2011-11-29T00:03:32 *** him has joined #aichallenge 2011-11-29T00:06:31 ChrisH: this next game should be the true test as it will pretty certainly include xathis :) 2011-11-29T00:06:45 *** bmh has joined #aichallenge 2011-11-29T00:06:54 *** delt0r_ has joined #aichallenge 2011-11-29T00:07:26 phew. Just set my bot on fire and stripped it down to a random walk. Time to build it again! 2011-11-29T00:08:09 *** him has quit IRC (Ping timeout: 265 seconds) 2011-11-29T00:09:15 a1k0n: instead of choosing who moves first when doing minimax, why not have a biased coin for each opponent and walk the coin based on performance? 2011-11-29T00:09:30 *** bmh has quit IRC (Client Quit) 2011-11-29T00:10:03 I feel so professional making unit test maps now 2011-11-29T00:10:18 lol, uni test... 2011-11-29T00:10:46 I gave up unit test after I had to rewrite them the 3rd time 2011-11-29T00:11:53 I'm still not sure how minmax is suitable for this 2011-11-29T00:12:07 minmax is based on the idea that your opponent can see your move 2011-11-29T00:12:09 that's not true here 2011-11-29T00:12:58 same thing, you just *assume* that your opponent can figure out how you are going to move 2011-11-29T00:13:52 here's a simple counterexample: 2011-11-29T00:13:58 2 ants vs 1 ant, 3 apart 2011-11-29T00:14:06 2 ants could advance and kill 1 if it didn't move 2011-11-29T00:14:15 but 1 could slide sideways forcing the 2 to lose one as well 2011-11-29T00:14:23 happens all the time in-game 2011-11-29T00:14:29 ya 2011-11-29T00:14:36 now if the 2 can see the 1 move it can guarantee it is safe 2011-11-29T00:14:40 i deliberately do that in fact 2011-11-29T00:14:44 it can kill if it has the opportunity, or just move 2011-11-29T00:14:58 does anybody want to see my pixel-drawings of combat situations that I used for studying the mechanics? 2011-11-29T00:15:02 similarly if the 1 can see the 2 move it has a lot more options 2011-11-29T00:15:12 sf17k: yes 2011-11-29T00:15:27 I filled up a lot of graph paper with those :) 2011-11-29T00:15:31 use at your own risk http://i.imgur.com/4d6c6.png 2011-11-29T00:15:46 color indicates ant status, players are not differentiated by color except with purple 2011-11-29T00:15:52 if you can prune all the possibilities, you can figure out the best result, which is possible for 2v1 situations 2011-11-29T00:15:59 although practically impossible for larger battles 2011-11-29T00:16:10 s/prune/evaluate 2011-11-29T00:16:38 the funny thing is I don't understand ant combat that well 2011-11-29T00:16:43 I just wrote a program that does it 2011-11-29T00:16:43 on the left are some interesting but useless scenarios where a large wall moves in all at once 2011-11-29T00:16:48 it does things I'd never think of! 2011-11-29T00:17:09 it's best for the 1 ant to either retreat or slide sideways (depends on if you prefer exchange or no-loss) 2011-11-29T00:17:19 (because that's unlikely to happen and you'd have to be able to predict it) 2011-11-29T00:17:20 so the 2-ant will move accordingly 2011-11-29T00:18:00 but of course, you don't know if your opponent prefers to retreat or exchange, or even have combat logic at all 2011-11-29T00:19:34 so a naive implementation might actually work better against stronger opponent (as their behaviours are more predictable) and worse against non-combat-aware bot 2011-11-29T00:20:13 *** Antimony has joined #aichallenge 2011-11-29T00:20:46 sf17k: I understand the 2 vs 1 drawings but the left ones with the bigger group is not clear. 2011-11-29T00:21:44 analyst74: opponent modeling but that is not allowed 2011-11-29T00:22:11 the very left ones? those are situations where each side has a vertical line of forces, 2-wide, in different formations, and they move forward at the same moment to meet each other 2011-11-29T00:22:14 *** antimatroid has joined #aichallenge 2011-11-29T00:22:43 besh, opponent modeling? 2011-11-29T00:23:31 basically the result of that was if you KNOW your opponent is going to move forward with a straight line, you can take it out with 3 to 4 losses if you use a clever formation 2011-11-29T00:23:44 yes like poker. You try to predict opponen't moves like in the week vs strong bot move example you gave. 2011-11-29T00:23:59 that's not allowed? 2011-11-29T00:24:07 *** Areks has joined #aichallenge 2011-11-29T00:24:29 someone was saying you can't have an anti-xathis code f.i 2011-11-29T00:24:37 it's not very useful yet, anyway 2011-11-29T00:25:02 we haven't nailed down basic mechanics and avoiding obvious mistakes 2011-11-29T00:25:28 you don't know who you are playing against, so I don't think pre-defined modeling is even possible 2011-11-29T00:25:39 sf17k: I think that is a good prediction. Most bots move their ants in a straight line 2011-11-29T00:26:04 besh: depends on the skill level of the bots, hardly any of the top bots do that 2011-11-29T00:26:56 I was saying that minimax works better against stronger bot, because the algorithm assumes enemy will choose the best move 2011-11-29T00:28:40 that is correct. So you need to model oppoent behaviour to make optimal moves in every situation. 2011-11-29T00:30:42 http://ants.fluxid.pl/player/strcat_foodfight since my fix at ~5:20 this seems to be doing well 2011-11-29T00:31:37 yeah, some sort of reinforced learning mechanism might help predict your opponent's next move, although not sure if that's an economical way to spend cpu time on 2011-11-29T00:32:41 thestinger, are you winning against A yet? :) 2011-11-29T00:33:02 only when I get lucky 2011-11-29T00:33:16 I haven't enabled my code to allow retreating yet 2011-11-29T00:33:27 makes the bugs too hard to find 2011-11-29T00:34:22 who is A? 2011-11-29T00:34:25 or is that unknown? 2011-11-29T00:34:25 lol 2011-11-29T00:34:27 FlagCapper 2011-11-29T00:34:31 _flag on irc iirc 2011-11-29T00:34:41 @seen _flag 2011-11-29T00:34:41 thestinger: _flag was last seen in #aichallenge 23 hours, 4 minutes, and 19 seconds ago: <_flag> So bots tend just to accept the stalemate 2011-11-29T00:35:00 ahhh, flag always does well 2011-11-29T00:35:16 do we know what he does? 2011-11-29T00:35:22 yes 2011-11-29T00:35:24 he wins 2011-11-29T00:35:25 :P 2011-11-29T00:35:33 lol 2011-11-29T00:35:34 irl :P 2011-11-29T00:35:36 he makes tcp game fast 2011-11-29T00:35:51 he's 16 2011-11-29T00:35:54 probably goes to highschool 2011-11-29T00:36:03 I eagerly await what top bots do once the tournament finished. 2011-11-29T00:36:27 what do you mean once it's finished? 2011-11-29T00:36:33 looking at source? 2011-11-29T00:36:41 yes 2011-11-29T00:36:46 is that not the case ? 2011-11-29T00:36:57 yes people share their code at the end (if they want) 2011-11-29T00:37:03 that's voluntary 2011-11-29T00:37:07 it was just a weird way of phrasing it :P 2011-11-29T00:37:19 you accidentally a word i think :P 2011-11-29T00:38:06 accidental not affect understanding 2011-11-29T00:38:09 he indeed accidentally a word 2011-11-29T00:38:41 I am new so I don't know much. But the contest organizers have a right to see the code right? 2011-11-29T00:38:55 besh: if cheating is suspected yes 2011-11-29T00:39:00 yep besh 2011-11-29T00:39:01 I was surprised to see that someone was reading it 2011-11-29T00:39:03 but we don't forcefully release everyones code 2011-11-29T00:39:06 was that amstan that was talking about that? 2011-11-29T00:39:07 so basically google wants your code 2011-11-29T00:39:12 cause they are going to harvest it 2011-11-29T00:39:19 I guess being a former sysadmin I'm ultra paranoid about looking at other people's files 2011-11-29T00:39:19 and use it to superpower their search engine 2011-11-29T00:39:20 lol, google are just a sponsor 2011-11-29T00:39:44 amazon on the other hand... 2011-11-29T00:39:53 amazon needs all the help they can get 2011-11-29T00:40:23 *** Kingpin13 has quit IRC (Ping timeout: 244 seconds) 2011-11-29T00:41:48 * BenJackson comes up with a concrete testcase where current combat sacrifices 2 to ensure the survival of 1 in a situation where the 1 is doomed on the next turn 2011-11-29T00:41:58 that will be hard to fix :) 2011-11-29T00:44:57 http://ants.fluxid.pl/player/strcat_foodfight \o/ 2011-11-29T00:45:13 fixed another bug xD 2011-11-29T00:46:59 some of the latest ai-class materials are actually quite applicable to this contest 2011-11-29T00:49:43 aichallenge: Janzert epsilon * r45ffb55 / (5 files in 3 dirs): 2011-11-29T00:49:44 aichallenge: Merge pull request #381 from stchang/epsilon 2011-11-29T00:49:44 aichallenge: add racket starter package - http://git.io/fI4XOQ 2011-11-29T00:50:06 oh? which ones? 2011-11-29T00:50:12 are their production values getting better? 2011-11-29T00:53:04 the production is not that bad, come on. :) 2011-11-29T00:53:27 I've seen the videos of the actual lectures 2011-11-29T00:53:30 they are far better 2011-11-29T00:53:39 high def, prepared slides, an audience pointing out silly mistakes 2011-11-29T00:53:52 not sure why they make special worse ones for ai-class 2011-11-29T00:54:12 cool, is the actual ones available some where? 2011-11-29T00:54:16 no 2011-11-29T00:54:20 I have a friend at Stanford 2011-11-29T00:54:38 :( 2011-11-29T00:56:55 I'm surprised I don't make my fluxid bot time out when I do these 10v10 tests 2011-11-29T00:57:26 er 10 player maps 2011-11-29T00:58:44 good for you, I'm constantly fighting against timeouts 2011-11-29T00:59:49 *** bmh has joined #aichallenge 2011-11-29T01:00:26 I never knew you could do minimax with probabilities 2011-11-29T01:01:07 maybe through a monte-carlo approach 2011-11-29T01:01:25 analyst74: my server is a Q6600 (2.4G) 2011-11-29T01:01:36 I suspect it's quite a bit slower than the true aichallenge servers 2011-11-29T01:02:14 BenJackson: heh, that's basically what I'm running my bots on atm 2011-11-29T01:02:28 *** Palmik has joined #aichallenge 2011-11-29T01:02:29 Linux arch 3.1.3-1-ARCH #1 SMP PREEMPT Sun Nov 27 21:08:51 CET 2011 x86_64 Intel(R) Core(TM)2 Quad CPU Q6700 @ 2.66GHz GenuineIntel GNU/Linux 2011-11-29T01:02:46 cell_maze_p09_02 sure can make some big multi-way bttles 2011-11-29T01:02:57 *** smiley1983 has quit IRC (Ping timeout: 258 seconds) 2011-11-29T01:03:16 I'm running 5 versions and compiling so there's some scheduling competition going on too :) 2011-11-29T01:03:29 *** smiley1983 has joined #aichallenge 2011-11-29T01:03:36 I got tired of my laptop overheating 2011-11-29T01:05:26 and there are no more TCP timeouts now that it's not connected over wifi 2011-11-29T01:05:41 *** Antimony has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-29T01:06:04 I have a mostly idle linode that I could run a server on (if running a server is idiot proof) 2011-11-29T01:06:31 I also tried running my bot on an ARM plug computer I have... didn't work out so well 2011-11-29T01:06:47 bmh: if you set up a "beginner" server and advertised it on the forum I'd participate 2011-11-29T01:07:09 BenJackson: are you being evil...? 2011-11-29T01:07:36 leftybot with combat 2011-11-29T01:07:41 I mean I'd put up a very old version of my bot 2011-11-29T01:07:45 or dumb one down specifically 2011-11-29T01:08:02 something better than the samples but that a newbie could hope to overcome with a few days of work 2011-11-29T01:08:05 how much work is involved? I'm running some archaic version of Debian 2011-11-29T01:08:09 I kind of want to write a silly bot in Go 2011-11-29T01:08:12 1 goroutine for each ant 2011-11-29T01:08:18 bmh: I don't know, sorry 2011-11-29T01:08:46 nothing beats the lolcat bot 2011-11-29T01:08:54 bmh, I believe running the tcp server is fairly trivial 2011-11-29T01:09:15 someone needs to write a bot in befunge or brainfuck :) 2011-11-29T01:10:17 duplicating actual contest server is a bit tricky 2011-11-29T01:10:24 *** u_ has quit IRC (Quit: u_) 2011-11-29T01:10:36 i would love to see a winning bot in brainfuck aha 2011-11-29T01:10:52 ... and released source code for all our benefits 2011-11-29T01:11:06 i imagine one would want to write an auto converter from another language 2011-11-29T01:11:10 antimatroid: it's doable. Write a winning bot in some language, write a brainfuck compiler and... 2011-11-29T01:11:20 yeah 2011-11-29T01:11:27 analyst74: the TCP server source is on git somewhere? 2011-11-29T01:11:33 would be pretty funny to see the leaderboard though :P 2011-11-29T01:11:44 bmh, yes, but I can't remember where 2011-11-29T01:11:45 bmh: https://github.com/berak/ants-tcp 2011-11-29T01:11:47 for the last ICFP contest, i ended up locating a scheme -> unlambda compiler which can also output brainfuck 2011-11-29T01:12:53 I think it's the "writing a winning bot" part that's some what complicated 2011-11-29T01:13:38 thestinger: I'll see what I can do. I forgot to setup DNS when I migrated my VM so it might be a while before I have a legitimate hostname 2011-11-29T01:13:57 http://www.hamsterfestdestiny.org/hamster/ LLVM output, lol 2011-11-29T01:14:24 I only ever did the first ICFP contest 2011-11-29T01:14:28 *** Magnify has joined #aichallenge 2011-11-29T01:14:53 it was fun but it turned out the winning team was like 6 grad students and 3 professors using a language they had happened to design that suited the problem 2011-11-29T01:15:17 I can't even find the pousse results online anymore 2011-11-29T01:15:50 thestinger: what's the correct way to launch this? run the server and disown the process? 2011-11-29T01:16:29 *** Garf has joined #aichallenge 2011-11-29T01:16:52 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-11-29T01:17:18 wow I really should have started making unit tests sooner 2011-11-29T01:18:31 I have some tests, but only because it helps me refactor/optimize 2011-11-29T01:18:46 I'm just making mini cases of the bugs I see 2011-11-29T01:18:48 *** bhasker has joined #aichallenge 2011-11-29T01:18:50 evening 2011-11-29T01:18:52 bmh, I believe you just run the 2 scripts using python 2011-11-29T01:18:54 Ive been doing some testing and I noticed that the input file is giving me all the ants regardless of whether they have spawned on the map. is there a way to distinguish if an ant is actually alive in the sim 2011-11-29T01:19:06 analyst74: will do. 2011-11-29T01:19:17 hopefully it'll be live on ch4n.org once DNS gets fixed 2011-11-29T01:20:08 yay cul-de-sac bug fixed 2011-11-29T01:26:41 ok, server's up 2011-11-29T01:27:01 http://74.207.246.79:2081/ 2011-11-29T01:27:34 ack, need to take it down briefly 2011-11-29T01:27:39 wait what's the web url? 2011-11-29T01:28:01 aichallenge: janzert epsilon * r8710159 / setup/worker_setup.py : Fix quotation on racket installation command - http://git.io/x5WVwA 2011-11-29T01:28:51 BenJackson: you can use bhickey.net:2081 2011-11-29T01:29:03 that's the game server 2011-11-29T01:29:20 argh. killed it by accident... 2011-11-29T01:29:22 * bmh is bad at zsh 2011-11-29T01:29:35 just put & on the end and disown %1 2011-11-29T01:29:56 if it has IO to stdout better to nohup it though 2011-11-29T01:31:42 webserver: http://74.207.246.79:2080/ game: http://74.207.246.79:2081/ 2011-11-29T01:31:47 I can't make any promises about stability (sorry) 2011-11-29T01:33:52 be sure to advertise in that "about xathis" thread 2011-11-29T01:34:16 BenJackson: I'll advertise it once DNS propagates for ch4n.org 2011-11-29T01:34:57 I'm connected as bj_v4 2011-11-29T01:35:14 are you sure you want to name yourself bj... :p 2011-11-29T01:35:26 indeed you are. I don't have a bot to throw at it other than the broken diffusion bot (which I didn't tag in git...) 2011-11-29T01:35:38 analyst74: been stuck with those initials all my life 2011-11-29T01:35:53 though I use the login bjj now because bj was taken when I got my first job in Oregon 2011-11-29T01:36:01 BenJackson: could be worse. As you might have gathered from an earlier URL, I was saddled with the surname "Hickey" 2011-11-29T01:36:06 yes 2011-11-29T01:36:12 hopefully you can't see my bot's password ;-) 2011-11-29T01:36:12 *** bhasker has left #aichallenge ("Killed buffer") 2011-11-29T01:36:21 *** bhasker has joined #aichallenge 2011-11-29T01:36:39 BenJackson: I probably could, but I'm redirecting output to /dev/null and I can't be buggered to try to capture it. 2011-11-29T01:37:33 wisecrack about your name 2011-11-29T01:37:47 in my defense it's hard to come up with so many passwords ;-) 2011-11-29T01:37:54 *** Magnify has quit IRC (Quit: Page closed) 2011-11-29T01:40:03 BenJackson: I just have a really crappy password for stuff like this that doesn't matter :P 2011-11-29T01:41:28 ut oh, I'm new to irssi and now it's saying something is "Act: 3" 2011-11-29T01:41:28 running my branches on tcp is taken care of by a Makefile target so my pw is in it 2011-11-29T01:41:36 I wonder how to ... go there 2011-11-29T01:41:37 *** b0rder has quit IRC (Read error: Connection reset by peer) 2011-11-29T01:41:43 BenJackson: in weechat you use alt+arrows 2011-11-29T01:41:57 *** JorgeB has quit IRC (Quit: Computer has gone to sleep.) 2011-11-29T01:42:01 maybe the same in irrsi 2011-11-29T01:42:03 irssi* 2011-11-29T01:42:41 haha I found out it means I have a msg 2011-11-29T01:42:48 however the doc didn't say how to switch to that window... 2011-11-29T01:43:35 alt-n 2011-11-29T01:43:40 ah 2011-11-29T01:43:43 alt-arrows probably works too 2011-11-29T01:43:45 which I'm kind of surprised works through screen 2011-11-29T01:43:49 it does 2011-11-29T01:43:53 I am surprised again 2011-11-29T01:44:20 I guess now that I have this fancy client I could hang out on #reprap and other 3d printing stuff 2011-11-29T01:44:24 and you can split windows, which is nice 2011-11-29T01:44:30 weechat is nicer :P 2011-11-29T01:45:15 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-29T01:45:35 hmm, does the trueskill system give preference to stable performance? 2011-11-29T01:45:41 I just googled "Freebsd irc client" and lo irssi was in the ports already 2011-11-29T01:45:44 *** b0rder has joined #aichallenge 2011-11-29T01:45:50 oh, so is weechat 2011-11-29T01:46:24 I will keep it in mind in case irssi pisses me off 2011-11-29T01:46:38 *** treeform_ has quit IRC (Remote host closed the connection) 2011-11-29T01:46:48 *** Chris_0076 has quit IRC (Read error: Connection reset by peer) 2011-11-29T01:47:18 *** Chris_0076 has joined #aichallenge 2011-11-29T01:47:58 BenJackson: irssi is very minimal 2011-11-29T01:48:08 weechat is a lot nicer out of the box 2011-11-29T01:48:28 and iirc, irssi can only be extended with perl 2011-11-29T01:48:57 both are nice clients though 2011-11-29T01:49:00 analyst74: analyst74: The sigma affects by how much your rating changes.Other than I think greater mu should always be higher ranked no matter what. 2011-11-29T01:49:23 besh: skill is a conservative estimate, it's 'mu - sigma*3' 2011-11-29T01:49:52 conservative as in "you are almost certainly better than this" 2011-11-29T01:49:53 so it is a lower bound 2011-11-29T01:50:21 99% confidence it seems 2011-11-29T01:51:09 http://aichallenge.org/visualizer.php?game=141274&user=3473 2011-11-29T01:51:12 you want to see something confusing 2011-11-29T01:51:17 go to turn 82 2011-11-29T01:51:26 look at the hill at 21,69 2011-11-29T01:51:32 THAT ANT IS A SPY 2011-11-29T01:52:41 *** b0rder has quit IRC (Read error: Connection reset by peer) 2011-11-29T01:52:58 your ant seems to behave rationally 2011-11-29T01:53:14 until one of them kills itself 2011-11-29T01:53:46 ah 2011-11-29T01:53:50 I'm talking about where my ants kill a defender and mine steps in 2011-11-29T01:53:55 my first version works! 2011-11-29T01:54:02 stands there defending 2011-11-29T01:54:43 *** b0rder has joined #aichallenge 2011-11-29T01:55:18 haha, only if it can change color 2011-11-29T01:56:02 that reminds me that I have to figure out how to make weechat use 256 colors for names 2011-11-29T01:56:29 but then I'll never remember who anyone is :P, it uses the same colors for people forever 2011-11-29T01:58:06 back to stigma, assuming 2 bots of same mu after same amount of games, but one arrived at that mu by bouncing around, the other stabilizes around that mu long time ago, is the later going to have a lower sigma, thus higher skill? 2011-11-29T02:00:04 *** bysin has quit IRC (Ping timeout: 245 seconds) 2011-11-29T02:00:30 I think sigma will drop a lot faster if you play as you're expected to play (consistently) 2011-11-29T02:00:42 BenJackson: you can also modify the url like that: http://aichallenge.org/visualizer.php?game=141274&turn=82&row=21&col=69 2011-11-29T02:01:33 neat 2011-11-29T02:01:39 although not as subtle with those colors 2011-11-29T02:01:42 maybe keep the user= 2011-11-29T02:02:06 http://aichallenge.org/visualizer.php?game=141274&turn=82&row=21&col=69&user=3473 2011-11-29T02:02:26 oh it is white and pale yellow :) 2011-11-29T02:03:08 Rank: Not Ranked Skill: No Skillz 2011-11-29T02:03:10 here we go again 2011-11-29T02:03:31 mleise, that's neat trick 2011-11-29T02:04:00 v9's first game on fluxid is a double win vs strcat variants ;-) 2011-11-29T02:04:17 thestinger: you might have so many versions of your bot on fluxid that you're polluting your own result 2011-11-29T02:04:21 trusekill : http://research.microsoft.com/en-us/projects/trueskill/details.aspx 2011-11-29T02:04:42 It seems sigma is slightly perturbed each game you play. 2011-11-29T02:04:47 BenJackson: yeah, there aren't enough people using tcp :( 2011-11-29T02:05:16 analyst74: there is more ^^: http://aichallenge.org/visualizer.php?game=141274&turn=82&row=21&col=69&user=3473&interactive=false&decorated=false&config={%22fullscreen%22:true} 2011-11-29T02:07:14 ok, re-up'd, now for some skyrim 2011-11-29T02:07:19 This reminds me that my rating on tcp servers maybe truly inflated given the thousands of games I play :( 2011-11-29T02:07:47 *** bysin has joined #aichallenge 2011-11-29T02:08:10 tcp-server ranking is useless 2011-11-29T02:08:24 it's the replay that's valuable 2011-11-29T02:09:01 also the ego boost 2011-11-29T02:09:39 thestinger: you were beating in fluxid 3558 by a guy with the suffix _blinded 2011-11-29T02:09:42 that's gotta hurt 2011-11-29T02:09:44 beating bj_v5 on tcpants has boosted my ego :) 2011-11-29T02:10:11 thanks for that btw 2011-11-29T02:11:07 BenJackson: done: http://forums.aichallenge.org/viewtopic.php?f=24&t=1996&p=12209#p12209 2011-11-29T02:11:10 v4 is on http://74.207.246.79:2080/ 2011-11-29T02:11:30 analys74: I don't know if it is applicable here but there is also bayeselo 2011-11-29T02:12:39 I believe at one point the contest (during Tron maybe?) used Elo, not sure if it's bayselo 2011-11-29T02:12:59 BenJackson: I'll join up with Diffusion bot :) 2011-11-29T02:13:06 halp...I'm on linux and my C++ bot timed out on the test uploaded, and the test_bot.sh script doesn't seem to work (python error) so I can't test locally 2011-11-29T02:13:11 wtf google, gmail is being shit 2011-11-29T02:13:19 when uploaded* 2011-11-29T02:13:38 it runs fine with play_one_game* scripts 2011-11-29T02:14:11 wow, rank 7151 2011-11-29T02:14:13 so many bots now 2011-11-29T02:14:39 yeah, and they can be tied for ranks 2011-11-29T02:14:51 *** roflmao has quit IRC (Quit: Leaving.) 2011-11-29T02:17:09 BenJackson: I think we're playing :) 2011-11-29T02:18:14 maybe I should start playing on these other servers 2011-11-29T02:18:27 bmh: Can we all play or Is that nOObs only server ? 2011-11-29T02:19:01 besh: go for it. I just don't want it to get flooded with strcat_[0..] and other uberbots 2011-11-29T02:20:40 well my bot is advanced now so I will let others play. But ready any time you need a bot :) 2011-11-29T02:21:03 besh: there's no one on it right now :) 2011-11-29T02:21:27 ok then 2011-11-29T02:22:11 diffusionBot is really bad. It runs away from its own hill. 2011-11-29T02:26:09 bmh, your server is fast. :) 2011-11-29T02:26:31 analyst74: because no one is on it. It's a Linode 512 2011-11-29T02:27:40 ping 80ms for me 2011-11-29T02:28:13 it's in Freemont, CA. My ping is ~25ms 2011-11-29T02:30:16 It seems trueskill is not enabled. 2011-11-29T02:30:30 besh: ack. Was I supposed to run something else? 2011-11-29T02:30:50 I don't know anything ,sorry 2011-11-29T02:31:01 it might just take a few games before it kicks in 2011-11-29T02:31:41 is mu just staying at 50 and sigma at like 14 or something? that happened to my bot when fluxid reset the server last time 2011-11-29T02:31:51 took ~10 games before it changed 2011-11-29T02:32:45 yes that may be how trueskill works 2011-11-29T02:32:54 it should update every game 2011-11-29T02:33:01 (makes no sense not to) 2011-11-29T02:34:07 well what will be sigma after one game, nitpicking 2011-11-29T02:34:13 undefined 2011-11-29T02:34:38 so it don't update every game then :) 2011-11-29T02:34:40 don't see why, there's a defined starting value and the first game gives information 2011-11-29T02:34:54 Garf: The variance is zero after one game. 2011-11-29T02:40:59 bmh: I have absolutely no idea how you get that. 2011-11-29T02:42:28 *** Kingpin13 has joined #aichallenge 2011-11-29T02:43:08 after 1 game of 2 new players, sigma should evolve from 8.3 to 7.15 2011-11-29T02:43:40 Garf: sorry. Thought you meant the *very* first game. (Before anyone is rated) 2011-11-29T02:43:47 there is no "amount of games played" in the update formulas, a a new player is forcedly initialized to some values 2011-11-29T02:43:53 ah 2011-11-29T02:43:57 no variance is never undefined 2011-11-29T02:44:27 *** ikaros has joined #aichallenge 2011-11-29T02:45:21 from the terms & conditions of an online casino: "4.4 WE DO NOT WARRANT THAT THE SOFTWARE WILL BE NON-INFRINGING OR THAT THE OPERATION OF THE SOFTWARE WILL BE ERROR FREE OR UNINTERRUPTED OR THAT ANY DEFECTS IN THE SOFTWARE WILL BE CORRECTED, OR THAT THE SOFTWARE OR THE SERVERS ARE VIRUS-FREE." 2011-11-29T02:46:17 o_o 2011-11-29T02:46:42 sounds inviting :D 2011-11-29T02:49:02 Makes me want to spend money on them. 2011-11-29T02:49:47 oh you can first use up those 400$ welcome bonus 2011-11-29T02:50:17 no risk no fun: http://www.virtualvegasclub.com/us/terms.html 2011-11-29T02:50:54 *** epicmonkey has joined #aichallenge 2011-11-29T02:54:47 https://gist.github.com/1403928 -- that's my 'who can reach what first' code :) 2011-11-29T02:55:16 BenJackson: you're really hammering on analyst74... 2011-11-29T02:56:21 my ants have no sense of self-preservation 2011-11-29T02:56:27 I taught them to retreat but they never do :( 2011-11-29T02:56:33 *** smiley1983 has quit IRC (Ping timeout: 252 seconds) 2011-11-29T02:56:52 thestinger: Maybe they're RED ants? 2011-11-29T02:57:30 *** smiley1983 has joined #aichallenge 2011-11-29T02:58:01 *** sigh has joined #aichallenge 2011-11-29T02:58:52 i look forward to knowing how many others noticed something i did for calculating one piece of information 2011-11-29T02:59:28 well, i calculate the information for the grid, but there was a little fact that made my information better than i expect most others to have 2011-11-29T03:00:17 bmh: were you the one working in a team? 2011-11-29T03:00:26 that asked a1k0n to be in said team? 2011-11-29T03:00:30 antimatroid: yeah :) 2011-11-29T03:00:36 (he wisely declined) 2011-11-29T03:00:43 do you still have a team? or is it just you now? 2011-11-29T03:00:58 i don't know how many team bots we end up getting 2011-11-29T03:01:03 my tron bot was a pair effort 2011-11-29T03:01:12 there are two of us. Eric wrote a lot of infrastructure, but he's losing steam 2011-11-29T03:01:30 tell him i said boooo 2011-11-29T03:01:38 noted. 2011-11-29T03:01:57 i got rid of generalised path finding 2011-11-29T03:02:00 that was a terrible terrible mistake 2011-11-29T03:02:05 I got rid of pathfinding. 2011-11-29T03:02:12 i use it lots :) 2011-11-29T03:02:17 pathfind all the moves 2011-11-29T03:02:21 except battle 2011-11-29T03:02:37 i don't actually get the paths, but i need to use the searches 2011-11-29T03:03:00 @eightball is pathfinding a good idea? 2011-11-29T03:03:00 thestinger: It shall be. 2011-11-29T03:03:06 :P 2011-11-29T03:03:20 @eightball: is pathfinding a bad idea? 2011-11-29T03:03:21 antimatroid: I'm sorry Dave, err antimatroid; I cannot 'eightball:'. 2011-11-29T03:03:34 @eightball is pathfinding a bad idea? 2011-11-29T03:03:34 antimatroid: What are you asking me for? 2011-11-29T03:03:42 * antimatroid is so confused 2011-11-29T03:03:43 haha 2011-11-29T03:04:01 @rainbow this fixes everything! 2011-11-29T03:04:01 antimatroid: this fixes everything! 2011-11-29T03:04:09 @azn this fixes everything 2011-11-29T03:04:09 thestinger: this fixes evelything 2011-11-29T03:04:12 *** besh has quit IRC (Quit: Page closed) 2011-11-29T03:04:15 @roullete 2011-11-29T03:04:16 antimatroid: Run as fast as you can and don't look back. 2011-11-29T03:04:21 @list Games 2011-11-29T03:04:21 thestinger: coin, dice, eightball, monologue, and roulette 2011-11-29T03:04:22 nawww 2011-11-29T03:04:31 nawww 2011-11-29T03:04:33 @roullete 2011-11-29T03:04:34 antimatroid: User error -- Replace user. 2011-11-29T03:04:41 @help roulette 2011-11-29T03:04:41 thestinger: (roulette [spin]) -- Fires the revolver. If the bullet was in the chamber, you're dead. Tell me to spin the chambers and I will. 2011-11-29T03:04:42 i think amstan stopped it again 2011-11-29T03:04:53 @roulette 2011-11-29T03:04:53 thestinger: *click* 2011-11-29T03:04:58 wtf :P 2011-11-29T03:05:15 @roulette 2011-11-29T03:05:15 antimatroid: *click* 2011-11-29T03:05:18 there we go 2011-11-29T03:05:26 i can't spell to kill myself :P 2011-11-29T03:05:41 @roulette 2011-11-29T03:05:41 antimatroid: *click* 2011-11-29T03:05:44 alright, I'm going to bed. Someone e-mail me if my server catches on fire. My address is easy to find. 2011-11-29T03:05:47 *** bmh has quit IRC (Quit: Page closed) 2011-11-29T03:06:00 he didn't even let contest bot kick him out :( 2011-11-29T03:06:06 I think this is not optimal strategy? 2011-11-29T03:06:13 Garf: what isn't? 2011-11-29T03:06:24 *** g0llum has joined #aichallenge 2011-11-29T03:06:25 The roulette you're playing :P 2011-11-29T03:06:33 @roulette 2011-11-29T03:06:33 antimatroid: *click* 2011-11-29T03:06:40 i don't think it's going to do it anyway 2011-11-29T03:06:44 i suspect amstan ruined the fun 2011-11-29T03:06:47 @roulette 2011-11-29T03:06:47 *BANG* Hey, who put a blank in here?! 2011-11-29T03:06:48 * contestbot reloads and spins the chambers. 2011-11-29T03:06:58 :( 2011-11-29T03:07:26 Unless I'm missing anything, you should spin after every click 2011-11-29T03:08:51 spin means reset 2011-11-29T03:08:58 @roulette 2011-11-29T03:08:58 antimatroid: *click* 2011-11-29T03:09:00 @roulette 2011-11-29T03:09:00 antimatroid: *click* 2011-11-29T03:09:03 @roulette 2011-11-29T03:09:03 antimatroid: *click* 2011-11-29T03:09:05 @roulette 2011-11-29T03:09:05 antimatroid: *click* 2011-11-29T03:09:07 @roulette 2011-11-29T03:09:07 *BANG* Hey, who put a blank in here?! 2011-11-29T03:09:08 * contestbot reloads and spins the chambers. 2011-11-29T03:09:11 nah, he ruined it 2011-11-29T03:09:20 i'll stop now :P 2011-11-29T03:11:18 Bot has to be opped to kill anyone ;] 2011-11-29T03:11:23 @list Web 2011-11-29T03:11:23 thestinger: doctype, fetch, headers, netcraft, size, title, urlquote, and urlunquote 2011-11-29T03:11:29 @list games 2011-11-29T03:11:29 magikmw: coin, dice, eightball, monologue, and roulette 2011-11-29T03:11:36 @monologue 2011-11-29T03:11:36 magikmw: Your current monologue is at least 2 lines long. 2011-11-29T03:13:49 *** smiley1983 has quit IRC (Ping timeout: 240 seconds) 2011-11-29T03:14:07 *** epicmonkey has quit IRC (Ping timeout: 248 seconds) 2011-11-29T03:15:40 @web title http://www.downforeveryoneorjustme.com/google.com 2011-11-29T03:15:42 thestinger: http://google.com Is Up -> Check if your website is up or down? 2011-11-29T03:15:45 neat 2011-11-29T03:17:14 *** besh has joined #aichallenge 2011-11-29T03:17:47 aichallenge: janzert epsilon * re7edca7 / setup/worker_setup.py : Correct racket symlink path - http://git.io/JyqLYA 2011-11-29T03:20:12 *** smiley1983 has joined #aichallenge 2011-11-29T03:20:55 wow, my combat wasn't working AT ALL 2011-11-29T03:21:03 stupid bug. 2011-11-29T03:21:17 wow, everything works now 2011-11-29T03:22:09 so I just imagined combat was working... :( 2011-11-29T03:23:17 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T03:24:06 strcat_foodfight should crush my other bots now... 2011-11-29T03:25:28 *** smiley1983 has quit IRC (Ping timeout: 244 seconds) 2011-11-29T03:26:32 *** smiley1983 has joined #aichallenge 2011-11-29T03:27:55 I am curious. Are most participants students (possibly computer science major( 2011-11-29T03:28:21 Most polish participants are from what I've seen. 2011-11-29T03:28:24 I'm not. 2011-11-29T03:28:39 But yeah, there's a lot of Uni students. 2011-11-29T03:28:47 *** kiv has joined #aichallenge 2011-11-29T03:28:55 i am a hobbist 2011-11-29T03:33:13 *** smiley1983 has quit IRC (Ping timeout: 258 seconds) 2011-11-29T03:33:57 *** smiley1983 has joined #aichallenge 2011-11-29T03:35:23 i'm in limbo land 2011-11-29T03:35:28 not sure if i'll get phd funding or not 2011-11-29T03:37:19 contestbot: 2011-11-29T03:37:43 @echo @echo 1 2011-11-29T03:37:43 @echo 1 2011-11-29T03:37:55 ignores itself :( 2011-11-29T03:38:02 nice try :P 2011-11-29T03:40:30 maybe you can do loops with regex somehow 2011-11-29T03:40:46 @random 2011-11-29T03:40:46 magikmw: Error: The command "random" is available in the Dict, Factoids, Quote, and QuoteGrabs plugins. Please specify the plugin whose command you wish to call by using its name as a command before "random". 2011-11-29T03:41:00 @quote random 2011-11-29T03:41:00 magikmw: Error: I have no quotes in my database for #. 2011-11-29T03:41:03 :o 2011-11-29T03:41:16 @quote grab thestinger 2011-11-29T03:41:16 magikmw: (quote [] ) -- Returns 's latest quote grab in . is only necessary if the message isn't sent in the channel itself. 2011-11-29T03:41:16 @help QuoteGrab 2011-11-29T03:41:17 thestinger: Error: There is no command "quotegrab". 2011-11-29T03:41:26 @grab thestinger 2011-11-29T03:41:26 magikmw: Job's done. 2011-11-29T03:41:31 @quote random 2011-11-29T03:41:31 magikmw: Error: I have no quotes in my database for #. 2011-11-29T03:41:38 @quote random thestinger 2011-11-29T03:41:38 thestinger: (quote random []) -- Returns a random quote from . is only necessary if the message isn't sent in the channel itself. 2011-11-29T03:41:43 I don't get this bot :p 2011-11-29T03:41:44 @quote random #aichallenge 2011-11-29T03:41:44 thestinger: Error: I have no quotes in my database for #. 2011-11-29T03:41:47 @quote random aichallenge 2011-11-29T03:41:47 thestinger: (quote random []) -- Returns a random quote from . is only necessary if the message isn't sent in the channel itself. 2011-11-29T03:41:49 @quote random #aichallenge 2011-11-29T03:41:49 magikmw: Error: I have no quotes in my database for #. 2011-11-29T03:42:00 Nevermind. 2011-11-29T03:42:07 maybe it's not enabled 2011-11-29T03:42:11 @quote random 2011-11-29T03:42:11 janzert: Error: I have no quotes in my database for #. 2011-11-29T03:42:24 @fortune 2011-11-29T03:42:24 thestinger: Error: The fortune command is not configured. If fortune is installed on this system, reconfigure the supybot.plugins.Unix.fortune.command configuration variable appropriately. 2011-11-29T03:42:35 heh, more like broken it appears 2011-11-29T03:43:04 like my bot :) 2011-11-29T03:43:19 The force is weak with this bot. 2011-11-29T03:43:21 :/ 2011-11-29T03:43:36 I intended to predict offensive enemy moves, and instead was making them run away 2011-11-29T03:43:45 so that totally screwed up my combat... 2011-11-29T03:44:17 Don't predict. Make a line. ADVANCE! 2011-11-29T03:44:28 use a tree :P 2011-11-29T03:44:42 Yeah, use a ram made from a tree. 2011-11-29T03:44:46 Good one. 2011-11-29T03:45:03 Scatter them! Murder every last one of them. 2011-11-29T03:46:00 antimatroid: I don't have enough time to use a tree for moderately sized battles with my horribly unoptimized code 2011-11-29T03:46:14 i use a tiny tree for each battle 2011-11-29T03:46:35 pick offensive and defensive moves for my ants and their ants in each paritition then use those as moves in a tree/matrix 2011-11-29T03:46:40 some battles do feel like real battles with all the confrontation at frontiers 2011-11-29T03:46:59 antimatroid: ah, I do a little 2x2 matrix with offensive/defensive for both myself and the "enemy" 2011-11-29T03:47:59 http://ants.fluxid.pl/replay.3680 yeah, combat sort of works now 2011-11-29T03:48:01 :) 2011-11-29T03:48:48 not sure why they run away sometimes 2011-11-29T03:49:21 I was starting to worry for a while that battle fronts might not develop so I'm happy to see them around :) 2011-11-29T03:49:26 *** Ashoka has joined #aichallenge 2011-11-29T03:52:29 *** mj41 has joined #aichallenge 2011-11-29T03:53:17 oh, more bugs! 2011-11-29T03:53:21 thestinger: which turn shows the new 2x2 matrix combat code in action 2011-11-29T03:54:08 it makes my ants form the battle lines at my own hill and the enemy hill near the end 2011-11-29T03:54:25 there was a bug (now fixed) that made them retreat for no reason tho 2011-11-29T03:55:30 thestinger: the thing is i implemented the second way 2011-11-29T03:55:46 thestinger: You mean they have known fear, but now they don't? ;] 2011-11-29T03:55:47 and pathfinding could take over 150ms 2011-11-29T03:56:01 Fluxid: really? 2011-11-29T03:56:07 *** ikaros has quit IRC (Remote host closed the connection) 2011-11-29T03:56:11 it goes really fast for me 2011-11-29T03:56:28 when I leave a tile with an ant, I do +3 instead of +1 2011-11-29T03:56:34 because shorter ways were overwriting the longer ones many times (i do height map) 2011-11-29T03:56:42 but i know what i need to do now 2011-11-29T03:56:49 do you use a multi-source BFS? 2011-11-29T03:57:02 it's 70x faster for me to do it that way 2011-11-29T03:57:09 compared to putting each target on the map one by one 2011-11-29T03:57:10 *** jon1 has joined #aichallenge 2011-11-29T03:57:17 thestinger: flood fill i told you earlier about 2011-11-29T03:57:33 yeah, but I do that with a multi-source BFS 2011-11-29T03:57:39 not sure how else to implement it :) 2011-11-29T03:57:50 i need to use sorted linked list instead of stack... 2011-11-29T03:58:09 do a BFS with a FIFO queue 2011-11-29T03:58:19 and stick all targets in the queue before you start 2011-11-29T03:58:44 well, i do put all the targets to queue right now 2011-11-29T03:59:28 *** Umsturz has joined #aichallenge 2011-11-29T03:59:33 it has to be a FIFO tho, not a stack 2011-11-29T04:00:23 i mean yeah, queue/fifo, not stack 2011-11-29T04:00:32 ah 2011-11-29T04:00:57 wow V9 has won 16 in a row on fluxid and gone to #1 2011-11-29T04:01:04 but still, it fails, it makes multiple passes over the same area, with the shorter and shorter paths 2011-11-29T04:01:05 I should play more skyrim ;-) 2011-11-29T04:01:17 so i takes 15x longer 2011-11-29T04:01:57 mine will do a bit of that, but it's still fast 2011-11-29T04:02:10 so i thought i'll use linked list linked list sorted by distance 2011-11-29T04:02:11 *** Kingpin13 has quit IRC (Quit: quit) 2011-11-29T04:02:26 but then putting element into it would be inefficient 2011-11-29T04:02:51 "multi source BFS" is still a true BFS it just has some imaginary root node whose children happen to be all of your ants or food or whatever 2011-11-29T04:02:54 but at least i would avoid multiple passes 2011-11-29T04:04:32 haha due to my fluke of winnin 16 in a row on fluxid even xathis will have trouble catching me 2011-11-29T04:04:39 I have 15 points on him in mu 2011-11-29T04:05:16 *** Ikol has joined #aichallenge 2011-11-29T04:05:51 wait till the day comes :) Pretty impressive though. 2011-11-29T04:06:07 ok, you could say what i do is BFS 2011-11-29T04:07:14 besh or just never play another agme with that id ;-) 2011-11-29T04:07:16 thestinger: how long your pathfinding does take? 2011-11-29T04:07:36 in tital by turn 2011-11-29T04:07:38 when I put ~7k unseen locations onto the queue and run it, less than 1ms 2011-11-29T04:07:39 *** ikaros has joined #aichallenge 2011-11-29T04:07:39 total 2011-11-29T04:07:43 well, I do a bunch of them 2011-11-29T04:07:59 ~4ms-10ms, but I do hundreds of searches for different things 2011-11-29T04:08:12 ok 2011-11-29T04:08:29 hmm strcat flock just brought you down 2011-11-29T04:08:32 nooo 2011-11-29T04:08:39 yeeeeeessssssssss 2011-11-29T04:08:44 hahaha dropping me to 3rd in one game 2011-11-29T04:08:56 thestinger: do you search for the same stuff every turn? 2011-11-29T04:09:07 yeah, I redo it every turn 2011-11-29T04:09:07 *** Rav3nsW0rd has quit IRC (Ping timeout: 260 seconds) 2011-11-29T04:09:17 Fluxid: what are your bfs data structures? 2011-11-29T04:09:21 that might make a big difference 2011-11-29T04:09:29 I do hundreds of BFSs every turn sometimes 2011-11-29T04:09:29 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T04:09:35 my searches just use std::queue, and I store distances in a 2d std::vector of ints 2011-11-29T04:09:36 and it's not my bottleneck by a long shot 2011-11-29T04:09:48 I don't keep track of visited nodes, I just make sure the new distance would be lower than the old one in a node 2011-11-29T04:10:07 BenJackson: a queue 2011-11-29T04:10:19 just structs with references to next node 2011-11-29T04:10:22 Fluxid: how do you track visited? 2011-11-29T04:10:30 ohhh no, that is not the right kind of queue 2011-11-29T04:10:34 what language? 2011-11-29T04:10:37 C 2011-11-29T04:10:49 use an array and read/write pointers 2011-11-29T04:11:10 you can just make the queue big enough for every possible node (rows*cols) 2011-11-29T04:11:11 *** Ikol has quit IRC (Ping timeout: 248 seconds) 2011-11-29T04:11:15 sys/queue.h 2011-11-29T04:11:18 a linked list with reference to last node, and store visited in 2D array 2011-11-29T04:11:22 is there a FIFO in that header? 2011-11-29T04:11:30 why should a queue be bad? push and pop is O(1) 2011-11-29T04:11:35 insert = q[wr++] = location, dequeue is q[rd++], empty is wr==rd, ... 2011-11-29T04:11:43 shouldn't it be a priority queue ? 2011-11-29T04:11:48 not for BFS 2011-11-29T04:11:52 besh: you're thinking of A* 2011-11-29T04:11:54 it should be first-in first-out 2011-11-29T04:12:06 so you pop off the front and push onto the end (or the other way) 2011-11-29T04:12:09 std::queue works great for my node list 2011-11-29T04:12:22 yeah, I use std::queue, which is like what I'm describing 2011-11-29T04:12:56 but the difference is, i don't stop, i "flood" whole map with distances to target, and then ants move by gradient descent 2011-11-29T04:13:01 ok but even if you have structs with pointers to next.. what would be bad about it? 2011-11-29T04:13:13 one such flood takes 10ms on 200x200 2011-11-29T04:13:16 Fluxid: yeah, that was my initial bot 2011-11-29T04:13:24 fix your queue and it won't 2011-11-29T04:13:50 fix? how can i fix it 2011-11-29T04:13:51 your queue has too much overhead, especially if you are malloc/freeing each node 2011-11-29T04:14:00 BenJackson: i cache nodes 2011-11-29T04:14:00 *** smiley1983 has quit IRC (Ping timeout: 260 seconds) 2011-11-29T04:14:18 i would be really surprised if that helps alot 2011-11-29T04:14:34 http://en.wikipedia.org/wiki/Circular_buffer 2011-11-29T04:14:39 and how that queue ca be better 2011-11-29T04:14:48 *** smiley1983 has joined #aichallenge 2011-11-29T04:15:21 I wrote a BFS *iterator* and tested *copying the iterator* to make sure it worked and it sure didn't take 10ms, so there's something wrong 2011-11-29T04:16:34 *** Rav3nsW0rd has quit IRC (Excess Flood) 2011-11-29T04:16:38 even if it did visit 40000 cells? 2011-11-29T04:16:52 yeah 2011-11-29T04:17:02 I can run a thousand of them in a turn visiting all the cells 2011-11-29T04:17:04 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T04:17:05 a loop to 40000 on a 2.5GHz processor does not take 10ms 2011-11-29T04:17:50 fluxid: if you want to paste your code somewhere I'll look at it 2011-11-29T04:19:25 it's a really horrible undocumented code 2011-11-29T04:19:36 but well, okay 2011-11-29T04:20:48 besh: you weren't kidding about strcat_flock 2011-11-29T04:21:01 that map kind of sucks 2011-11-29T04:21:09 uh 2011-11-29T04:21:12 no route host 2011-11-29T04:21:21 to host* 2011-11-29T04:21:25 ? 2011-11-29T04:21:31 did you try some irc thing? I'm not irc savvy 2011-11-29T04:21:45 i have the code on my home computer and i can't connect 2011-11-29T04:22:49 huh, I guess cell_maze_p06_02 is a legal map by the new rules 2011-11-29T04:23:13 is it the one with thick walls? 2011-11-29T04:23:23 * BenJackson makes a note to panic if ants show up in his vision on turn 24 2011-11-29T04:23:29 http://ants.fluxid.pl/replay.3699 2011-11-29T04:24:07 definitely need a special case for a single-hill map with a neighbor that close. 2011-11-29T04:25:58 Very close hill placements will be avoided in the final tournament. 2011-11-29T04:27:31 I think my bot can defend in that kind of situation. Problem is that is all it will do, zero exploration. 2011-11-29T04:29:17 my bot X get clobbered by strcat on that one. needs to improve my combat skills. it just suicides currently 2011-11-29T04:29:46 besh: very close was something like 25 squares 2011-11-29T04:29:49 besh: those hills are far enough apart though under the current proposed distances we'll use 2011-11-29T04:29:51 those hills are actually more than 25 apart 2011-11-29T04:29:54 36 is the cutoff 2011-11-29T04:30:44 still doesn't seem like a very good map for ranking bots 2011-11-29T04:30:45 or maybe I miscounted and they would fall just inside the cutoff 2011-11-29T04:33:55 Hills must be 100 to 250 steps away from each other 2011-11-29T04:34:19 why didn't i know there's sys/queue.h... 2011-11-29T04:34:21 36 to 150 is what was decided earlier today 2011-11-29T04:34:22 BenJackson: thanks 2011-11-29T04:34:46 Fluxid: it's non-standard 2011-11-29T04:34:55 it's from the BSDs iirc, linux has an old version of it 2011-11-29T04:35:21 Oh I didn't know that. I just read what was here yesterday http://forums.aichallenge.org/viewtopic.php?f=27&t=1973 2011-11-29T04:35:29 besh: you can go back and look at the conversation mcleopold and I had maybe 12-14 hours ago 2011-11-29T04:35:31 but it's just a header, right? 2011-11-29T04:35:34 yeah 2011-11-29T04:35:38 you can count on it being there on linux 2011-11-29T04:35:47 it's all done with macros in a header 2011-11-29T04:35:48 *I* didn't know about sys/queue.h 2011-11-29T04:35:49 the spec page hasn't been updated yet either 2011-11-29T04:36:09 you can also steal the linux code for heaps, queues, etc. :P 2011-11-29T04:36:14 yeah 2011-11-29T04:36:30 and you know max possible q len here so you can just allocate it static 2011-11-29T04:36:41 ok I gotta quit watching my fluxid wins and go to bed :) 2011-11-29T04:37:37 BenJackson: Then why don't we start sentences with capital letter and end it with a dot? 2011-11-29T04:37:44 :) 2011-11-29T04:45:47 *** jon1 has quit IRC (Quit: Leaving.) 2011-11-29T04:46:42 *** grwip has joined #aichallenge 2011-11-29T04:47:53 *** Kingpin13 has joined #aichallenge 2011-11-29T04:50:16 *** antimatroid has quit IRC (Quit: Leaving.) 2011-11-29T04:50:28 *** antimatroid has joined #aichallenge 2011-11-29T04:52:33 anyone using minimax? 2011-11-29T04:54:53 *** antimatroid has quit IRC (Read error: No route to host) 2011-11-29T04:55:07 *** antimatroid has joined #aichallenge 2011-11-29T04:55:48 a1k0n's tron code is relevant here as well 2011-11-29T04:55:55 *** Umsturz has left #aichallenge 2011-11-29T05:01:36 not really 2011-11-29T05:01:49 the state space for tron was tiny compared to ants 2011-11-29T05:01:55 even once you partition ants up into battles 2011-11-29T05:02:03 unless you do 1v1 of course :P 2011-11-29T05:02:11 even then you don't have ants old moves blocking 2011-11-29T05:02:50 voronoi 2011-11-29T05:04:54 antimatroid : did u finish your assignments and stuff ? hope u will have enough time to compete :-) 2011-11-29T05:06:02 I hate random walks so much 2011-11-29T05:09:23 2207 stingis on one file, ideotistic opesource 2011-11-29T05:10:47 jast delete this sthing 2011-11-29T05:11:54 *** Zao has left #aichallenge 2011-11-29T05:14:03 *** b0rder has quit IRC (Quit: 离开) 2011-11-29T05:14:05 *** Rav3nsW0rd has quit IRC (Read error: Connection reset by peer) 2011-11-29T05:14:26 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T05:22:05 *** Rav3nsW0rd has quit IRC (Max SendQ exceeded) 2011-11-29T05:22:20 Ashoka: yep, been working on a bot last two weeks or so 2011-11-29T05:22:30 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T05:28:03 *** foRei has joined #aichallenge 2011-11-29T05:33:36 antimatroid : good to hear. 2011-11-29T05:36:57 *** NotABug has joined #aichallenge 2011-11-29T05:40:13 *** speakeasy has joined #aichallenge 2011-11-29T05:43:49 *** ikaros has quit IRC (Ping timeout: 245 seconds) 2011-11-29T05:44:31 *** speakeasy has quit IRC (Ping timeout: 248 seconds) 2011-11-29T05:52:46 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-29T05:56:47 *** sorki_sleeps is now known as sorki 2011-11-29T05:59:59 *** jon1 has joined #aichallenge 2011-11-29T06:00:18 *** liberforce has joined #aichallenge 2011-11-29T06:08:23 *** smiley1983 has quit IRC (Ping timeout: 244 seconds) 2011-11-29T06:09:20 *** smiley1983 has joined #aichallenge 2011-11-29T06:14:37 *** jon1 has quit IRC (Quit: Leaving.) 2011-11-29T06:16:39 *** smiley1983 has quit IRC (Ping timeout: 244 seconds) 2011-11-29T06:17:15 *** ikaros has joined #aichallenge 2011-11-29T06:17:38 *** smiley1983 has joined #aichallenge 2011-11-29T06:18:55 *** shifu has joined #aichallenge 2011-11-29T06:24:52 *** Conorach has joined #aichallenge 2011-11-29T06:26:51 *** Blkt has joined #aichallenge 2011-11-29T06:52:20 *** AntDroid has joined #aichallenge 2011-11-29T07:02:12 *** liberforce has quit IRC (Quit: Leaving.) 2011-11-29T07:12:48 *** Areks has quit IRC (Read error: Connection reset by peer) 2011-11-29T07:20:58 for historical reasons, one match I was looking forward to : http://aichallenge.org/visualizer.php?game=139943&user=66 :-) 2011-11-29T07:35:56 *** olexs has joined #aichallenge 2011-11-29T07:48:51 *** Ashoka has quit IRC (Quit: Page closed) 2011-11-29T07:50:28 *** goffrie_ has joined #aichallenge 2011-11-29T07:50:37 *** goffrie has quit IRC (Disconnected by services) 2011-11-29T07:54:44 *** liberforce has joined #aichallenge 2011-11-29T07:55:01 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-29T07:56:47 *** smiley1983 has quit IRC (Ping timeout: 248 seconds) 2011-11-29T07:57:30 *** smiley1983 has joined #aichallenge 2011-11-29T08:00:19 *** u_ has joined #aichallenge 2011-11-29T08:01:57 *** u__ has joined #aichallenge 2011-11-29T08:04:46 *** u_ has quit IRC (Ping timeout: 260 seconds) 2011-11-29T08:04:46 *** u__ is now known as u_ 2011-11-29T08:07:24 *** delt0r_ has joined #aichallenge 2011-11-29T08:07:55 *** mviel_ has joined #aichallenge 2011-11-29T08:10:59 *** mviel__ has quit IRC (Ping timeout: 252 seconds) 2011-11-29T08:16:01 *** jon has joined #aichallenge 2011-11-29T08:21:29 *** akarukov has joined #aichallenge 2011-11-29T08:23:00 *** akarukov has quit IRC (Client Quit) 2011-11-29T08:25:34 *** jon1 has joined #aichallenge 2011-11-29T08:25:34 *** jon has quit IRC (Read error: Connection reset by peer) 2011-11-29T08:25:39 *** jon1 has quit IRC (Client Quit) 2011-11-29T08:29:18 *** avdg has joined #aichallenge 2011-11-29T08:32:44 *** u_ has quit IRC (Quit: u_) 2011-11-29T08:35:25 *** u_ has joined #aichallenge 2011-11-29T08:36:18 *** rwest_ has left #aichallenge 2011-11-29T08:36:45 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-29T08:37:01 quiet night 2011-11-29T08:37:04 everyone busy programming? :P 2011-11-29T08:37:27 it's the middle of the day where I am :) 2011-11-29T08:38:32 it's 12:38am here :P 2011-11-29T08:39:04 and summer :D 2011-11-29T08:39:05 8:39am :) 2011-11-29T08:39:12 i rode to a friends in a shirt before 2011-11-29T08:39:58 *turns head to window* 4°C outside right now, and raining... 2011-11-29T08:40:17 yeah, raining here too 2011-11-29T08:40:24 6C :( 2011-11-29T08:40:55 27C in my room but I have ants to thank for that... 2011-11-29T08:41:24 using the cpu instead of central heating? :) 2011-11-29T08:41:42 pretty much, have the air vent in my bedroom closed 2011-11-29T08:42:06 :p 2011-11-29T08:42:19 hehe 2011-11-29T08:42:27 i haven't raped my cpu too much yet 2011-11-29T08:42:30 planet wars did 2011-11-29T08:42:34 *** smiley1983 has quit IRC (Ping timeout: 258 seconds) 2011-11-29T08:43:16 I'm running 5 bots on tcp :P 2011-11-29T08:43:53 I'm currently running one on tcpants and one on fluxid, avg cpu load is <10%... 2011-11-29T08:44:17 I guess an i7-860 is good for something 2011-11-29T08:44:34 *** smiley1983 has joined #aichallenge 2011-11-29T08:45:12 load average: 0.16, 0.17, 0.24 2011-11-29T08:45:23 not that bad, but it's just heated up the room over time 2011-11-29T08:45:34 and that heats your room? is that an AMD by any chance? :) 2011-11-29T08:45:41 < antimatroid> quiet night ← i'm busy at work ;P 2011-11-29T08:45:49 olexs: nope 2011-11-29T08:45:56 quad-core sandy bridge in this desktop 2011-11-29T08:46:06 i'm getting my bot back to what it was before i tried to generalise path finding 2011-11-29T08:47:11 what are others thoughts on terranova? 2011-11-29T08:48:44 *** TheLinker has joined #aichallenge 2011-11-29T08:53:19 *** TheLinker has quit IRC (Ping timeout: 248 seconds) 2011-11-29T08:56:05 *** ThiagoRRamos has joined #aichallenge 2011-11-29T08:56:20 *** ThiagoRRamos has joined #aichallenge 2011-11-29T08:56:34 Hi 2011-11-29T08:57:41 I looked for this information in the game specifications, but I didn't find it 2011-11-29T08:58:03 what kind of information? 2011-11-29T08:58:10 If an ant moves to a location where there is another ant 2011-11-29T08:58:22 they just collide 2011-11-29T08:58:28 unless the other ant moves too 2011-11-29T08:58:36 but this second ant is moving to somewhere else 2011-11-29T08:58:40 and collision means that both ants are killed 2011-11-29T08:58:41 do they collide? 2011-11-29T08:58:50 ThiagoRRamos: no, if the other ant moves too they won't collide 2011-11-29T08:59:03 well, unless the other ant gets moved as wel 2011-11-29T08:59:13 does it depends in what other my bot issues the movements? 2011-11-29T08:59:17 on what* 2011-11-29T08:59:17 so you can swap 2 ants from place without collision 2011-11-29T08:59:49 just be sure you don't allocate 2 ants on the same tile ;-) 2011-11-29T08:59:54 *** TheLinker has joined #aichallenge 2011-11-29T09:00:06 Hmm 2011-11-29T09:00:07 *** Israfel has quit IRC (Quit: Bye!) 2011-11-29T09:00:07 (dunno from where that word is coming from) 2011-11-29T09:00:08 Ok 2011-11-29T09:00:20 Thanks, avdg 2011-11-29T09:00:46 Then there is really something wrong in my code :P 2011-11-29T09:01:09 oh we all have bugs, I guess 2011-11-29T09:01:18 * avdg is doing pointer math 2011-11-29T09:01:57 Yeah 2011-11-29T09:02:15 But this one I thought I had already checked :P 2011-11-29T09:02:20 *** goffrie_ has quit IRC (Ping timeout: 244 seconds) 2011-11-29T09:02:25 *** goffrie_ has joined #aichallenge 2011-11-29T09:02:26 *** goffrie_ is now known as goffrie 2011-11-29T09:08:58 *** bob__ has joined #aichallenge 2011-11-29T09:13:01 Yeah, I found it :D :P 2011-11-29T09:13:29 *** praveen has joined #aichallenge 2011-11-29T09:14:19 *** NotABug has quit IRC (Ping timeout: 265 seconds) 2011-11-29T09:20:54 ThiagoRRamos: this game had horrible collision rules: http://queue.acm.org/icpc/ 2011-11-29T09:21:49 *** avdg has quit IRC (Read error: Connection timed out) 2011-11-29T09:22:34 *** avdg has joined #aichallenge 2011-11-29T09:23:14 antimatroid: what are they? doesn't seem to be much info posted there 2011-11-29T09:23:28 oh, http://queue.acm.org/icpc/2011/main.html this? 2011-11-29T09:23:49 yeah 2011-11-29T09:24:30 http://queue.acm.org/icpc/2011/rules.html 2011-11-29T09:24:36 it begins to explain it at the bottom of that page 2011-11-29T09:25:01 after playing around for a bit and seeing just how unintuitive when two things did or did not collide i gave up on that gmae 2011-11-29T09:25:46 sounds weird 2011-11-29T09:25:59 so you set paths and then they travel on them til they hit something? 2011-11-29T09:26:23 *** smiley1983 has quit IRC (Ping timeout: 248 seconds) 2011-11-29T09:26:36 *** smiley1983 has joined #aichallenge 2011-11-29T09:27:04 they did have an awesome visualiser though 2011-11-29T09:27:20 i want to see a 3d ants one with a torus i can shift around :P 2011-11-29T09:27:26 although the current one is awesome too 2011-11-29T09:29:22 I fixed a few bugs in my bot and now it's so much better than it was 2011-11-29T09:29:26 http://ants.fluxid.pl/replay.4120 2011-11-29T09:29:58 I really need to write tests... 2011-11-29T09:30:15 *** treeform has joined #aichallenge 2011-11-29T09:30:22 *** analyst74 has quit IRC (*.net *.split) 2011-11-29T09:30:22 *** Accoun has quit IRC (*.net *.split) 2011-11-29T09:30:22 *** JamesMG has quit IRC (*.net *.split) 2011-11-29T09:30:22 *** pairofdice has quit IRC (*.net *.split) 2011-11-29T09:30:22 *** sorki has quit IRC (*.net *.split) 2011-11-29T09:30:22 *** krishna has quit IRC (*.net *.split) 2011-11-29T09:31:26 *** bob__ has quit IRC (Quit: Page closed) 2011-11-29T09:33:08 I really wish the tcpservers gave more info back to the client at the end of a match 2011-11-29T09:34:50 rwest: you mean scores? 2011-11-29T09:34:55 yea 2011-11-29T09:34:59 for every bot 2011-11-29T09:35:00 i fixed a client some time ago 2011-11-29T09:35:05 it should do it now 2011-11-29T09:35:09 redownload client 2011-11-29T09:35:52 at least it should be fixed on my server, dunno what about tcpants.com 2011-11-29T09:36:08 but fix went to break's upstream 2011-11-29T09:36:18 *** analyst74 has joined #aichallenge 2011-11-29T09:36:18 *** Accoun has joined #aichallenge 2011-11-29T09:36:18 *** JamesMG has joined #aichallenge 2011-11-29T09:36:19 *** pairofdice has joined #aichallenge 2011-11-29T09:36:19 *** krishna has joined #aichallenge 2011-11-29T09:36:19 *** sorki has joined #aichallenge 2011-11-29T09:36:46 I had the latest 2011-11-29T09:37:00 I meant, I would like to see more details 2011-11-29T09:37:16 like scores from ever bot 2011-11-29T09:37:20 every 2011-11-29T09:37:47 http://ants.fluxid.pl/replay.4141 lol. 2011-11-29T09:37:50 look at the end of this game 2011-11-29T09:37:52 >.< 2011-11-29T09:39:09 LOL 2011-11-29T09:39:29 rwest: ah, this way. dunno, client gets what engine sends to it, nothing else is added 2011-11-29T09:40:03 thestinger: haha 2011-11-29T09:40:04 thestinger: LOL 2011-11-29T09:40:08 roflmao 2011-11-29T09:40:34 3 turns away 2011-11-29T09:40:59 does the engine not give scores? 2011-11-29T09:41:07 i added it to the specs 2011-11-29T09:41:12 and input etc. 2011-11-29T09:41:17 but i didn't actually write the engine 2011-11-29T09:42:09 I will have to train it locally :-/ 2011-11-29T09:42:17 I wish I could train it against real opponents 2011-11-29T09:42:33 but I need to know the maximum scores to be able to weight the scores right 2011-11-29T09:42:46 thestinger: if strcat_next sees enemy near hill it attacks it? what made strcat_next stand there? 2011-11-29T09:43:00 Fluxid: yeah 2011-11-29T09:43:11 it was blocking the enemy from taking the hill 2011-11-29T09:43:24 my defense code is kinda buggy 2011-11-29T09:43:35 it uses all the ants to defense for some reason, even though that's not what I tried to code 2011-11-29T09:43:39 to defend* 2011-11-29T09:43:44 :) 2011-11-29T09:43:49 http://ants.fluxid.pl/replay.4148 need to improve my defense here. 2011-11-29T09:44:05 sometimes it doesn't use all the ants to defend 2011-11-29T09:44:10 * thestinger sighs 2011-11-29T09:44:38 my bot is so inconsistant 2011-11-29T09:45:48 my bot is approaching the level where I wish it was written in another language 2011-11-29T09:46:27 I reached that point 2011-11-29T09:46:32 and then I realized C++11 had lambdas :P 2011-11-29T09:47:11 I'm seriously incapable of code reuse via classes :\ 2011-11-29T09:47:25 AntDroid: haha what's green doing 2011-11-29T09:47:26 this is the first time I have used Java 2011-11-29T09:47:36 and I really wish I used C/C++ 2011-11-29T09:47:39 wtb pointers 2011-11-29T09:48:01 c++ <3 2011-11-29T09:48:11 c++ is popular again for ants 2011-11-29T09:48:18 it wasn't so popular for planet wars :\ 2011-11-29T09:48:34 people want the performance 2011-11-29T09:48:52 I would never have done this in Java if I was using A* 2011-11-29T09:48:57 or bfs 2011-11-29T09:48:58 I've got a strategy that I think is gonna be stupidly fast 2011-11-29T09:49:10 sf17k: my bot X4 just watch its hill being massacred 2011-11-29T09:49:16 i use a* to collect food but that's it 2011-11-29T09:49:38 bfs all the other things 2011-11-29T09:49:56 my running time is (rows * cols) ^ 2 2011-11-29T09:49:59 AntDroid: he was just teasing you by spawnkilling your hill lol 2011-11-29T09:50:01 always 2011-11-29T09:50:04 based on rather 2011-11-29T09:50:18 and thats wrong lol 2011-11-29T09:50:23 my time is always 4ms before combat on 200x200 maps now 2011-11-29T09:50:41 I should add a check for my time to see it 2011-11-29T09:50:42 doesn't matter how many ants I have, but it does go up to ~10ms if I bot times out and the map fills up with food 2011-11-29T09:51:05 if a bot times out* 2011-11-29T09:51:11 *** shifu has quit IRC (Quit: Lost terminal) 2011-11-29T09:51:12 rwest: why not in java? You know best formaer xathis uses it. 2011-11-29T09:51:14 figuring out a 6 vs 7 combat situation takes time :( 2011-11-29T09:51:33 besh: because I am not as efficient as others hah 2011-11-29T09:51:43 and I am too lazy to optimize 2011-11-29T09:52:01 I am setting up JGAP right now to get my weights for me 2011-11-29T09:52:12 because I am sick of trying to figure this out 2011-11-29T09:52:46 *** Rav3nsW0rd has quit IRC (Excess Flood) 2011-11-29T09:53:09 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T09:54:04 and I know my problems should just be an issue of incorrect weights just by the gains I get when I get a new one right 2011-11-29T09:57:15 *** sigh has quit IRC (Remote host closed the connection) 2011-11-29T10:00:13 the stinger: 4ms for the whole move ? 2011-11-29T10:00:57 *** Rav3nsW0rd has quit IRC (Max SendQ exceeded) 2011-11-29T10:01:00 that's unlikely :) 2011-11-29T10:01:00 depends on language I guess 2011-11-29T10:01:17 with c++ you can compile code thats better optimized with the hardware 2011-11-29T10:01:24 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T10:01:37 but who knows what he does "before combat" 2011-11-29T10:02:32 what would you do with the rest of the time :) 2011-11-29T10:02:32 any tutorial or starting points for combat(attack/defense)? 2011-11-29T10:03:08 praveen: http://aichallenge.org/specification_battle.php 2011-11-29T10:03:24 you can turn the math/pseudo-code/python at the bottom into a function for evaluating combat 2011-11-29T10:03:36 besh: I would update the internal state. the path finder, the symmetry detector and so on 2011-11-29T10:03:51 Ya I understood the spec, but expecting starting points for algorithms which people use 2011-11-29T10:03:52 My bot keeps on usuing A* for unseen locations until it runs out of time. Gotta come up with some better ideas 2011-11-29T10:04:26 it takes me 1ms to do a multi-source BFS from every unseen location across the map 2011-11-29T10:04:27 besh: do you cache a path once it is found 2011-11-29T10:04:44 thestinger: I remember you were suggesting some link about game trees - Can you point me to that now 2011-11-29T10:04:52 No. I wanted to ask about it here though. How do you do it ? 2011-11-29T10:05:01 praveen: that wasn't me :P 2011-11-29T10:05:04 the ant using it can keep it until you find a water tile on the path 2011-11-29T10:05:18 finally decided to look at how much time I use 2011-11-29T10:05:28 160ms max per turn 2011-11-29T10:05:29 thestinger: some tron tutorial - I could not find that 2011-11-29T10:05:40 avg 60 2011-11-29T10:06:10 make that 50ms 2011-11-29T10:06:13 besh: so I would store a path object with the ant and also keep all the paths in a list, so i can go through all of them when i find a new water tile 2011-11-29T10:06:35 if a path is blocked it is recalculated 2011-11-29T10:07:34 water tiles may get dropped in the middle of the game, so you recalculate i guess 2011-11-29T10:08:39 what I I just use a hash table with a key (source,destination) and value of (direcetion) 2011-11-29T10:08:49 what if* 2011-11-29T10:09:18 you can do that, the memory is enough 2011-11-29T10:11:09 But when water gets dropped, it can become messy 2011-11-29T10:12:19 the whole path is needed to see if the path is affected so it may not work. 2011-11-29T10:12:31 probably just use a map (as in 2d array in the size of the board) 2011-11-29T10:13:26 each cell contains a pointer to another map with the directions (if it isn't a water tile) 2011-11-29T10:14:19 new water tiles set the pointer to null and you would have to recalculate all the paths that went through that tile previously 2011-11-29T10:14:36 or something like that ^^ 2011-11-29T10:15:02 how many maps are needed for that ? 2011-11-29T10:15:22 anyway 2 bits * 200 * 200 * 200 * 200 = 381 MB 2011-11-29T10:15:32 how many targets do you have at a time? 2011-11-29T10:16:02 I just use distance maps (usually with multiple targets per map) 2011-11-29T10:16:07 besh: better use a BFS. i looks like it will work pretty well 2011-11-29T10:16:35 distance map calculated through BFS ;) 2011-11-29T10:16:37 http://www.sifflez.org/misc/tronbot/index.html 2011-11-29T10:16:40 I read about that in the forum. 2011-11-29T10:16:40 that's the game tree link 2011-11-29T10:16:52 also my planet wars bot had a very broken game tree 2011-11-29T10:17:12 besh: http://sprunge.us/bQZU little example 2011-11-29T10:17:47 8x8 one with a target at row 1, col 3 and another at row 5, col 0 2011-11-29T10:18:00 antimatroid: Thanks for that 2011-11-29T10:18:25 *** treeform has quit IRC (Remote host closed the connection) 2011-11-29T10:18:39 thestinger: I use a multi source A* and it works well. About 150ms per turn under "normal" circumstances 2011-11-29T10:18:45 that's a lot 2011-11-29T10:18:58 for how many ants/targets? 2011-11-29T10:19:13 *** treeform has joined #aichallenge 2011-11-29T10:19:25 haha.. will never be able to make it 4ms thats for sure 2011-11-29T10:19:37 what does multi source mean? 2011-11-29T10:19:41 besh: do you use multi targets? 2011-11-29T10:19:54 mleise: I use a multi-source BFS for generating the distance maps 2011-11-29T10:19:59 mleise: add multiple sources to the search queue 2011-11-29T10:20:09 yeah, that ^ 2011-11-29T10:20:14 multi target requires a different heuristic function 2011-11-29T10:20:20 for food search , I start from all ants at once to the specific food target. 2011-11-29T10:20:31 i do all ants to all food with a* 2011-11-29T10:20:42 the iteratively collect food 2011-11-29T10:20:47 antimatroid: ah, with all ants as sources and all food as targets? 2011-11-29T10:20:52 until they're all taken 2011-11-29T10:20:54 and reuse an ant if it's not going to battle an enemy on its path 2011-11-29T10:20:57 but... A* can only handle one source and multiple targets if you check them all in the exit condition 2011-11-29T10:21:01 thestinger: yeah 2011-11-29T10:21:02 damn, I rebuilt my bot and it synced the files via dropbox and now my tcpclient at home is timing out over and over lol 2011-11-29T10:21:10 mleise: not true :P 2011-11-29T10:21:17 i don't think 2011-11-29T10:21:51 besh: well, for attacking an enemy hill - do you do A* for each ant to the hill? 2011-11-29T10:22:04 you can do a single BFS from a hill (or all enemy hills) and make a distance map 2011-11-29T10:22:16 mleise; right, just invert the whole process. The problem is with multi-source and multi-target 2011-11-29T10:22:17 then you know the distance and shortest paths from every tile on the map to the target(s) 2011-11-29T10:22:20 *** tobym_ has joined #aichallenge 2011-11-29T10:22:26 if i turned it around and search from food to ants, it works as well, but i cannot see how to search in one go from multiple sources to multiple targets 2011-11-29T10:22:45 an ants path length starts as their available time (0 if unmoved), then add those locations to the search queue, then hdist(x) = min manhatten distance from x to a food location 2011-11-29T10:22:52 thestinger: yes I use A* for everything as long as time allows it. 2011-11-29T10:23:02 http://tcpants.com/replay.11432 <-- no A*, no bfs, no pathfinding whatsoever (rossxwest) 2011-11-29T10:23:13 and if you reuse an ant their availability is the path length to their target square (adjacent to food in this case) 2011-11-29T10:23:15 no battle code 2011-11-29T10:23:49 where x is a location in the map that you want a minimum path distance to a target for 2011-11-29T10:24:00 rwest: ok, that map doesn't need pathfinding ;) 2011-11-29T10:24:57 but it is still remarkable that you won 2011-11-29T10:25:02 mleise: there is the floyd-warshal algorithm people talk about. Don't know if it is worith implementing it. 2011-11-29T10:26:48 mleise: http://ants.fluxid.pl/replay.3893 betteR? 2011-11-29T10:27:11 rwest: do you use diffusion ? 2011-11-29T10:27:17 yesw 2011-11-29T10:28:11 I wonder if I can put that on top of A* to use when I run out of time 2011-11-29T10:28:31 do the distance map thing :P 2011-11-29T10:28:50 besh uses around 50ms and I diffuse (rows + cols)/2 times 2011-11-29T10:29:13 150ms 2011-11-29T10:30:07 I have a lot of weird things that get weighted 2011-11-29T10:30:20 is anyone here using floyd warshall? 2011-11-29T10:30:20 after I add a couple more 2011-11-29T10:30:20 thestinger: I never timeout before i send ants to food or hills. So its use is sending ants to unseen locations. 2011-11-29T10:30:29 i wonder how fast that is, it sounds rather useful! 2011-11-29T10:30:42 I am gonna train it to find good weights 2011-11-29T10:30:54 antimatroid: iirc, a1k0n was talking about it and decided it would use too much memory 2011-11-29T10:31:10 *** sf17k has quit IRC (Ping timeout: 265 seconds) 2011-11-29T10:32:11 thestinger: http://en.wikipedia.org/wiki/Floyd%E2%80%93Warshall_algorithm#Path_reconstruction even with that? 2011-11-29T10:32:49 *** g0llum has joined #aichallenge 2011-11-29T10:33:06 you need a 200x200 map for every square though, right? 2011-11-29T10:33:25 assuming 200 rows and 200 cols 2011-11-29T10:33:29 dunno, it says "Therefore, information to reconstruct all paths can be stored in a single N×N matrix" 2011-11-29T10:33:44 I think it reconstructs them from distances to every tile from every tile 2011-11-29T10:33:54 so you need 200^4 integers 2011-11-29T10:34:26 @calc 200 ** 4 * 4 / 1000 / 1000 2011-11-29T10:34:26 thestinger: Error: The command "calc" is available in the Google and Math plugins. Please specify the plugin whose command you wish to call by using its name as a command before "calc". 2011-11-29T10:34:31 @Math calc 200 ** 4 * 4 / 1000 / 1000 2011-11-29T10:34:31 thestinger: 6400 2011-11-29T10:34:41 that's not that bad? 2011-11-29T10:34:48 6.4GB 2011-11-29T10:34:53 oh :P 2011-11-29T10:35:06 lol 2011-11-29T10:35:06 if I'm thinking about it correctly 2011-11-29T10:35:09 I think that's what it does though 2011-11-29T10:35:47 rwest: your ants seem to avoid water. you do *some* sort of obstacle avoidance 2011-11-29T10:36:29 nope 2011-11-29T10:36:35 I just don't diffuse over water 2011-11-29T10:36:45 antimatroid: if you use int16_t that would get you down to 3.2GB :P 2011-11-29T10:36:46 water blocks the values from spreading 2011-11-29T10:36:51 interesting. they look more intelligent 2011-11-29T10:38:00 avoid enemy ants unless I outnumber them, group to ant near enemy 2011-11-29T10:39:55 it tends to do better on my open maps, but I am assuming I don't diffuse enough for mazed 2011-11-29T10:39:58 mazes 2011-11-29T10:40:08 I mat add code to recognize a maze 2011-11-29T10:40:13 and up the diffusions 2011-11-29T10:40:22 s/mat/may 2011-11-29T10:41:26 @math convert 6400 MB to GB 2011-11-29T10:41:26 mleise: 6.25 2011-11-29T10:41:34 cool stuff, supybot! 2011-11-29T10:42:07 @math convert 3 inch to cm 2011-11-29T10:42:07 mleise: 7.62 2011-11-29T10:43:40 @math convert 3m to lightyear 2011-11-29T10:43:40 pairofdice: Error: 3m is not a valid unit. 2011-11-29T10:44:00 @math convert 3 m to lightyear 2011-11-29T10:44:00 pairofdice: 0 2011-11-29T10:44:12 Bad accuracy 2011-11-29T10:44:19 lol 2011-11-29T10:44:38 @math convert 3000000 m to lightyear 2011-11-29T10:44:38 rwest: 3.17100250207e-10 2011-11-29T10:46:17 @eightball should I merge my combat branch now? 2011-11-29T10:46:17 thestinger: One would be wise to think so. 2011-11-29T10:47:07 supybot can make all my decisions for me now :) 2011-11-29T10:47:29 @eighball Should I take a shower? 2011-11-29T10:47:30 pairofdice: I'm sorry Dave, err pairofdice; I cannot 'eighball'. 2011-11-29T10:47:35 @eightball Should I take a shower? 2011-11-29T10:47:35 pairofdice: Come again? 2011-11-29T10:48:06 Hmph 2011-11-29T10:50:29 *** olexs has quit IRC (Ping timeout: 252 seconds) 2011-11-29T10:50:36 *** olexs has joined #aichallenge 2011-11-29T10:52:33 owned 2011-11-29T10:54:47 @math convert nanolightyear to attoparsec per microfortnight 2011-11-29T10:54:47 Minthos: Error: nanolightyear is not a valid unit. 2011-11-29T10:54:59 @math convert nc to attoparsec per microfortnight 2011-11-29T10:54:59 Minthos: Error: nc is not a valid unit. 2011-11-29T10:55:13 ah crap 2011-11-29T10:55:20 @math convert c to attoparsec per microfortnight 2011-11-29T10:55:21 Minthos: Error: c is not a valid unit. 2011-11-29T10:55:28 @math convert lightspeed to attoparsec per microfortnight 2011-11-29T10:55:28 Minthos: Error: attoparsecpermicrofortnight is not a valid unit. 2011-11-29T10:55:50 @math convert nanolightspeed to furlong per fortnight 2011-11-29T10:55:50 Minthos: Error: nanolightspeed is not a valid unit. 2011-11-29T10:55:55 @match convert ly to attoparsec 2011-11-29T10:55:55 p_l: Error: 'convert' is not a valid integer. 2011-11-29T10:56:03 @math convert ly to attoparsec 2011-11-29T10:56:03 p_l: Error: ly is not a valid unit. 2011-11-29T10:56:06 heh 2011-11-29T10:56:16 @help convert 2011-11-29T10:56:16 thestinger: (convert [] to ) -- Converts from to . If number isn't given, it defaults to 1. For unit information, see 'units' command. 2011-11-29T10:56:19 @units distance 2011-11-29T10:56:19 mleise: valid types: mass, length, time, scheduling, temperature, temp. diff., current, charge, potential, resistance, conductance, capacitance, magn. flux, inductance, flux density, molecular qty, size of a mol, lum. intens., luminous flux, illuminance, luminance, angle, solid angle, data, data transfer, quantity, interest rate, concentration, force, area, volume, velocity, rot. velocity, fluid (1 more message) 2011-11-29T10:56:30 @math convert mass to time 2011-11-29T10:56:30 Minthos: Error: mass is not a valid unit. 2011-11-29T10:56:40 @units length 2011-11-29T10:56:40 p_l: length units: m, meter, metre, decimeter, cm, centimeter, mm, millimeter, micrometer, micron, nanometer, nm, dekameter, hectometer, km, kilometer, megameter, angstrom, fermi, inch, in, inches, mil, microinch, microinches, foot, ft, feet, yard, yd, mile, mi, nautical mile, nmi, league, chain, fathom, rod, furlong, hand, cubit, point, pica, caliber, football field, marathon, mil Swedish, au, (1 more message) 2011-11-29T10:56:45 @more 2011-11-29T10:56:45 thestinger: [syn: {line roulette}] 2: a wheel with teeth for making a row of perforations [syn: {toothed wheel}] 3: a gambling game in which players bet on which compartment of a revolving wheel a small ball will come to rest in; moby-thes: 24 Moby Thesaurus words for "roulette": bingo, card games, chuck and toss, chuck farthing, chuck-a-luck, crack-loo, crap game, crap shooting, craps, fan-tan, (3 more messages) 2011-11-29T10:56:47 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-29T10:56:51 what? 2011-11-29T10:56:57 rofl 2011-11-29T10:56:58 @more 2011-11-29T10:56:58 p_l: astronomical unit, light year, light minute, light second, parsec, kiloparsec, megaparsec, screw size, AWG, American Wire Gauge, standard gauge, zinc gauge, ring size, shoe size mens, shoe size womens 2011-11-29T10:57:14 this thing should send PMs 2011-11-29T10:57:14 with help 2011-11-29T10:57:20 @math convert light year to parsec 2011-11-29T10:57:21 p_l: Error: light is not a valid unit. 2011-11-29T10:57:35 @math convert 1 light year to parsec 2011-11-29T10:57:35 p_l: Error: light is not a valid unit. 2011-11-29T10:57:41 heh 2011-11-29T10:57:44 @math convert parsec to shoesizewomens 2011-11-29T10:57:44 Minthos: 3644501080275590144 2011-11-29T10:57:51 hahahahaha 2011-11-29T10:57:57 hahaaa good one 2011-11-29T10:58:12 @math convert 1 lightyear to parsec 2011-11-29T10:58:12 p_l: 0.306601393805 2011-11-29T10:58:24 @math convert 1 lightsecond to nauticalmile 2011-11-29T10:58:24 p_l: 161874.977322 2011-11-29T10:58:56 @ math convert 1 lightyear to footballfield 2011-11-29T10:58:57 rwest: 1.03463806568e+14 2011-11-29T10:59:09 *** treeform has quit IRC (Remote host closed the connection) 2011-11-29T10:59:39 @math convert 1 foot to hand 2011-11-29T10:59:39 rwest: 3 2011-11-29T10:59:44 rofl 2011-11-29T11:00:55 @convert township to barn 2011-11-29T11:00:55 mleise: 932395719720959924276838429351739392 2011-11-29T11:01:09 seriously? 2011-11-29T11:01:29 @math convert 1 km to fathom 2011-11-29T11:01:30 pairofdice: 546.806649169 2011-11-29T11:02:24 barn is the size of an electron or something ? 2011-11-29T11:02:39 @url stats 2011-11-29T11:02:39 thestinger: I have 11967 URLS in my database. 2011-11-29T11:02:44 :0 2011-11-29T11:03:01 not sure what the point of that plugin is... 2011-11-29T11:03:09 @math convert chain to rod 2011-11-29T11:03:09 rwest: 4 2011-11-29T11:03:16 @list url 2011-11-29T11:03:16 thestinger: last and stats 2011-11-29T11:03:20 @help last 2011-11-29T11:03:20 thestinger: (last [--{from,in,on,with,without,regexp} ] [--nolimit]) -- Returns the last message matching the given criteria. --from requires a nick from whom the message came; --in requires a channel the message was sent to; --on requires a network the message was sent on; --with requires some string that had to be in the message; --regexp requires a regular expression the message must (1 more message) 2011-11-29T11:03:35 oh. 2011-11-29T11:04:30 /^1?$|^(11+?)\1+$/ 2011-11-29T11:04:47 Best regexp ever: /^1?$|^(11+?)\1+$/ 2011-11-29T11:04:55 determines if a number is prime 2011-11-29T11:05:58 *** praveen has quit IRC (Ping timeout: 265 seconds) 2011-11-29T11:06:14 *** besh has quit IRC (Quit: Page closed) 2011-11-29T11:06:32 *** TheLinker has quit IRC (Quit: Bye) 2011-11-29T11:06:56 *** besh has joined #aichallenge 2011-11-29T11:07:43 @match rossxwest killingtime 2011-11-29T11:07:43 mleise: Error: 'rossxwest' is not a valid integer. 2011-11-29T11:08:37 @match 12139 7526 2011-11-29T11:08:38 mleise: An error has occurred and has been logged. Please contact this bot's administrator for more information. 2011-11-29T11:08:42 !re ^1?$|^(11+?)\1+$ 3 2011-11-29T11:08:52 @re ^1?$|^(11+?)\1+$ 3 2011-11-29T11:08:52 thestinger: Error: '^1?$|^(11+?)\\1+$ 3' is not a valid regular expression. 2011-11-29T11:09:02 @re '^1?$|^(11+?)\1+$' 3 2011-11-29T11:09:03 thestinger: Error: I tried to send you an empty message. 2011-11-29T11:09:17 it doesnt work -.- 2011-11-29T11:09:22 @help re 2011-11-29T11:09:23 thestinger: (re ) -- If is of the form m/regexp/flags, returns the portion of that matches the regexp. If is of the form s/regexp/replacement/flags, returns the result of applying such a regexp to . 2011-11-29T11:09:56 @re s/^1?$|^(11+?)\1+$/ 3 2011-11-29T11:09:56 thestinger: Error: 's/^1?$|^(11+?)\\1+$/ 3' is not a valid regular expression. 2011-11-29T11:10:02 @re 's/^1?$|^(11+?)\1+$/' 3 2011-11-29T11:10:03 thestinger: Error: I tried to send you an empty message. 2011-11-29T11:11:18 perl -wle 'print "Prime" if (1 x shift) !~ /^1?$|^(11+?)\1+$/' 3 2011-11-29T11:11:22 better way to test it 2011-11-29T11:12:24 forgot the beginning ^^ 2011-11-29T11:12:36 *** rofer is now known as rowfur 2011-11-29T11:12:44 rwest: I tried it in python and it works there too 2011-11-29T11:12:45 *** rowfur is now known as rofer 2011-11-29T11:13:04 and you would think that supybot uses python's re module... 2011-11-29T11:13:20 well 2011-11-29T11:13:26 @re 's/^1?$|^(11+?)\1+$/' 111 2011-11-29T11:13:27 rwest: Error: I tried to send you an empty message. 2011-11-29T11:13:41 @re 'm/^1?$|^(11+?)\1+$/' 111 2011-11-29T11:13:42 rwest: Error: I tried to send you an empty message. 2011-11-29T11:13:47 @re 'm/^1?$|^(11+?)\1+$/111' 2011-11-29T11:13:47 rwest: (re ) -- If is of the form m/regexp/flags, returns the portion of that matches the regexp. If is of the form s/regexp/replacement/flags, returns the result of applying such a regexp to . 2011-11-29T11:14:21 @re m/^1?$|^(11+?)\1+$/ 11 2011-11-29T11:14:22 rwest: Error: I tried to send you an empty message. 2011-11-29T11:14:24 @re m/^1?$|^(11+?)\1+$/ 111 2011-11-29T11:14:26 rwest: Error: I tried to send you an empty message. 2011-11-29T11:14:28 fail 2011-11-29T11:15:49 rwest: prime number test regex? 2011-11-29T11:16:02 @re '^1?$|^(11+?)\\1+$' 3 2011-11-29T11:16:03 thestinger: Error: I tried to send you an empty message. 2011-11-29T11:16:16 >>> re.match('^1?$|^(11+?)\\1+$', '1'*10) #matches 2011-11-29T11:16:18 <_sre.SRE_Match object at 0x302b990> 2011-11-29T11:16:21 works in python 2011-11-29T11:16:24 oh well 2011-11-29T11:16:38 you need to create as many 1 as the number is, so 3 won't work 2011-11-29T11:16:51 I tried 111 2011-11-29T11:16:59 111 is 3 in unary 2011-11-29T11:17:04 oh, whoops 2011-11-29T11:17:07 @re ^1?$|^(11+?)\1+$ 111 2011-11-29T11:17:07 rwest: Error: '^1?$|^(11+?)\\1+$ 111' is not a valid regular expression. 2011-11-29T11:17:09 @re '^1?$|^(11+?)\\1+$' 111 2011-11-29T11:17:10 thestinger: Error: I tried to send you an empty message. 2011-11-29T11:17:13 @re /^1?$|^(11+?)\1+$/ 111 2011-11-29T11:17:14 rwest: Error: I tried to send you an empty message. 2011-11-29T11:17:16 @re '^1?$|^(11+?)\1+$' 111 2011-11-29T11:17:17 thestinger: Error: I tried to send you an empty message. 2011-11-29T11:17:20 nope :P 2011-11-29T11:17:28 @re /^1?$|^(11+?)\1+$/ 1111 2011-11-29T11:17:29 jix: 1111 2011-11-29T11:17:37 it matches composite and unit numbers 2011-11-29T11:17:41 so it matches all nonprime numbers 2011-11-29T11:17:45 lol 2011-11-29T11:17:50 >.< 2011-11-29T11:17:51 it it was ok 2011-11-29T11:18:20 @last --regexp /^1?$|^(11+?)\1+$/ 2011-11-29T11:18:20 Minthos: Error: I couldn't find a message matching that criteria in my history of 1000 messages. 2011-11-29T11:19:02 @last --regexp m/^1?$|^(11+?)\1+$/ 2011-11-29T11:19:02 Minthos: Error: I couldn't find a message matching that criteria in my history of 1000 messages. 2011-11-29T11:19:16 @re /^1?$|^(11+?)\1+$/ 111111 2011-11-29T11:19:17 rwest: 111111 2011-11-29T11:19:20 @re /^1?$|^(11+?)\1+$/ 11111 2011-11-29T11:19:21 rwest: Error: I tried to send you an empty message. 2011-11-29T11:19:26 @re /^1?$|^(11+?)\1+$/ 111111111 2011-11-29T11:19:27 rwest: 111111111 2011-11-29T11:19:40 @re /^1?$|^(11+?)\1+$/ 21 2011-11-29T11:19:41 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:19:43 @re /^1?$|^(11+?)\1+$/ 22 2011-11-29T11:19:44 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:19:54 @re /^1?$|^(11+?)\1+$/ 21*'1' 2011-11-29T11:19:55 rwest: Error: I tried to send you an empty message. 2011-11-29T11:20:45 @re '/^1?$|^(11+?)\1+$/' 7 2011-11-29T11:20:47 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:20:50 @re '/^1?$|^(11+?)\1+$/' 9 2011-11-29T11:20:51 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:20:54 @re '/^1?$|^(11+?)\1+$/' 1 2011-11-29T11:20:55 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:20:58 @re '/^1?$|^(11+?)\1+$/' 111 2011-11-29T11:20:59 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:20:59 Minthos: you need to do it in 1's 2011-11-29T11:21:02 @re '/^1?$|^(11+?)\1+$/' 1111 2011-11-29T11:21:03 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:21:08 *** galdor has left #aichallenge ("ERC Version 5.3 (IRC client for Emacs)") 2011-11-29T11:21:18 and no double delimeters 2011-11-29T11:21:21 remove quotes 2011-11-29T11:21:23 @re /^1?$|^(11+?)\1+$/ 11 2011-11-29T11:21:23 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:21:26 @re /^1?$|^(11+?)\1+$/ 111 2011-11-29T11:21:27 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:21:35 @re /^1?$|^(11+?)\1+$/ 1111111 2011-11-29T11:21:36 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:21:38 @re /^1?$|^(11+?)\1+$/ 1111 2011-11-29T11:21:39 rwest: 1111 2011-11-29T11:21:52 @re /^1?$|^(11+?)\1+$/ 1111 2011-11-29T11:21:54 Minthos: 1111 2011-11-29T11:21:56 *** JorgeB has joined #aichallenge 2011-11-29T11:21:57 *** treeform has joined #aichallenge 2011-11-29T11:22:10 @re /^1?$|^(11+?)\1+$/ 1111ii 2011-11-29T11:22:11 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:22:17 @re /^1?$|^(11+?)\1+$/ 1111iii 2011-11-29T11:22:18 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:22:22 @re /^1?$|^(11+?)\1+$/ 1111111 2011-11-29T11:22:23 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:22:25 @re /^1?$|^(11+?)\1+$/ 111111 2011-11-29T11:22:26 Minthos: 111111 2011-11-29T11:22:31 @re /^1?$|^(11+?)\1+$/ 7 2011-11-29T11:22:32 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:22:34 @re /^1?$|^(11+?)\1+$/ 6 2011-11-29T11:22:35 Minthos: Error: I tried to send you an empty message. 2011-11-29T11:22:40 :( 2011-11-29T11:23:23 @re m/^1?$|^(11+?)\1+$/ 11111111111 2011-11-29T11:23:24 mleise: Error: I tried to send you an empty message. 2011-11-29T11:23:34 @re m/^1?$|^(11+?)\1+$/ 111111111111 2011-11-29T11:23:35 mleise: 111111111111 2011-11-29T11:23:42 it works as expected 2011-11-29T11:23:51 12 is even, so i get it back 2011-11-29T11:23:51 *** smiley1983 has quit IRC (Read error: Operation timed out) 2011-11-29T11:26:20 @last /user=4823/ 2011-11-29T11:26:20 Minthos: (last [--{from,in,on,with,without,regexp} ] [--nolimit]) -- Returns the last message matching the given criteria. --from requires a nick from whom the message came; --in requires a channel the message was sent to; --on requires a network the message was sent on; --with requires some string that had to be in the message; --regexp requires a regular expression the message must match; (1 more message) 2011-11-29T11:26:42 *** smiley1983 has joined #aichallenge 2011-11-29T11:26:45 *** Antimony has joined #aichallenge 2011-11-29T11:28:34 < Minthos> @last --regexp /^1?$|^(11+?)\1+$/ ← there are ^ and $ so it tries to match whole message, not just numbers. you'd need to replace ^ and $ with something like (^|[^\d]) and {$|[^\d]) or just \s 2011-11-29T11:29:43 @last --regex /^1?($|\s)|(^|\s)(11+?)\1+($|\s)/ 2011-11-29T11:29:43 Fluxid: Error: I couldn't find a message matching that criteria in my history of 1000 messages. 2011-11-29T11:29:58 bah 2011-11-29T11:30:01 whatever 2011-11-29T11:31:02 @rainbow [error] 2011-11-29T11:31:03 thestinger: I have no idea what you mean. 2011-11-29T11:31:44 @last --regexp /.*^1?$|^(11+?)\1+$.*/ 2011-11-29T11:31:44 Minthos: Error: I couldn't find a message matching that criteria in my history of 1000 messages. 2011-11-29T11:31:51 @rainbow [apply utilities shuffle [error]] 2011-11-29T11:31:52 thestinger: 'error'. sorry thestinger; Dave, cannot I err I'm 2011-11-29T11:32:16 @last --regexp /.*/ 2011-11-29T11:32:16 Minthos: [11:31:51] @rainbow [apply utilities shuffle [error]] 2011-11-29T11:33:03 @last --regexp /.*4823.*/ 2011-11-29T11:33:03 Minthos: [11:26:20] @last /user=4823/ 2011-11-29T11:34:03 *** ruuhkis has joined #aichallenge 2011-11-29T11:35:03 wow, my ants really seem to make progress :3 I really need to work on defense now :( my poor ants don't got any idea that their home is about to be destroyed 2011-11-29T11:38:23 *** sofuture has quit IRC (Read error: Operation timed out) 2011-11-29T11:40:05 *** kire has joined #aichallenge 2011-11-29T11:41:19 *** smiley1983 has quit IRC (Ping timeout: 245 seconds) 2011-11-29T11:42:09 *** nickjohnson has quit IRC (Ping timeout: 244 seconds) 2011-11-29T11:42:10 *** smiley1983 has joined #aichallenge 2011-11-29T11:43:50 *** sofuture has joined #aichallenge 2011-11-29T11:47:53 *** nickjohnson has joined #aichallenge 2011-11-29T11:48:01 *** amstan has joined #aichallenge 2011-11-29T11:48:01 *** ChanServ sets mode: +o amstan 2011-11-29T11:50:21 *** kire has quit IRC (Remote host closed the connection) 2011-11-29T11:54:07 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-11-29T11:58:44 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-29T11:58:53 *** smiley1983 has quit IRC (Read error: Operation timed out) 2011-11-29T11:58:57 *** grwip has quit IRC (Remote host closed the connection) 2011-11-29T11:59:45 *** smiley1983 has joined #aichallenge 2011-11-29T12:00:31 *** Antimony has joined #aichallenge 2011-11-29T12:09:20 *** rofer is now known as rowfur 2011-11-29T12:10:05 *** mviel_ has quit IRC (Remote host closed the connection) 2011-11-29T12:11:37 amstan: can you sticky the tcp servers list on the forum? 2011-11-29T12:11:46 BenJackson: link? 2011-11-29T12:11:48 *** rowfur is now known as rofer 2011-11-29T12:11:59 http://forums.aichallenge.org/viewtopic.php?f=24&t=2007 2011-11-29T12:12:10 *** rofer is now known as rowfur 2011-11-29T12:12:51 BenJackson: hmm 2011-11-29T12:12:59 BenJackson: can you make it an intro topic instead? 2011-11-29T12:13:10 I didn't even post it, I just added tcpants 2011-11-29T12:13:19 darn, k 2011-11-29T12:13:30 *** rowfur is now known as rofer 2011-11-29T12:13:43 *** rofer is now known as rowfur 2011-11-29T12:15:16 why tcp server times me out ? :( 2011-11-29T12:15:43 ruuhkis: i don't know 2011-11-29T12:15:50 perhaps you should contact the tcp server admin 2011-11-29T12:15:53 how should i run java bot? 2011-11-29T12:15:55 ruuhkis: it does matchmaknig on a first come, first serve basis 2011-11-29T12:16:07 if a match is running when you connect I've seen it time out several times and reconnect 2011-11-29T12:16:19 the match starts 2011-11-29T12:16:23 but after first turn 2011-11-29T12:16:24 it times out :o 2011-11-29T12:16:27 (but when it time out it closes its connection to your bot which can cause the C++ starter to crash) 2011-11-29T12:16:46 it does start your bot pretty early, so I don't think it's startup 2011-11-29T12:16:58 i added debug on loop 2011-11-29T12:17:08 and it goes thru the loop once and then on second loop it qutis 2011-11-29T12:18:30 *** rowfur is now known as rofer 2011-11-29T12:18:54 *** besh has quit IRC (Quit: Page closed) 2011-11-29T12:21:57 http://ants.fluxid.pl/replay.4390 \o/ 2011-11-29T12:23:18 cant get my bot to run :((( 2011-11-29T12:24:39 woot it works :D 2011-11-29T12:25:01 *** treeform has quit IRC (Ping timeout: 240 seconds) 2011-11-29T12:26:07 oh man, my aichallenge submit had a bit of a senile moment there against rank 1000 bots 2011-11-29T12:26:10 arrr 2011-11-29T12:29:17 *** mj41 has quit IRC (Ping timeout: 255 seconds) 2011-11-29T12:29:53 BenJackson: is bj_v9 your best version atm? 2011-11-29T12:43:08 *** Antimony_ has joined #aichallenge 2011-11-29T12:45:19 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-11-29T12:45:33 *** Antimony_ is now known as Antimony 2011-11-29T12:48:07 *** liberforce has left #aichallenge 2011-11-29T12:52:40 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-29T12:54:07 *** Antimony has joined #aichallenge 2011-11-29T12:55:48 thestinger: yes, just re-up'd to aichallenge 2011-11-29T12:56:39 Nooo, my bot dropped from 130 to 175 2011-11-29T12:57:12 I want to make an on-the-fly brain transplant 2011-11-29T12:57:25 meh @ the official site :P 2011-11-29T12:57:38 bah my current version gets shit on by my previous version lol 2011-11-29T12:58:28 thestinger: v9 won 16 in a row on fluxid and had an untouchable score despite a high sigma 2011-11-29T12:58:37 I should have retired befoer you beat me in the 17th game 2011-11-29T12:58:49 haha 2011-11-29T12:59:04 *** Antimony_ has joined #aichallenge 2011-11-29T12:59:13 strcat_jedimindtricks and strcat_foodfight are doing well now that I fixed a dozen bugs 2011-11-29T12:59:18 i still wonder WHY my bot is 25th 2011-11-29T12:59:31 v9 is my "bugfix patch" release of v8 2011-11-29T12:59:48 my combat wasn't even working due to bugs 2011-11-29T13:00:25 but the debugging output was misleading so I thought it was working 2011-11-29T13:00:42 so it was disappointing that my new versions sucked compared to the old ones :P 2011-11-29T13:01:14 *** Antimony has quit IRC (Ping timeout: 255 seconds) 2011-11-29T13:01:15 *** Antimony_ is now known as Antimony 2011-11-29T13:01:27 my recent fluxid experience makes me think they will have to play a LOT of games to settle the ranks in the challenge 2011-11-29T13:01:37 yeah 2011-11-29T13:01:51 the two bots with combat were @ rank 50 with low sigma due to it being broken 2011-11-29T13:02:05 it has taken so long for them to climb up 2011-11-29T13:02:11 hopefully someone will do an analysis of the current data to see which setups produce meaningful output and which ones are random 2011-11-29T13:02:23 yeah my v8 only got to 60 2011-11-29T13:03:20 http://aichallenge.org/visualizer.php?game=144011&user=757 in this game tictac is more interested in running chasing my ants than exploring and doing "stuff" 2011-11-29T13:04:02 oops, lost a game to xathis on aichallenge 2011-11-29T13:04:09 but it is not bad, i need to implement chasing too 2011-11-29T13:07:24 I don't want to reupload to the main site because it amuses me that my old bot is still at rank 20 2011-11-29T13:08:08 it seems like you just can't drop in the rankings once you get up... 2011-11-29T13:08:36 haha i'm in 5th on aichallenge thanks to resubissions 2011-11-29T13:09:56 ugh version 2 of my bot beats version 4 in mazes 2011-11-29T13:10:08 but gets trounced in anything slightly open 2011-11-29T13:11:05 I read on the forums the tournament maps are gonna be closer tothe cell maze, so I guess it's not that bad 2011-11-29T13:11:22 they will be a mix of the different types 2011-11-29T13:11:27 all maps should be cell mazes :P 2011-11-29T13:15:02 ug, my bot gets stuck because too many ants clump together and it is unable to caluclate a movement set for them 2011-11-29T13:15:15 too many ants spoil the soup 2011-11-29T13:16:09 http://ants.fluxid.pl/replay.4461 my bots still do this :( 2011-11-29T13:17:01 I adjust the losses they're willing to take, but not much when they can only see 1/4 of the map 2011-11-29T13:19:43 *** Rav3nsW0rd has quit IRC (Excess Flood) 2011-11-29T13:20:07 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T13:25:02 *** goffrie_ has joined #aichallenge 2011-11-29T13:25:02 *** goffrie has quit IRC (Disconnected by services) 2011-11-29T13:25:03 *** goffrie_ is now known as goffrie 2011-11-29T13:28:24 hmm 2011-11-29T13:28:26 this is interesting 2011-11-29T13:29:00 that's a scary thing to say 2011-11-29T13:29:16 if you do sin(x+sin(2x)) you get a very good approx of a square wave 2011-11-29T13:29:41 better than sin(x)+1/3sin(3x)(second step of the furier series) 2011-11-29T13:29:59 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-11-29T13:34:56 *** Antimony has joined #aichallenge 2011-11-29T13:36:54 *** Apophis has joined #aichallenge 2011-11-29T13:37:30 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-11-29T13:39:20 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-11-29T13:39:53 *** Antimony has joined #aichallenge 2011-11-29T13:40:32 *** amstan has joined #aichallenge 2011-11-29T13:40:36 *** ChanServ sets mode: +o amstan 2011-11-29T13:40:52 *** treeform has joined #aichallenge 2011-11-29T13:40:55 *** jacob_strauss has joined #aichallenge 2011-11-29T13:43:22 hmm. The problem with reducing food gathering when you have more atns than enemies 2011-11-29T13:43:30 is that sometimes you just can't see the enemies 2011-11-29T13:44:15 Great. it's not algorithm what is slow 2011-11-29T13:44:16 *** dvladim has joined #aichallenge 2011-11-29T13:44:28 but calling function itself takes 5ms 2011-11-29T13:44:49 using additional python stuff another 5ms 2011-11-29T13:46:12 *** Hexren has joined #aichallenge 2011-11-29T13:46:21 *** Hexren has left #aichallenge 2011-11-29T13:46:34 *** olexs has left #aichallenge 2011-11-29T13:46:37 *** olexs has joined #aichallenge 2011-11-29T13:46:55 python function calling is slow, but THAT slow? 2011-11-29T13:47:36 yeah that can't possibly be true 2011-11-29T13:47:59 *** amstan has quit IRC (Ping timeout: 245 seconds) 2011-11-29T13:48:01 a1k0n: well it could be true for C extensions if some huge amount of state is copied for some reason 2011-11-29T13:48:06 i suppose 2011-11-29T13:48:15 *** kire has joined #aichallenge 2011-11-29T13:48:23 it takes my bot <1us to move an ant, update combat, evaluate position 2011-11-29T13:48:56 python -m timeit -s "def f(): return 1" "y = f()" 2011-11-29T13:48:56 1000000 loops, best of 3: 0.207 usec per loop 2011-11-29T13:49:28 10000000 loops, best of 3: 0.142 usec per loop 2011-11-29T13:49:29 wait, i did something wrong 2011-11-29T13:49:35 on my Q6600 @ 2.4G 2011-11-29T13:49:39 :) 2011-11-29T13:49:56 ah, i see 2011-11-29T13:49:59 no, it was my fault 2011-11-29T13:50:04 *** bhasker has quit IRC (Ping timeout: 260 seconds) 2011-11-29T13:50:05 i didn;t recompile 2011-11-29T13:50:06 was it the usleep(5000)? 2011-11-29T13:50:13 heh 2011-11-29T13:50:18 I did find a bug like that once 2011-11-29T13:50:26 no, i removed the body of function to see how fast it calls 2011-11-29T13:50:32 and forgot to recompile 2011-11-29T13:50:37 someone put a sleep in a function call because some caller of it was spinning calling it 2011-11-29T13:50:43 BenJackson: me too 2011-11-29T13:50:46 only it turned out some other function needed to call it a lot 2011-11-29T13:50:57 we thought we had some horrible performance problem 2011-11-29T13:51:00 *** besh has joined #aichallenge 2011-11-29T13:51:33 use native code and statistical compilers :p 2011-11-29T13:51:33 Fluxid: feel free to paste your BFS code 2011-11-29T13:51:46 *statistical profilers 2011-11-29T13:52:28 I liked the statistical compiler better, 80% chance of making the code faster and 40% chance of inserting bugs :P 2011-11-29T13:52:31 I always have an OProfile target in my Makefile template :) 2011-11-29T13:52:41 oprofile is nice 2011-11-29T13:53:00 I guess I could work on my bot in a Linux VM or something 2011-11-29T13:53:03 I do miss oprofile on FreeBSD 2011-11-29T13:53:15 *** praveen has joined #aichallenge 2011-11-29T13:56:12 how can i quickly zero an array of doubles? 2011-11-29T13:56:17 At turn 146, row=58,col=33 red ant is killed by two ants although they are not in range in this turn. Turn 147 red ant is not there, so it got killed in the previous turn. But clearly the two enemy ants are not in range of that red ant at 58,33 2011-11-29T13:56:20 memset 2011-11-29T13:56:21 fluxid: you can memset it to 0 2011-11-29T13:56:37 http://ants.fluxid.pl/replay.4489 2011-11-29T13:56:39 I do 99% of my bot work in int, though 2011-11-29T13:56:54 and quickly set to given value? 2011-11-29T13:56:55 praveen: are you using arrow keys or arrow BUTTONS to advance? 2011-11-29T13:57:04 you need to see the half-steps with the arrow BUTTONS to see what happened 2011-11-29T13:57:12 BenJackson: yes 2011-11-29T13:57:16 rofl 2011-11-29T13:57:18 fluxid: have to write your own loop 2011-11-29T13:57:30 ok, i thought there is some other way than loop ;) 2011-11-29T13:57:31 in C++, std::fill 2011-11-29T13:57:41 no, memset works on char 2011-11-29T13:57:54 praveen, use the buttons in the ui, not the ones on your keyboard 2011-11-29T13:58:08 pairofdice: Ya will do and check 2011-11-29T13:58:34 ui buttons show sub-turns 2011-11-29T13:58:39 also someone was demonstrating last night that you can amend those replay urls with row, col, turn 2011-11-29T13:58:42 if you want to show someone something 2011-11-29T13:59:13 Fluxid: you could have a copy of it somewhere and memcpy it over the array you want to reset if you really need to reset an array to some initial state a lot of times. 2011-11-29T13:59:24 *** kurti has quit IRC (Ping timeout: 260 seconds) 2011-11-29T13:59:35 bqf: probably faster to write your own loop 2011-11-29T13:59:58 half the memory accesses and doubles are probably as wide as the loop's copy anyway 2011-11-29T14:00:00 BenJackson: will check that too 2011-11-29T14:00:06 http://ants.fluxid.pl/replay.4489?row=58&col=33&turn=146 2011-11-29T14:00:26 (same as praveen posted, but with coords and turn embedded) 2011-11-29T14:00:30 nice 2011-11-29T14:00:32 BenJackson: memcpy is pretty damn fast; how would manually iterating over all elements and copying them one by one be faster than copying a large block of memory? 2011-11-29T14:00:35 yep, red charges in 2011-11-29T14:00:49 thanks 2011-11-29T14:00:51 *** jacob_strauss_ has joined #aichallenge 2011-11-29T14:01:00 bqf: no need to read the source memory, and the compiler is going to do similar things to your loop as memcpy does 2011-11-29T14:01:01 bgf: you don't have to read the originating array out of memory 2011-11-29T14:01:08 Yup buttons show the half-steps. Thanks now I proceed to the combat 2011-11-29T14:01:19 "now I proceed to the combat" 2011-11-29T14:01:23 it's like lambs to the slaughter 2011-11-29T14:01:42 "how hard could combat be?" (one week later) "aiigh my ants" 2011-11-29T14:02:32 :) 2011-11-29T14:02:34 yeah I don't do combat resolution, I just have an overall fight-or-flight heuristic and support for charging in formation 2011-11-29T14:02:36 BenJackson / janzert: i'd be tempted to benchmark this if that wouldn't get me dangerously close to actually starting my own bot. once i start coding anything even remotely related to this challenge, i'll probably get sucked in ;) 2011-11-29T14:03:02 works pretty well but it fails against the really good bots 2011-11-29T14:03:54 Minthos: Do you check whether the next move is a suicide? 2011-11-29T14:04:07 nope 2011-11-29T14:04:16 but it usually isn't 2011-11-29T14:04:32 *** heinrich5991 has quit IRC (Ping timeout: 260 seconds) 2011-11-29T14:04:48 *** jacob_strauss has quit IRC (Ping timeout: 265 seconds) 2011-11-29T14:05:03 bgf: one thing to keep in mind is that any memory access that misses the cpu cache(s) is going to be relatively dreadfully slow on a modern cpu 2011-11-29T14:05:19 Oh, fuck 2011-11-29T14:05:28 4.00s memcpy, 2.55 memset, 2.66 loop 2011-11-29T14:05:58 Now, after 10 versions of code trying to minimize the time I spent on excessive A star searches 2011-11-29T14:06:00 and that's not counting the influence of memcpy on the cache and how it would be worse if the source was ejected between uses 2011-11-29T14:06:10 (and lots of losses from timeout) 2011-11-29T14:06:26 I realized viewradius2 is the radius SQUARED... 2011-11-29T14:06:32 loop using MMX instructions: 2.10 2011-11-29T14:07:52 *** JustStarted has joined #aichallenge 2011-11-29T14:08:06 using MMX and forcing it to unroll: 2.06 2011-11-29T14:08:12 less improvement than I thought 2011-11-29T14:08:14 Hi, is there anybody here that would like to answer a question for me ? 2011-11-29T14:08:32 JustStarted: you just ask and if people want they'll answer 2011-11-29T14:08:32 I just did 2011-11-29T14:08:39 oh no I didn't 2011-11-29T14:08:43 no 2011-11-29T14:08:43 Ok thanks:) 2011-11-29T14:08:44 *I* did 2011-11-29T14:08:47 there, now I did 2011-11-29T14:08:55 but I contradicted myself 2011-11-29T14:09:00 *** heinrich5991 has joined #aichallenge 2011-11-29T14:09:00 *** heinrich5991 has joined #aichallenge 2011-11-29T14:09:16 Minthos: Could you not answer this question? 2011-11-29T14:09:29 I just finished the tutorial in java, and my bot behaves a bit different from the example in the tutorial. In the tutorial it says that it might differ because of the order of my ants 2011-11-29T14:09:31 BenJackson: possibly, but now it's too late for that 2011-11-29T14:09:53 my question: where can i change this order, so it will be the same as the tutorial 2011-11-29T14:10:10 it's due to the settings being wrong in the tutorial 2011-11-29T14:10:15 there's a forum post on this somewhere 2011-11-29T14:10:28 something is wrong. viewradius? 2011-11-29T14:10:44 BenJackson: yeah, but I don't remember what the setting was the tutorial was made with 2011-11-29T14:10:49 77 maybe? 2011-11-29T14:10:53 is it in the replay? 2011-11-29T14:10:58 Isn't that something related to the player_seed? 2011-11-29T14:11:01 ahh, probably 2011-11-29T14:11:05 *** kurti has joined #aichallenge 2011-11-29T14:11:15 hmm so something is wrong with the settings? 2011-11-29T14:11:26 ill look for the post on the forums :) thanks! 2011-11-29T14:11:29 JustStarted: the tutorial was written before the final contest settings were chosen 2011-11-29T14:11:33 JustStarted: well the default changed after the tutorial was made 2011-11-29T14:11:39 you're getting the default for playgame which doesn't match what the default was when the tutorial was written 2011-11-29T14:11:43 Ok:) where can i find the right default settings 2011-11-29T14:11:46 but you can pass the option 2011-11-29T14:12:31 janzert / BenJackson: thanks, you just opened my mind a little bit. also, damn you for making me want to start fiddling with this myself now! i don't know why, but i always assumed that a memcpy would be leaps and bounds faster. and i have done FPGA programming, so it is not like i shouldn't have known better ;) 2011-11-29T14:12:43 Juststarted: I had same problem when I started. Just make sure the order is correct. And even after that do not expect to get exact result for steps 4 & 5, due to random seeds. 2011-11-29T14:13:07 Ok thanks all, but now i'm not sure how to get the order correct 2011-11-29T14:13:56 ahh, there's a note on the first tutorial page about it. Use --viewradius 55 to match the tutorial settings 2011-11-29T14:14:06 Every step of the tutorial, it tells you where to place each piece of code. I had some confusion there. 2011-11-29T14:14:52 does anyone else here test with player_seed 42 exclusively due to never bothering to change it? 2011-11-29T14:14:57 BenJackson: http://paste.pocoo.org/show/514560/ here's part of my horrible code 2011-11-29T14:16:50 quick easy one: don't use fmod 2011-11-29T14:16:54 use integer mod (%) 2011-11-29T14:16:55 function in question is cstuff_DirectionMap_fill. it takes a list of points on map from which i start flooding (i put them to queue at line 101 while iterating over python objects) 2011-11-29T14:16:57 @janzert Thanks! thats the solution to my problem 2011-11-29T14:16:58 JustStarted: No! 2011-11-29T14:17:07 oh wait 2011-11-29T14:17:15 #define fmod(a,b) (a<0?((a%b)+b)%b:a%b) 2011-11-29T14:17:31 ohh 2011-11-29T14:17:38 i shoudl rename it to FMOD, sorry 2011-11-29T14:17:42 yes 2011-11-29T14:17:59 it's because i want -1 modulo 100 == 99 2011-11-29T14:19:22 1000 iterations of running this function without block of code from 112 to 141 take 1ms. 100 iterations woth everything and one element put into stack: 5s 2011-11-29T14:19:27 @BenJackson : I didn't change it, is it 42? 2011-11-29T14:19:28 ThiagoRRamos: No! 2011-11-29T14:19:52 Thanks, contestbot! 2011-11-29T14:20:00 why the branch? is the branch less expensive than the second %? 2011-11-29T14:20:29 Fluxid: if your limit check causes you to "continue" you leak the stack element 2011-11-29T14:20:32 so flooding 1000x1000 with it takes 50ms. maybe it just can't go faster, can't say 2011-11-29T14:20:37 leaking memory in a loop like this can make it verrry slow 2011-11-29T14:20:42 ah, right 2011-11-29T14:21:01 * Fluxid fixes 2011-11-29T14:21:07 still, i've run it without limit 2011-11-29T14:22:12 ah, and usenear is 0 2011-11-29T14:23:51 well one quick thing I'd do 2011-11-29T14:23:55 *** jstrong has joined #aichallenge 2011-11-29T14:24:01 not knowing for sure how you're using float distances 2011-11-29T14:24:09 is to put a counter in while (stack) 2011-11-29T14:24:15 and make sure you're not iterating way more than rows*cols times 2011-11-29T14:24:22 BFS should be linear in the number of grid squares 2011-11-29T14:24:38 if your value function is causing you to revisit a lot that could cause it 2011-11-29T14:25:45 BenJackson: because usenear causes paths to be slightly longer by fractions... actually usenear did cause revisiting many times 2011-11-29T14:26:08 BenJackson: do you think using double could make it slower? 2011-11-29T14:26:25 I'm sure it does 2011-11-29T14:26:27 but not radically 2011-11-29T14:27:14 do you clear out value elsewhere? 2011-11-29T14:27:48 i clear values before filling it again 2011-11-29T14:28:33 with negative numbers? 2011-11-29T14:29:21 yeah 2011-11-29T14:30:26 exactly rows*cols iterations 2011-11-29T14:31:37 well, BenJackson thanks 2011-11-29T14:31:45 looks fine 2011-11-29T14:31:52 I see small things you could do, but nothing radically wrong 2011-11-29T14:31:56 other than that leak 2011-11-29T14:32:04 i fixed it already 2011-11-29T14:32:33 *** Regis has joined #aichallenge 2011-11-29T14:32:41 you could use ints instead of floats, you could use a simpler queue (circular list) and you could make each queue element simpler by removing "value" 2011-11-29T14:32:42 i'll simplify it by removing usenear, and do second function which would do the same with usenear, as it would need priority queue to function correctly 2011-11-29T14:33:11 i would then read values from map? 2011-11-29T14:33:14 but none of those things is going to explain why 5ms 2011-11-29T14:33:25 Fluxid: no the trick is that you put your initial locations on the queue, then a marker 2011-11-29T14:33:34 when you pop the marker off, increment distance, push a new marker 2011-11-29T14:33:44 ah 2011-11-29T14:33:59 (make sure you don't infinite loop with markers at the end :) 2011-11-29T14:34:13 you're not destroying that object between uses, right? 2011-11-29T14:34:22 no 2011-11-29T14:34:23 Fluxid, aren't you loosing refs in line 98 ? 2011-11-29T14:34:44 g0llum: I was assuming he wasn't calling it erroneously, but I agree 2011-11-29T14:35:40 which one? 2011-11-29T14:36:11 oh, maybe not 2011-11-29T14:36:23 hmm, sorry got that wrong :( 2011-11-29T14:36:26 so how would I build this into a python module? is that all there is to it? 2011-11-29T14:36:30 i decref item and iterator, dunno if i should decref anything else 2011-11-29T14:36:30 I've never extended python 2011-11-29T14:36:33 only perl 2011-11-29T14:36:57 BenJackson: i've pasted only parts, whole module is 690 lines long 2011-11-29T14:37:08 *** ikaros has joined #aichallenge 2011-11-29T14:37:14 do you want it complete to play around? 2011-11-29T14:40:05 Fluxid, all in all, so you outsourced the bfs job from python to c ? 2011-11-29T14:40:19 yeah 2011-11-29T14:40:21 wow 2011-11-29T14:40:37 but it apparently got only 4x speedup 2011-11-29T14:40:46 *only* 2011-11-29T14:40:50 lol 2011-11-29T14:41:00 i thought it would get much faster 2011-11-29T14:42:08 so it failed my expectations a bit 2011-11-29T14:42:35 *** dvladim has quit IRC (Ping timeout: 252 seconds) 2011-11-29T14:45:18 fluxid if you do alot of bfs you could try caching the maps.. i do that 2011-11-29T14:46:17 is anyone here using the A* algorithm for pathfinding ? 2011-11-29T14:46:17 ikaros, your scheme, when to update them, and when to throw them away ? 2011-11-29T14:46:58 well i have two things to check 1) is the max distance larger than the old one 2011-11-29T14:47:15 JustStarted: pretty much every one 2011-11-29T14:47:22 and 2) is it already permanent.. means: did i have tiles that werent explored 2011-11-29T14:47:30 ok, stupid question.. 2011-11-29T14:47:30 in those cases i have to recalculate 2011-11-29T14:47:41 besh, i dont think so 2011-11-29T14:47:49 JustStarted: I don't think it's a stupid question 2011-11-29T14:47:56 I'm not using A*, although I did play with it 2011-11-29T14:47:57 alot using only bfs or similar 2011-11-29T14:48:11 It is not a stupid question. 2011-11-29T14:48:21 ok, im totally new to this and i tought of using A* but not sure how to create the graph 2011-11-29T14:48:25 JustStarted, a* vs bfs is an ongoing flamewar here 2011-11-29T14:48:28 *** epicmonkey has joined #aichallenge 2011-11-29T14:48:52 *** sofuture has quit IRC (Ping timeout: 260 seconds) 2011-11-29T14:48:58 na its just a matter of purpose id say :) 2011-11-29T14:48:58 the graph is sort of imaginary here 2011-11-29T14:49:02 every square is a node 2011-11-29T14:49:10 and the squares all ahve edges to their cardinal neighbors 2011-11-29T14:49:17 you really don't want to construct a special "graph" of that 2011-11-29T14:49:20 it's just implied 2011-11-29T14:49:29 Ok :) Thanks again 2011-11-29T14:49:47 JustStarted: you started a flamewar here :) 2011-11-29T14:49:59 i seem to be good at that.. lol 2011-11-29T14:50:12 i havent seen any real flamewar in here :) 2011-11-29T14:50:46 *** nickjohnson has quit IRC (Max SendQ exceeded) 2011-11-29T14:52:02 But seriously what you can do with breadth first search, you can do better with a* 2011-11-29T14:53:27 a* will show you the way to ne next whikey bar, but all it does is 1on1, while bfs can be 1on many, or many on many 2011-11-29T14:53:40 like calculating all best routes to one or more targets? =) 2011-11-29T14:53:45 a* can do one on many and many on 1 and many on many just fine 2011-11-29T14:54:18 *** Palmik has quit IRC (Remote host closed the connection) 2011-11-29T14:54:43 but needs slightly more complex data structuring than bfs. my bot utilizes both, switching to a* when heuristics are available and search radius is sufficiently large 2011-11-29T14:55:11 yea i will add an a* too when i reach optimization 2011-11-29T14:55:26 have different tasks where i do bfs for a single target which is crap 2011-11-29T14:55:31 Any language with a binary heap implementation eases implmentation of a* 2011-11-29T14:55:31 Why does my new bot keep losing to really old versions? 2011-11-29T14:55:33 well im totally new to AI programming so i'm just going to try implementing A*, if that works ill see what i can optimize 2011-11-29T14:55:38 but so far its fast enough so no need to waste time 2011-11-29T14:55:39 *** Accoun has quit IRC () 2011-11-29T14:55:59 JustStarted, probably coding a simple bfs will be much easier 2011-11-29T14:56:33 JustStarted: java is a good choice 2011-11-29T14:57:59 *** AntDroid has joined #aichallenge 2011-11-29T14:58:04 hmm, but im using a book and that suggests A*, and i would like to implement it :P 2011-11-29T14:58:08 im using java :) 2011-11-29T14:58:32 i've never used it before but i know C# and found it easier to do the tutorial using java 2011-11-29T14:58:37 go for it if you want to. but keep in mind that it definitely is more complicated than bfs 2011-11-29T14:58:41 well its all up to you. implementing a* is surely no mistake :) 2011-11-29T14:58:46 good point to start using java too, for now it's not a lot different 2011-11-29T14:59:21 JustStarted, still 3 weeks to go, so whatever gives you a good start, take that. 2011-11-29T14:59:55 I'll do my best implementing A*, if i cant manage to make that work i'll try bfs first 2011-11-29T15:02:20 Fluxid: ok, I ran it locally 2011-11-29T15:02:24 without python at all 2011-11-29T15:02:37 ikaros: the problem is that ants standing on a path make the path longer. so as ants move i need to update the map 2011-11-29T15:02:40 takes 3.5ms, with -O3 it takes 1.4ms 2011-11-29T15:02:53 for how big map? 2011-11-29T15:02:57 200x200 2011-11-29T15:03:37 fluxid: do you consider ants unpassable ? 2011-11-29T15:03:39 you mean own ants or enemy ants? 2011-11-29T15:03:56 own 2011-11-29T15:03:59 because i ignore them when calculating distance maps 2011-11-29T15:04:43 besh: if there are many own ants on the path then ants farther away can go around 2011-11-29T15:05:08 thats true but you move most ants each turn anyways id guess 2011-11-29T15:05:14 *** sofuture has joined #aichallenge 2011-11-29T15:05:17 and this spreads ants a bit if there are many possible paths 2011-11-29T15:05:22 and you "push" them away with priorities 2011-11-29T15:05:51 hm ok yea i handle the spreading seperately 2011-11-29T15:06:07 separately? 2011-11-29T15:06:18 http://paste.aichallenge.org/sq8OH/ 2011-11-29T15:06:54 BenJackson: damn, i need to measure it more 2011-11-29T15:07:25 fluxid: that is exactly my problem. My ants always follow the best possible path, even though there are other sub-optimal that avoid congestion. 2011-11-29T15:07:38 ikaros: in that paste, making path longer make ants go to sides 2011-11-29T15:08:07 but are too spread though.. 2011-11-29T15:08:22 yea i understand. i just dont handle the spreading via my distance maps but thats probably just a design choice 2011-11-29T15:08:56 BenJackson: thank you for your time 2011-11-29T15:10:00 FLuxid: avoiding real work :) 2011-11-29T15:10:10 fluxid: That looks great. Can you explain a bit more of the spreading ? 2011-11-29T15:10:48 BenJackson: work on your bot or some other real-real work? 2011-11-29T15:11:05 Fluxid: looking at some profiling I have laying around suggests I get similar results to your code when compiled with optimization 2011-11-29T15:11:13 real-real work 2011-11-29T15:11:42 after my bugfix release last night I have to wait a while to rank up on aichallenge 2011-11-29T15:11:51 game 5 I started next to xathis, ow 2011-11-29T15:12:16 his defense against me was pretty sweet though 2011-11-29T15:12:19 I need to emulate that 2011-11-29T15:12:31 besh: ant standing on the path make the path longer. that's all :) in that case i was adding +1 to path length under the ant, +0.6 by it and +0.3 one cell away from ant 2011-11-29T15:12:45 what are people using for spreading out their ants over the "controlled" area? the process looks a lot like molecular diffusion, but I can't think of any efficient algorithms to implement that... 2011-11-29T15:12:46 BenJackson: link to game? 2011-11-29T15:12:52 also, what do you do for real fork? 2011-11-29T15:12:54 work* 2011-11-29T15:13:18 fluxid: smart idea. I will try it. 2011-11-29T15:13:23 and a spoon 2011-11-29T15:14:13 http://aichallenge.org/visualizer.php?game=144100&user=3473 2011-11-29T15:14:27 olexs: i'd like to know a nice solution too... i just go to part which isn't in visible area 2011-11-29T15:15:03 olexs: I also couldn't think of any efficient ways to do it so I did it inefficiently and that was fast enough 2011-11-29T15:15:06 olexs, perhaps it would be possible to do some kind of mass-spring system but i think it would be computationally to expensive 2011-11-29T15:15:06 much to my surprise 2011-11-29T15:15:18 BenJackson: that was a fast game 2011-11-29T15:15:21 yep 2011-11-29T15:15:51 you weren't really able to do anything... 2011-11-29T15:16:11 he did a nice job of killing off my small attack party, though 2011-11-29T15:16:21 oh, i've got a game with you, but with version 8 2011-11-29T15:16:24 which I clearly could have avoided with a lot more lookahead 2011-11-29T15:16:49 *** Palmik has joined #aichallenge 2011-11-29T15:16:51 olexs: so is it molecular diffusion ? 2011-11-29T15:17:19 *** gorosan has joined #aichallenge 2011-11-29T15:17:44 besh: well, the principle is very similar, but I don't know how do implement it properly 2011-11-29T15:17:54 fluxid: btw your ants have that feel of the molecular diffusion too. 2011-11-29T15:17:56 Fluxid: yeah that game was kind of random 2011-11-29T15:18:04 *** bhasker has joined #aichallenge 2011-11-29T15:18:04 ;) 2011-11-29T15:18:28 typical on massive multihill maps 2011-11-29T15:18:28 given an area, I want the ants to spread out as equally as possible, leaving no invisible areas and no tight groups 2011-11-29T15:18:56 every ant has 4 hills and on turn 36 just before most of the taking happens the average bot has 10 ants 2011-11-29T15:19:03 barely 2/hill 2011-11-29T15:19:31 winner is determined on about turn 47 2011-11-29T15:19:52 Here is a video of what I did for spreading out ants some time ago http://www.youtube.com/watch?v=YmIlhMSGX_c 2011-11-29T15:19:55 the exploration/spread I have so far is based on a completely different principle and works, but not really very well 2011-11-29T15:20:03 I've considered it but strategies to optimize for that case would probably lose you most games 2011-11-29T15:20:03 olexs: i do two things: 1. pathfind to unvisible area 2. for every ant i calculate vectors to all other ants and go in direction where are less ants 2011-11-29T15:20:32 the closer are other ants the stronger is the repelling force 2011-11-29T15:20:35 something like that 2011-11-29T15:20:38 With a dummy opponent though... I try to put the ants view_radius apart to maximize view 2011-11-29T15:20:59 how are you guys handling combat? 2011-11-29T15:21:08 my bot seems to have trouble with combat a lot:( 2011-11-29T15:21:08 first I bring more ants to the party 2011-11-29T15:21:12 then I kill the enemy ants 2011-11-29T15:21:19 unless they bring MORE to the party 2011-11-29T15:21:33 Fluxid: that sounds reasonable... build vectors to all ants in visible radius and just sum them 2011-11-29T15:21:50 bhasker: after calculating new ants posiotions i check if any ant may be killed. then i modify weights for those ants and recalculate moves. and then again. and again. until looking good. 2011-11-29T15:22:07 fucking inefficient 2011-11-29T15:22:11 interesting 2011-11-29T15:22:12 and not really useful 2011-11-29T15:22:55 hmm i was thinking of doing a small minmax when entering combat 2011-11-29T15:23:01 but it seems too complicated 2011-11-29T15:23:12 bhasker: min-max search with fairly complex move generator working on ant clusters 2011-11-29T15:23:22 because ants are too scared and this causes deadlocks like here: http://ants.fluxid.pl/replay.2000 2011-11-29T15:23:30 yes something like that just that my code isn't really structured for minmax very well 2011-11-29T15:23:53 mine was from the beginning, so it wasn't too complicated implementing 2011-11-29T15:24:07 how does it work, is it doing the right thing? 2011-11-29T15:24:25 just look at fights by olexs_* bots on tcp 2011-11-29T15:24:31 i didn't know how to apply minimax for lots of ants so i just used my intuition... 2011-11-29T15:24:32 hmm k 2011-11-29T15:24:36 *** Accoun has joined #aichallenge 2011-11-29T15:25:29 it does work ok. sometimes though the bot will opt for a safe move locally, but globally being pushed back is not a good thing 2011-11-29T15:26:29 sometimes it's really smart, like when it can corner an enemy into water and kill them safely, it will do that 2011-11-29T15:26:43 Ha! The lone ant strikes again! http://tcpants.com/replay.11804 2011-11-29T15:26:55 *** jstrong has quit IRC (Ping timeout: 248 seconds) 2011-11-29T15:26:56 or run over an enemy hill even if that means losing a few ants 2011-11-29T15:30:54 hmm k 2011-11-29T15:31:05 my ant can't stop committing suicide, stupid move generator sucks 2011-11-29T15:31:20 and for some reason it makes moves when it will clearly lose an ant 2011-11-29T15:31:22 that'd rather be evaluation that's bad 2011-11-29T15:31:25 need to debug my combat 2011-11-29T15:31:52 *** praveen has quit IRC (Quit: Page closed) 2011-11-29T15:32:07 move generator should only create possible moves, then you simulate them and evaluate the resulting situation... even if the move gen sucks, bad moves it creates should get bad evaluations 2011-11-29T15:32:40 unless it only creates suicides moves, in which case you don't have much choice :) 2011-11-29T15:33:22 i think there is a bug in my combat resolution 2011-11-29T15:33:29 it seems to undercount enemies or something i think 2011-11-29T15:33:37 unit-test it 2011-11-29T15:33:41 yea going to do that 2011-11-29T15:33:47 need to generate a stupid map 2011-11-29T15:33:49 and test it on that 2011-11-29T15:33:55 I found out the hard way that unit testing helps... a lot... 2011-11-29T15:35:34 *** Palmik has quit IRC (Remote host closed the connection) 2011-11-29T15:40:55 Is there a way I can print something in the console other than the bot output? 2011-11-29T15:41:09 ThiagoRRamos: use stderr output 2011-11-29T15:44:43 *** jstrong has joined #aichallenge 2011-11-29T15:46:18 olexs: It doesn't work 2011-11-29T15:47:16 ThiagoRRamos: to print it in the console when running in a game, run playgame.py with -E param, and specified --log_dir. it will then print your stderr output to a log file 2011-11-29T15:47:46 ThiagoRRamos, try plagame.py whthout any params, you'll see the nessecary options there 2011-11-29T15:50:32 then edit whatever shell/cmd script you're using 2011-11-29T15:51:22 Lol, I timeout half way through the game and somehow still win! http://ants.fluxid.pl/replay.4663 2011-11-29T15:51:48 your ants became zombies? 2011-11-29T15:52:06 *** mj41 has joined #aichallenge 2011-11-29T15:52:24 Poor strcat 2011-11-29T15:52:36 Almsot 800 turns agaisnt a timed out opponent and it sitll couldn't win 2011-11-29T15:53:03 nice antimony!! 2011-11-29T15:53:09 i can see the food fix 2011-11-29T15:53:11 made your bot way way better :) 2011-11-29T15:53:26 what food fix? 2011-11-29T15:53:37 you weren't picking up food 2011-11-29T15:53:42 oh that 2011-11-29T15:53:44 had some bug where you were hills would be preferred 2011-11-29T15:53:47 That was just a stupid bug 2011-11-29T15:53:49 yeah 2011-11-29T15:53:57 yeah, I really cannot do the logging :P 2011-11-29T15:53:58 obviously, a bot that collects food is better 2011-11-29T15:54:29 yeah but one that drastically affected the effectiveness of your bot, no? 2011-11-29T15:54:41 it's always the stupid bugs that make a huge difference in the performance of your hive 2011-11-29T15:54:54 *** besh has quit IRC (Quit: Page closed) 2011-11-29T15:55:11 *** delt0r_ has quit IRC (Ping timeout: 248 seconds) 2011-11-29T15:55:23 haha you surrounded his hill and then timed out. cruel. 2011-11-29T15:56:43 *** Rav3nsW0rd has quit IRC (Excess Flood) 2011-11-29T15:57:13 *** Rav3nsW0rd has joined #aichallenge 2011-11-29T15:58:31 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-29T15:59:37 To be fair, his bot could have easily taken care of it 2011-11-29T15:59:54 I think my bot would break out in that circumstance 2011-11-29T16:01:10 you'd have to break in 2011-11-29T16:02:40 Is there really a way of making a bot output something? 2011-11-29T16:02:47 I mean, do you guys do it? 2011-11-29T16:03:10 ThiagoRRamos: I do what I described... -E to playgame.py and output my debug info into stderr, that gets saved to a file 2011-11-29T16:03:18 *** Akranis has joined #aichallenge 2011-11-29T16:03:42 afterwards I can see the replay and read in that file what my bot was thinking 2011-11-29T16:06:35 *** choas has joined #aichallenge 2011-11-29T16:07:40 *** delt0r_ has joined #aichallenge 2011-11-29T16:08:54 *** JustStarted has left #aichallenge 2011-11-29T16:09:59 olexs: you mean sys.stderr.write ? 2011-11-29T16:10:12 -e or -E is probably what you are missing 2011-11-29T16:10:39 ThiagoRRamos: for python, probably yes. mine is in java, so it is System.err.println() 2011-11-29T16:14:32 *** ruuhkis has quit IRC (Quit: Page closed) 2011-11-29T16:14:40 I have both -e and -E 2011-11-29T16:17:42 I wouldn't need this if my bot just stopped putting all my ants in a diagonal 2011-11-29T16:17:44 *** ltriant has joined #aichallenge 2011-11-29T16:18:18 with no reason... 2011-11-29T16:18:27 lol 2011-11-29T16:18:56 it could be worse, your bot could move back and forth in place 2011-11-29T16:19:12 so i recovered my partitions 2011-11-29T16:19:19 because your weights make it change target over and over 2011-11-29T16:19:21 every move 2011-11-29T16:19:23 and can boot into linux, can mount ntfs but still can't boot into windows. 2011-11-29T16:20:02 who really wants to boot windows? 2011-11-29T16:20:19 rwest: i do 2011-11-29T16:21:01 *** trinnity has joined #aichallenge 2011-11-29T16:22:37 yeah, it could be :D 2011-11-29T16:22:47 sys.stderr.write() doesn't seem to work for me either 2011-11-29T16:23:45 Printing to a file from the bot is a bit trickier than usual aswell.. 2011-11-29T16:24:22 crap. wrong channel. sorry fellas :) 2011-11-29T16:24:43 10 lashes of the whip 2011-11-29T16:27:03 pairofdice: it's simple in java :) 2011-11-29T16:27:48 Yeah, it's simple. It just doesn't work. 2011-11-29T16:28:20 if you write to stderr you have to pass an option to playgame to allow it 2011-11-29T16:28:40 I have the options set 2011-11-29T16:29:19 --log_dir game_logs -E -e 2011-11-29T16:31:49 pairofdice: what 2011-11-29T16:31:57 's wrong with print()? 2011-11-29T16:33:15 works for me 2011-11-29T16:33:32 *** bhasker has left #aichallenge ("Killed buffer") 2011-11-29T16:33:37 *** bhasker has joined #aichallenge 2011-11-29T16:34:21 *** trinnity has quit IRC (Quit: Page closed) 2011-11-29T16:36:26 *** opti-free has joined #aichallenge 2011-11-29T16:36:47 Could anyone help me start up implementing a pathfinding algorithm ? i have no idea how and where to start... 2011-11-29T16:37:03 Wikipedia 2011-11-29T16:38:32 http://en.wikipedia.org/wiki/Breadth-first_search 2011-11-29T16:40:03 *** peter___ has joined #aichallenge 2011-11-29T16:41:09 --capture_errors... 2011-11-29T16:43:32 *** ThiagoRRamos has quit IRC (Ping timeout: 255 seconds) 2011-11-29T16:44:22 aichallenge: janzert epsilon * r535b4a8 / website/starter_packages.php : Add Racket to starter package page - http://git.io/3791IQ 2011-11-29T16:45:28 pairofdice: did you figure out stderr yet? 2011-11-29T16:45:40 Nope 2011-11-29T16:45:46 what language? 2011-11-29T16:45:52 python 3 2011-11-29T16:46:04 you'll need to call sys.stderr.flush() after writing 2011-11-29T16:46:39 you can call it once just before finish_turn() 2011-11-29T16:47:00 stderr should be line buffered anyway? 2011-11-29T16:47:18 I can't explain it technically, I just know I need it. 2011-11-29T16:47:29 hmm, ok 2011-11-29T16:47:40 Still nothing... I'm puzzled 2011-11-29T16:48:15 is the 0.bot.error file blank? 2011-11-29T16:48:24 Yes 2011-11-29T16:48:39 try with ErrorBot 2011-11-29T16:49:29 That one printed traceback 2011-11-29T16:49:42 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-29T16:49:46 and it's in the 0.bot.error file? 2011-11-29T16:49:59 Yeah 2011-11-29T16:50:08 okay, so it's not your playgame command, it's your bot 2011-11-29T16:50:30 Interesting. You slowly build up forces and then suddenly go for the kill. http://ants.fluxid.pl/replay.4748 2011-11-29T16:51:17 Could someone tell me how to use cProfile with the bots? 2011-11-29T16:51:44 sure, run a bot input file back through your bot 2011-11-29T16:52:01 python -m cProfile MyBot.py < 0.bot.input 2011-11-29T16:52:21 unfortunately I found that FreeBSD and whatever glibc the contest is running don't agree on random seeds 2011-11-29T16:52:35 could be that bug I filed against glibc a while ago 2011-11-29T16:52:58 whoa, cool game 2011-11-29T16:53:21 Antimony: I think he has some kind of "number of opponents" estimate 2011-11-29T16:53:29 maybe it finally kicked over to "1" and thus suicide was ok 2011-11-29T16:53:49 I 2011-11-29T16:53:58 I'm beginning to wonder if my bot has recently gotten worse 2011-11-29T16:54:06 and those beautiful spawnkills at the end 2011-11-29T16:54:17 just looking at that game makes me think I really need to change over from diffusion 2011-11-29T16:54:24 or at least go hybrid for battles 2011-11-29T16:55:10 Fluxid: that was one of the bugs I fixed in V9 2011-11-29T16:55:17 I accidentally incentivized stepping on occupied hills 2011-11-29T16:56:17 *** TheLinker has joined #aichallenge 2011-11-29T16:56:23 *** gorosan has quit IRC (Ping timeout: 265 seconds) 2011-11-29T16:56:23 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-29T17:03:37 *** Ikol has joined #aichallenge 2011-11-29T17:04:51 *** ikaros has joined #aichallenge 2011-11-29T17:06:36 *** Ikol has quit IRC (Read error: Connection reset by peer) 2011-11-29T17:07:02 *** Ikol has joined #aichallenge 2011-11-29T17:07:27 *** Rav3nsW0rd has quit IRC (Ping timeout: 260 seconds) 2011-11-29T17:07:30 hmm, is it normal that cprofile takes more time then the match? 2011-11-29T17:09:09 *** peter____ has joined #aichallenge 2011-11-29T17:10:59 *** Ikol has quit IRC (Excess Flood) 2011-11-29T17:11:02 *** Israfel has joined #aichallenge 2011-11-29T17:11:31 *** Ikol has joined #aichallenge 2011-11-29T17:12:20 *** peter___ has quit IRC (Ping timeout: 265 seconds) 2011-11-29T17:18:07 *** Ikol has quit IRC (Excess Flood) 2011-11-29T17:21:00 *** Ikol has joined #aichallenge 2011-11-29T17:29:35 *** Garf has quit IRC (Quit: Make a new plan, Stan!) 2011-11-29T17:32:08 *** Ikol has quit IRC (Excess Flood) 2011-11-29T17:32:36 *** Ikol has joined #aichallenge 2011-11-29T17:34:14 *** smiley1983 has quit IRC (Ping timeout: 245 seconds) 2011-11-29T17:35:15 *** smiley1983 has joined #aichallenge 2011-11-29T17:37:26 *** Akranis has quit IRC (Quit: Lmnar) 2011-11-29T17:40:30 *** opti-free has quit IRC (Quit: Page closed) 2011-11-29T17:40:41 *** Ikol has quit IRC (Excess Flood) 2011-11-29T17:42:06 *** epicmonkey has quit IRC (Ping timeout: 244 seconds) 2011-11-29T17:42:30 *** ThiagoRRamos has joined #aichallenge 2011-11-29T17:42:41 *** ThiagoRRamos has joined #aichallenge 2011-11-29T17:42:54 :D 2011-11-29T17:46:07 *** Ikol has joined #aichallenge 2011-11-29T17:46:10 Ug, my bot has apparently forgotten about water 2011-11-29T17:49:40 *** kire has quit IRC (Remote host closed the connection) 2011-11-29T17:50:09 okay, my vector-based ants spreading sorta works, except ants get hung up on water... and if I add water to things they should keep away from they never make it far because of narrow passes. hm. 2011-11-29T17:52:40 finally won a game on aichallenge without anything "interesting" happening 2011-11-29T17:53:26 *** Ikol has quit IRC (Max SendQ exceeded) 2011-11-29T17:55:20 *** Ikol has joined #aichallenge 2011-11-29T17:57:11 I just found a nasty bug in my bfs 2011-11-29T17:57:23 If I add a square to teh queue while it is unknown, and I alter find out it is water 2011-11-29T17:57:27 I expand it anyway 2011-11-29T17:58:10 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-11-29T17:58:19 you must be doing it very differently than I 2011-11-29T17:59:11 I first update the map, then I do stuff that might require pathfinding.. there's no way my map is going to get updated in the middle of a bfs search 2011-11-29T17:59:12 *** amstan has joined #aichallenge 2011-11-29T17:59:12 *** ChanServ sets mode: +o amstan 2011-11-29T17:59:24 no, this is a special bfs that takes place over multiple turns 2011-11-29T17:59:32 why? 2011-11-29T17:59:37 to calculate distances to the hill 2011-11-29T17:59:42 *** peter____ has quit IRC (Ping timeout: 265 seconds) 2011-11-29T17:59:50 huh? 2011-11-29T17:59:53 It stores every known distance from a square to the hill 2011-11-29T18:00:01 has more squares become known, the bfs continues 2011-11-29T18:00:15 aha 2011-11-29T18:00:28 that makes sense 2011-11-29T18:00:35 so my bot had incorrect distances to my hill 2011-11-29T18:00:53 but I still don't see why you'd bother 2011-11-29T18:01:10 when you can do that bfs each turn instead for very little cost 2011-11-29T18:01:27 or rather, do it for the ants you need to do it for 2011-11-29T18:01:59 but it's your bot, I'm sure you have your reasons 2011-11-29T18:02:00 You only need to do it 'once', so why would you do it every turn 2011-11-29T18:02:02 I need it for every square on the map 2011-11-29T18:02:31 *** jstrong has quit IRC (Quit: Leaving.) 2011-11-29T18:02:39 simplicity, but I'm not going to argue this like I said I'm sure he has his reasons 2011-11-29T18:02:43 because it takes a few ms and simplifies the implementation? 2011-11-29T18:06:46 *** jacob_strauss_ has quit IRC (Quit: Page closed) 2011-11-29T18:08:24 On a 2 palyer map, my bot figures out the enemy hill location on turn 113 by symmetry, but it doesn't get around to razing it until turn 600... 2011-11-29T18:09:53 when I did that I found my bots all attacked the hill before I was ready 2011-11-29T18:10:01 python: can't open file 'sample_bots/python/LeftyBot.py': [Errno 2] No such file or directory 2011-11-29T18:10:03 wtf? 2011-11-29T18:10:19 I should probably detect it and then have some trigger for revealing the info to my algorithm 2011-11-29T18:15:37 I wonder if I should try to make my bot more aggressive at razing hills it knows about. 2011-11-29T18:15:56 At the very least, it could send a single ant incase they have no defences 2011-11-29T18:20:33 *** Jak_o_Shadows has joined #aichallenge 2011-11-29T18:27:03 *** bhasker has quit IRC (Ping timeout: 244 seconds) 2011-11-29T18:27:15 *** mj41 has quit IRC (Ping timeout: 260 seconds) 2011-11-29T18:31:44 My problem occurs when 3 or more ants of mine can see an enemy hill :P 2011-11-29T18:33:02 *** olexs has quit IRC (Ping timeout: 252 seconds) 2011-11-29T18:43:35 *** choas has quit IRC (Ping timeout: 244 seconds) 2011-11-29T18:44:08 *** ThiagoRRamos has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-29T18:53:06 *** Antimony has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-11-29T18:54:58 *** Ikol has quit IRC (Excess Flood) 2011-11-29T18:55:21 *** Ikol has joined #aichallenge 2011-11-29T18:57:29 *** besh has joined #aichallenge 2011-11-29T18:57:35 *** Conorach has quit IRC (Ping timeout: 260 seconds) 2011-11-29T19:08:53 *** jstrong1 has joined #aichallenge 2011-11-29T19:09:21 *** rwest_ has joined #aichallenge 2011-11-29T19:09:32 http://aichallenge.org/visualizer.php?game=144830&user=12139 ant orgy in the corner 2011-11-29T19:10:38 rwest_: why? 2011-11-29T19:11:16 a bug I fixed like 4 versions ago hah 2011-11-29T19:11:27 I added real explore code and it never does that anymore 2011-11-29T19:11:51 after enough time of not seeing his hill, my ants would have wanted to see it more than anything 2011-11-29T19:12:15 now if I could only fix the jitteryness 2011-11-29T19:14:15 *** NoirSoldats has joined #aichallenge 2011-11-29T19:15:14 I am experiencing an issue with the PHP starter pack in attempting to dump out debug information locally.. Even opening and writing to a file doesn't work... ideas? 2011-11-29T19:15:48 use the error stream? 2011-11-29T19:15:54 NoirSoldats: no idea, maybe just do an exec 2011-11-29T19:16:17 @seen fluxid 2011-11-29T19:16:17 BenJackson: fluxid was last seen in #aichallenge 2 hours, 22 minutes, and 11 seconds ago: and those beautiful spawnkills at the end 2011-11-29T19:16:18 exec("cat error shit to < file") 2011-11-29T19:16:28 (the playgame executable can be set to store information from the stream) 2011-11-29T19:16:37 or that 2011-11-29T19:16:41 ^^ 2011-11-29T19:17:05 s/ 2011-11-29T19:19:02 Okie, figured that out, next issue in the PHP framework, doSetup isn't being called? I though it was called once initially and then it was doTurn that's called... each turn. 2011-11-29T19:21:47 *** bhasker has joined #aichallenge 2011-11-29T19:22:23 *** Jak_o_Shadows1 has joined #aichallenge 2011-11-29T19:24:39 *** Jak_o_Shadows has quit IRC (Ping timeout: 244 seconds) 2011-11-29T19:28:11 *** Regis has quit IRC (Ping timeout: 244 seconds) 2011-11-29T19:32:26 *** replore has joined #aichallenge 2011-11-29T19:40:06 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-29T19:43:36 *** twisted has joined #aichallenge 2011-11-29T19:45:40 *** Blkt has quit IRC (Quit: good night...) 2011-11-29T19:52:55 *** Jak_o_Shadows has joined #aichallenge 2011-11-29T19:56:05 *** Jak_o_Shadows1 has quit IRC (Ping timeout: 258 seconds) 2011-11-29T19:58:07 *** twisted has quit IRC (Ping timeout: 265 seconds) 2011-11-29T20:09:04 *** JorgeB has quit IRC (Quit: Computer has gone to sleep.) 2011-11-29T20:12:45 *** treeform has quit IRC (Remote host closed the connection) 2011-11-29T20:16:10 *** analyst74 has quit IRC (Quit: I love my HydraIRC -> http://www.hydrairc.com <-) 2011-11-29T20:22:57 anyone aspiring to learn BFS here? 2011-11-29T20:23:01 I'm working on a helpful animation 2011-11-29T20:23:08 I need beta viewers 2011-11-29T20:32:19 BenJackson: try the forums :P 2011-11-29T20:32:30 yep, posting 2011-11-29T20:32:36 I'll look at a webpage if you have a link :P 2011-11-29T20:32:40 otherwise, mehffort :P 2011-11-29T20:33:55 *** b0rder has joined #aichallenge 2011-11-29T20:34:11 http://forums.aichallenge.org/viewtopic.php?f=24&t=2010 2011-11-29T20:34:46 by the way, gifsicle is awesome 2011-11-29T20:36:44 that is cool. 2011-11-29T20:36:59 :-) 2011-11-29T20:37:38 BenJackson: i imagine it'd be way more useful if people could click throw at their own page 2011-11-29T20:37:59 haha I should put it up as 2000 ad supported pages 2011-11-29T20:38:03 2011-11-29T20:38:05 2011-11-29T20:38:31 I suppose I could have tried to write a flash applet 2011-11-29T20:38:39 in theory I learned how to do that a few weeks ago 2011-11-29T20:38:40 other ideas would be multi source and a* obviously 2011-11-29T20:38:55 I could make another illustrating multi-source 2011-11-29T20:40:12 * avdg wonders if it would fit in the visualizer 2011-11-29T20:40:27 even more awesome, if you made an applet, let people put targets in, pick a path finding method then have it animate finding the path 2011-11-29T20:40:31 and show the found path at the end 2011-11-29T20:43:19 *** dlila has joined #aichallenge 2011-11-29T20:44:29 http://forums.aichallenge.org/viewtopic.php?f=24&t=1996 2011-11-29T20:44:30 that's awesome 2011-11-29T20:45:20 :p my js to c++ switch 2011-11-29T20:45:34 *** rajanaresh has joined #aichallenge 2011-11-29T20:45:43 (I'm lucky to have c++ with me atm) 2011-11-29T20:46:24 everyone knows about the new "easy mode" tcp server, right? 2011-11-29T20:46:55 what? 2011-11-29T20:47:06 i'm not testing in the open till i finish my bot anyway 2011-11-29T20:47:12 people putting up weak-ish bots on http://bhickey.net:2080 2011-11-29T20:47:22 to give new players something to test that's not the slaughterfest of fluxid 2011-11-29T20:47:28 that's cool 2011-11-29T20:47:32 is there a crazy maps tcp? 2011-11-29T20:47:42 fluxid has the cell maps 2011-11-29T20:47:51 like ridiculous numbers of players, large grid sizes, 1 width mazes? 2011-11-29T20:47:56 cell maps aren't crazy :P 2011-11-29T20:48:04 someone put up a replay where every square was a hill 2011-11-29T20:48:05 i want to see like a 200x200 actual maze map 2011-11-29T20:48:10 oh, fluxid put it on his server briefly 2011-11-29T20:48:12 the battling over water will make it ridiculous 2011-11-29T20:48:18 yeah i saw that 2011-11-29T20:48:20 *** Ikol has quit IRC (Excess Flood) 2011-11-29T20:48:33 i want a tcp server dedicated to ridiculous maps like that :P 2011-11-29T20:48:33 my bot would solve an actual maze although probably not that efficiently 2011-11-29T20:48:42 so would mine :P 2011-11-29T20:48:48 *** Ikol has joined #aichallenge 2011-11-29T20:48:59 it would be funny to see what bots did though 2011-11-29T20:49:19 my planet wars bot was awful on maps with heaps of planets 2011-11-29T20:49:36 it's funny, I still have no idea how to approach planet wars 2011-11-29T20:49:40 that's why I never got into it 2011-11-29T20:49:56 I wouldn't know how to play it if *I* played it 2011-11-29T20:53:28 I would probably play more stupidly :p 2011-11-29T20:53:58 (well, I'm learning things slower than I want to, but I'm at least learning some things, so I won't be able to write a bot back then) 2011-11-29T20:56:05 *** Ikol has quit IRC (Excess Flood) 2011-11-29T20:56:44 *** Ikol has joined #aichallenge 2011-11-29T21:03:02 *** pguillory has joined #aichallenge 2011-11-29T21:03:50 BenJackson: it was really tough to improve your bot after a while 2011-11-29T21:05:12 Fluxid: did you end up speeding up your flood fill? 2011-11-29T21:05:22 *** bmh has joined #aichallenge 2011-11-29T21:08:01 oh nice, strcat_foodfight is doing really well :) 2011-11-29T21:08:59 *** Ikol has quit IRC (Excess Flood) 2011-11-29T21:09:29 *** Ikol has joined #aichallenge 2011-11-29T21:09:34 *** Jak_o_Shadows1 has joined #aichallenge 2011-11-29T21:11:36 *** Jak_o_Shadows has quit IRC (Ping timeout: 258 seconds) 2011-11-29T21:13:53 cool. my server hasn't melted down 2011-11-29T21:15:27 *** Ikol has quit IRC (Excess Flood) 2011-11-29T21:17:22 thanks for running that server, bmh 2011-11-29T21:17:38 and everyone else who has one going 2011-11-29T21:19:03 pguillory: you're welcome. The box was otherwise idle. 2011-11-29T21:20:57 *** Ikol has joined #aichallenge 2011-11-29T21:25:47 *** Ikol has quit IRC (Excess Flood) 2011-11-29T21:26:18 *** Ikol has joined #aichallenge 2011-11-29T21:28:02 *** Ikol has quit IRC (Excess Flood) 2011-11-29T21:28:28 *** Ikol has joined #aichallenge 2011-11-29T21:29:04 bmh: I think I'll make a new bot to run on there, in Go :) 2011-11-29T21:29:34 thestinger: *sunglasses* go for it. 2011-11-29T21:29:43 with each ant having a goroutine, and they'll independently decide on moves and improve them until time runs out 2011-11-29T21:30:04 *** besh has quit IRC (Quit: Page closed) 2011-11-29T21:31:25 thestinger: sounds sort of like the fight code I want to write 2011-11-29T21:32:58 *** Ikol is now known as Rav3nSw0rd 2011-11-29T21:33:10 *** jstrong1 is now known as roflmao 2011-11-29T21:34:54 *** tcoppi has joined #aichallenge 2011-11-29T21:38:12 *** dlila has quit IRC (Quit: Leaving) 2011-11-29T21:42:13 *** sf17k has joined #aichallenge 2011-11-29T21:43:29 *** bmh has quit IRC (Ping timeout: 265 seconds) 2011-11-29T21:52:16 *** b0rder_ has joined #aichallenge 2011-11-29T21:54:44 I think it's funny that pguillory put up his "weak" bot on bmh's server and it started kicking my ass 2011-11-29T21:54:54 it wasn't obvious from the log but the ones I win are where it times out 2011-11-29T21:54:58 which is surprisingly often 2011-11-29T21:55:41 *** b0rder has quit IRC (Ping timeout: 258 seconds) 2011-11-29T21:58:04 ha 2011-11-29T21:58:41 heh i tried to strip it down 2011-11-29T21:59:02 I'm fully expecting us to get into some kind of "one arm tied behind our backs" style contest there 2011-11-29T21:59:04 but the ants, they're so hungry 2011-11-29T21:59:08 *** AntDroid has joined #aichallenge 2011-11-29T22:00:15 that would be great right? 2011-11-29T22:01:08 *** AntDroid_ has joined #aichallenge 2011-11-29T22:01:39 you can only take hills using ants in the shape of a smiley face 2011-11-29T22:02:35 You can only move your ants in the glider formation 2011-11-29T22:02:45 *** bhasker has quit IRC (Ping timeout: 244 seconds) 2011-11-29T22:03:01 did you guys play Civ? 2011-11-29T22:03:09 do you remember keeping one enemy city alive so you could keep on building? 2011-11-29T22:03:18 *** NoirSoldats has quit IRC (Ping timeout: 265 seconds) 2011-11-29T22:03:23 I don't remember doing that, no :) 2011-11-29T22:03:37 so surround the last ant, don't let any harm come to it 2011-11-29T22:03:47 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-29T22:03:47 *** sf17k has quit IRC (Ping timeout: 265 seconds) 2011-11-29T22:03:47 then do crazy things with your ants 2011-11-29T22:04:45 *** pguillory has quit IRC (Ping timeout: 265 seconds) 2011-11-29T22:08:25 2 7 1.0 2011-11-29T22:08:25 3 15 1.0 2011-11-29T22:08:25 4 16 1.6875 2011-11-29T22:08:25 5 26 1.5 2011-11-29T22:08:25 6 22 2.40909090909 2011-11-29T22:08:28 7 24 1.91666666667 2011-11-29T22:08:30 8 15 1.73333333333 2011-11-29T22:08:33 9 16 2.25 2011-11-29T22:08:35 10 11 2.54545454545 2011-11-29T22:08:38 reprise of stats from yesterday 2011-11-29T22:08:39 http://en.cppreference.com/w/cpp/numeric/valarray what the hell? 2011-11-29T22:08:47 that's bj_v9 vs 2, 3, 4, ... players, number of games, average place 2011-11-29T22:08:51 I still suck at 6?! 2011-11-29T22:09:03 BenJackson: maybe the 6-player map is a terrible one 2011-11-29T22:09:36 std::valarray reminds me of Matlab 2011-11-29T22:09:58 reminds me of numpy :P 2011-11-29T22:10:00 if I parse that right, though, I won all of my 2 and 3 player games 2011-11-29T22:10:46 http://ants.fluxid.pl/map/cell_maze_p06_03.map\ 2011-11-29T22:10:49 http://ants.fluxid.pl/map/cell_maze_p06_03.map * 2011-11-29T22:11:00 Yes, and TrueSkill takes gradually less ranking info from games with more players 2011-11-29T22:11:42 one of my recent aichallenge games I lost due to starting next to xathis 2011-11-29T22:11:57 at that moment I really felt it should take WHO took your hill into account and not just who else was in the game :) 2011-11-29T22:14:33 *** bmh has joined #aichallenge 2011-11-29T22:18:09 *** dvladim has joined #aichallenge 2011-11-29T22:18:12 well I won't paste in my ranks by map :) 2011-11-29T22:18:27 *** JorgeB has joined #aichallenge 2011-11-29T22:18:40 4.0 2 cell_maze_p05_06.map 2011-11-29T22:18:40 4.33333333333 3 random_walk_09p_02.map 2011-11-29T22:18:40 4.5 2 random_walk_06p_01.map 2011-11-29T22:18:44 those are the ones I suck at though 2011-11-29T22:18:53 random walks suck 2011-11-29T22:19:08 the 'random' part of their name holds true 2011-11-29T22:19:45 thestinger: any of your bots you care to know about? 2011-11-29T22:19:53 http://ants.fluxid.pl/player/strcat_foodfight 2011-11-29T22:19:56 well 2011-11-29T22:19:58 actually no 2011-11-29T22:20:07 since it was broken until early today :P 2011-11-29T22:20:25 if you can tell me the last n pages of results to use I can 2011-11-29T22:20:50 I've played 152 games on fluxid overnight, I'm sure you have lots of data 2011-11-29T22:20:56 http://ants.fluxid.pl/replay.5222 WTF? 2011-11-29T22:21:11 BenJackson: ah, try the first page 2011-11-29T22:21:19 BenJackson: stop parking four ants around your hill. It makes me sad 2011-11-29T22:21:32 on your server? 2011-11-29T22:21:40 bmh: they used to massacre me before I had prediction of enemy movement 2011-11-29T22:21:46 nah, on fluxid 2011-11-29T22:21:49 they give the illusion of static defense :P 2011-11-29T22:21:52 *** treeform has joined #aichallenge 2011-11-29T22:21:56 they're really just like assigned dynamic defenders 2011-11-29T22:21:59 they are just my static defense code never removed 2011-11-29T22:22:16 mainly they're there to ensure I have hill vision 2011-11-29T22:22:33 although the recent game I didn't win outright on aichallenge was due to really crap defending 2011-11-29T22:27:01 *** Rav3nSw0rd has quit IRC (Excess Flood) 2011-11-29T22:27:41 *** Rav3nSw0rd has joined #aichallenge 2011-11-29T22:28:42 http://tcpants.com/replay.12187 2011-11-29T22:28:44 note how purple got taken 2011-11-29T22:28:53 big defense vs me, blue walks in the back door 2011-11-29T22:29:22 thats because my bot is the lame :) 2011-11-29T22:29:36 bugnuts: my defense weights enemies by distance to my hill 2011-11-29T22:29:43 it sucks for other reasons but that part of defending works :) 2011-11-29T22:30:10 yeah, my problem is I am greedy on patching to defence points which means it won't go back to defend the hill. 2011-11-29T22:30:43 my problem is that I prioritize ants closer to my hill as defenders 2011-11-29T22:30:51 so they all end up defending despite the # being limited 2011-11-29T22:31:02 you should look at xathis's defense vs me on aichallenge 2011-11-29T22:31:35 you can see (at least I think) some really early movement to encircle my sortie 2011-11-29T22:32:01 I also over allocate to defense since I don;t pay enough attention to attacker distance 2011-11-29T22:32:18 BenJackson: it would be awesome to get that map data stuff onto a stats page for the tcp server 2011-11-29T22:32:40 does the tcp server use a real db? 2011-11-29T22:32:42 so in a lot of matches I will get starved by a small number of attackers 2011-11-29T22:32:50 *** treeform_ has joined #aichallenge 2011-11-29T22:33:00 bugnuts: I think that's why I'm getting poached more 2011-11-29T22:33:05 mostly the real problem is I have no combat :( 2011-11-29T22:33:10 V7 would definitely collapse onto the base in defense 2011-11-29T22:33:16 and after that I'd never recover 2011-11-29T22:33:28 BenJackson: yeah, it uses sqlite 2011-11-29T22:33:33 new theory is that the home ants better hold out because if I stop foraging to save home I'll just lose later 2011-11-29T22:33:35 McLeopold: what do you think about this? http://forums.aichallenge.org/viewtopic.php?f=21&t=2002 2011-11-29T22:33:40 BenJackson: https://github.com/berak/ants-tcp/blob/master/sql.py 2011-11-29T22:33:41 contestbot: seen McLeopold 2011-11-29T22:33:41 amstan: McLeopold was last seen in #aichallenge 5 hours, 41 minutes, and 40 seconds ago: python -m cProfile MyBot.py < 0.bot.input 2011-11-29T22:34:28 *** treeform has quit IRC (Read error: Operation timed out) 2011-11-29T22:35:35 *** avdg has quit IRC (Quit: Leaving.) 2011-11-29T22:35:54 BenJackson: I don't assign static defenders for that reason 2011-11-29T22:36:01 because if I'm not foraging, I'm screwed anyway 2011-11-29T22:36:11 I need an ant to stay there and keep vision of the area around my hill though 2011-11-29T22:36:23 I'll just always make a square around my hill a forage target 2011-11-29T22:36:37 BenJackson: awesome visualizations 2011-11-29T22:36:53 BenJackson: have a sticky! 2011-11-29T22:37:02 thanks! 2011-11-29T22:37:08 *** AntDroid_ has quit IRC (Ping timeout: 265 seconds) 2011-11-29T22:37:29 http://tcpants.com/replay.12109 2011-11-29T22:37:39 I wonder if "ants not razing hills" could have an exception for combat 2011-11-29T22:37:44 not like I wasn't trying! 2011-11-29T22:38:18 BenJackson: did you see that replay where my bot was 3 steps from the hill and that happened? 2011-11-29T22:38:20 xD 2011-11-29T22:38:28 3 steps from an undefended hill 2011-11-29T22:40:23 BenJackson: what kind of exception? 2011-11-29T22:40:36 enemy ants near hill? 2011-11-29T22:40:47 I assume it's just to avoid wasting server time on starter bots or forage-only bots 2011-11-29T22:40:58 or standoffs where ants aren't killing each other 2011-11-29T22:41:04 we can waste time on that 2011-11-29T22:41:05 if ants are actually dying, it shouldn't end 2011-11-29T22:51:52 I hate that map... 2011-11-29T22:52:23 *** cuademon has joined #aichallenge 2011-11-29T22:52:38 BenJackson: why are your ants running into the corner? 2011-11-29T22:52:53 *** ljerez has joined #aichallenge 2011-11-29T22:53:20 bmh where? 2011-11-29T22:53:30 *** ljerez has left #aichallenge 2011-11-29T22:53:40 *** Ox86 has joined #aichallenge 2011-11-29T22:54:49 *** pairofdice has quit IRC (Ping timeout: 252 seconds) 2011-11-29T22:54:53 BenJackson: that one game where your ants were going in the dead end 2011-11-29T22:55:01 or was that you? 2011-11-29T22:55:18 wait.. nvm, that's another game i think, nvm 2011-11-29T22:56:05 i still do not understand why ants constantly vibrate 2011-11-29T22:56:41 wow, -flto makes my pathfinding even faster 2011-11-29T22:57:15 30ms for 5 targets on a 2000x2000 map :), can't really measure it on 200x200 maps anymore 2011-11-29T22:57:27 lol 2011-11-29T22:57:41 i think my mapgen would take hours to generate maps that big 2011-11-29T22:58:39 lets see what PGO does to it now xD 2011-11-29T22:59:28 nice 2011-11-29T22:59:31 2x as fast with PGO 2011-11-29T22:59:50 *** TheLinker has quit IRC (Quit: Bye) 2011-11-29T22:59:55 just -fprofile-generate, run my little test benchmark and recompile with -fprofile-use 2011-11-29T23:00:01 *** smiley1983 has quit IRC (Ping timeout: 244 seconds) 2011-11-29T23:00:36 *** AntDroid has joined #aichallenge 2011-11-29T23:00:46 oh 2011-11-29T23:00:57 it's just unrolling a loop 2011-11-29T23:01:09 ah, it unrolls the inner loop for going in each of the 4 distances 2011-11-29T23:01:30 and does some weird assembly magic I don't understand 2011-11-29T23:02:11 *** smiley1983 has joined #aichallenge 2011-11-29T23:04:49 *** pguillory has joined #aichallenge 2011-11-29T23:07:50 amstan: ants vibrate because evaluation functions are independent between turns and something about one move tips the balance in favor of another 2011-11-29T23:08:25 thestinger: of course youc an't -fprofile-use on the real contest site... 2011-11-29T23:08:43 BenJackson: I could precompile it :) 2011-11-29T23:08:59 also I can just figure out what PGO is doing and use the gcc attributes and stuff 2011-11-29T23:09:02 *** bmh has quit IRC (Ping timeout: 265 seconds) 2011-11-29T23:09:13 likely() and unlikely() macros etc. 2011-11-29T23:18:23 *** Jak_o_Shadows1 has quit IRC (Ping timeout: 248 seconds) 2011-11-29T23:18:52 *** dvladim has quit IRC (Ping timeout: 258 seconds) 2011-11-29T23:24:23 *** rajanaresh has quit IRC (Ping timeout: 244 seconds) 2011-11-29T23:24:30 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-29T23:26:47 *** Jak_o_Shadows has joined #aichallenge 2011-11-29T23:31:57 *** Rav3nSw0rd has quit IRC (Ping timeout: 260 seconds) 2011-11-29T23:49:31 *** u_ has quit IRC (Quit: u_) 2011-11-29T23:51:46 *** smiley1983 has quit IRC (Quit: Lost terminal) 2011-11-29T23:55:09 *** sf17k has joined #aichallenge 2011-11-29T23:55:11 *** delt0r_ has quit IRC (Ping timeout: 248 seconds) 2011-11-29T23:58:35 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-11-29T23:58:37 *** chris__0076 has joined #aichallenge 2011-11-29T23:59:14 *** Chris_0076 has quit IRC (Read error: Connection reset by peer)