ? 2011-12-17T07:32:54 IM GOING BALLISTIC 2011-12-17T07:32:55 ikaros: ahh yea - I have an ant object for each one 2011-12-17T07:33:00 mcstar: I wont, just kidding. thanks for the info 2011-12-17T07:33:09 im kidding too 2011-12-17T07:33:12 i canceled that.. each turn brings "fresh" ants now :) 2011-12-17T07:33:24 *** Kurnevsky has left #aichallenge 2011-12-17T07:34:29 ikaros: I don't really use the ant objects for much - they track when they've moved so ants can move on to each other's squares if they know the other will move out of the way 2011-12-17T07:35:16 lol, i remember when i tried to solve that problem in a nice way 2011-12-17T07:35:22 hehe 2011-12-17T07:35:33 i made it more un-understandable with each try 2011-12-17T07:35:46 finally i went with a simple alg. that just works 2011-12-17T07:35:49 sounds familiar 2011-12-17T07:35:51 even if not optimally 2011-12-17T07:35:54 my newest version does that by sorting all ants commands by scores 2011-12-17T07:36:07 previously i just recursively pushed them away 2011-12-17T07:37:14 *** stilkin has quit IRC (Quit: bye!) 2011-12-17T07:37:27 mcstar: did you check the visualizer paste? not exactly awesome but more interesting than some of them up there 2011-12-17T07:37:41 yep 2011-12-17T07:37:54 it looks nice, but i dont find it useful :D 2011-12-17T07:38:00 *** pairofdice has joined #aichallenge 2011-12-17T07:38:07 theres no really important info up on it 2011-12-17T07:38:15 *** mj41 has joined #aichallenge 2011-12-17T07:39:11 i mean mine for example show exactly how the battle is resolved 2011-12-17T07:39:19 and why the ants act the way they do 2011-12-17T07:39:27 but i dont want to upload that 2011-12-17T07:39:43 but i dont have fancy yellow stars 2011-12-17T07:39:54 i used your star to draw little hexagons :D 2011-12-17T07:40:40 *** Zaphus has quit IRC (Quit: Page closed) 2011-12-17T07:41:19 @seen delineate 2011-12-17T07:41:19 UncleVasya: I have not seen delineate. 2011-12-17T07:41:50 jab_bott: the only problem is that on normal sized maps, it takes too much time to stream the debug info 2011-12-17T07:41:55 *** Anilm3 has joined #aichallenge 2011-12-17T07:42:03 Hi! 2011-12-17T07:42:15 *** flowenol has joined #aichallenge 2011-12-17T07:42:27 sometimes it takes >2 secs to write out a hue-map 2011-12-17T07:42:33 mcstar: yeah - you have to keep disabling large parts of it if you want anything that doesn't take forever 2011-12-17T07:42:40 and its not even a full board 2011-12-17T07:43:00 for the next contest, i opt for a binary interface 2011-12-17T07:43:16 so you wont have to parse test 2011-12-17T07:43:18 text* 2011-12-17T07:44:03 mcstar: that's the price of using the existing visualizer though - I already spent long enough on the ai-state visualization, I'm sure it helped loads of people though so I don't resent the time.. pity I didn't have long left to work on my bot though 2011-12-17T07:44:07 and maybe you should add a full array visualizer 2011-12-17T07:44:33 i.e. expect values to be rgba numbers, that span a nxm matrix 2011-12-17T07:45:00 mcstar: why do you choose Haskell instead of OCaml? 2011-12-17T07:45:02 im just giving you ideas for the next contest 2011-12-17T07:45:14 i understand you spent much time with this, which is appreciated 2011-12-17T07:45:30 (but those ideas, you probably knew already) 2011-12-17T07:45:43 hi Anilm3 2011-12-17T07:46:00 Hi mcstar 2011-12-17T07:46:03 Hi UncleVasya 2011-12-17T07:46:18 Does anybody know who is the owner of delineate (rank 4)? He submited in very November and still sitting so high. 2011-12-17T07:46:19 mcstar: yeah - well maybe we can do something really useful for the next challenge and get it incorporated into the game code and visualizer from early on 2011-12-17T07:46:23 Anilm3: hi 2011-12-17T07:46:56 *very early November 2011-12-17T07:47:15 *** goffrie has quit IRC (Ping timeout: 244 seconds) 2011-12-17T07:47:18 *** goffrie_ has joined #aichallenge 2011-12-17T07:47:19 *** goffrie_ is now known as goffrie 2011-12-17T07:48:05 UncleVasya: if i want a functional language, why not go with a strict one, and haskell has monads, which i find interesting, and has MANY implementations, ocaml has only one, and the type system is better, and i dont want to put up with those kind of numerical operator bs that ocaml has, and the haskell community is growing, and there are some very smart people in there, and it is lazy, which i find 2011-12-17T07:48:07 also interesting, and 2011-12-17T07:48:19 cant think of anything else right now 2011-12-17T07:48:32 but there are a couple thosand reasons left 2011-12-17T07:49:08 *** g0llum has joined #aichallenge 2011-12-17T07:50:59 oh 2011-12-17T07:51:01 concurrency 2011-12-17T07:51:16 what has ocaml to offer there? 2011-12-17T07:51:27 You need to talk with Smilry. He will motivate you to make the right choise :) 2011-12-17T07:51:34 in what part are ocaml obejcts immutable? 2011-12-17T07:51:38 *Smiley 2011-12-17T07:51:47 UncleVasya: i made the right choice already 2011-12-17T07:52:21 Haskell is too obvious... 2011-12-17T07:52:28 Speaking of the visualizer... I found this cut down the delay between the game finishing and the browser launching: https://gist.github.com/1490137 2011-12-17T07:52:30 pfff 2011-12-17T07:52:36 i wont even comment on your comment 2011-12-17T07:52:44 Too obvious? :D' 2011-12-17T07:52:53 UncleVasya: you are an unpurposeful commenter 2011-12-17T07:53:15 he means haskell isnt obscure enough 2011-12-17T07:53:24 but has i wanted obscure i would learn J 2011-12-17T07:53:40 that language has a VOCABULARY 2011-12-17T07:53:47 Sounds like hipster talk 2011-12-17T07:54:29 J has 2 tables of 1 character operator vocabulary, each character has 3 different meaning depending on context and punctuation 2011-12-17T07:54:38 matthewd: what's a gist? 2011-12-17T07:54:46 matthewd: and what file is that in? 2011-12-17T07:54:51 gist is the Git pastebin 2011-12-17T07:54:59 github* 2011-12-17T07:55:06 (I think) 2011-12-17T07:55:20 UncleVasya: please, go with this instead of ocaml http://www.jsoftware.com/help/dictionary/vocabul.htm 2011-12-17T07:55:33 matthewd: I don't really understand the ants code much - I never did python or javascript before, so I just found the releavnt bits and modified those without trying to understand the rest of it 2011-12-17T07:55:40 mcstar: yeahhh, vocabulary is something I need to have... 2011-12-17T07:55:58 jab_bott: The filename is in the diff there -- ants/visualizer/visualize_locally.py 2011-12-17T07:56:04 go where? 2011-12-17T07:56:16 *** choas has joined #aichallenge 2011-12-17T07:56:25 UncleVasya: go to the coffee shop 2011-12-17T07:56:29 i dont know :D 2011-12-17T07:56:41 "go with this" == use this 2011-12-17T07:56:54 choose this 2011-12-17T07:56:57 whatever 2011-12-17T07:56:57 idiom 2011-12-17T07:57:05 I understand but 2011-12-17T07:57:09 I didn't particularly prove it, but I think the problem is that the .sub() call means after swapping in the massive debug output into the HTML, the regexp then has to scan through all of it 2011-12-17T07:57:13 choose for what, use for what? 2011-12-17T07:57:24 you want an obscure language right? 2011-12-17T07:57:25 matthewd: is that modification from my vis overlay branch/fork? 2011-12-17T07:57:40 pls explain then what did you mean by "Haskell is too obvious..." 2011-12-17T07:57:53 and what quality did you find in ocaml that made you choose that language 2011-12-17T07:57:58 I wanted when I was choosing languahe for Ants. 2011-12-17T07:58:12 I do not choose language for something right now. 2011-12-17T07:58:17 jab_bott: I don't think your branch differs there, but yes -- I needed it because with a bunch of vis (and esp. info) was just making the output enormous 2011-12-17T07:58:28 matthewd: I noticed some dodgy escaping in the JSON that gets embedded in the HTML - it doesn't handle spaces in strings well at all 2011-12-17T07:58:30 pairofdice: can you understand me? 2011-12-17T07:58:50 jab_bott: Yeah, the space stripping is there too. 2011-12-17T07:59:08 mcstar, possibility is there 2011-12-17T07:59:36 mcstar: By too obvious I mean that when somebody start thinking about functional way, the first thing he hear is Haskell. 2011-12-17T07:59:41 jab_bott: I suspect it only exists to avoid useless bytes in the output, though I didn't investigate 2011-12-17T07:59:56 matthewd: probably a bit late to worry about that now anyway :/ - personally I never made much use of the info command, though I did use it a couple of times very sparingly due to the slow speed 2011-12-17T08:00:00 UncleVasya: theres a reason for that probably 2011-12-17T08:00:18 but its not logical to choose ocaml just because you hear too much of haskell 2011-12-17T08:00:38 pairofdice: it was just a check to see if i make sense 2011-12-17T08:01:09 I think he just wanted a language he had no experience with for Ants 2011-12-17T08:01:25 there are thousands of such languages 2011-12-17T08:01:33 he must have a reson 2011-12-17T08:01:33 But you can only pick one 2011-12-17T08:01:47 or, must have had 2011-12-17T08:02:48 UncleVasya: was you motivation to win a category, in which you suspected wont be many submissions? 2011-12-17T08:02:50 r 2011-12-17T08:03:24 anyway 2011-12-17T08:03:47 tell me something about mutability in ocaml 2011-12-17T08:03:53 mcstar: no :) 2011-12-17T08:04:29 @rankings 2011-12-17T08:04:30 pairofdice: Top 10 players: xathis(98.8), a1k0n(91.4), pguillory(91.1), delineate(89.4), lazarant(89.4), GreenTea(89.0), Memetix(88.6), ThisIsNotABug(88.4), cumbuz(88.3), protocolocon(88.0) 2011-12-17T08:04:34 I was inspired to learn something about functional programming. 2011-12-17T08:05:32 I did some googling and found that F# is most suitable for me. 2011-12-17T08:05:53 Then I came here and asked about F# support. 2011-12-17T08:06:09 f# is multi paradigm 2011-12-17T08:06:29 Smiley said there is no support for F# but if I wanna touch the functional way I can use OCaml. 2011-12-17T08:06:37 what? 2011-12-17T08:06:41 no support for it? 2011-12-17T08:06:47 i find that hard to believe 2011-12-17T08:07:04 its not different than c# in support 2011-12-17T08:07:32 its been awhile since I have tried, but using F# on mono has been a pain so far. 2011-12-17T08:08:05 *** bhasker has quit IRC (Ping timeout: 240 seconds) 2011-12-17T08:08:09 speed was compromised? 2011-12-17T08:08:19 momobot still on version 1 and #11 2011-12-17T08:09:59 *lol* I was just in the top 500 for four minutes! 2011-12-17T08:10:45 i wish you back there 2011-12-17T08:10:57 mcstar: btw, I think this is very silly to choose non-popular language and be happy to be #1 on it :) 2011-12-17T08:11:02 do you get picked for more games if you rank higher 2011-12-17T08:11:03 ? 2011-12-17T08:11:48 UncleVasya: sure, but i dont REALLY know you so it could have been your reason as far as im concerned 2011-12-17T08:11:59 *** Jacob_Strauss has quit IRC (Quit: Page closed) 2011-12-17T08:12:38 *** dvladim has quit IRC (Ping timeout: 252 seconds) 2011-12-17T08:12:54 my bot seems really clever when it plays against olv versions of itself, but quite dumb when pitted against other bots 2011-12-17T08:13:09 *** NotABug has joined #aichallenge 2011-12-17T08:13:17 it is time for me to stop meaningless talks and go impement something before deadline. 2011-12-17T08:13:23 *** zyberkiddy has joined #aichallenge 2011-12-17T08:13:24 :D 2011-12-17T08:13:28 i second that 2011-12-17T08:13:37 *** UncleVasya has quit IRC () 2011-12-17T08:13:39 but its good to relax a bit 2011-12-17T08:13:45 the situation is very tense 2011-12-17T08:13:50 it inhibits productivity 2011-12-17T08:13:51 I should do this ai-class final exam 2011-12-17T08:14:36 mcstar: what's your bot called? 2011-12-17T08:14:54 agentmsith 2011-12-17T08:14:57 smith 2011-12-17T08:15:58 ooh - so close to top100 - a little tweaking and you can get there I'm sure :) 2011-12-17T08:16:34 maybe this v12 will get there - it's annoying that it takes so long to get your ranking stabilised though 2011-12-17T08:17:01 I wish I had more than a week time with this one. :| 2011-12-17T08:17:26 even with as many servers as there currently are, there are too many submissions now that the deadline is approaching 2011-12-17T08:18:32 *** lawmax has joined #aichallenge 2011-12-17T08:18:55 as a result - not enough games... I think I'll have one more upload tonight, and then the final optimized one shortly before the deadline 2011-12-17T08:19:31 mcstar: your bot is a bit like mine in that it doesn't know what to do when it gets too crowded 2011-12-17T08:19:46 but neither has enough time to stabilize the ranking, I just hope that it will make top100... 2011-12-17T08:19:51 *** lawmax_ has joined #aichallenge 2011-12-17T08:19:57 *** lawmax has quit IRC (Client Quit) 2011-12-17T08:20:36 jab_bott: you mean when it is sorrounded on all sides? 2011-12-17T08:20:51 *** lawmax_ has quit IRC (Client Quit) 2011-12-17T08:20:52 mcstar: when you have too many of your own ants, they just jibber around 2011-12-17T08:20:58 oh 2011-12-17T08:21:02 yeah 2011-12-17T08:21:17 they probably should have some specific target 2011-12-17T08:21:33 *** law-max has joined #aichallenge 2011-12-17T08:21:34 really they should all pile in towards the enemy - I need to work on that (after I've done this exam) 2011-12-17T08:21:59 i think im gonna redirect them to the closest enemy hill or something 2011-12-17T08:22:13 mcstar: are you gonna share your battle resolution after the contest? it looks pretty good.. better than mine for sure 2011-12-17T08:22:27 but i would need a calculattor to estimate the defcon level of my own hills 2011-12-17T08:22:37 sure 2011-12-17T08:22:40 I might not have time to do anything clever as I don't have an A* search implemented or a priority queue 2011-12-17T08:22:43 but its too simple 2011-12-17T08:22:50 you might not learn anything from it 2011-12-17T08:24:54 i really wish i had come up with that battle logic a long time ago 2011-12-17T08:24:55 mcstar: heh - mine si very dumb - it just works out enemy possible moves based on water, 'draws' a map of where they can go, runs an edge-detection on it to get safe spots surrounding the enemy ants, BFS to find my ants towards those spots... then from the enemy ant POV I calculate if it's scared from a particular direction and erase the coverage map from that direction so my ants pile in 2011-12-17T08:25:06 i spent too much time on enumeration/full battle resolution and evaluation 2011-12-17T08:25:45 no comment XD 2011-12-17T08:25:58 mcstar: it kind of works when my ants are lined up, but fron an angle it's really bad 2011-12-17T08:26:10 *** law-max has quit IRC (Ping timeout: 258 seconds) 2011-12-17T08:26:46 mcstar: you mocking? I know it's rubbish, but I wrote it in about 4 hours last night... the thing I had before was insanely complex and I scrapped it because it didn't work :P 2011-12-17T08:27:09 no mocking 2011-12-17T08:27:21 im just not comfortable saying that mine works in a similar way 2011-12-17T08:27:52 cause you would draw the conclusion that "hm, i can do as good as he is, with little extra effort" 2011-12-17T08:28:14 heh - maybe I could... I will keep trying 2011-12-17T08:28:26 ok, but remember i didnt say that :D 2011-12-17T08:28:50 aikon has, admittedly a simple bot 2011-12-17T08:29:00 and he is #3 2011-12-17T08:29:03 or was at least 2011-12-17T08:29:30 Is #2 2011-12-17T08:29:52 ok, i cant keep up with rankings 2011-12-17T08:30:24 mcstar: sorry, had to walk away. Re: F#, yes it was slower, tail calls weren't right (but might be now), and the interactive shell had lots of problems. It would crash on valid code. 2011-12-17T08:31:10 it may be much better now, and Microsoft's lead F# guy said that they are working hard to make sure its ok on Mono. 2011-12-17T08:32:02 That being said... if there was as starter pack for it a month ago, I might have used it. 2011-12-17T08:32:04 later. 2011-12-17T08:32:16 ok 2011-12-17T08:35:25 i think im going offline 2011-12-17T08:35:37 need to work out some issues 2011-12-17T08:35:40 bb 2011-12-17T08:35:44 *** mcstar has left #aichallenge ("WeeChat 0.3.6") 2011-12-17T08:38:49 *** sigh has quit IRC (Remote host closed the connection) 2011-12-17T08:42:01 *** zyberkiddy has left #aichallenge 2011-12-17T08:46:33 *** f4hy has quit IRC (Ping timeout: 255 seconds) 2011-12-17T08:48:56 *** praveen has joined #aichallenge 2011-12-17T08:50:33 *** Redgis has quit IRC (Read error: Connection reset by peer) 2011-12-17T08:51:14 *** Redgis has joined #aichallenge 2011-12-17T08:57:11 *** TheLinker has joined #aichallenge 2011-12-17T08:57:39 !timeleft 2011-12-17T08:57:39 1 day, 15:01:20.664279 until cutoff 2011-12-17T08:58:25 *** Kurnevsky has joined #aichallenge 2011-12-17T09:05:15 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-17T09:05:16 *** Murashka has quit IRC (Ping timeout: 258 seconds) 2011-12-17T09:09:37 *** besh has joined #aichallenge 2011-12-17T09:14:10 *** GeorgeSebastian has joined #aichallenge 2011-12-17T09:17:23 *** Anilm3 has quit IRC (Ping timeout: 252 seconds) 2011-12-17T09:28:14 !timeleft 2011-12-17T09:28:15 1 day, 14:30:45.029801 until cutoff 2011-12-17T09:38:30 *** likedeeler has joined #aichallenge 2011-12-17T09:54:34 *** Redgis has quit IRC (Ping timeout: 252 seconds) 2011-12-17T10:01:01 *** dvladim has joined #aichallenge 2011-12-17T10:05:14 *** avdg has joined #aichallenge 2011-12-17T10:05:50 *** yahim91 has quit IRC (Ping timeout: 258 seconds) 2011-12-17T10:07:12 I sooo did not want to be making big changes at this point 2011-12-17T10:09:47 *** u_ has joined #aichallenge 2011-12-17T10:11:35 *** likedeeler has quit IRC (Quit: Page closed) 2011-12-17T10:18:09 *** mleise has joined #aichallenge 2011-12-17T10:24:48 *** Redgis has joined #aichallenge 2011-12-17T10:32:32 derp accidentally made the cost of moving free 2011-12-17T10:32:37 that makes for some odd behavior 2011-12-17T10:33:04 lol the tutorial map is on tcpants 2011-12-17T10:34:25 tcpclient needs a graceful shutdown 2011-12-17T10:36:50 BenJackson: you mean killing the bot process? 2011-12-17T10:37:04 then you timeout/lose 2011-12-17T10:37:59 would be nice to have a way to tell tcpclient 'shutdown when current game ends' 2011-12-17T10:38:12 but tbh it's not important 2011-12-17T10:38:22 a few timeouts are no biggie 2011-12-17T10:38:43 last time I did it my bot dropped from 1st to 3rd on tcpants because it dropped 3 quick games 2011-12-17T10:38:47 my vanity can't take it!! 2011-12-17T10:38:52 :x 2011-12-17T10:41:27 *** cyphase has quit IRC (Ping timeout: 240 seconds) 2011-12-17T10:48:34 *** besh has quit IRC (Quit: Page closed) 2011-12-17T10:49:08 *** xcvxcvx has joined #aichallenge 2011-12-17T10:53:48 *** moises has joined #aichallenge 2011-12-17T10:55:06 *** cyphase has joined #aichallenge 2011-12-17T10:55:38 *** Murashka has joined #aichallenge 2011-12-17T10:57:43 *** besh has joined #aichallenge 2011-12-17T10:58:03 ok, so my v7 bot got to 460th in 9 games - just uploaded v8, let's hope it does a bit better/faster as I'm runningout of time :P 2011-12-17T10:58:23 sorry, it takes a long time to rank up on the official server 2011-12-17T10:58:33 and how fast you go is kind of random 2011-12-17T10:58:50 let's suppose I'm top 10. I could win a game vs #100 (probably) from position 7800 2011-12-17T10:58:53 but it won't give me that game 2011-12-17T10:59:06 one time I played a bunch in a row against #1000+ 2011-12-17T10:59:10 and you just don't get points for that 2011-12-17T10:59:17 yeah - I want to fight some of the better bots so I can see how my combat code does (I only wrote it last night after scrapping a previous attampt at a more complex system - and improved it today) 2011-12-17T11:01:14 9 games overnight isn't bad I guess - I just wish I could fight top 500 bots to start with - not to get ranking, but to see how my fights go in the real world (since testing against my old bots is prone to over-fitting specific cases) 2011-12-17T11:01:32 I assume you're on tcp too 2011-12-17T11:01:33 *** airapuru has joined #aichallenge 2011-12-17T11:01:36 *** choas has quit IRC (Ping timeout: 240 seconds) 2011-12-17T11:01:53 no - I've not looked into that... is it someone else's server? 2011-12-17T11:02:05 you need to look into that immediately :) 2011-12-17T11:02:10 see the forum post 2011-12-17T11:02:12 *** airapuru has quit IRC (Client Quit) 2011-12-17T11:02:25 in terms of competition fluxid > tcpants >> bhickey 2011-12-17T11:02:36 bhickey is intentionally aiming at a lower bar 2011-12-17T11:02:43 I was planning to do this ai-class final exam and let my v8 bot have some battles 2011-12-17T11:02:46 I will adjust what I'm running there if you have requests 2011-12-17T11:03:23 do those servers grab bot binaries from the official server or something? 2011-12-17T11:04:04 no, you run on your machine 2011-12-17T11:04:16 bot interaction is over tcp 2011-12-17T11:04:29 it's like 'play_one_game' but with remote participants 2011-12-17T11:04:37 here's a game I just watched: http://tcpants.com/replay.2883 2011-12-17T11:04:50 I am super pleased with what my bot did because that map has that weird dead space 2011-12-17T11:04:56 and it handled that perfectly 2011-12-17T11:06:56 isn't dead-space against the spec.? 2011-12-17T11:08:24 oh I see - you mean no targets reachable rather than the space itself not reachable 2011-12-17T11:09:14 right 2011-12-17T11:09:28 I know the first time I saw that map I put way too many ants in there 2011-12-17T11:09:42 at this point my bot even knows that enemie can't come from there 2011-12-17T11:09:56 wow - impressive 2011-12-17T11:10:05 very effective use of the ants you do have 2011-12-17T11:10:36 I notice at move 592, a cluster of ants around BenJackson (16, 102) are discovering that it's dead and turn back immediately 2011-12-17T11:11:26 your battle code looks amazing too - you seem to come out better off most times 2011-12-17T11:12:24 if I have any advantage it's that my implementation is totally different than everyone else's 2011-12-17T11:13:04 why's that? 2011-12-17T11:13:38 well I'm not using min-max and I'm not using the "simple combat" from the forum 2011-12-17T11:13:49 I doubt anyone cloned a1k0n successfully :) 2011-12-17T11:14:43 when I first read a1k0n's post I promised myself to fully understand his post in the future, but that never happened... 2011-12-17T11:15:06 oh - I just made some stuff up - dunno what the 'simple combat' does - so I might be doing something similar 2011-12-17T11:15:07 *** UncleVasya has joined #aichallenge 2011-12-17T11:16:18 I hate to resubmit at this point because there's no way I'll settle by tomorrow night 2011-12-17T11:16:23 but I have to fix that exploration bug 2011-12-17T11:16:51 my combat is calculate enemy-attackable squares for next turn, edge-detect to find initial frontier for BFS, charge my ants to those locations, then some crappy logic to check if the enemy ant is in danger from my ants and if so kill his attack field so my ants pile in 2011-12-17T11:17:19 managed to impelement a1k0n's method, same with memetix. but it needs a lot of tuning and i don't have time anymore. even implemented a technique described by benjackson. 2011-12-17T11:17:23 *** amstan has joined #aichallenge 2011-12-17T11:17:23 *** ChanServ sets mode: +o amstan 2011-12-17T11:17:24 jab_bott: pretty reasonable 2011-12-17T11:17:24 I gave up on fixing things at this point, It's a little risky right now 2011-12-17T11:18:04 itzkow: I wanted to but then: http://aichallenge.org/visualizer.php?game=256849&user=3473 2011-12-17T11:18:25 I fixed the bugs that led to that and now my new version beats my old version soundly 2011-12-17T11:18:39 BenJackson: so that post lists tcp servers - how do I run a game against one of them? 2011-12-17T11:18:51 go to one of them and see "Howto" 2011-12-17T11:18:57 basically download a tcpclient.py 2011-12-17T11:19:17 run tcpclient.py host port YOUR_BOT user pass 2011-12-17T11:19:30 host part include: fluxid.pl 2081, tcpants.com 2081, bhickey.net 2081 2011-12-17T11:19:41 the latter was intended for newcomers because fluxid got too tough 2011-12-17T11:19:53 fluxid is now overloaded, everything is timing out, db is getting corrupted 2011-12-17T11:21:31 I have a known bug where if a food pops next to my ant, it may cause collision between two of my ants, but that's quite rare I believe 2011-12-17T11:22:03 yeah, I had to special case that 2011-12-17T11:22:44 BenJackson: ahh - duh! I was thinking it literally just talked ant-speak and didn't even think of going there from a browser (despite the separate link!) 2011-12-17T11:23:56 so much for the ai-class final! 2011-12-17T11:24:34 Mmh 2011-12-17T11:24:43 Half-way done 2011-12-17T11:28:08 *** deltree_ has joined #aichallenge 2011-12-17T11:29:48 Ben, I see you uploaded your new bot, good luck 2011-12-17T11:30:03 thanks 2011-12-17T11:30:16 it's doing well on tcpants right now 2011-12-17T11:31:06 BenJackson: ok - I think I'm fighting online atm 2011-12-17T11:31:11 output is blank 2011-12-17T11:31:20 it has a random startup sleep 2011-12-17T11:31:24 to try to shuffle the matchups 2011-12-17T11:31:30 then you'll see the raw io 2011-12-17T11:31:49 *** ikaros has joined #aichallenge 2011-12-17T11:31:51 BenJackson: re ai-class final - I read through the questions at work yesterday - looks pretty easy, so I'll do it later :P 2011-12-17T11:32:14 BenJackson: yeah I'm fighting 2011-12-17T11:32:37 jab_bott: if you're on bhickey I will tune the level of competition for you 2011-12-17T11:32:50 when bmh put it up I ran my version 4 2011-12-17T11:32:55 why didn't I know of this before ?? :P I wish I had enough time to spend on stuff like this properly then I might get something really goods 2011-12-17T11:32:59 BenJackson: I'm on tcpants 2011-12-17T11:33:02 and as people levelled up we went thruogh 5, ... and it's up to 10 2011-12-17T11:33:09 BenJackson: I want to play some harder bots 2011-12-17T11:33:09 *** jmcarthur has quit IRC (Read error: Connection reset by peer) 2011-12-17T11:34:39 *** jmcarthur has joined #aichallenge 2011-12-17T11:34:55 !timeleft 2011-12-17T11:34:55 1 day, 12:24:04.639103 until cutoff 2011-12-17T11:36:22 BenJackson: my fight must be going really bad - I only see about 15 lines of 'o' commands :D 2011-12-17T11:36:37 it's like that sometimes 2011-12-17T11:36:41 *** praveen has quit IRC (Ping timeout: 258 seconds) 2011-12-17T11:36:50 seems to be hanging in there though 2011-12-17T11:37:02 *** Surya has joined #aichallenge 2011-12-17T11:37:24 nearly dead... 4 left 2011-12-17T11:38:36 *** dom7b5 has quit IRC (Ping timeout: 240 seconds) 2011-12-17T11:39:10 weird, I played the exact same guy in my first post-submit match as last time 2011-12-17T11:39:18 *** dvladim has quit IRC (Ping timeout: 248 seconds) 2011-12-17T11:39:37 I only remember because I was confused last time by two sequential games where I played people named "6"-something 2011-12-17T11:39:47 jab_bott: it's time to call and back them to reality. 2011-12-17T11:39:57 awesome!! BenJackson: if I specify num_rounds will it do that many games then finish? So I could say like 5 rounds then come back later? 2011-12-17T11:40:24 jab_bott: you can use -1 2011-12-17T11:40:33 if you don't say it will go forever 2011-12-17T11:40:36 but yes 2011-12-17T11:40:39 for continuing testing 2011-12-17T11:40:47 if you don't put it, it does 1 game 2011-12-17T11:40:49 *continious 2011-12-17T11:41:06 or something like that, when there is no game limith 2011-12-17T11:41:41 I don't see my game in the games list? 2011-12-17T11:42:22 *** SMJ has joined #aichallenge 2011-12-17T11:42:59 arrr I still fail to take hills when they're offered super early 2011-12-17T11:43:26 ahh there it is.. BenJackson http://tcpants.com/replay.2916 2011-12-17T11:44:50 ahhh - the game was still going on after I died so I was disconnected 2011-12-17T11:44:56 right 2011-12-17T11:45:06 good news is you will connect and start another game at that point 2011-12-17T11:45:07 if it's looping 2011-12-17T11:45:21 so you can actually lose games faster than you can win them :) 2011-12-17T11:46:16 um... I notice some people are running multiple bots - are they running them all locally or can you upload some? 2011-12-17T11:46:39 they're running them all locally 2011-12-17T11:46:42 multiple invocations of the client 2011-12-17T11:47:03 I'm running about 4 right now, one on tcpants, I think one on fluxid (if it hasn't died) and 2 on bhickey 2011-12-17T11:47:06 it's really cool! thanks for telling me to look into it immediately! 2011-12-17T11:47:10 the latter two are for other people to play 2011-12-17T11:47:11 old versions 2011-12-17T11:47:35 I'll try one against bhickey - see how I get on 2011-12-17T11:48:45 you'd be the only person "developing" there so if you're getting flattened I can dial it back 2011-12-17T11:49:16 http://aichallenge.org/visualizer.php?game=259634&user=757 HOLY SHIT 2011-12-17T11:49:20 1000 turns, rank stabilized » 2011-12-17T11:49:20 Won at 1000 » 2011-12-17T11:49:36 thestinger: ↑ 2011-12-17T11:49:56 lucky 2011-12-17T11:50:03 wow, looking at the graph I wouldn't expect it to be close 2011-12-17T11:50:38 ah, your surround of his base was a bit leaky 2011-12-17T11:50:43 what happened with flag? he's way down 2011-12-17T11:51:34 wow .. do you have code that favours attacking when turn limit gets close ? 2011-12-17T11:52:08 someone does pay attention to that 2011-12-17T11:52:24 I've seen ChrisH's bot go catatonic when it realizes it can't get meaningful ants to the battle 2011-12-17T11:52:26 because there's no time 2011-12-17T11:52:46 besh: no. my ants only go kamikaze when all hills are down 2011-12-17T11:53:02 *** praveen has joined #aichallenge 2011-12-17T11:53:08 *** dvladim has joined #aichallenge 2011-12-17T11:53:28 i think it started to choke because of turn time limit, if that happens combat slowly starts to malfunction (so by paradox it starts to be more offensive, lol) 2011-12-17T11:54:13 nice game 2011-12-17T11:54:30 I see iterative refinement of combat moves gets poorer 2011-12-17T11:54:37 @timeleft 2011-12-17T11:54:37 I had a similar game where I could have won if I had 2 more turns 2011-12-17T11:54:37 BenJackson: Run as fast as you can and don't look back. 2011-12-17T11:54:43 @contest time 2011-12-17T11:54:44 BenJackson: User error -- Replace user. 2011-12-17T11:54:47 haha 2011-12-17T11:54:53 !timeleft 2011-12-17T11:54:54 1 day, 12:04:06.547402 until cutoff 2011-12-17T11:55:00 oh that explains it 2011-12-17T11:55:03 ok, 36 hours 2011-12-17T11:55:10 Fluxid: when you can see alost 100% of the map and you know you've got like 10x more ants than them - you can probably afford to lose a few 2011-12-17T11:55:14 game rate about 82 2011-12-17T11:55:16 26 more games 2011-12-17T11:55:17 this contestbot drops some funny lines from time to time ;) 2011-12-17T11:55:18 so just pile in 2011-12-17T11:55:22 so nope, no way any submit at this point will settle 2011-12-17T11:55:36 * avdg wonders how many people will send their bot in the last 12 hours 2011-12-17T11:55:39 jab_bott: i wanted to implement that logic 2011-12-17T11:55:48 @Hello 2011-12-17T11:55:49 besh: I'm sorry Dave, err besh; I cannot 'Hello'. 2011-12-17T11:55:51 avdg: I didn't even want to do it today, but I had to fix a bug 2011-12-17T11:55:56 when mine blob is bigger than enemy blob: attack 2011-12-17T11:56:04 I did that (blob) 2011-12-17T11:56:11 but I didn't use my knowledge that I see all enemies 2011-12-17T11:56:23 I fear too many special cases, no matter how clever they are 2011-12-17T11:56:32 but i got problems with combat in general and then got crazy 2011-12-17T11:56:34 lazy* 2011-12-17T11:56:52 the latest code uploaded will be used for the trounament right? and I believe I still have 1.5 days to upload another one 2011-12-17T11:57:39 yes 2011-12-17T11:57:40 yeah 2011-12-17T11:57:47 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-17T11:57:47 ok 2011-12-17T11:58:00 my bot gets aggressive when most of the map is explored, which is kinda good with 1on1 battles 2011-12-17T11:58:04 but personally I think I will only just see enough games to feel comfortable 2011-12-17T11:58:24 if I subbed in the final hours I'd be sleepless until the finals 2011-12-17T11:58:27 meh, I tried to find the worst bot, it looks like that bot is a starter bot as well (and its even not timing out), oh well... 2011-12-17T11:58:36 http://aichallenge.org/profile.php?user=10043 2011-12-17T11:58:48 .12 rank and rising 2011-12-17T11:58:54 avdg: Fluxid though he saw signs of life around rank 2000 2011-12-17T11:59:22 i think i stabilized around skill 84.5 2011-12-17T11:59:45 why would someone upload the starter bot? 2011-12-17T11:59:50 well, one of my previous bots dropped from 1500 to 2k at a certain time (not even a few weeks ago) 2011-12-17T11:59:56 itzkow: it tells you to do that 2011-12-17T12:00:06 the tutorial 2011-12-17T12:00:11 oh right 2011-12-17T12:00:17 well 2011-12-17T12:00:31 how did i end up ahead of pguillory? haha 2011-12-17T12:00:32 since there are 6000 starter bots out of 8000 2011-12-17T12:00:38 my last two bots finished with identical mu 2011-12-17T12:00:43 but that's certainly not "true" 2011-12-17T12:00:52 anyway, there were cases of bots worse than starter bots... because they timeouted on gathering food which caused them to loose points andfail 2011-12-17T12:01:12 i wonder how many more times flagcapper is going to resubmit.. sheesh 2011-12-17T12:01:12 maybe pguillory is prepearing something, he hasn't uploaded a new bot in a while 2011-12-17T12:01:23 i haven't either, but i'm not preparing anything 2011-12-17T12:01:49 i'm afraid to touch it at this stage as i have no idea why it's working so well 2011-12-17T12:01:49 at some point pguillory said he has something special which he didn't enabled 2011-12-17T12:01:51 ChrisH was the one trying everything for an advantage 2011-12-17T12:02:04 he has enemy behavior prediction 2011-12-17T12:02:09 rankings seem a bit out of control when lots of players keep resubmitting 2011-12-17T12:02:11 but i think that's as likely to hurt as it is to help 2011-12-17T12:02:29 *** GeorgeSebastian has quit IRC (Ping timeout: 244 seconds) 2011-12-17T12:02:34 i wonder if the finals will be very different than the current rankings ( i think so) 2011-12-17T12:02:45 my new bot has enemy prediction behaviour 2011-12-17T12:02:46 so it takes almost exactly 80 skill to get in the top 100 now 2011-12-17T12:02:57 ok, who bets xathis will be able to get skill 100 before finals? 2011-12-17T12:03:24 itzkow: so does mine, as long as you only include moving (somewhere) and not moving 2011-12-17T12:03:32 with a slight bias against expecting them to suicide 2011-12-17T12:03:39 BenJackson: I just got my ass handed to me by bj_v10 2011-12-17T12:03:44 interesting 2011-12-17T12:03:50 jab_bott: I can go back to v7 2011-12-17T12:04:23 Q is as good as xathis on fluxid but the latter didnt play games recently. 2011-12-17T12:04:29 or I guess 7 is up? 5 and 7 then 2011-12-17T12:04:32 BenJackson: sure - lemmie have a god :P 2011-12-17T12:04:40 who is Q? 2011-12-17T12:04:45 BenJackson: Q == FlagCapper 2011-12-17T12:04:50 besh: ↑ 2011-12-17T12:05:01 aka. A aka. _flag 2011-12-17T12:05:03 fluxid rank is pretty meaningless now 2011-12-17T12:05:10 BenJackson: i know buuuuut 2011-12-17T12:05:12 you have to look at every game to know if anything useful happened 2011-12-17T12:05:30 it's kind of fun watching how he climbs up with mu over 100 2011-12-17T12:05:45 my new bot has won 8 in a row on tcpants, woo woo 2011-12-17T12:06:32 I'm not sure how good is xathis on 1v1 battles 2011-12-17T12:06:37 i wonder if playing same games with different order results with the same skill 2011-12-17T12:06:46 Fluxid: I was wondering the same thing 2011-12-17T12:06:57 when I played 10+ games practically at once due to a matchup glitch 2011-12-17T12:07:04 i guess not, because one can start to improve after some time 2011-12-17T12:07:22 I have to think order matters 2011-12-17T12:12:22 sometimes I lose, but I can't blame my bot 2011-12-17T12:12:43 I get angry on other bots who act wierd 2011-12-17T12:13:06 :/ 2011-12-17T12:13:12 itzkow: I've found it especially rewarding to reliably win games that early on I would have disregarded as unwinnable 2011-12-17T12:13:15 *** GeorgeSebastian has joined #aichallenge 2011-12-17T12:13:49 partly I did that by setting up tests at home whre I gave my old bots extra advantages 2011-12-17T12:13:53 and focused them on my new test bot 2011-12-17T12:14:48 how can you give advantage to a bot? 2011-12-17T12:15:02 lots of ways 2011-12-17T12:15:10 play a map where bots are paired up, give one a weak opponent 2011-12-17T12:15:27 I have a hack in ants.py that lets me tell all bots where the hills of a given target are 2011-12-17T12:15:30 on every turn 2011-12-17T12:15:47 delay the start of one by a few turns 2011-12-17T12:16:08 tune an old bot to be suicidally aggressive and put it next to the test bot 2011-12-17T12:16:18 interesting 2011-12-17T12:16:55 pretty much every iteration of bot I've submitted can soundly beat the previous bot and mostly beat any combination of all prior versions 2011-12-17T12:17:08 !timeleft 2011-12-17T12:17:09 1 day, 11:41:51.113098 until cutoff 2011-12-17T12:17:10 and more than once I didn't think I could make improvements that big 2011-12-17T12:17:46 v12 was #1 on tcpants but apparently v13 can't lose 2011-12-17T12:17:51 muahah 2011-12-17T12:18:15 heh 2011-12-17T12:18:24 *** rebelxt has joined #aichallenge 2011-12-17T12:18:27 I hope the next challenge will be atleast as awesome as this one 2011-12-17T12:18:40 this one sure was awsome 2011-12-17T12:18:53 I haven't been to the previous ones 2011-12-17T12:19:04 wasn't aware of them 2011-12-17T12:19:26 this one is pretty awesome 2011-12-17T12:19:48 I fell like a team manager 2011-12-17T12:20:01 fill* 2011-12-17T12:20:06 feel* 2011-12-17T12:20:06 feel* 2011-12-17T12:20:09 :) 2011-12-17T12:20:43 @seen thestinger 2011-12-17T12:20:43 BenJackson: thestinger was last seen in #aichallenge 12 hours, 55 minutes, and 37 seconds ago: not really sure how to solve it in a better way 2011-12-17T12:20:51 agreed ants was as good or even better than planetwars. wish i had time to give it a go 2011-12-17T12:21:36 planet wars was 1v1? 2011-12-17T12:22:06 yes 2011-12-17T12:22:14 itzkow: yep 2011-12-17T12:22:20 tron and planet wars could have been more than just 1v1, I wonder why it hasn't 2011-12-17T12:22:30 Ants is the first multiplayer game 2011-12-17T12:22:51 multiplayer is fun to watch 2011-12-17T12:23:07 but I think we're burning a lot of cpu on games that don't tell us much about the relative strengths of the players 2011-12-17T12:23:09 yeah, but 1v1 ants is quite diffrent 2011-12-17T12:23:30 I suppose the uncertainty does mean we need combat 2011-12-17T12:23:30 itzkow: concern was that multi-player games in planetwars would result in deadlocks when noone wants to attack first and lose 2011-12-17T12:23:34 in 1v1 you could trade 2011-12-17T12:23:41 Maybe because organizers only started to do this contest and were scary but multiplayer complications. 2011-12-17T12:23:57 I see 2011-12-17T12:24:31 planetwars 2v2 team games would be awesome though 2011-12-17T12:24:45 I knew the game tron as a multiplayer so I was surprised to see they only did 1v1 2011-12-17T12:25:00 Murashka: nice one! http://aichallenge.org/visualizer.php?game=259582&user=3186 2011-12-17T12:25:50 I'm still used to thinking of ifdef as pretty strong 2011-12-17T12:26:11 slipped to 44, wow 2011-12-17T12:29:36 UncleVasya it's a luck 2011-12-17T12:30:59 that one went bad for me :) 2011-12-17T12:31:34 FlagCapper is only 16, that's impressive 2011-12-17T12:32:10 years old I mean, not rank 2011-12-17T12:32:54 I guess programmers just keeps getting younger 2011-12-17T12:32:56 yep he just killed me :) 2011-12-17T12:34:44 kids today... 2011-12-17T12:36:02 meh, I wonder what would happen if I had better parents :p 2011-12-17T12:36:40 * avdg is 21 2011-12-17T12:37:19 thats a strong statement.. =) 2011-12-17T12:37:44 wonder if there is anyone older than 60yo participating. hopefully so since i would def be doing it 2011-12-17T12:39:00 *** JorgeB has joined #aichallenge 2011-12-17T12:39:09 I wonder if someone is actually checking the google analystic data 2011-12-17T12:39:13 next aichallenge they should make a voluntary field "age" in the personal information mask lol 2011-12-17T12:42:41 meh, another ant that attacks my hill and then time out 2011-12-17T12:43:20 *** jab_bott has quit IRC (Quit: Page closed) 2011-12-17T12:43:47 I don't know if I have to laugh about it or take it serious 2011-12-17T12:49:26 *** xcvxcvx has quit IRC (Quit: Page closed) 2011-12-17T12:50:47 *** dvladim has quit IRC (Ping timeout: 252 seconds) 2011-12-17T12:51:16 *** sashaSochka has joined #aichallenge 2011-12-17T12:51:48 i'm interested if google suggests work to the best bot creators? 2011-12-17T12:52:55 *** TheLinker2 has joined #aichallenge 2011-12-17T12:54:07 *** TheLinker has quit IRC (Ping timeout: 240 seconds) 2011-12-17T12:57:04 *** Jacob_Strauss has joined #aichallenge 2011-12-17T13:00:17 *** bmh has joined #aichallenge 2011-12-17T13:00:32 janzert: can you add another language? 2011-12-17T13:03:10 *** Israfel has quit IRC (Read error: Operation timed out) 2011-12-17T13:03:51 *** replore has quit IRC (Read error: Connection reset by peer) 2011-12-17T13:04:33 *** replore_ has joined #aichallenge 2011-12-17T13:07:10 *** olexs1 has joined #aichallenge 2011-12-17T13:07:10 *** olexs has quit IRC (Read error: Connection reset by peer) 2011-12-17T13:10:05 *** amstan has joined #aichallenge 2011-12-17T13:10:05 *** ChanServ sets mode: +o amstan 2011-12-17T13:10:16 hiya amstan :-D 2011-12-17T13:10:22 hi 2011-12-17T13:10:52 who's in charge of the server config? 2011-12-17T13:11:03 depends 2011-12-17T13:11:06 what config? 2011-12-17T13:11:14 it's mostly janzert though 2011-12-17T13:11:38 I didn't realize that you were using an archaic haskell compiler 2011-12-17T13:11:53 *** rebelxt has quit IRC (Quit: Page closed) 2011-12-17T13:12:00 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-12-17T13:12:06 that was the only one available at the time for natty, short of compiling it from scratch 2011-12-17T13:12:12 and `haskell-platform` doesn't seem to be installed 2011-12-17T13:12:50 yeah, there were some issues, the one from the repo actually had a broken dependency for a while, but then they downgraded and it worked 2011-12-17T13:12:55 idk why it's not installed 2011-12-17T13:13:40 *** Surya has quit IRC (Ping timeout: 258 seconds) 2011-12-17T13:13:58 oh. I know how to fix this -- I'll just submit a batch script that invoked a binary I compile 2011-12-17T13:14:06 %s/batch/bash 2011-12-17T13:14:08 *** GeorgeSebastian has quit IRC (Ping timeout: 240 seconds) 2011-12-17T13:14:52 you could 2011-12-17T13:15:04 it's cool if my bot is 15M? 2011-12-17T13:15:24 might not be, i think the limit is 10 2011-12-17T13:15:27 but i'm sure we can change it 2011-12-17T13:15:44 try it and see 2011-12-17T13:15:50 if it doesn't work ask janzert 2011-12-17T13:15:51 sashaSochka: Hi. You are one of 2 guys who is responsible for me being kicked out from Ukrain's Top 3 :) 2011-12-17T13:15:52 *** deltree_ has quit IRC (Quit: ChatZilla 0.9.87-rdmsoft [XULRunner 1.9.0.17/2009122204]) 2011-12-17T13:16:56 *** Israfel has joined #aichallenge 2011-12-17T13:17:42 UncleVasya: apparently sochka is 16 :) 2011-12-17T13:18:27 :) 2011-12-17T13:18:38 How does memetrix's combat technique avoid double counting strength? 2011-12-17T13:19:25 amstan, he asked i'm interested if google suggests work to the best bot creators? 2011-12-17T13:19:54 idk, it's up to jeff to put in a good word with the top ppl 2011-12-17T13:20:07 Yes, i asked) I'd like to work there 5-6 years later if it's possible 2011-12-17T13:21:05 sashaSochka: a1k0n got a job there because of this too, so there's good chances 2011-12-17T13:21:06 sashaSochka: Will you submit another version? 2011-12-17T13:21:35 *** Antimony has joined #aichallenge 2011-12-17T13:21:58 And i suppose, someone have lost their jobs (whifes) because of this challenge :D 2011-12-17T13:21:59 ugh, I hope the ai class final doesn't lower my average :x 2011-12-17T13:22:46 Fortunately I had no job or wife when this started. (?) 2011-12-17T13:23:42 I'd like and I know what to improve (dynamical defense) and how to. But my last form in school doesn't give me enough time.. So my answer is no 2011-12-17T13:24:49 technical school? 2011-12-17T13:24:58 no 2011-12-17T13:25:22 simple school but I'm in math form 2011-12-17T13:25:40 11th? 2011-12-17T13:25:48 yes 2011-12-17T13:26:09 what's a form? 2011-12-17T13:26:14 amstan: it was a fun joke... until we figuret out it is true :( 2011-12-17T13:26:19 11th 2011-12-17T13:26:44 I am a loser :( I am 20 years old 2011-12-17T13:26:48 *crying* 2011-12-17T13:27:14 *** GeorgeSebastian has joined #aichallenge 2011-12-17T13:27:32 what is your nickname on the contest server? 2011-12-17T13:30:03 http://aichallenge.org/profile.php?user=53 2011-12-17T13:31:33 you're next to me. My last games weren't succesfull, I moved from 20 to 33, so you will beat me 2011-12-17T13:32:57 hi guys 2011-12-17T13:33:09 hi 2011-12-17T13:33:24 uncle, it's another responsible guy 2011-12-17T13:33:34 GreenTea missing :) 2011-12-17T13:34:36 *** yoden-cloud has joined #aichallenge 2011-12-17T13:35:21 Coll story by a yang GreenTea: http://brunneng.blogspot.com/2011/12/blog-post.html 2011-12-17T13:35:27 *Cool 2011-12-17T13:36:28 *Pojarniki otkrili ogon' vodoy* is my favourite moment. 2011-12-17T13:37:08 *** cyphase has quit IRC (Ping timeout: 240 seconds) 2011-12-17T13:41:33 *** GeorgeSebastian has quit IRC (Quit: popped out) 2011-12-17T13:41:37 greentea has chanses to be on the 2nd place 2011-12-17T13:41:49 chances* 2011-12-17T13:42:35 after his sigma will be less than 1.8 2011-12-17T13:44:10 ...can you have your movement blocked by spawning food? 2011-12-17T13:44:36 yep but very rare 2011-12-17T13:44:42 you can't move on food 2011-12-17T13:44:46 gah 2011-12-17T13:45:10 been trying to figure out why I randomly get can't move errors, and have ants suiciding 2011-12-17T13:45:11 now i see 2011-12-17T13:45:12 Q on fluxid ..looks like he intend to get 2-nd or 1-st place :) 2011-12-17T13:45:13 T_T 2011-12-17T13:45:47 who is Q on the official server? 2011-12-17T13:46:01 it's a good question 2011-12-17T13:46:04 flag 2011-12-17T13:46:30 then 2-nd ;) 2011-12-17T13:46:57 and who is A? 2011-12-17T13:47:07 _flag too 2011-12-17T13:47:30 so many flags 2011-12-17T13:47:36 *** Darhuuk has quit IRC (Ping timeout: 255 seconds) 2011-12-17T13:49:50 yey very good game for me 2011-12-17T13:51:04 sashaSochka: By any chance don't you know Alexander Polozov (Skiminok)? 2011-12-17T13:51:17 no, I don't. who is he? 2011-12-17T13:51:32 *** Darhuuk has joined #aichallenge 2011-12-17T13:52:08 One guy with the life story same to yours: 2011-12-17T13:52:27 *** cyphase has joined #aichallenge 2011-12-17T13:53:01 math class --> olympiads --> KPI --> ACM, topcoder, etc --> Microsoft intership in Silicon valley... 2011-12-17T13:53:05 life story? 2011-12-17T13:53:08 ahhh 2011-12-17T13:53:29 sashaSochka: how old are you? 2011-12-17T13:53:35 16 2011-12-17T13:53:58 yikes, all these 16 year olds 2011-12-17T13:54:02 * antimatroid is 23 2011-12-17T13:54:06 sashaSochka: He has an interesting blog. 2011-12-17T13:54:47 how did you write this "* antimatroid is 23" without your nickname? 2011-12-17T13:54:58 ./me is a silly bugger 2011-12-17T13:55:02 without the . 2011-12-17T13:55:10 * sashaSochka 2011-12-17T13:55:18 * sashaSochka is 16 2011-12-17T13:55:27 thanks 2011-12-17T13:55:28 huh? [13:55] [CTCP] Received unknown CTCP-ACTION request from sashaSochka to Channel #aichallenge. 2011-12-17T13:55:31 np 2011-12-17T13:55:37 lol 2011-12-17T13:55:42 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-17T13:55:48 what does that mean? 2011-12-17T13:55:56 no idea, it could be your client 2011-12-17T13:55:58 if I had a lawn, I'd tell these damn kids to get off of it. 2011-12-17T13:56:08 bmh: lol, why? 2011-12-17T13:56:13 oh.. right 2011-12-17T13:56:14 * bmh is an old man at 25 2011-12-17T13:56:24 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-17T13:56:30 _flag: You can talk with sashaSochka. So many common between you... 2011-12-17T13:56:31 _flag: what's going on with your bot? 2011-12-17T13:57:08 <_flag> amstan: It lost some games, so I tried to make some improvements and resubmitted 2011-12-17T13:57:19 <_flag> UncleVasya: How so? 2011-12-17T13:57:26 he's 16 too 2011-12-17T13:57:40 ah, _flag is flagCapper 2011-12-17T13:57:46 i didn't even write hello world till i was 18 2011-12-17T13:57:56 _flag: your combat is horrifyingly good. 2011-12-17T13:57:58 it looks like my defence method doesn't work with multiple hills 2011-12-17T13:58:07 <_flag> bmh: Thanks 2011-12-17T13:58:27 _flag: feel free to tell me if you see testbot do something retarded :) 2011-12-17T13:58:29 _flag: my favorite is this one game with you and a1k0n where you kill all of his ants rather than killing his kill. 2011-12-17T13:58:35 other than time out 2011-12-17T13:58:41 er, hill. 2011-12-17T13:58:45 <_flag> antimatroid: You don't control explore for food enough 2011-12-17T13:58:56 <_flag> Uhh.. that was badly worded 2011-12-17T13:59:02 UncleVasya, i just have loosen 2 games, so difference between our skills is minimal.. just 0.05 2011-12-17T13:59:08 <_flag> See the games you play against me and turn on your vision versus mine 2011-12-17T13:59:11 in just battles, or exploration generally? 2011-12-17T13:59:18 ahh yeah okay 2011-12-17T13:59:24 i'll try and explore a bit more 2011-12-17T14:00:22 *** Accoun has quit IRC () 2011-12-17T14:00:38 i've been trying to stop the ants from dancing around too much on their way out to targets 2011-12-17T14:01:03 do people explicitly search towards their hills? 2011-12-17T14:01:10 i have tried to avoid that 2011-12-17T14:01:41 eh....1000 turn limit...again http://aichallenge.org/visualizer.php?game=260434&user=3186 2011-12-17T14:02:37 _flag just got robbed by the 750 turn limit on tcpants http://tcpants.com/replay.3042 2011-12-17T14:02:51 <_flag> antimatroid: The 750 turn limit tends to do that 2011-12-17T14:03:04 <_flag> The 1000 turn limit does as well though, so it's not just there 2011-12-17T14:04:28 yeah, a turn limit will always have that problem 2011-12-17T14:04:38 but finite resources and all 2011-12-17T14:07:44 don't try to avoid the real reason for turning limit on: it is fun :) 2011-12-17T14:07:52 My new version is performing better than my old version in tcp server. But in local, my old version always beats the new version. now which one to pick 2011-12-17T14:08:27 praveen: do you have 50 cent? 2011-12-17T14:08:45 UncleVasya: nice game 2011-12-17T14:08:56 yup 2011-12-17T14:09:31 with bucash? I was lucky that strong opponents have killed each other. 2011-12-17T14:10:08 *** nha has quit IRC (Ping timeout: 252 seconds) 2011-12-17T14:14:58 _flag: made an alteration to exploration 2011-12-17T14:15:05 shall see how that does :) 2011-12-17T14:15:46 <_flag> antimatroid: Okay, I'll be watching :) 2011-12-17T14:16:21 *** Accoun has joined #aichallenge 2011-12-17T14:17:56 *** capcap has joined #aichallenge 2011-12-17T14:18:15 http://aichallenge.org/visualizer.php?game=260418&user=4513 2011-12-17T14:19:26 _flag: so what are you planning to do at uni? 2011-12-17T14:20:50 wow 2011-12-17T14:21:03 <_flag> antimatroid: computer science, but I'm not ruling out something physics related 2011-12-17T14:21:13 not maths? :( 2011-12-17T14:21:34 what university are you going to enter? 2011-12-17T14:21:35 maths/comp sci is a good combo 2011-12-17T14:21:39 <_flag> antimatroid: Both those things involve math, but probably not pure math 2011-12-17T14:21:45 <_flag> So there will be math in it 2011-12-17T14:21:52 i went crazy in undergrad and did cs, economics and pure maths majors 2011-12-17T14:22:12 <_flag> sashaSochka: I don't even know; probably something nearby and nothing too special 2011-12-17T14:22:19 <_flag> So for me that would be Western or Waterloo 2011-12-17T14:22:24 _flag: not waterloo? it has a good rep 2011-12-17T14:22:31 well, i had heard of it before this 2011-12-17T14:22:32 Waterloo = something special 2011-12-17T14:22:48 *** capcap has quit IRC (Client Quit) 2011-12-17T14:22:50 ahh 2011-12-17T14:23:07 _flag: consider doing a maths major too :P 2011-12-17T14:23:12 algebra is awesome 2011-12-17T14:23:25 *** nha has joined #aichallenge 2011-12-17T14:23:29 and statistics. 2011-12-17T14:23:38 I took too much biology. I should have taken more math. 2011-12-17T14:23:51 i should have taken more stats, but meh 2011-12-17T14:24:06 i did econometrics through economics and 1 stats unit through maths 2011-12-17T14:24:24 I'm reading donald knut now.. and I understand my math is to bad for understanding everything well.. 2011-12-17T14:24:26 nah, I'm afraid I'll lose my stipend because of math 2011-12-17T14:24:40 too* 2011-12-17T14:26:40 sashaSochka: Have you read CLRS or The Algorithm Design Manual? 2011-12-17T14:26:52 no 2011-12-17T14:27:24 I haven't even heared about it yet 2011-12-17T14:27:47 CLRS is the MIT algo book yeah? 2011-12-17T14:27:57 http://en.wikipedia.org/wiki/Introduction_to_Algorithms 2011-12-17T14:27:59 antimatroid: yeah. It's what I used as an undergrad. Blew my mind. 2011-12-17T14:28:03 i have a copy of that across my room somewhere 2011-12-17T14:28:33 my uni would never use a book like that :( 2011-12-17T14:28:48 the maths school rocks, everything else is pretty crap 2011-12-17T14:29:06 like TSAP ot CLRS? 2011-12-17T14:29:08 oh cool, the contest has a netherland's sponsor now 2011-12-17T14:29:37 taop*(the art of programming) 2011-12-17T14:29:47 haven't looked at TAOP, but they wouldn't touch CLRS 2011-12-17T14:30:23 they did have an advanced algo's course for people who hd'd algo's which was pretty decent 2011-12-17T14:30:27 but still used minimal maths 2011-12-17T14:31:48 well, I'm leaving you. Need to do my school math) 2011-12-17T14:32:16 I hope janzert comes around. I'm not going to try implementing combat until I know if my bot can compile. 2011-12-17T14:32:49 gl 2011-12-17T14:33:39 bmh: do it anyway :P 2011-12-17T14:34:20 *** foRei has joined #aichallenge 2011-12-17T14:34:26 *** jimi_hendrix has joined #aichallenge 2011-12-17T14:34:40 antimatroid: on it. I'm not going to implement Memetix's method. Monte carlo here we come. 2011-12-17T14:34:46 hi, my bot is failing test cases when i upload it. is there a way to check exactly what it is failing? 2011-12-17T14:36:01 check an isolated ant, I'm sure that is the problem 2011-12-17T14:36:07 jimi_hendrix: use the test_bot.sh thing in tools 2011-12-17T14:36:18 isolated ant? 2011-12-17T14:36:18 especially if your bot fails on the first turn 2011-12-17T14:36:29 yes 2011-12-17T14:36:34 what do you mean? 2011-12-17T14:36:41 antimatroid, alright 2011-12-17T14:37:03 try test_bot script 2011-12-17T14:37:22 and you will see on test_map isolated square with an ant 2011-12-17T14:38:15 coords 1:18 2011-12-17T14:39:44 *** Jacob_Strauss has quit IRC (Quit: Page closed) 2011-12-17T14:53:51 *** treeform has joined #aichallenge 2011-12-17T14:55:21 *** ccc has joined #aichallenge 2011-12-17T14:57:07 *** mleise has quit IRC (Ping timeout: 240 seconds) 2011-12-17T15:03:40 Fluxid: wow, that game. 2011-12-17T15:04:01 i know, right? :D 2011-12-17T15:14:36 *** replore_ has quit IRC (Remote host closed the connection) 2011-12-17T15:16:59 *** ccc has quit IRC (Quit: Computer has gone to sleep.) 2011-12-17T15:17:50 *** TheLinker2 has quit IRC (Remote host closed the connection) 2011-12-17T15:20:17 bmh: you were wondering something? 2011-12-17T15:24:39 *** praveen has quit IRC (Quit: Page closed) 2011-12-17T15:27:03 *** goffrie has quit IRC (Read error: Connection reset by peer) 2011-12-17T15:27:41 *** goffrie has joined #aichallenge 2011-12-17T15:29:25 janzert: given the old haskell compiler you're using and the lack of the standard haskell libs, I can't build my bot on prod :) 2011-12-17T15:29:39 :( 2011-12-17T15:29:47 can I give you a binary? 2011-12-17T15:30:30 I should have tested earlier, but I didn't want to place load on the servers and then forgot.. 2011-12-17T15:30:30 the library package was broken at the time (don't know if it's fixed now or not, but can't add it at this point) 2011-12-17T15:30:32 hmm 2011-12-17T15:31:45 You could use apt pinning and source ghc from the latest version of ubuntu 2011-12-17T15:31:50 are you familiar with the contest compiler system? 2011-12-17T15:31:58 not in the slightest. 2011-12-17T15:33:03 ok, first the hint if you were; take a look at how scheme language support is implemented 2011-12-17T15:33:11 and since you're not, a more exlicit hint 2011-12-17T15:34:06 but first double checking that I have it right ;) 2011-12-17T15:34:53 ok, if a MyBot.ss file is found it will chmod *.ss files and then run ./MyBot 2011-12-17T15:35:30 so are you suggesting that I write a scheme wrapper and hand you a binary blob? ('cause that ain't fun) 2011-12-17T15:35:34 so basically for whatever reason and whoever implemented "Scheme" support is really just binary support 2011-12-17T15:35:42 don't even need any scheme 2011-12-17T15:35:43 ha. 2011-12-17T15:35:53 that's broken :) 2011-12-17T15:36:00 I'll give it a shot later today 2011-12-17T15:36:12 just make a MyBot.ss file with anything in it and include a binary MyBot to run 2011-12-17T15:37:15 but you'll be listed as using Scheme of course :/ 2011-12-17T15:38:00 derp 2011-12-17T15:38:10 oh and probably need to upload as a tar file with correct permissions set on the binary already 2011-12-17T15:38:40 and of course I've not tested this, but I'm pretty sure it'll work :) 2011-12-17T15:39:25 error while loading shared libraries: libgmp.so.10: cannot open shared object file: No such file or directory turn 0 bot 0 crashed 2011-12-17T15:40:00 looks like it works 2011-12-17T15:40:03 (Why the hell it's trying to use GMP is beyond my comprehension. I guess the haskell runtime uses it) 2011-12-17T15:40:18 can haskell statically compile? 2011-12-17T15:40:30 that'd be the easiest solution probably 2011-12-17T15:40:44 bmh: try using -static 2011-12-17T15:40:55 I thought that was the default though 2011-12-17T15:40:57 janzert: the binary will be *huge* 2011-12-17T15:41:21 so strip and then upx compress it :P 2011-12-17T15:41:32 thestinger: what? 2011-12-17T15:41:41 strip -s MassiveBinary 2011-12-17T15:41:45 gets rid of tons of crap 2011-12-17T15:41:51 then 'upx -9 MassiveBinary' 2011-12-17T15:41:51 neat 2011-12-17T15:42:06 upx packs it and then it unpacks at runtime 2011-12-17T15:42:36 *** BlindSniper has joined #aichallenge 2011-12-17T15:42:57 *** BlindSniper is now known as BlindSniper5 2011-12-17T15:45:14 Fluxid: if only I had finished my next gen defense, my bot could have survived an extra turn :P 2011-12-17T15:45:46 (Will it blend?) 2011-12-17T15:46:20 thestinger: here's how I'm compiling ghc --make -O3 -outputdir bin -fforce-recomp -rtsopts -static -Wall -Werror -o bin/Skynet MyBot || exit 1 2011-12-17T15:46:25 it's still choking on libgmp 2011-12-17T15:47:20 *** treeform has quit IRC (Remote host closed the connection) 2011-12-17T15:47:39 -static -optl-static 2011-12-17T15:47:42 try thaty 2011-12-17T15:47:43 that* 2011-12-17T15:48:41 you can use 'ldd MyBot' to see if it's actually a static executable 2011-12-17T15:49:21 or `file' ofc 2011-12-17T15:52:47 damn, fails 2011-12-17T15:53:52 it doesn't compile? 2011-12-17T15:54:07 it doesn't link 2011-12-17T15:54:33 https://gist.github.com/1491350 2011-12-17T15:55:39 add -lpthread to the end 2011-12-17T15:55:47 *** jimi_hendrix has quit IRC (Ping timeout: 240 seconds) 2011-12-17T15:56:25 not sure about that first error 2011-12-17T15:57:45 actually you can probably just add '-Wl,--as-needed' 2011-12-17T15:58:02 that will just stop it from linking to anything that's not being used 2011-12-17T15:58:11 definitely worth adding anyway since it will make it smaller :P 2011-12-17T15:58:42 *** itzkow has quit IRC (Quit: Page closed) 2011-12-17T15:58:50 sure about those compile flags? 2011-12-17T15:58:58 ghc chokes on them as unrecognised 2011-12-17T15:59:05 oh, well they're gcc flags 2011-12-17T15:59:33 I guess you have to pass them to it somehow 2011-12-17T16:01:31 oh 2011-12-17T16:01:38 anything prefixed by -optl is passed to gcc 2011-12-17T16:01:59 -static -optl-static -optl-pthread 2011-12-17T16:02:01 so try that 2011-12-17T16:02:33 thestinger: I hope you get to use haskell for your day job 2011-12-17T16:02:34 *** Kingpin13 has quit IRC (Quit: quit) 2011-12-17T16:02:51 xD 2011-12-17T16:03:15 this is more a gcc thing anyway 2011-12-17T16:03:29 sweet. compiled with a single warning 2011-12-17T16:03:48 anyway, check the size 2011-12-17T16:03:55 then 'strip -s MyBot' and check again 2011-12-17T16:03:58 1.1m compressed 2011-12-17T16:04:12 oh, that's not bad then 2011-12-17T16:04:13 I'm stripping in my build script. It's 3.7M stripped 2011-12-17T16:04:40 are you using upx? 2011-12-17T16:04:55 anyway, it's small enough 2011-12-17T16:05:04 I'm not using upx 2011-12-17T16:06:21 sweet. It's running, though it explodes because I forgot to turn off logging which tries to open files :) 2011-12-17T16:06:45 anyway, upx could get you a 30-60% compression ratio if it ends up being too big 2011-12-17T16:06:53 and it still loads pretty much just as fast 2011-12-17T16:06:54 *** Dlayne has joined #aichallenge 2011-12-17T16:08:38 my relationship with C really blows in the wind. Sometimes I love the bare metal. Other times I just want to write WebGL and have things work 2011-12-17T16:10:10 so you're going to be on top of the Scheme rankings? xD 2011-12-17T16:10:50 guess I shouldn't bother implementing combat. 2011-12-17T16:11:22 *** cyphase has quit IRC (Read error: Connection reset by peer) 2011-12-17T16:11:28 *** dvladim has joined #aichallenge 2011-12-17T16:11:58 bmh: you can get something basic done really fast 2011-12-17T16:12:06 *** Kurnevsky has left #aichallenge 2011-12-17T16:12:31 just implement a function for simulating what would happen with ants at certain positions 2011-12-17T16:13:01 then it's trivial to deal with unmoving enemies, and you can sort of deal with moving ones just by generating at least one possible set of moves 2011-12-17T16:13:01 thestinger: I will, once I get it to run in prod. 2011-12-17T16:13:32 whois parasprites 2011-12-17T16:13:36 It's dying somewhere in the bowels of GHC IO -- mkTextEncoding 2011-12-17T16:13:42 Fluxid: Antimony I think? 2011-12-17T16:14:15 yep 2011-12-17T16:14:16 i remember someone telling he's parasprites, maybe that was Antimony, right... anyway 2011-12-17T16:14:32 http://aichallenge.org/visualizer.php?game=261084&user=757 Antimony why did you hold back when defending your hill? 2011-12-17T16:14:50 janzert: could you try pulling in ghc 7.x and the standard haskell lib? 2011-12-17T16:14:52 at turn 813 your ants go back two steps 2011-12-17T16:15:16 Can anyone tell me what the Mu tells us? Say my skill is 43.1869, Mu 48.0786 and Sigma 1.6306, after 47 games? 2011-12-17T16:15:25 I can't figure it out. 2011-12-17T16:15:38 Vaenom: Mu is the standard deviation 2011-12-17T16:15:46 er 2011-12-17T16:15:47 mean 2011-12-17T16:15:51 isn't sigma sd? 2011-12-17T16:15:54 I though Sigma was 2011-12-17T16:15:58 mu is predicted skill 2011-12-17T16:16:07 sigma is how unsure system is about your skill 2011-12-17T16:16:10 bmh: sorry, much too high a chance of breaking one or more of the other haskell submissions 2011-12-17T16:16:14 skill is mu-3*sigma 2011-12-17T16:16:24 Sigma is sd 2011-12-17T16:16:37 janzert: what are the specs on your box? 32-bit? 64-bit? 2011-12-17T16:16:46 64bit 2011-12-17T16:16:50 ubuntu 11.04 2011-12-17T16:17:30 http://aichallenge.org/visualizer.php?game=261087&user=757 this map is bad.. three enemies killed, little effort is needed 2011-12-17T16:17:35 at the beginning 2011-12-17T16:17:36 ok, so Mu tells you the current estimates of your future skills when it gets stable? 2011-12-17T16:17:38 I have no idea why that happened 2011-12-17T16:17:40 harsh 2011-12-17T16:18:09 Vaenom: pretty much... but that would take million years 2011-12-17T16:18:49 Oh, I have plenty of time 2011-12-17T16:19:01 !timeleft 2011-12-17T16:19:01 1 day, 7:39:58.866745 until cutoff 2011-12-17T16:19:04 uh oh 2011-12-17T16:19:06 ;) 2011-12-17T16:19:31 thanks Fluxid 2011-12-17T16:19:49 !source 2011-12-17T16:19:56 aw :( 2011-12-17T16:20:13 sadly contestbot doesn't know how to grab quotes 2011-12-17T16:20:16 actually not quite, because your sigma gets bumped up a little bit each game 2011-12-17T16:20:26 at least that's the TrueSkill default 2011-12-17T16:20:36 so sigma does not go down to 0 in the limit 2011-12-17T16:20:50 janzert: sweet -- my girlfriend's laptop is running ubuntu 11.04 i686 2011-12-17T16:20:52 maybe this will work 2011-12-17T16:21:01 7:39:58 ? 2011-12-17T16:21:24 aaa EST 2011-12-17T16:21:35 bmh: oh, I think glibc isn 2011-12-17T16:21:37 nha: sigma bumped up? 2011-12-17T16:21:42 isn't backwards compatible with itself* 2011-12-17T16:21:46 not sure if that would be a problem 2011-12-17T16:21:56 sigma shoul get smaller every game 2011-12-17T16:22:09 Fluxid: the default of TrueSkill is to increase your sigma by a small amount for each game, *before* the usual sigma update 2011-12-17T16:22:11 sure 2011-12-17T16:22:21 that's the "normal update" 2011-12-17T16:22:29 Fluxid: not always 2011-12-17T16:22:42 is there currently any kind of wisdom about the Fluxid.pl ranks and the aichallenge.org ranks? what would a Fluxid rank 200 bought you on Aichallenge? Any comparison any of you can make? 2011-12-17T16:22:46 and it's possible that the organizers took the sigma increase out, because it doesn't really apply here 2011-12-17T16:22:55 so if xathis lost to 7 starter bots, his sigma would go up? :) 2011-12-17T16:23:20 thestinger: yes, though that's unrelated to what I meant 2011-12-17T16:24:06 Fluxid: TrueSkill comes from evaluating the skill of humans, and the skill of humans can change over time. To account for that, TrueSkill increase the sigma of every player by some constant before the real update step 2011-12-17T16:24:20 I should probably close irc so I stop using it as an excuse from getting the things needed for finals implemented :P 2011-12-17T16:24:21 of course it would make sense not to do that for the AI challenge 2011-12-17T16:24:39 and I haven't checked the TrueSkill implementation used here 2011-12-17T16:25:10 *** dvladim has quit IRC (Ping timeout: 268 seconds) 2011-12-17T16:25:18 bmh: does your bot run locally? 2011-12-17T16:25:42 thestinger: it runs brilliantly locally. It runs on OSX Lion and Debian 5.0 64bit 2011-12-17T16:26:04 do bots get killed if they try to spawn a thread? 2011-12-17T16:26:17 *** bgs100 has joined #aichallenge 2011-12-17T16:26:17 haskell might be using threads for I/O or something like that... 2011-12-17T16:26:29 nha: okay 2011-12-17T16:26:51 Vaenom: it's not really comparable, because when one bots are offline other climb up 2011-12-17T16:26:58 it's blowing up on mkTextEncoding which appears to stem from probability issues with iconv 2011-12-17T16:27:16 there is not matchmaking and you can't play with offline bots to check if you're better 2011-12-17T16:27:16 %s/probability/portability 2011-12-17T16:27:17 *** Dlayne has left #aichallenge 2011-12-17T16:27:33 *** moises has quit IRC (Remote host closed the connection) 2011-12-17T16:28:15 back in a while 2011-12-17T16:28:58 *** Jacob_Strauss has joined #aichallenge 2011-12-17T16:29:52 *** GeorgeJ has joined #aichallenge 2011-12-17T16:29:52 *** GeorgeJ has joined #aichallenge 2011-12-17T16:30:08 right, I think I've implemented everything I wanted... now it's only cleaning up and optimizing 2011-12-17T16:32:13 *** Palmik has quit IRC (Remote host closed the connection) 2011-12-17T16:32:15 Fluxid: I get it, but is there any kind of tendencies? is it usually harder to get some rank-range on your server because of the high-level bots duplicates and the no-matchmaking? or is there really no tendencies at all? 2011-12-17T16:32:44 well, fluxid's server is packed with bots that don't play anymore 2011-12-17T16:32:53 so their skill is set in stone 2011-12-17T16:33:38 really, the tcp servers should increment everyone's (the whole population) sigma after each game 2011-12-17T16:37:12 *** ztfw has joined #aichallenge 2011-12-17T16:41:54 *** mj41 has joined #aichallenge 2011-12-17T16:43:10 what about just excluding bots from the ranking, that did not play in the last 24v hs, but keep their skill for the moment, they show up again ? 2011-12-17T16:43:11 am I the only one already a little sad that the contest will soon be over? I really wish we would have an official permanent X times per year contest web site. 2011-12-17T16:44:29 *** ztfw has quit IRC (Ping timeout: 240 seconds) 2011-12-17T16:46:09 Vaenom: that probably wouldn't attract as much interest 2011-12-17T16:46:43 unless it's for a game that is popular for some other reason (e.g. StarCraft) 2011-12-17T16:46:57 on the contrary since the contest would get known with times 2011-12-17T16:47:22 yes, but if it's always the same thing, the novelty wears off 2011-12-17T16:47:31 that's why the AI challenge was run with different games each time 2011-12-17T16:47:58 *** Antimony has quit IRC (Ping timeout: 245 seconds) 2011-12-17T16:48:09 so in that sense, your wish has already come to pass: there will certainly be another iteration of the contest 2011-12-17T16:48:13 just not with ants 2011-12-17T16:48:30 13,000 contestants is far from being a significant portion of all the people who could possibly enter the contest. 2011-12-17T16:49:04 but how many of those 13,000 are really serious about it anyway? 2011-12-17T16:49:11 how many new software engineers are formed _every years_ in universities worldwide? 2011-12-17T16:49:16 sure 2011-12-17T16:49:23 that's why you have a regular contest, but with *different* games 2011-12-17T16:49:50 exactly! so, it's not even 13,000.. 2011-12-17T16:49:59 so that it's equally interesting for both oldtimers and newcomers 2011-12-17T16:50:49 indeed, but why not accumulate them rather than letting them rot like Tron or Planet War? 2011-12-17T16:51:47 well, since all the infrastructure is open source, you could always run an "eternal ants challenge" yourself if you feel so inclined 2011-12-17T16:52:24 Ants require lots of CPU-time tho 2011-12-17T16:52:53 you always have the latest new "main" contest and you have other permanent tournaments where you can go at any time and learn.. 2011-12-17T16:53:32 looks like it's a good thing I resubmitting this morning 2011-12-17T16:53:46 my updaed version on tcpants has a mu 4 higher 2011-12-17T16:54:23 nha: right :) not sure I'd do that, but I'm thinking of helping with the next challenges. 2011-12-17T16:54:58 *** McLeopold has joined #aichallenge 2011-12-17T16:55:01 pairofdice: indeed 2011-12-17T16:55:31 BenJackson: tcpants = Fluxid.pl ? 2011-12-17T16:58:04 different servers 2011-12-17T16:58:17 tcpants.com 2011-12-17T17:00:00 *** Gnome has joined #aichallenge 2011-12-17T17:00:23 Hey I got a question about the system. does food() return all food on the map? Or all food visible to my ants? 2011-12-17T17:00:36 should return only visible food 2011-12-17T17:00:39 *** flowenol has quit IRC (Ping timeout: 255 seconds) 2011-12-17T17:00:40 what starter package? 2011-12-17T17:00:43 python 2011-12-17T17:01:06 yep, just visible 2011-12-17T17:01:20 Some starter packages return all foods? ;) 2011-12-17T17:01:42 okay thanks. Second question. If I use the distance(loc1, loc2) does the system take into account the fact that the map wraps? 2011-12-17T17:01:48 yes 2011-12-17T17:01:51 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-12-17T17:02:06 you can verify that by looking for the modulo operator 2011-12-17T17:02:20 awesome. I appreciate the help. 2011-12-17T17:03:04 *** UncleVasya has joined #aichallenge 2011-12-17T17:03:11 *** rwsads has joined #aichallenge 2011-12-17T17:03:18 aichallenge: janzert epsilon * r864c1c0 / (sql/2_generate_matchup.sql sql/opponent.sql): Implement pairing cutoff - http://git.io/NcgyRQ 2011-12-17T17:03:25 *** rwsads has quit IRC (Client Quit) 2011-12-17T17:04:37 pairofdice: who runs it? 2011-12-17T17:04:39 *** Migi32 has joined #aichallenge 2011-12-17T17:05:13 Vaenom: tcpants is a different server 2011-12-17T17:05:14 same idea 2011-12-17T17:05:45 I gave up on fluxid because there are so many timeouts that it's hard to tell if you won or lost a game 2011-12-17T17:05:47 ok, but who runs it? 2011-12-17T17:05:53 and I'm not really analyzing micro stuff like combat anymore 2011-12-17T17:06:00 I'd have to check the forum post 2011-12-17T17:06:22 I finally had an idea of how to make an acrylic "trophy" for ants 2011-12-17T17:06:24 janzert: I'm working on the countdown timer 2011-12-17T17:06:39 and I want to include the people who provided tcp servers in the list of people who get free ones 2011-12-17T17:06:45 I'll try it. What's the turntime at tcpants.com? 2011-12-17T17:06:54 McLeopold: great, thank you 2011-12-17T17:06:55 500ms 2011-12-17T17:07:04 only strange thing about tcpants is a 750 turn limit 2011-12-17T17:07:08 janzert: next, I'll do the script for generating final maps 2011-12-17T17:07:11 but then fluxid is on 1500 instead of 1000 so 2011-12-17T17:07:45 sounds good 2011-12-17T17:07:45 http://tcpants.com/replay.3217 this map is srsly awful 2011-12-17T17:07:59 thestinger: hey I was watching games on tcpants 2011-12-17T17:08:19 I noticed several cases where you got into "deadlocks" 2011-12-17T17:08:26 yeah 2011-12-17T17:08:29 and I later busted you out by altering the local area 2011-12-17T17:08:35 causing you to spring to life 2011-12-17T17:08:36 I wish fluxid.pl would set the same settings as the contest. That'd give a better idea of "real world scenarios" lol 2011-12-17T17:08:36 xD 2011-12-17T17:09:43 Vaenom: only turn limit is different 2011-12-17T17:09:57 *** rajanaresh has joined #aichallenge 2011-12-17T17:10:03 oh nice! scheduler error in my favor! 2011-12-17T17:10:20 http://aichallenge.org/visualizer.php?game=260934&user=3473 2011-12-17T17:10:24 turntime is different too because of network latency, but to client it sends 500 2011-12-17T17:10:31 considering my opponents there are all ranked about 1000 the competition really has heated up 2011-12-17T17:10:38 Fluxid: ah! and why the 1500 turn limit? wouldn't there be less timeouts and such if you would reduce it to 1000? 2011-12-17T17:10:41 I remember with 1000th place was still basically starter-bot-esque 2011-12-17T17:10:55 *** ace1010 has joined #aichallenge 2011-12-17T17:11:31 *** rajanaresh has quit IRC (Client Quit) 2011-12-17T17:11:32 Vaenom: dunno, i don't remember why... 2011-12-17T17:12:04 probably because the actual server settings are a turn limit of 1500, just every maps limit is set to 1000 2011-12-17T17:12:06 1000? What's your latest "high" on Aichallenge? 2011-12-17T17:12:11 BenJackson: yeah, the bots around rank 2k start to have reasonable logic 2011-12-17T17:12:41 *** Migi32 has quit IRC (Quit: bye) 2011-12-17T17:12:53 Vaenom: my overall high is 5th but that wasn't a recent sub 2011-12-17T17:13:02 currently I'm probably >5 <15 2011-12-17T17:13:55 *** cyphase has joined #aichallenge 2011-12-17T17:14:18 Fluxid: well, that's a simple and quick idea to improve things on your server I guess. Or maybe I just don't realize the full implications of changing it to 1000 2011-12-17T17:14:37 not only didi I get 4 quick games, the latest 3 were against players in the 100s so I'm up to 289 after only 7 games 2011-12-17T17:15:27 *** Boscop has left #aichallenge 2011-12-17T17:16:14 Vaenom: i don't want to kill server because client is buggy and after disconnection it eats memory and crashes and there are 25 games, 51 players online 2011-12-17T17:16:31 you were only seed in the first one btw, so no scheduler error involved 2011-12-17T17:16:43 BenJackson: but if you're 5 < p < 15, then who are those higher in the top 1000 on tcpants? please don't tell me there's hundreds of secret elite bots not yet on aichallenge :P 2011-12-17T17:16:57 janzert: then I feel sorry for the seed ;-) 2011-12-17T17:17:18 Vaenom: I just resubmitted so I played a bunch of 1000s on the way up 2011-12-17T17:17:28 but those guys at 1000 are better than I remember from previous climbs 2011-12-17T17:17:32 Fluxid: indeed not a good idea then, yes 2011-12-17T17:17:48 I wish I'd known abuot the tcpclient/server problems earlier 2011-12-17T17:17:55 you did get rather lucky with your first pairing but you were actually highest mu in the second two 2011-12-17T17:17:57 I would have fixed them 2011-12-17T17:18:08 so if anything the others were too low of rank 2011-12-17T17:18:27 so, my bot is great on huge maps 2011-12-17T17:18:31 but it sucks on the little tiny ones 2011-12-17T17:18:31 janzert: I am assuming that games show up in the list in the order started 2011-12-17T17:18:37 but that mu is recomputed in the order finished? 2011-12-17T17:18:38 nope 2011-12-17T17:18:50 i have huge disgust to this code, so doing anything in it makes me sad and suicidal 2011-12-17T17:19:00 games are in order they finish 2011-12-17T17:19:07 thank god it's the last day 2011-12-17T17:19:14 Fluxid: your bot or tcpserver? 2011-12-17T17:19:20 tcpserver 2011-12-17T17:19:25 i'll try to make tcp server from scratch my way 2011-12-17T17:19:29 but you're right the mu doesn't necessarily reflect what it was at pairing time 2011-12-17T17:19:30 next challenge* 2011-12-17T17:19:37 Fluxid: stackless python! 2011-12-17T17:19:53 janzert: I assume they all started with the same (lowest) mu because of the tight spacing of the games 2011-12-17T17:19:57 Ben: yes, I remember when getting in the 400s on aichallenge was only a matter of collecting food in mobs and "snaking" to hills.. 1 hour and you'd have one such bot. Now, not so much. 2011-12-17T17:20:04 I guess you have the matchups stored somewhere? 2011-12-17T17:20:09 or is that lost when games start? 2011-12-17T17:20:37 yeah, you're probably right 2011-12-17T17:21:05 yeah, combat used to get you into the top 25 2011-12-17T17:21:06 so first and last games you were paired with a higher mu 2011-12-17T17:21:15 middle game a little lower 2011-12-17T17:21:17 now it seems like everyone has combat :P 2011-12-17T17:21:20 thestinger: i don't see any profits in stackless here 2011-12-17T17:21:27 my bot makes a strategic error in the blue direction but recovers: http://aichallenge.org/visualizer.php?game=261667&user=3473 2011-12-17T17:21:29 but i thought of twisted 2011-12-17T17:21:38 Fluxid: well, I like the Go/Erlang way of doing servers 2011-12-17T17:21:40 I really like how much more reliable my bot is now 2011-12-17T17:21:41 matchups information is deleted right after the game result is stored 2011-12-17T17:22:13 that's not going to sound good in front of the judge ;-) 2011-12-17T17:22:28 thestinger: i mean, stackles is a bit outdated, isn't it? 2011-12-17T17:22:36 Fluxid: no, you can use pypy 2011-12-17T17:22:42 it has the stackless features 2011-12-17T17:22:53 microthreads/channels are awesome 2011-12-17T17:22:59 well, yeah, continuations 2011-12-17T17:23:28 i really like how you can serialize continuations and resume them later 2011-12-17T17:23:45 in different proces 2011-12-17T17:23:53 dimkadimon almost seems to predict my hill in that game 2011-12-17T17:24:06 he doesn't see it before his big sortie 2011-12-17T17:24:25 he actually doesn't see it until that sortie is nearly dead 2011-12-17T17:25:08 *** NotABug has quit IRC (Ping timeout: 258 seconds) 2011-12-17T17:26:24 Fluxid: tasklets/channels in stackless python get you same thing as goroutines/channels in Go 2011-12-17T17:26:39 it's really nice compared to traditional threading 2011-12-17T17:26:57 you don't have to worry about locks, etc - you share data by communicating over channels, not by sharing memory 2011-12-17T17:27:09 yeah yeah 2011-12-17T17:27:10 i know 2011-12-17T17:27:34 *** cyphase has quit IRC (Ping timeout: 248 seconds) 2011-12-17T17:27:37 thestinger: I think haskell's model is smarter. Immutability is parallelism. 2011-12-17T17:28:45 bmh: well, Haskell has all the nice types of concurrency :) 2011-12-17T17:29:43 thestinger: is there some kind of deepseq monad that prohibits laziness? 2011-12-17T17:29:48 it has lightweight threads and... http://hackage.haskell.org/packages/archive/base/4.3.0.0/doc/html/Control-Concurrent-Chan.html 2011-12-17T17:29:56 *** Gnome has quit IRC (Ping timeout: 240 seconds) 2011-12-17T17:30:42 thestinger: I think this static linking is a doomed proposition. 2011-12-17T17:31:13 aichallenge: janzert epsilon * rf64c742 / website/server_stats.php : Have server_stats display the pairing cutoff - http://git.io/jzxLiw 2011-12-17T17:31:40 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-17T17:32:36 *** nha has quit IRC (Ping timeout: 240 seconds) 2011-12-17T17:33:10 *** Antimony has joined #aichallenge 2011-12-17T17:34:16 !timeleft 2011-12-17T17:34:16 1 day, 6:24:43.862944 until cutoff 2011-12-17T17:34:33 *** Migi32 has joined #aichallenge 2011-12-17T17:34:42 *** UncleVasya has quit IRC () 2011-12-17T17:34:43 McLeopold: see, you can get !timeleft from irc for the web page ;-) 2011-12-17T17:37:06 bmh: http://biohaskell.org/Installation 2011-12-17T17:37:18 Downloading static binaries <- go to that section 2011-12-17T17:37:28 looks like a known problem due to glibc devs being stupid 2011-12-17T17:37:42 I blame the glibc maintainer. 2011-12-17T17:37:55 (drepper) 2011-12-17T17:38:02 yeah, apparently he has killed off proper static linking 2011-12-17T17:38:07 maybe you can use a different C library... 2011-12-17T17:38:23 oh 2011-12-17T17:38:27 so you need the same glibc version 2011-12-17T17:38:33 as the server 2011-12-17T17:40:10 *** cyphase has joined #aichallenge 2011-12-17T17:40:44 bmh: maybe try with '-optl-liconv' 2011-12-17T17:42:19 thestinger: I should just submit a starter bot and win the scheme category 2011-12-17T17:43:57 oh right, I still don't have a category to win 2011-12-17T17:43:59 no dice on -optl-iconv 2011-12-17T17:44:17 hmm... 2011-12-17T17:45:28 *** yoden-cloud has quit IRC (Quit: Page closed) 2011-12-17T17:45:45 bmh: install ubuntu 11.04server in a vm and compile there? 2011-12-17T17:46:06 ^ that sounds like a good idea 2011-12-17T17:46:24 submit the entire compiler 2011-12-17T17:46:30 heh 2011-12-17T17:46:30 compile on your first turn 2011-12-17T17:46:40 someone tried to do that in planetwars 2011-12-17T17:46:52 the upload size limit was quickly added 2011-12-17T17:47:01 something like a 300MB upload 2011-12-17T17:47:03 scipy lets you do inline C :) 2011-12-17T17:47:07 *** Kingpin13 has joined #aichallenge 2011-12-17T17:47:08 my submission is just 'telnet giant_server 2081' and I run my bot there on a 64-way cluster 2011-12-17T17:47:11 there's no way it would compile in time though 2011-12-17T17:47:53 it may have also been why a compile time limit was added 2011-12-17T17:48:03 I don't remember for sure what submission broke that 2011-12-17T17:48:42 janzert: what would you suggest for a vm provider? I don't have vmware installed 2011-12-17T17:48:47 Anyone know if "Q" is flag_'s latest or just something he's messing with? 2011-12-17T17:49:19 I'm surprised I'm beating him on tcpants (by rank, we're trading games) 2011-12-17T17:49:29 EC2 free tier? :-D 2011-12-17T17:49:32 bmh: is ok 2011-12-17T17:49:33 hmm, virtualbox is what I use but under windows as the host OS 2011-12-17T17:49:46 somehow I accidentally "virtualbox" in my statement 2011-12-17T17:50:13 bmh: just get someone who doesn't know Haskell to build it for you 2011-12-17T17:50:23 that's a lot of people 2011-12-17T17:50:37 BenJackson: volunteering? :) 2011-12-17T17:50:56 I don't have the contest env 2011-12-17T17:51:08 I test on freebsd and gcc 4.2.1! 2011-12-17T17:51:16 if I upgraded freebsd it'd be clang! 2011-12-17T17:51:48 * BenJackson rebuilds with g++45 to see if any other warnings pop 2011-12-17T17:52:37 BenJackson: clang 3.0 is awesome though 2011-12-17T17:52:54 g++45 warned about some () in an &&|| 2011-12-17T17:52:59 luckily it was already ok 2011-12-17T17:53:16 clang -Weverything 2011-12-17T17:53:18 :) 2011-12-17T17:53:48 2:53PM up 508 days, 9:03, 17 users, load averages: 0.19, 0.23, 0.24 2011-12-17T17:53:53 this always happens 2011-12-17T17:54:01 lol? is that your desktop? 2011-12-17T17:54:03 my server will have 2+ year uptime and then I'm waaay out of date 2011-12-17T17:54:07 that's my server 2011-12-17T17:54:49 thestinger: what do you do and how old are you? 2011-12-17T17:54:59 Fluxid: 19, unemployed :) 2011-12-17T17:55:00 woot, 17:54:32 up 901 days, 1:27, 1 user, 2011-12-17T17:55:07 BenJackson: Have you ever thought of selling your domain and retiring to some place nice? 2011-12-17T17:55:13 duh 2011-12-17T17:55:13 haha 2011-12-17T17:55:16 people ask me all the time 2011-12-17T17:55:22 * Fluxid needs to learn mooooore 2011-12-17T17:55:59 bmh: I've never had a credible offer more than about $25k 2011-12-17T17:56:08 most people want me to start the bidding, but I won't 2011-12-17T17:56:20 BenJackson: I don't get out of bed for less than $100k 2011-12-17T17:56:22 *** sashaSochka has quit IRC (Quit: Page closed) 2011-12-17T17:56:44 can we assume that all land are reachable from each other? 2011-12-17T17:56:49 BenJackson: you could always just switch to ben.xxx 2011-12-17T17:57:04 I'd need to develop new content 2011-12-17T17:57:20 Antimony: I thikn that's in the spec but even the test map (that they run when you re-upload) has a trapped hill 2011-12-17T17:57:48 Antimony: yes, "Maps must not contain islands" 2011-12-17T17:58:26 m %%%%%%%%%%%%%%%%%%0 2011-12-17T17:58:26 m .................%% 2011-12-17T17:58:26 m %..%.0.%....*...%*0 2011-12-17T17:58:26 m ....%%%..*.......%% 2011-12-17T17:58:26 m *..%.%.%........... 2011-12-17T17:58:33 spammer! 2011-12-17T17:58:33 that's a segment of submission_test/test.map 2011-12-17T17:58:39 janzert: what version of glibc are you running? 2011-12-17T17:58:41 so don't crash when you see that trapped hill 2011-12-17T17:59:13 bmh: bleh, I don't know :) 2011-12-17T17:59:35 janzert: sure it's 11.04? Ubuntu is offering me 10.04 and 11.10 2011-12-17T17:59:45 10.04 is the LTS release 2011-12-17T17:59:53 yep, 11.04 they just offer the LTS and most recent by default 2011-12-17T18:00:05 you mean 10.04? :P 2011-12-17T18:00:05 but you can still get it 2011-12-17T18:00:11 nope 2011-12-17T18:00:13 * BenJackson laughs at "11.04" 2011-12-17T18:00:27 oh, that's not just an accidental combining of both options? 2011-12-17T18:00:33 got it 2011-12-17T18:00:53 http://releases.ubuntu.com/natty/ 2011-12-17T18:01:42 aichallenge: McLeopold epsilon * r838c764 / (website/css/typo.css website/header.php): add countdown timer - http://git.io/kILD8A 2011-12-17T18:02:56 looks like glibc 2.13-0ubuntu13 2011-12-17T18:03:16 at least according to ldd --version 2011-12-17T18:03:33 okay everyone, check the countdown timer and let's make sure different timezones agree 2011-12-17T18:03:46 yay 2011-12-17T18:03:52 janzert: is the server utc? 2011-12-17T18:03:53 the site knows my timezone? 2011-12-17T18:04:07 except it looks like their is text underneath getting cutoff 2011-12-17T18:04:11 McLeopold: yes it is 2011-12-17T18:04:18 McLeopold: neat! but there's something in white below it 2011-12-17T18:04:20 which is cut off 2011-12-17T18:04:21 okay, I'll need to set it back 5 hours 2011-12-17T18:04:24 so the current countdown time is wrong 2011-12-17T18:04:36 what browser? 2011-12-17T18:04:58 oh "until submission deadline" if I cut'n'paste 2011-12-17T18:05:02 firefox 9 2011-12-17T18:05:03 firefox 9 (?) 2011-12-17T18:05:28 crap on a stick 2011-12-17T18:05:46 *** yoden-cloud has joined #aichallenge 2011-12-17T18:05:48 and what does it do when it gets to zero? 2011-12-17T18:05:54 I certainly hope virtualbox works 2011-12-17T18:06:20 janzert: amstan: McLeopold: you may remember I offered to lase acrylic plaques/trophies for the winners 2011-12-17T18:06:27 I was sketching design ideas 2011-12-17T18:06:38 very cool looking in any case, much nicer than what I did 2011-12-17T18:06:43 and I was wondering if you had a preference about including sponsor information on them 2011-12-17T18:06:51 and what sort of emphasis 2011-12-17T18:07:30 hmm, we would need to get permission from the sponsors before putting them on 2011-12-17T18:07:31 1day, 0 hours, 52 Minutes, 15 Seconds 2011-12-17T18:07:55 other than that I don't have a strong opinion either way 2011-12-17T18:07:55 janzert: I was more concerned about the reverse: accidentally spurning supporters 2011-12-17T18:08:21 yes, it would be nice but we definitely still have to ask first :( 2011-12-17T18:08:46 so far I haven't really figured out how to work text into my design anyway :) 2011-12-17T18:08:49 especially and understandably Google is very concerned about where and how there name is used 2011-12-17T18:09:02 ok :) 2011-12-17T18:11:15 *** ebloo has joined #aichallenge 2011-12-17T18:12:39 aichallenge: McLeopold epsilon * rbabc5da / (website/css/typo.css website/header.php): update countdown timer - http://git.io/oYtzDg 2011-12-17T18:13:02 oops, we just hit the deadline :( 2011-12-17T18:13:13 but the text is looking correct now here 2011-12-17T18:14:19 we just gained 5h but the text is still hiding 2011-12-17T18:14:20 *** Antimony has quit IRC (Ping timeout: 268 seconds) 2011-12-17T18:14:32 hmm yes, now the time is correct but the text isn't again 2011-12-17T18:15:05 stupid text 2011-12-17T18:15:37 what if instead of the
the text is wrapped in a

2011-12-17T18:15:52 don't really know if it would help 2011-12-17T18:16:30 I wonder how to stop the javascript from running so I can play with the markup in firebug 2011-12-17T18:16:47 does everyone show 1day, 5 hours? 2011-12-17T18:17:06 1 day 5 hours 43 minutes here 2011-12-17T18:17:14 oh god, why does that render so horribly? 2011-12-17T18:17:32 http://tcpants.com/replay.3287?turn=70 2011-12-17T18:17:44 play from there and check out light yellow's ant on the right 2011-12-17T18:17:49 leads me on a merry chase and gets a bunch of food 2011-12-17T18:17:52 tcpants makes me snicker -- I read it as "TC pants" 2011-12-17T18:17:59 *** Jak_o_Shadows has joined #aichallenge 2011-12-17T18:18:01 never have I forced an ant to gorge itself so much 2011-12-17T18:18:01 every browser but firefox is fine 2011-12-17T18:18:59 BenJackson: have you tried leading enemy ants to other enemy ants' hills? 2011-12-17T18:19:35 generally that's bad 2011-12-17T18:19:46 I already see games where hills are "sniped" 2011-12-17T18:19:51 I get there first and cause the defense to face me 2011-12-17T18:20:00 the hill is lost to some other bot from the other side 2011-12-17T18:20:37 *** Antimony has joined #aichallenge 2011-12-17T18:22:30 *** TheLinker has joined #aichallenge 2011-12-17T18:23:22 Distract :) 2011-12-17T18:26:03 *** bugnuts has quit IRC (Read error: Connection reset by peer) 2011-12-17T18:26:26 *** bugnuts has joined #aichallenge 2011-12-17T18:28:09 aichallenge: McLeopold epsilon * r68e0437 / (website/css/typo.css website/header.php): update countdown timer style - http://git.io/kjMqwA 2011-12-17T18:28:40 okay, firefox monkeys, is that better? 2011-12-17T18:28:46 yes 2011-12-17T18:29:27 yes, but I think the minutes/seconds should get a leading 0 when <10 2011-12-17T18:29:38 Looks the same on chrome and FF 2011-12-17T18:29:40 blarg, broke opera 2011-12-17T18:29:57 Browsers.... 2011-12-17T18:30:18 oh, no, I didn't 2011-12-17T18:31:28 *** sigh has joined #aichallenge 2011-12-17T18:33:09 *** HaraKiri has joined #aichallenge 2011-12-17T18:33:42 *** kire has quit IRC (Remote host closed the connection) 2011-12-17T18:35:12 *** besh has quit IRC (Quit: Page closed) 2011-12-17T18:39:40 hm, watching Q play that version is more willing to suicide 1v1 than mine 2011-12-17T18:39:59 perhaps that's more expedient than sending multiple ants to try to get rid of the loner 2011-12-17T18:43:39 when to trade is the most complicated part of this, i think... 2011-12-17T18:43:48 especially because you can't know the number of players 2011-12-17T18:45:15 *** ace1010 has quit IRC (Ping timeout: 258 seconds) 2011-12-17T18:45:39 *** bboona has joined #aichallenge 2011-12-17T18:45:51 *** pairofdice has quit IRC (Quit: When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl.) 2011-12-17T18:47:02 *** bboona has quit IRC (Client Quit) 2011-12-17T18:47:11 if you don't move an ant... 2011-12-17T18:47:18 will the simulation not remind you that it's there the next turn? 2011-12-17T18:49:20 *** ebloo has quit IRC (Quit: Page closed) 2011-12-17T18:53:03 thestinger, janzert: I'm dead in the water. GHC7x relies on libc > 2.7 2011-12-17T18:54:34 ouch 2011-12-17T18:54:59 tcp server (bhickey) just gave me an invalid (I think) map 2011-12-17T18:55:09 bmh: ouch :( 2011-12-17T18:55:11 was ~ 31k elements in my array 2011-12-17T18:55:28 dragonfyre13: too damn bad :) 2011-12-17T18:55:28 thought max was 25k 2011-12-17T18:55:41 bmh: I fixed my code... 2011-12-17T18:55:42 :D 2011-12-17T18:55:53 I was using a 16 bit int for a couple things, now using 32 2011-12-17T18:56:28 it overflowed (signed) with the high values based on map size, so I totally just sat by my hill and bounced back and forth 2011-12-17T18:56:32 hehe 2011-12-17T18:56:34 cool 2011-12-17T18:57:10 but thanks for that, wouldn't have thought about overflowing that unless I saw that map, and I found a few other places I could have overflowed. 2011-12-17T18:57:15 thestinger: I think this falls under "fatal error" 2011-12-17T18:57:37 I actually considered using haskell but the version is so ancient 2011-12-17T18:57:39 *** mj41 has quit IRC (Ping timeout: 255 seconds) 2011-12-17T18:57:46 so I learned C++ instead 2011-12-17T18:57:51 *** Surya has joined #aichallenge 2011-12-17T18:57:57 bmh: you should make a bot in tcl :) 2011-12-17T18:58:05 thestinger: in a day? 2011-12-17T18:58:08 yeah! 2011-12-17T18:58:27 oh god, everyone needs to stop making everything in TCL. That language needs to die, and then be taken out and shot. Twice. 2011-12-17T18:58:29 I have better things to do... like go to the gym 2011-12-17T18:58:38 or a colonoscopy 2011-12-17T18:58:56 yeah, tcl is awful 2011-12-17T18:58:59 dragonfyre13: they really aren't that bad 2011-12-17T18:59:00 I have to deal with TCL at work, and let's just say it's a very special brand of hell. 2011-12-17T18:59:07 thestinger: what are you doing for combat? 2011-12-17T18:59:17 bmh: way better than TCL. 2011-12-17T18:59:32 first, I partition the map into battles 2011-12-17T18:59:46 bmh: how about a colonoscopy where they keep you awake, no numbing, and they use a garden hose? Still better than TCL. 2011-12-17T18:59:49 well, this is also after I've come up with all my moves basically 2011-12-17T19:00:21 I do 'come up with good/neutral moves' -> 'do moves that won't affect combat' -> combat -> do rest of moves 2011-12-17T19:00:25 thestinger: you doing some memetix magic? 2011-12-17T19:00:27 so I have a hash table with orders 2011-12-17T19:00:41 well, I originally started with something like a payoff matrix 2011-12-17T19:00:48 and then threw in some of the memetix magic type stuff 2011-12-17T19:00:52 it' 2011-12-17T19:00:55 it's alright* 2011-12-17T19:01:29 I use the memetix stuff to pick out an offensive set of moves for each battle 2011-12-17T19:01:41 I either do offense or defense (retreating/positioning always happen) 2011-12-17T19:02:19 I pick offense/defense based on whichever maximizes the worst case (I test against unmoving enemies, and a few offensive sets of moves - the payoff matrix part) 2011-12-17T19:02:36 amstan: ping 2011-12-17T19:02:42 then I have more code to position ants well, to retreat, etc. 2011-12-17T19:02:46 it's not at all ideal 2011-12-17T19:02:48 McLeopold: pong 2011-12-17T19:03:01 amstan: which script generates your maps? 2011-12-17T19:03:10 mapgen should do it 2011-12-17T19:03:12 I started on combat way too late and I didn't feel like rewriting my bot to make it better 2011-12-17T19:03:16 it has command line options 2011-12-17T19:03:27 most of the commands output something for testing purposes though 2011-12-17T19:03:43 will I get a good variety if I just run it? 2011-12-17T19:05:12 amstan: will these be called caves? 2011-12-17T19:05:18 yes 2011-12-17T19:05:36 you might have to play around with the translation_factor if they look weird 2011-12-17T19:05:43 but other than that, it should be fine 2011-12-17T19:06:02 bmh: the big flaw in my bot is that it still piles up ants in stalemates 2011-12-17T19:06:39 thestinger: I guess I should just open up my source. The bot works, but it doesn't work :) 2011-12-17T19:06:42 I have a fix for it (just changing weighting more) but it doesn't actually improve my bot, and there's no time left to test stuff until it works :P 2011-12-17T19:06:57 *** Murashka has quit IRC (Quit: Page closed) 2011-12-17T19:07:17 bmh: wait anyway :) 2011-12-17T19:08:29 *** sigh has quit IRC (Remote host closed the connection) 2011-12-17T19:10:06 I've already shared all my secrets/ideas anyway 2011-12-17T19:10:28 my code isn't anything special - it's just ugly C++ 2011-12-17T19:10:35 *** bluegaspode has joined #aichallenge 2011-12-17T19:10:35 *** Gnome has joined #aichallenge 2011-12-17T19:10:39 yeah, but at this point code could still be submitted 2011-12-17T19:10:51 just on the off chance someone got bmh's to run on the server somehow 2011-12-17T19:11:12 I'm trying to run my bot on my machine, to make sure it's working okay. But I can't seem to get the environment setup. 2011-12-17T19:11:39 when I run the ./tutorial.sh I get an invalid syntax error on line 241, which is a print statement in playgame.py 2011-12-17T19:15:32 *** sigh has joined #aichallenge 2011-12-17T19:18:03 antimatroid: how many sets of moves do you make per battle now? 2011-12-17T19:18:08 for the enemy 2011-12-17T19:18:35 mine actually works best with just 1 set of moves 2011-12-17T19:18:47 makes my combat more aggressive/reckless 2011-12-17T19:18:58 thestinger: i am atm doing 4 from them 2011-12-17T19:19:02 i'll try out less after 2011-12-17T19:19:09 i'm doing 2 lots of offensive, and 2 lots of neutral 2011-12-17T19:19:14 bbl, gotta go out for a bit 2011-12-17T19:21:47 alright, the towel is thrown in 2011-12-17T19:21:48 *** bluegaspode has quit IRC (Quit: Page closed) 2011-12-17T19:22:04 at least I've managed to serve up 17k games 2011-12-17T19:25:08 mostly to me 2011-12-17T19:26:39 yeah... 2011-12-17T19:26:49 overeager git ;) 2011-12-17T19:27:34 *** TheLinker has quit IRC (Quit: Bye) 2011-12-17T19:28:02 *** TheLinker has joined #aichallenge 2011-12-17T19:28:23 *** besh has joined #aichallenge 2011-12-17T19:28:26 well if I was redoing the tcp server I'd make an option for people to flag their bots "as needed" 2011-12-17T19:28:32 basically like "deactivateD" on the server 2011-12-17T19:28:33 *** sigh has quit IRC (Remote host closed the connection) 2011-12-17T19:29:14 BenJackson: run it next year 2011-12-17T19:37:41 *** k333es has joined #aichallenge 2011-12-17T19:38:04 *** HaraKiri has quit IRC () 2011-12-17T19:41:33 *** Jacob_Strauss has quit IRC (Quit: Page closed) 2011-12-17T19:47:26 Gnome: try python2.7 2011-12-17T19:47:46 you're probably using python 3.x 2011-12-17T19:47:48 Fluxid yeah thanks. I thought I had python2.7 but I forgot my harddrive crashed last month and everything needed to be updated. 2011-12-17T19:49:45 now it's crashing because class Ants() does not have attribute run() 2011-12-17T19:50:01 when MyBot.py tries to do Ants.run(MyBot()) at the bottom. 2011-12-17T19:54:12 *** Jak_o_Shadows has quit IRC (Read error: Connection reset by peer) 2011-12-17T19:56:04 BenJackson: well the tcp server could have "house bots" and let you schedule your own games with them 2011-12-17T19:59:09 oh, _flag reuploaded again 2011-12-17T19:59:32 xathis needs some competition at the top :P 2011-12-17T20:07:14 *** xathis has joined #aichallenge 2011-12-17T20:08:48 jeez my bot suicides.. probably due to some last minute change. 2011-12-17T20:10:51 yeah, I'm not making any changes or merging anything :P 2011-12-17T20:11:01 I'm just aiming for top 50 2011-12-17T20:11:09 so I don't want to break what I have already 2011-12-17T20:11:45 anyone doing iterative memetix? 2011-12-17T20:12:02 iterative? 2011-12-17T20:13:08 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-17T20:13:21 yoden-cloud: run memetix, now look at where each ant can step -- run memetix again 2011-12-17T20:13:29 *** amstan has joined #aichallenge 2011-12-17T20:13:29 *** ChanServ sets mode: +o amstan 2011-12-17T20:13:52 what, doing more than one turn look ahead? 2011-12-17T20:14:18 yoden-cloud: no -- so you generate the influence graph and now you know where ants can't go (because they would die) 2011-12-17T20:14:23 so you regenerate the influence graph 2011-12-17T20:14:42 sounds similar to what i'm trying 2011-12-17T20:14:44 you should be able to iterate this to build up a probabilitiy distribution of the safety of moves 2011-12-17T20:15:05 hmm, maybe not 2011-12-17T20:15:14 probability distributions too complicated for me~~ 2011-12-17T20:16:37 yoden-cloud: don't sell yourself short. All you need to do is take the number of iterations and the number of times that square was safe 2011-12-17T20:16:43 thestinger: It was leading by 40 ants but lost all of them suciding at the same spot. Found the bug though 2011-12-17T20:16:53 i don't think it's that simple though 2011-12-17T20:17:02 because not all configurations are equal probability 2011-12-17T20:17:44 sure 2011-12-17T20:19:02 enemy move prediciton is hard... if you're wrong, you tend to always lose ants :( 2011-12-17T20:19:28 *** k333es has quit IRC (Quit: Page closed) 2011-12-17T20:20:47 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-17T20:21:20 yoden-cloud: I just assume they will come towards me atm 2011-12-17T20:21:38 before, I did ~9 other sets of moves for them 2011-12-17T20:21:59 but this makes my ants able to push around the cautious bots 2011-12-17T20:22:55 well not really "towards me" but "into places they can attack me" 2011-12-17T20:23:05 <_flag> who runs tcpants? 2011-12-17T20:24:12 right, my current movement assumption is basically "enemies won't go where they will for sure die" 2011-12-17T20:24:23 *** Gnome has quit IRC (Quit: Gnome) 2011-12-17T20:25:11 yoden-cloud: yeah, before I made an aggressive offensive move, a cautious offensive move, etc. 2011-12-17T20:25:58 I couldn't really get it working well 2011-12-17T20:25:59 and weighted them equally? 2011-12-17T20:26:21 well, then I would maximize the worst case 2011-12-17T20:26:48 I still do that, but only against 1 set of defensive moves and 1 set of offensive moves + the memetix combat stuff 2011-12-17T20:26:52 *** Scryer has joined #aichallenge 2011-12-17T20:29:35 tcpants is really slow :\ 2011-12-17T20:32:28 *** Redgis has quit IRC (Ping timeout: 268 seconds) 2011-12-17T20:33:15 thestinger: run on my server 2011-12-17T20:33:41 bmh: what's the url for that? 2011-12-17T20:36:16 oh, bhickey.net - my bot is running on there now :) 2011-12-17T20:37:16 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-17T20:38:00 *** Antimony has joined #aichallenge 2011-12-17T20:41:13 *** treeform has joined #aichallenge 2011-12-17T20:42:52 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-17T20:46:56 *** Gnome has joined #aichallenge 2011-12-17T20:49:31 bleh 2011-12-17T20:49:38 lack of breadth first search coming back to bite me~~ 2011-12-17T20:54:37 the submission deadline display is very broken in Chrome 2011-12-17T20:54:37 *** treeform has quit IRC (Remote host closed the connection) 2011-12-17T20:55:01 *** treeform has joined #aichallenge 2011-12-17T20:55:58 *** bmh has quit IRC (Ping timeout: 258 seconds) 2011-12-17T20:56:16 *** Antimony has joined #aichallenge 2011-12-17T21:00:46 olexs1:try reloading the page 2011-12-17T21:01:40 besh: hm, must have been some cached css or something. it displayed n plaintext in between blocks, breaking layout, now it's nice and clean in the header 2011-12-17T21:05:00 *** olexs1 has quit IRC (Quit: Leaving.) 2011-12-17T21:06:54 *** amstan has joined #aichallenge 2011-12-17T21:06:54 *** ChanServ sets mode: +o amstan 2011-12-17T21:19:01 Ug, how am I supposed to get any useful data from tcp when I timeout every game? 2011-12-17T21:19:23 Antimony: which server are you using? 2011-12-17T21:19:43 tcpants 2011-12-17T21:20:08 well, stop timing out then :P 2011-12-17T21:20:17 * Zannick is helpful 2011-12-17T21:22:45 *** foRei has quit IRC (Quit: Bye) 2011-12-17T21:34:24 Is it normal to timeout when running your bot locally? Is there some trick to getting around this or does it mean my code sucks? python. 2011-12-17T21:37:31 It's not normal. Your bot might not be aware of the time limit 2011-12-17T21:44:13 ok 2011-12-17T21:45:38 *** Paradoxial has joined #aichallenge 2011-12-17T22:11:58 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-17T22:17:44 GreenTea seems pretyy strong. 2011-12-17T22:23:58 *** hutch has joined #aichallenge 2011-12-17T22:33:20 *** Antimony has quit IRC (Ping timeout: 268 seconds) 2011-12-17T22:36:59 *** Surya has quit IRC (Quit: Page closed) 2011-12-17T22:42:47 *** Antimony has joined #aichallenge 2011-12-17T22:44:07 *** hutch has quit IRC (Quit: Page closed) 2011-12-17T22:48:50 *** besh has quit IRC (Quit: Page closed) 2011-12-17T22:50:51 *** ChrisH_ has joined #aichallenge 2011-12-17T22:59:01 *** treeform_ has joined #aichallenge 2011-12-17T23:02:16 *** treeform has quit IRC (Ping timeout: 244 seconds) 2011-12-17T23:04:22 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-17T23:08:23 hi _flag ready for finals? 2011-12-17T23:16:21 I'm not 2011-12-17T23:17:25 I'm surprsied the channel isn't more active now 2011-12-17T23:17:31 *** Migi32 has quit IRC (Remote host closed the connection) 2011-12-17T23:18:11 Everyone is busy doing some last minute changes perhaps. At least i do. :) 2011-12-17T23:22:47 <_flag> ChrisH_: I certainly hope so 2011-12-17T23:22:47 i'm done changing for today 2011-12-17T23:22:48 I'm just trying to get my bot finished. :| 2011-12-17T23:22:52 going to let it run over night 2011-12-17T23:22:57 and try to fix up any obvious problems 2011-12-17T23:23:22 <_flag> ChrisH_: Unfortunately I have trouble getting games with top players on tcp, so I have no idea where I stand in relation to everyone else 2011-12-17T23:23:29 http://www.youtube.com/watch?v=yvCxqv2bOc8 2011-12-17T23:24:18 I'm pretty sure the version I submitted a couple hours ago is my final version. There is one thing I added recently that I may remove if it isn't proving itself tomorrow, but otherwise I'm done. 2011-12-17T23:24:32 I'm still trying to debug my bot 2011-12-17T23:24:35 _flag: I'm running my bot on tcpants tonight if you want to try there 2011-12-17T23:24:40 it'd be so much more fun if the main server had more power 2011-12-17T23:24:52 by the way, I think your bot misestimates spawning time Chrish 2011-12-17T23:25:04 these are algorithms... i selfishly feel they should be ranked in minutes, not days~~ 2011-12-17T23:25:05 it thinks ants will spawn 1 turn too eaerly 2011-12-17T23:25:09 Antimony: How so? 2011-12-17T23:25:22 just look at the replay you posted 2011-12-17T23:25:23 <_flag> ChrisH_: Okay, I'll play some games there then :) 2011-12-17T23:25:47 <_flag> ChrisH_: Ombedor1, right? 2011-12-17T23:25:52 Antimony: are you talking about early in the game or later? 2011-12-17T23:25:56 _flag: yes 2011-12-17T23:26:20 early in the game 2011-12-17T23:26:59 Antimony: The spawn control is disabled early in the game to allow expansion. Otherwise the initial ants would never step off the hills to get some food. 2011-12-17T23:27:35 no I mean your ants step off the hill one turn too early 2011-12-17T23:27:40 it looks like they're expecting a spawn 2011-12-17T23:27:48 or they step onto the hill one turn too early 2011-12-17T23:28:05 *** avdg has quit IRC (Quit: Leaving.) 2011-12-17T23:28:15 remember, there's a 1 turn delay between collecting food and spawning 2011-12-17T23:28:46 Antimony: i've got the replay up, can you show me a specific hill and turn # where this happens? 2011-12-17T23:29:16 do you have a link handy? 2011-12-17T23:29:24 http://ants.fluxid.pl/replay.35482 2011-12-17T23:30:06 The first time is turns 8-10 2011-12-17T23:30:56 The feature isn't enabled yet on turn 8. 2011-12-17T23:31:42 Turn 15 is the first time an ant steps off a hill after the feature is on. 2011-12-17T23:32:17 In that case it is because the ant in question is going for some food, which will replace itself on hill, so that is ok. 2011-12-17T23:32:46 oh, I thought you turned on the feature at turn 8. I guess that was coincidence 2011-12-17T23:32:54 anyway it happens later too 2011-12-17T23:33:00 On turn 33 an ant steps off without food in the hive because it is particpating in a combat. 2011-12-17T23:33:00 *** mleise has joined #aichallenge 2011-12-17T23:33:13 yeah 2011-12-17T23:33:20 I think so 2011-12-17T23:34:03 I did discover a bug in my food collection tracking during development of this feature that I had to fix to get it working right in some cases. 2011-12-17T23:34:39 *** mleise has quit IRC (Client Quit) 2011-12-17T23:35:06 The bug was that if one of my ants moved next to a food (so I eat it), but a new food spawns in the same place on that same turn by bot used to think it had not eaten the first food, but then a new ant would spawn unexpectedly and I would think I had -1 food in the hive. 2011-12-17T23:35:38 But that's fixed now, so I'm pretty sure it's working as intended. 2011-12-17T23:35:50 Antimony: thanks for checking up on me though. :) 2011-12-17T23:36:20 _flag: are you Q on tcpants? 2011-12-17T23:36:31 <_flag> ChrisH_: yes 2011-12-17T23:37:19 ok, we just had a couple games, the first I timed out (not sure why that happens on the TCP servers, but the second is a legit game. you won both 2011-12-17T23:37:39 It happens again on turn 66 2011-12-17T23:38:08 <_flag> ChrisH_: Yeah, I saw both, you had me the first one :) 2011-12-17T23:38:24 I'm just starting to watch them. 2011-12-17T23:39:23 The second one had you me and BenJackson all in a row on a translational symetry, tough section of the board for BJ. 2011-12-17T23:39:46 ug, I've got a nasty bug in my bot that I can't figure out 2011-12-17T23:40:19 <_flag> Yeah, one of the tough things about multiplayer games is that not all positions are created equal 2011-12-17T23:41:35 especially on the tcp servers where the matches involve a wider range of bot skill 2011-12-17T23:42:21 apparently, my bot thinks enemy ants are jesus 2011-12-17T23:42:31 It expects them to walk on water 2011-12-17T23:44:10 _flag: the game where I timed out was a really good test of my newest code feature, and it requires a top bot to test against, so thanks for that 2011-12-17T23:45:15 <_flag> ChrisH_: Is it pushing through battle disadvantages when at a positional advantage? I had a tricky time getting something like that to work 2011-12-17T23:45:18 wow, what a stupid bug. I had an incorrect else if for enemy movement 2011-12-17T23:46:08 _flag: I wouldn't describe it like that, but that can be the behavior. 2011-12-17T23:46:13 *** Scryer has quit IRC (Quit: Page closed) 2011-12-17T23:46:16 so do you usually play on fluxid, tcpants, or what? 2011-12-17T23:46:42 <_flag> I usually play on fluxid, but the game quality has really gone down there recently, so I'm on tcpants now 2011-12-17T23:47:02 *** xathis has quit IRC (Quit: Page closed) 2011-12-17T23:47:31 _flag: it is a bit of a hack and it could end up really biting me in the rear, but I think it could work, and I'll know tomorrow after my latest submission has gotten back up the ladder on the main servers and I'll decide to keep it or take it out. 2011-12-17T23:47:50 <_flag> ChrisH_: I see 2011-12-17T23:48:32 _flag: but it turns out that this "bit of a hack" was very easy to incorporate into my battle code, so I'm giving it a shot. 2011-12-17T23:49:46 so what happened on turn 66, ChrisH? 2011-12-17T23:50:22 <_flag> ChrisH_: Everything's worth a shot at this point :) 2011-12-17T23:50:28 <_flag> Now here is an interesting way to lose: http://aichallenge.org/visualizer.php?game=264039&user=31 2011-12-17T23:50:29 Antimony: of the game I linked earlyer? 2011-12-17T23:50:33 yeah 2011-12-17T23:51:16 _flag: the dreaded three player game against hyper aggressive bots with no defense, I hate those. 2011-12-17T23:51:30 wow 2011-12-17T23:51:33 Antimony: What do you see on turn 66 that you are asking about? 2011-12-17T23:51:51 You moved off the hill at 41,62 for no apparent reason 2011-12-17T23:52:59 Antimony: I don't see that. On turn 66 all my hills are covered and there is one food in the hive. On turn 67 I step off that hill. 2011-12-17T23:53:26 and a new ant spawns in it's place 2011-12-17T23:53:40 when I veiw the replay, it says you step off on turn 66 2011-12-17T23:54:05 oh wait, I see 2011-12-17T23:54:07 nevermind 2011-12-17T23:54:21 *** Paradoxial has quit IRC (Quit: :) 2011-12-17T23:54:40 ok, I'm glad you see, because if the visualizer isn't consistent I'd be very worried 2011-12-17T23:55:24 _flag: yes, everything is worth a shot as long as you have time to test it and take it out if it doesn't work ;) 2011-12-17T23:55:57 sorry to bother you, I'm just curious about hwo your bot works 2011-12-17T23:56:13 <_flag> ChrisH_: Right, you wouldn't want to sabotage your own entry. That's why I'm scared to make any major modifications, especially without knowing how I preform against xathis and the others 2011-12-17T23:56:30 it looks like you could have sped up your second ants spawn one turn at turn 80 2011-12-17T23:57:36 _flag: absolutely, which is why it was nice to get a game against you on tcpants. :) 2011-12-17T23:58:20 Antimony: no problem, if you really had found a problem I would have owed you big time. 2011-12-17T23:58:56 Antimony: and I'm going to try and do a thorough writeup of my bot next week. 2011-12-17T23:59:13 I planned to implement spawn blocking today, but I spent the day just fixing stupid bugs instead 2011-12-17T23:59:27 Also, http://www.youtube.com/watch?v=yvCxqv2bOc8 2011-12-17T23:59:50 I was thinking about starting a thread on the forums and working on the description a little each night either by posting replies to my initial post or doing edits and adding more.