2011-12-15T00:00:13 I never thought to add special code to fk around in games vs starter ants 2011-12-15T00:00:57 *** kevlar_work has quit IRC (Ping timeout: 255 seconds) 2011-12-15T00:01:15 oh I see, if you do that in every game 2011-12-15T00:01:17 *** kevlar_work has joined #aichallenge 2011-12-15T00:01:19 you'll ALWAYS play starter ants 2011-12-15T00:03:17 *** TheLinker has quit IRC (Quit: Bye) 2011-12-15T00:06:04 *** dom7b5 has quit IRC (Quit: Ex-Chat) 2011-12-15T00:06:11 *** dom7b5 has joined #aichallenge 2011-12-15T00:09:18 *** jstemmer has quit IRC (Ping timeout: 255 seconds) 2011-12-15T00:10:16 *** dom7b5 has quit IRC (Ping timeout: 240 seconds) 2011-12-15T00:12:29 *** Paradoxial has quit IRC (Quit: :) 2011-12-15T00:15:45 *** jstemmer has joined #aichallenge 2011-12-15T00:16:59 *** bqf has quit IRC (Changing host) 2011-12-15T00:16:59 *** bqf has joined #aichallenge 2011-12-15T00:18:27 *** AlliedEnvy has quit IRC (Ping timeout: 240 seconds) 2011-12-15T00:21:46 *** UncleVasya has joined #aichallenge 2011-12-15T00:22:46 *** Areks has joined #aichallenge 2011-12-15T00:23:19 *** AlliedEnvy has joined #aichallenge 2011-12-15T00:23:25 amstan, hi. Greedy bot now in the 3072th place. 2011-12-15T00:24:54 original greedybot? 2011-12-15T00:25:34 so it looks like on PlanetWars the precentage of real contestants was a little higher 2011-12-15T00:25:37 yes 2011-12-15T00:26:08 BenJackson: http://aichallenge.org/profile.php?user=12359 2011-12-15T00:26:12 McLeopold: Query idea: average score of everything that appears to be a starter kit (by language) 2011-12-15T00:26:28 to see if some starter kit got the spec wrong 2011-12-15T00:27:16 Althought it gets timeout often 2011-12-15T00:27:32 *** Antimony has joined #aichallenge 2011-12-15T00:27:57 greedybot gets better in the endgame when there's a ton of unconsumed food 2011-12-15T00:28:02 until it times out 2011-12-15T00:29:00 need to go to the study-place 2011-12-15T00:29:04 *** UncleVasya has quit IRC () 2011-12-15T00:32:03 whats the process for coming up with the next ai challenge? 2011-12-15T00:32:18 talking in the suggestion area of the forums 2011-12-15T00:32:36 whens the next one scheduled for? 2011-12-15T00:32:46 or is it pretty ad-hoc? 2011-12-15T00:32:57 I only heard someone comment offhand "at least 3 months away" 2011-12-15T00:34:10 *** nplus has quit IRC (Quit: ChatZilla 0.9.87-rdmsoft [XULRunner 1.9.0.17/2009122204]) 2011-12-15T00:34:25 *** nek has joined #aichallenge 2011-12-15T00:36:57 *** alc has quit IRC (Ping timeout: 252 seconds) 2011-12-15T00:41:15 *** nplus has joined #aichallenge 2011-12-15T00:42:19 *** justin_pdx has joined #aichallenge 2011-12-15T00:43:36 *** Fandekasp has quit IRC (Quit: Lost terminal) 2011-12-15T00:58:16 *** justin_pdx has quit IRC (Quit: justin_pdx) 2011-12-15T01:13:18 wow I don't know why exactly but my last test game at home shows my dev bot is super meticulous 2011-12-15T01:13:29 it built up a lead and then took like 600 turns to finish the enemy 2011-12-15T01:13:48 *** scribble has joined #aichallenge 2011-12-15T01:14:24 *** goldcaddy77 has quit IRC (Quit: Page closed) 2011-12-15T01:14:42 *** GeorgeSebastian has joined #aichallenge 2011-12-15T01:16:55 *** ltriant has quit IRC (Quit: Get MacIrssi - http://www.sysctl.co.uk/projects/macirssi/) 2011-12-15T01:20:46 *** Yexo has quit IRC (Read error: Connection reset by peer) 2011-12-15T01:20:59 holy shit, I may have just found The Missing Heuristic 2011-12-15T01:21:10 *** Palmik has joined #aichallenge 2011-12-15T01:28:05 *** bhasker has joined #aichallenge 2011-12-15T01:28:15 hi 2011-12-15T01:40:04 hi 2011-12-15T01:52:24 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-12-15T01:59:25 heaven help me I'm messing with my food gathering 2011-12-15T02:01:08 *** codetiger has joined #aichallenge 2011-12-15T02:03:18 *** Kurnevsky has joined #aichallenge 2011-12-15T02:05:14 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-15T02:10:16 BenJackson: :P 2011-12-15T02:12:06 antimatroid :P 2011-12-15T02:12:28 codetiger: you joined too late :P 2011-12-15T02:12:36 he said before he's messing with his food gathering code 2011-12-15T02:13:40 I was messing with diffusion until a couple of days back 2011-12-15T02:13:54 Just added some combat 2011-12-15T02:13:56 i'm still messing with everything 2011-12-15T02:14:09 and you guys on tcp kick by bots butt 2011-12-15T02:14:09 i noticed some friendly collisions before so i'm about to work out how the fuck that happened 2011-12-15T02:14:45 i've re-upload my bot for 6 days in a row, and every new upload seems to have a lower skill than the last for almost the same number of games. :( 2011-12-15T02:14:53 meanwhile, I just found running on to food is ignored 2011-12-15T02:15:18 TL_: you are not alone 2011-12-15T02:16:53 Mine once had 75 skills and now every update is struggling to reach 72 and my new bot is still struggling to reach 70 2011-12-15T02:17:29 I just feel sick, every other player on TCP is better than me 2011-12-15T02:17:34 finally broke into top 30 on my latest submit 2011-12-15T02:17:42 codetiger: tcp is where the top bots test 2011-12-15T02:17:42 codetiger: tcp is tough 2011-12-15T02:18:25 tcp has less than 1% the bots of the challenge but msot of the players in the top 10 2011-12-15T02:18:41 and some people are playing more than one entry 2011-12-15T02:18:54 now i just need to find one of these collisions while debugging is turned on :\ 2011-12-15T02:18:54 codetiger: you know about bhickey.net, right? 2011-12-15T02:18:56 many people are playing more than 10 entries 2011-12-15T02:19:55 yeah, my bot came first in bhickey.net 2011-12-15T02:20:07 so I thought it was not a good place for my bot 2011-12-15T02:20:46 and there was only one player bj_v4 and bj_v5 to compete. So I was not able to judge my bots performance 2011-12-15T02:21:10 I will switch to v7 if you want 2011-12-15T02:21:26 sure 2011-12-15T02:21:38 but isn't that server down couple of days 2011-12-15T02:21:52 well my bots were playing games 2011-12-15T02:21:58 but you're right that I can't get to it via the web 2011-12-15T02:22:05 @seen bmh 2011-12-15T02:22:05 BenJackson: bmh was last seen in #aichallenge 1 day, 5 hours, 57 minutes, and 50 seconds ago: BenJackson: can you tell me if bhickey.net:2080 is alive? I have a really slow internet connectoin at the moment 2011-12-15T02:22:50 its not live.. the web end is down 2011-12-15T02:23:06 the server was playing games until just a minute ago at least 2011-12-15T02:23:42 yes, playing games is still possible, but I want to check my bot's performance. using visualizer. 2011-12-15T02:23:44 :( 2011-12-15T02:24:11 codetiger: on aichallenge or tcp severs (skill drop)? 2011-12-15T02:24:14 I emailed him 2011-12-15T02:24:44 BenJackson: Thanks for that... 2011-12-15T02:25:09 GarfTop: aichallenge, but with every new submit 2011-12-15T02:25:33 and I guess that was my fault with some strategy tuning 2011-12-15T02:26:05 even tcpants.com is very tough 2011-12-15T02:26:26 I submitted a bot that was about 3 mu better on fluxid, and it dropped :-/ 2011-12-15T02:27:07 fluxid, it happens 2011-12-15T02:27:21 well it dropped on aichallenge, I mean 2011-12-15T02:27:49 I'm confused now whether it was really better and skills are being pushed down (which would be somewhat expected), or fluxid is not representative 2011-12-15T02:28:02 fluxid skills are pretty wild too 2011-12-15T02:28:10 there are so many more games 2011-12-15T02:28:15 and the population changes 2011-12-15T02:28:20 I always run 2 bots simultanously on tcp, so drift shouldn't affect me there 2011-12-15T02:28:30 my bot will get to 3rd or something and I'll quit playing and a few hours later it's at 10 2011-12-15T02:29:22 BJ, how do you guys test your bots. I know testing it against my old version is not good idea 2011-12-15T02:29:46 codetiger: most of my testing is done with very specific situations i set up myself 2011-12-15T02:29:47 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-15T02:29:51 I use my old versions, but I have strong old versions 2011-12-15T02:30:03 then i may run a couple of games or something on tcp just to look for flaws etc. 2011-12-15T02:30:17 BenJackson: the problem with that is often your old bots and new bots think a lot alike 2011-12-15T02:30:26 ie. with combat they often consider similar sets of moves etc. 2011-12-15T02:30:38 I've had 3 top 20 bots that are pretty significantly different 2011-12-15T02:30:49 fair enough 2011-12-15T02:30:55 plus I have even older ones that represent players that blunder around with no combat 2011-12-15T02:31:11 i try to think my strategy through before implementing, although details change as i discover what's feasible and think of other things 2011-12-15T02:31:12 also I set things up to give old bots an advantage 2011-12-15T02:31:19 tell them where my new bot's hill is 2011-12-15T02:31:30 use a map with close neighbors and give an old one a weak opponent 2011-12-15T02:31:43 i can't even find the old game with collisions now :\ 2011-12-15T02:31:55 there was one part of a game where it just started suiciding bots in one region :\ 2011-12-15T02:32:04 and generally I don't play on fluxid or resubmit until my new version is reliably killing all my old versions 2011-12-15T02:32:13 *** GeorgeSebastian has joined #aichallenge 2011-12-15T02:32:24 The biggest problem with my bot right now is playing with a close neighbhour 2011-12-15T02:32:47 on my local it does 2011-12-15T02:32:48 yeah results seem quite random to me on those maps? 2011-12-15T02:32:52 I've also used setups like extremely low food 2011-12-15T02:33:23 may be I should try, giving advantage to old versions like you said 2011-12-15T02:33:47 BenJackson: do you timeout on tcp much? 2011-12-15T02:33:49 you could also try having your new version wait a few turns before moving 2011-12-15T02:33:57 antimatroid: I don't think so 2011-12-15T02:34:03 but I haven't been looking at fluxid results today 2011-12-15T02:34:04 i get a few timeouts, yet if debugging is one it usually seems to think it returned go well within the time limit 2011-12-15T02:34:10 on* 2011-12-15T02:34:10 http://aichallenge.org/visualizer.php?game=239093&user=757 I like how SDIL behaves 2011-12-15T02:34:17 180ms rtt to fluxid.pl for me 2011-12-15T02:34:36 ~340ms for me 2011-12-15T02:35:15 but that should still stay well under 5s or whatever it is? 2011-12-15T02:35:17 hmmmm 2011-12-15T02:35:22 Fluxid: is that an intentonal "windshield wiper"? 2011-12-15T02:36:04 *** bmh has joined #aichallenge 2011-12-15T02:36:09 BenJackson: fixed. 2011-12-15T02:36:24 codetiger: 2011-12-15T02:36:37 Fluxid: it's like a more robotic xathis 2011-12-15T02:36:38 yes, am coming there 2011-12-15T02:37:01 sorry for being the least attentive sysadmin ever. 2011-12-15T02:37:24 codetiger: putting up v5 and v7 2011-12-15T02:37:36 Fluxid: also pretty flower at the end 2011-12-15T02:37:52 BenJackson: can you pull v4 (and possibly v5) off my server? I don't think it's producing useful data 2011-12-15T02:38:04 v4 is gone 2011-12-15T02:38:11 I can take down 5 too 2011-12-15T02:38:19 v5 accounts for half of all the games played :) 2011-12-15T02:39:24 so, is it just me or is it taking a lot longer for new versions to make their way back up to a stable skill level than it did before? 2011-12-15T02:39:37 kevlar: I see the same 2011-12-15T02:39:42 BenJackson: yeah, i like how he creates 1 ant ant thin walls, if he had more attackers than ants who wander around, and make wanderers more distributed... 2011-12-15T02:39:49 anyway, off to work 2011-12-15T02:40:06 BenJackson: thanks for the e-mail. I'm going to bed 2011-12-15T02:40:08 *** bmh has quit IRC (Client Quit) 2011-12-15T02:40:25 kevlar: more games but less time 2011-12-15T02:42:20 BJ: am on tcp now 2011-12-15T02:42:30 *** GarfTop has quit IRC (Quit: Make a new plan, Stan!) 2011-12-15T02:42:46 i'm waiting for a tcp game to end so i can look for friendly collisions :) 2011-12-15T02:43:09 damn they're slow! :P 2011-12-15T02:43:24 fluxid is slow 2011-12-15T02:43:30 lol 2011-12-15T02:43:39 yay maxing out the turn time 2011-12-15T02:43:55 well, here goes V12 on fluxid 2011-12-15T02:44:07 Fluxid: try not to trash the db tonight :) 2011-12-15T02:44:25 shhh, no giving him ideas. 2011-12-15T02:44:40 BJ or antimatrid, if its not too much to ask. Can you give me a review about my bot on TCP Bhickey? 2011-12-15T02:45:10 well you're #1 2011-12-15T02:45:32 but if you lose to bj_v4 that's bad 2011-12-15T02:45:38 thats because, I left it to play 100 games in a row 2011-12-15T02:46:05 the key to beating bj_v4 is minimal combat 2011-12-15T02:46:13 basically anything that ensures you have 2 ants where they have 1 2011-12-15T02:46:56 mine does good with bj_v4 if its not close neighbour multihill maze 2011-12-15T02:47:08 about all v5 has over v4 is minimal combat 2011-12-15T02:47:12 and it slaughters v4 2011-12-15T02:47:32 V7 tunes the combat and adds better food gathering and exploration 2011-12-15T02:48:12 *** Kurnevsky has left #aichallenge 2011-12-15T02:48:18 my V4 strategy and still a big part through V7 was "have more ants, try to run over people" 2011-12-15T02:48:32 which works against similar ants 2011-12-15T02:48:39 but against better combat it's suicide 2011-12-15T02:48:50 mine crossed 850 turns with your v7 so am happy 2011-12-15T02:49:31 what factors do I consider before choosing suicide? 2011-12-15T02:49:33 i'm at turn 740 :\ 2011-12-15T02:49:41 don't choose suicide 2011-12-15T02:49:44 suicide is usually bad 2011-12-15T02:49:56 in > 1v1 if you suicide everyone else benefits 2011-12-15T02:50:35 better to retreat and send another ant to form a wall 2011-12-15T02:50:45 BenJackson: is your evaluation of battle states secret? 2011-12-15T02:51:06 I think one flaw in this game is that there's no "easy" combat (like diffusion is "easy" strategy) 2011-12-15T02:51:10 when i get to comparing ant deaths, atm i tiebreak off enemyDeaths-friendDeaths then other stuff 2011-12-15T02:51:24 yeah, I do eventually trade ants 2011-12-15T02:51:28 but i'm wondering if 1 death for 2 kills, or n deaths for n+kills is worthwhile? 2011-12-15T02:51:29 but almost never 1 on 1 2011-12-15T02:51:36 maybe just minimise friend deaths then maximise enemy deaths? 2011-12-15T02:51:40 for me that depends on board state, location, etc 2011-12-15T02:51:48 how is that a flaw? 2011-12-15T02:51:55 combat is what makes this game imo 2011-12-15T02:52:06 lots of strong bots participated just this December. protocolocon is climbing. that SDIL may make it to the top15 2011-12-15T02:52:09 antimatroid: would be better if the AI writing had an easier learning curve for new participants 2011-12-15T02:52:44 peopel manage to gather food and get stuck trying to do any combat 2011-12-15T02:53:27 antimatroid: no, not a state secret 2011-12-15T02:53:40 probably more complex than necessary now that I've seen other ideas 2011-12-15T02:53:42 hmmm, i like the games to be simple enough for beginners to be able to participate and compete with each other but have enough depth that people like us are still interested in competing 2011-12-15T02:54:05 other ideas? for battle? 2011-12-15T02:54:12 codetiger: http://bhickey.net:2080/replay.14678 nice work 2011-12-15T02:54:12 i was a big proponent for this battle resolution 2011-12-15T02:54:28 I thikn the bug in V7 there is that super long paths cause it to not bother to explore 2011-12-15T02:54:58 most other candidates just favoured spreading your ants out and not getting more than one ant involved in a battle etc. 2011-12-15T02:55:01 which is very boring 2011-12-15T02:55:10 antimatroid: anyway my original battle plan was to (conceptutally) make a matrix of all possible enemy moves vs all my possible moves 2011-12-15T02:55:19 lol :P 2011-12-15T02:55:19 then try variations of my moves against all enemy moves simultaneously 2011-12-15T02:55:22 and maximize a score function 2011-12-15T02:55:32 but not variations of *their* moves 2011-12-15T02:55:43 you do realise how large that matrix would be right? 2011-12-15T02:55:52 it's not very big 2011-12-15T02:55:56 only participating ants 2011-12-15T02:56:05 and it's 5*n, not 5^n 2011-12-15T02:56:15 how? 2011-12-15T02:56:30 take a piece of graph paper 2011-12-15T02:56:31 do you guys think hill defense, is worth it? I removed it few versions back and now mine does better. But I think my bot still needs it. 2011-12-15T02:56:33 if you have {N,E} x {W,S} there's 4 combinations 2011-12-15T02:56:47 across the top write "ant 0" and for ant 0 have 5 moves (nsew, still) 2011-12-15T02:56:52 for each ant 2011-12-15T02:56:55 ohhhh 2011-12-15T02:57:03 down the side put the locations of everywhere an enemy could go 2011-12-15T02:57:07 yeah okay 2011-12-15T02:57:09 where they are now +4 2011-12-15T02:57:12 what are you doing now for battle? 2011-12-15T02:57:15 put a '1' anywhere they're in range 2011-12-15T02:57:24 do you use possible battle enemy/friend information when picking moves? 2011-12-15T02:57:38 then maximize the set of my moves that avoids having any enemy threatened exactly once 2011-12-15T02:57:45 0 times is ok, >=2 is good 2011-12-15T02:57:47 1 is bad 2011-12-15T02:58:03 it's basically perfect for 2 on 1 but trades 2 for 2 and above 2011-12-15T02:58:33 that's all V5 and V7 do, with some misc side factors so that "don't care" moves do something useful 2011-12-15T02:59:07 do you minimax your possible move sets? 2011-12-15T02:59:25 i pick out like 5 sets of moves for my ants and 3-4 for them and run minimax on that 2011-12-15T02:59:37 no 2011-12-15T02:59:43 it's just a matter of picking good heuristics for picking out those sets of moves 2011-12-15T02:59:55 I try a few thousand combinations of moves and take the highest scoring one 2011-12-15T03:00:04 simulated annealing 2011-12-15T03:00:32 that's cooler 2011-12-15T03:00:58 i wanted to try adding in new sets of moves where dying ants move more defensively, but it's too messy to set up in the remaining days 2011-12-15T03:00:58 http://ants.fluxid.pl/replay.34397 2011-12-15T03:01:11 I think carlose_guia_v31_rc5 is standing on hills to control spawning there 2011-12-15T03:01:36 antimatroid: if *I* had it to do again I'd do cheap combat as described in the forum 2011-12-15T03:01:42 and spend my time optimizing strategy 2011-12-15T03:01:52 my combat is very cheap :) 2011-12-15T03:02:06 running minimax on like a 5x4 tree isn't very expensive 2011-12-15T03:02:20 although i need to do it for each battle partition 2011-12-15T03:03:03 BJ: saw that v12 razing hills crazy... Congrates 2011-12-15T03:03:15 I hope that game is evidence that I finally found a working anti-rotation code 2011-12-15T03:03:20 er heuristic 2011-12-15T03:03:34 blah i timed out again :\ 2011-12-15T03:03:39 antimatroid: at some point I started splitting combat into islands due to nonlinearities in my combat eval 2011-12-15T03:03:52 bot thinks it returned go in 79ms though 2011-12-15T03:03:55 also gives nice local information like relative strength 2011-12-15T03:05:12 * antimatroid waits for the game to turn up on fluxid.pl 2011-12-15T03:05:42 wow rank 1314 on fluxid 2011-12-15T03:05:52 BenJackson: you need donations to cover your disk space? :) 2011-12-15T03:06:31 bah, i debugged 79ms after outputting go and still timed out :\ 2011-12-15T03:06:51 either i'm starting my timer in the wrong place or my network really is that bad 2011-12-15T03:06:59 BenJackson: when do you start your timer? 2011-12-15T03:07:05 i start it when I receive "turn" 2011-12-15T03:07:20 I don't actually use a timer for anything 2011-12-15T03:07:28 but the C++ starter starts a timer and prints it in the debug 2011-12-15T03:07:34 i wrote that bot :P 2011-12-15T03:07:38 it starts it in a terrible place 2011-12-15T03:07:58 it starts it when it receives "go" 2011-12-15T03:08:02 the only concession I make to timing is to become increasingly suicidal over 200 ants 2011-12-15T03:08:05 if you do much while reading input that's bad 2011-12-15T03:08:16 although i think it's still acceptable for the starter bot 2011-12-15T03:08:32 (i would do it differently if doing it again though obviously :P) 2011-12-15T03:09:08 do you guess what type of map is it? 2011-12-15T03:09:10 I'm amazed I won that last aichallenge.org game 2011-12-15T03:09:15 damn red poached my kill 2011-12-15T03:09:27 still waiting on this game :( 2011-12-15T03:09:40 antimatroid: when you lose on fluxid you ahve to wait for someone to win 2011-12-15T03:09:44 i know 2011-12-15T03:09:57 still sucks :P 2011-12-15T03:10:02 i miss live tcp games 2011-12-15T03:10:15 wowww red nearly poached my last kill too 2011-12-15T03:10:20 it was cool watching planet wars tcp games as they played 2011-12-15T03:10:32 I might need to use the same code that protects my hill to protect enemy hills against other enemies 2011-12-15T03:10:43 of course you'd only really be able to do it with the information your bot receives 2011-12-15T03:11:04 I won that game by about 1 step 2011-12-15T03:11:06 BenJackson: hmmm, i don't protect enemy hills 2011-12-15T03:11:17 that might be a pita to set up information for 2011-12-15T03:11:17 look at the final move of http://aichallenge.org/visualizer.php?game=239612&user=3473 2011-12-15T03:11:24 red got an early lead by poaching purple 2011-12-15T03:12:25 my protection code just does a bfs out from hills to find enemies, then does an ordered bfs to an ant and moves it to meet the enemy on one of the enemies shortest paths to my hill 2011-12-15T03:12:26 do you do that? 2011-12-15T03:13:37 *** Garf has joined #aichallenge 2011-12-15T03:14:00 as soon as possible too, so it'll move ants along the shortest path towards the enemy 2011-12-15T03:14:29 so basically searches within the perimeter of my hill for an ant that can meet it at the hill if necessary, but as soon as possible 2011-12-15T03:16:21 antimatroid: what tcp server you are on? 2011-12-15T03:16:29 fluxid 2011-12-15T03:16:37 name? 2011-12-15T03:16:50 http://ants.fluxid.pl/replay.34373 2011-12-15T03:16:52 got my replay 2011-12-15T03:16:56 i'm SkyCaptain 2011-12-15T03:17:01 it's still a work in progrss 2011-12-15T03:17:14 some things work really well, i'm trying to fix the things that don't before submissions close 2011-12-15T03:17:32 i had the most ants when i timed out, but was way down in the rankings 2011-12-15T03:17:49 antimatroid: I was thinking along those lines today 2011-12-15T03:18:00 but the shortest path between two corners of a rectangle is... anywhere in the rectangle 2011-12-15T03:18:02 which lines? 2011-12-15T03:18:16 oh intercepting 2011-12-15T03:18:29 BenJackson: what do you mean by rectangle? pastebin ascii example? 2011-12-15T03:18:39 just pick two points in a grid 2011-12-15T03:18:41 they make a rectangle 2011-12-15T03:18:47 opposite corners 2011-12-15T03:18:52 there is the problem of when you step too far in one direction and their chosen shortest path is now shorter than yours 2011-12-15T03:18:57 all the manhattan paths through that rectangle are equal length 2011-12-15T03:19:15 yeah sure, just go onto one of the shortest paths 2011-12-15T03:19:34 but imagine one corner is your hill 2011-12-15T03:19:43 if you have closest friend hill information it's fairly easy to just do an ordered bfs 2011-12-15T03:19:44 the enemy ant can take either side of the rectangle first 2011-12-15T03:19:53 rather than actually generating the enemy ants shortest path to friend hill info 2011-12-15T03:19:58 yeah i know 2011-12-15T03:20:05 anyawy my V11 does something like you describe 2011-12-15T03:20:06 i haven't resolved that 2011-12-15T03:20:16 but i haven't seen it being much of a problem 2011-12-15T03:20:20 V12 I'm working on is more sophisticated 2011-12-15T03:20:30 I'm inspired by xathis 2011-12-15T03:21:09 V12 tries to make it a goal to already have ants in the way when enemies appear 2011-12-15T03:21:10 if you do it right, your ants will push enemies away from your hills 2011-12-15T03:21:20 *** ikaros has joined #aichallenge 2011-12-15T03:21:25 based on possible enemy info? 2011-12-15T03:21:32 i've considered that, but seems like a large use of ants 2011-12-15T03:21:47 if you already have heaps of ants it's unlikely they'll just pop up near your hill 2011-12-15T03:21:51 *** jacob_strauss has joined #aichallenge 2011-12-15T03:22:36 the other modification of V12 is maybe a successful attempt to avoid having my combat formations rotate sideways and let enemies slip by 2011-12-15T03:23:04 my combat is currently more offensive that it had been previously 2011-12-15T03:23:11 but i think it might actually be working better for it :\ 2011-12-15T03:23:21 I forgot to run my unit test for that, but it passed, yay 2011-12-15T03:23:47 when I tried to move beyond the "dumb" combat of V7 the lost "accidental aggressiveness" made V8 weaker for a long time 2011-12-15T03:23:53 I had to do a lot of work on that 2011-12-15T03:27:15 i plan to look at battle again once i work out what caused friends to collide before 2011-12-15T03:27:19 that's BAD :) 2011-12-15T03:28:12 my ants are very good at mopping up tmied out ants 2011-12-15T03:28:22 but sometimes I wonder if I should leave them as a defensive screen 2011-12-15T03:28:47 nah, claim the territory 2011-12-15T03:28:58 it would just be a pita the navigate around them 2011-12-15T03:29:46 ARRRRRRRRR I beat GreenTea... BUT he's ranked #175 right now 2011-12-15T03:30:15 *** jacob_strauss has quit IRC (Remote host closed the connection) 2011-12-15T03:30:42 *** aerique has joined #aichallenge 2011-12-15T03:30:47 *** AlliedEnvy has quit IRC (Ping timeout: 240 seconds) 2011-12-15T03:33:02 *** AlliedEnvy has joined #aichallenge 2011-12-15T03:34:11 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-15T03:34:13 bah, i want to find a collision :( 2011-12-15T03:34:32 just use the "bot in" from a game where you saw it 2011-12-15T03:38:24 *** Blkt has joined #aichallenge 2011-12-15T03:38:27 true 2011-12-15T03:38:31 now i just need to find that game :P 2011-12-15T03:39:02 <_flag> antimatroid: Nice bot btw 2011-12-15T03:39:16 _flag: some of it's nice, some of it sucks :P 2011-12-15T03:39:17 but cheers 2011-12-15T03:39:39 I'm getting psyched again. current aichallenge submit is on a trajectory for maybe 8th or 9th 2011-12-15T03:39:44 and I think I have something better going 2011-12-15T03:39:52 as long as everyone else takes a break... 2011-12-15T03:39:55 :P 2011-12-15T03:39:58 <_flag> Well, that's kind of like all bots, but I'm pretty sure if it stops timing out its top 10 quality 2011-12-15T03:40:00 *** mj41 has joined #aichallenge 2011-12-15T03:40:01 good luck with that last part 2011-12-15T03:40:20 my secret to not timing out is ignoring the timer 2011-12-15T03:40:30 i don't know why it's even timing out 2011-12-15T03:40:37 debug info claims it's returning "go" in under 100ms 2011-12-15T03:40:55 <_flag> tcp times out a lot for me just from latency 2011-12-15T03:41:03 that's what i think it is 2011-12-15T03:41:07 i'm glad you get the same 2011-12-15T03:41:13 good morning everyone 2011-12-15T03:41:14 i think i saw a1k0n time out too the other day 2011-12-15T03:41:14 _flag: still hard to believe because it's actual timeout is 5s 2011-12-15T03:41:18 <_flag> Well it happens when I have a lot of ants but I still have short turns 2011-12-15T03:41:29 its* 2011-12-15T03:41:33 <_flag> So I think the computational time of the engine contributes the timeout 2011-12-15T03:41:37 oops sorry :P 2011-12-15T03:41:39 <_flag> But that's just an untested theory 2011-12-15T03:41:44 i thought i was correcting myself 2011-12-15T03:42:03 I'm looking at Flag's profile 2011-12-15T03:42:09 _flag: when do you start your timer? 2011-12-15T03:42:13 skill 81.91 is only 53rd now? 2011-12-15T03:42:14 i start it when i receive "turn" 2011-12-15T03:42:14 wow 2011-12-15T03:42:35 _flag: I've read the engine and it shouldn't 2011-12-15T03:42:37 <_flag> antimatroid: after all the game information is initialized 2011-12-15T03:42:44 <_flag> But that time should be negligable 2011-12-15T03:42:52 i update information as i read the state info 2011-12-15T03:42:53 <_flag> BenJackson: Have you read the tcp source code? 2011-12-15T03:43:02 yes 2011-12-15T03:43:11 <_flag> Well I have no idea what it is then 2011-12-15T03:43:18 oh you think it's contributing to it for tcp? 2011-12-15T03:43:22 I've read the engine for the main contest 2011-12-15T03:43:36 <_flag> The tcp source code has a history of being buggy 2011-12-15T03:43:36 if tcp has a bug where it's sitting ona line or something I don't know about that 2011-12-15T03:43:43 yeah I found a bug the first time I looked at it :) 2011-12-15T03:44:03 <_flag> Yeah, there were some security issues, etc. 2011-12-15T03:44:21 <_flag> It was kind of released as an alpha and not fully tested and completed 2011-12-15T03:44:32 <_flag> And the developer of it disappeared 2011-12-15T03:44:45 *** sigh has joined #aichallenge 2011-12-15T03:45:09 v12 has won 4 in a row on fluxid. SHIP IT 2011-12-15T03:47:13 every replay I watch these days is a map I've never seen 2011-12-15T03:47:23 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-15T03:48:31 janzert: can those of us in the top 100 petition to get our profile pages in memcached? :) 2011-12-15T03:49:43 what i should do is 2 ply minimax 2011-12-15T03:50:01 I still don't see how min-max applies 2011-12-15T03:50:08 their moves aren't a response to your moves 2011-12-15T03:50:24 i know, but i try to pick their moves as a best response to my moves 2011-12-15T03:51:15 then i assume they do what's worst for me 2011-12-15T03:51:34 then i pick my moves which is the best out of those worst outcomes 2011-12-15T03:51:59 yeah but I've constructed examples where that gives the wrong answer 2011-12-15T03:52:03 it's no different to treating it like a payoff matrix then picking your strategy which maximises the worst outcome 2011-12-15T03:52:17 such as? 2011-12-15T03:52:22 I'd simplify it down to: let's write a min-max AI to play rock-paper-scissors 2011-12-15T03:52:36 ok, let's say I choose rock as default 2011-12-15T03:52:39 what would they do? 2011-12-15T03:52:40 paper! 2011-12-15T03:52:44 ok, my best move is then scissors 2011-12-15T03:53:12 scissors isn't *wrong* it's just based on your original assumption that you'd start with rock 2011-12-15T03:53:16 sure, but i don't think other bots are analysing my bot's moves to realise it's always going "rock" and i'm suspect of how many "rock paper scissor" scenarios actually exist? 2011-12-15T03:53:17 <_flag> Yes, but not all options are equal 2011-12-15T03:53:19 the same is true of ant combat, it's just harder to see 2011-12-15T03:53:49 _flag: yes, that's the interesting thing, and I found basically 0 literature on simultaneous move games like that 2011-12-15T03:54:11 take tron for example, that was simultaneous at each turn, yet minimax still reined supreme 2011-12-15T03:54:28 BenJackson: games like what? 2011-12-15T03:54:33 did it? I read a1k0n's writeup 2011-12-15T03:54:43 i just did my honours thesis on finite normal (simultaneous) form games 2011-12-15T03:54:48 antimatroid: the list of simultaneous move games fits on like 1 page 2011-12-15T03:54:53 rps, prisoner's dilemma 2011-12-15T03:55:05 but most of it was to do with structural properties preserved with equivalence of games and various definitions of symmetric games 2011-12-15T03:55:10 some similar games to prisoner's dilemma 2011-12-15T03:55:27 BenJackson: there's 144 strategically different (ordinally) 2x2 games :P 2011-12-15T03:55:35 yeah, there you go 2011-12-15T03:55:44 the literature is about Nash equilibria 2011-12-15T03:55:50 http://www.amazon.com/Topology-Games-Routledge-Advances-Theory/dp/0415336090 2011-12-15T03:56:00 that's just 2x2 2011-12-15T03:56:02 *** scribble has left #aichallenge 2011-12-15T03:56:26 you can *acquire* that book if you look 2011-12-15T03:57:10 BenJackson: do you know what a symmetric simultaneous game is? 2011-12-15T03:57:25 not the "symmetric" part 2011-12-15T03:57:31 damn 2011-12-15T03:57:43 most people who say yes to that have a very naive understanding 2011-12-15T03:58:30 I did look into simultaneous games, but finding only the kind that book is about I was not that interested for this contest 2011-12-15T03:58:49 I think there's an element of that in this game 2011-12-15T03:59:03 we all actually went through a phase where we became "rational" and quit suiciding each other 2011-12-15T03:59:39 I think the game rates on fluxid and aichallenge are going to converge 2011-12-15T04:02:13 so, the runtime actually kills bots that time out? 2011-12-15T04:03:38 I have been chasing down what I thought was my bot crashing when I think it's actually the runtime getting impatient and killing it :) 2011-12-15T04:04:07 *** GeorgeSebastian has joined #aichallenge 2011-12-15T04:04:09 (it occasionally spews nulls into the logs when that happens, which is what had me thinking it was coring or something) 2011-12-15T04:07:50 it does kill the timeout ones 2011-12-15T04:07:55 it also kills the exterminated ones 2011-12-15T04:08:00 but with a short shutdown grace period 2011-12-15T04:08:31 this fluxid game has been going on for 25 minutes 2011-12-15T04:09:42 according to my debug I'm winning it handily but turns are taking 2+ seconds each 2011-12-15T04:14:54 must be M24 2011-12-15T04:14:59 when he timed out the game finished fast 2011-12-15T04:16:30 *** ikaros has joined #aichallenge 2011-12-15T04:17:05 yay, i found the collisions 2011-12-15T04:17:14 i think i might have an idea of what caused it too 2011-12-15T04:17:58 http://ants.fluxid.pl/replay.33663 turn 853 loc (43, 50), what are you doing buddy? 2011-12-15T04:18:06 it does it quite a bit there 2011-12-15T04:18:10 but that's a specific example 2011-12-15T04:18:34 what i really don't get is why they're even walking over to that side of the passge 2011-12-15T04:18:38 that makes no friggin sense 2011-12-15T04:18:43 you know you can put &turn= ?row= ?col= 2011-12-15T04:19:11 nope :P 2011-12-15T04:19:37 looks like you block your guy from going S so he doesn't move but the guy next to him goes E anyway 2011-12-15T04:20:13 yeah, i'll run the bot in stuff and see where it's making the moves then go investigate 2011-12-15T04:20:25 should be easy to track down now, just requires some mehffort :P 2011-12-15T04:20:45 I don't know who chu is but he sure was patient while you surrounded him 2011-12-15T04:21:47 I'm put in mind of a little girl putting bows in the hair of a giant dog 2011-12-15T04:22:35 BenJackson: how're you watching your debug? 2011-12-15T04:22:50 what's mean to happen when i click ok to get bot input? 2011-12-15T04:22:51 my fluxid bot is just logging to a file 2011-12-15T04:22:53 it seems to do nothing? :\ 2011-12-15T04:23:03 antimatroid: takes a while if it's a long game 2011-12-15T04:23:04 BenJackson: are you using my bug thing? :P 2011-12-15T04:23:09 yes 2011-12-15T04:23:11 oh, i did it twice :P 2011-12-15T04:23:19 would you change anything to the bug thing? 2011-12-15T04:23:25 BenJackson: we can access the files that our bots write? 2011-12-15T04:23:37 I made the bug thing log to debug.pid.txt 2011-12-15T04:23:41 so they get unique names 2011-12-15T04:23:41 kevlar: sure, just keep reloading 2011-12-15T04:23:48 ah fair enough 2011-12-15T04:23:52 kevlar: you can access any files anything is writing 2011-12-15T04:23:59 O_o 2011-12-15T04:24:00 well, ok, maybe not on windows 2011-12-15T04:24:02 i should do that 2011-12-15T04:24:03 where? 2011-12-15T04:24:10 oh, you're not talking about live games 2011-12-15T04:24:19 * kevlar was very confused 2011-12-15T04:24:19 I'm talking about the tcp server 2011-12-15T04:24:20 fluxid.pl 2011-12-15T04:24:28 I don't have debug from my aichallenge games 2011-12-15T04:24:43 unless I download the game and run it through again 2011-12-15T04:24:44 what's fluxid.pl? 2011-12-15T04:24:56 BenJackson: how long is a while? :P 2011-12-15T04:25:03 what browser? 2011-12-15T04:25:08 chrome 2011-12-15T04:25:08 probably a lot faster on chrome than anything else 2011-12-15T04:25:14 is it in a tab? 2011-12-15T04:25:16 yeah 2011-12-15T04:25:23 I mean is your result in the next tab? 2011-12-15T04:25:31 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-15T04:25:36 nope 2011-12-15T04:25:43 but now i know what i;m looking for 2011-12-15T04:25:44 cheers 2011-12-15T04:25:48 i'll just wait a while 2011-12-15T04:26:14 did you get the two prompts? 2011-12-15T04:26:25 nope, just one 2011-12-15T04:26:25 firefox offers to block the second one 2011-12-15T04:26:29 maybe chrome is helping you 2011-12-15T04:26:35 1 for the name, 1 for the turn range 2011-12-15T04:26:41 oh yeah, sorry aha 2011-12-15T04:26:53 http://en.wikipedia.org/wiki/Symmetric_game#The_general_case that definition is wrong :P 2011-12-15T04:28:01 *** kaemo has joined #aichallenge 2011-12-15T04:28:49 *** 16SAADD24 <16SAADD24!~replore@203.152.213.161.static.zoot.jp> has quit IRC (Remote host closed the connection) 2011-12-15T04:30:31 kevlar: ants.fluxid.pl 2011-12-15T04:30:58 nooo it's too busy already 2011-12-15T04:31:28 if you're new try bhickey.net 2011-12-15T04:31:33 22 games, 46 players online 2011-12-15T04:31:37 wow 2011-12-15T04:31:46 http://bhickey.net:2080/ 2011-12-15T04:32:16 fluxid: I wasn't kidding when I said that soon aichallenge would be faster than fluxid.pl 2011-12-15T04:32:43 I think either M24 has bad net or he's ignoring the time limit 2011-12-15T04:32:56 game 34472 was terribly slow until he timed out 2011-12-15T04:33:31 he's also pretty strong and I've got no idea who he is on official which might also suggest flouting the time limit 2011-12-15T04:34:01 <_flag> Fluxid: A fair number of my games timeout and on my computer I never go over around 0.2 seconds 2011-12-15T04:34:23 <_flag> eg. http://ants.fluxid.pl/replay.34272 2011-12-15T04:35:01 _flag: is this just recently? 2011-12-15T04:35:06 cause my bot is loving to timeout today 2011-12-15T04:35:28 _flag: i'm afraid i can't do much about it 2011-12-15T04:35:47 i'm just glad it's not just me 2011-12-15T04:35:55 i'm still reasonably worried, but not as worried 2011-12-15T04:36:16 if you don't use all the limit, it's either network or you crash... 2011-12-15T04:36:55 <_flag> antimatroid: It has been more frequent recently 2011-12-15T04:37:02 can the tcp client not detect a crash? 2011-12-15T04:37:18 <_flag> Fluxid: I'm entirely sure that it's either of those things 2011-12-15T04:37:38 <_flag> I've run pings side by side with games and they still seem to end abruptly 2011-12-15T04:37:47 <_flag> It may be a problem with the server 2011-12-15T04:37:59 Fluxid: would it be possible to put like T next to peoples names when listing games to indicate they timed out? 2011-12-15T04:38:08 might make it easier to check the frequency of bots timing out 2011-12-15T04:38:24 <_flag> It just happened to me again :( 2011-12-15T04:38:24 i don't care about the timeouts provided it's not my bot crashing 2011-12-15T04:38:53 <_flag> And I've been talking and reading irc this entire time, so it's unlikely to be my connection 2011-12-15T04:39:31 *** Palmik_ has joined #aichallenge 2011-12-15T04:39:49 *** Palmik has quit IRC (Ping timeout: 252 seconds) 2011-12-15T04:40:48 antimatroid: not really, there is no such data in db, only in replay json 2011-12-15T04:41:02 *** mviel__ has joined #aichallenge 2011-12-15T04:41:28 _flag: and pings are all okay? could you run mtr to schranz.io for a while? 2011-12-15T04:42:28 or maybe tcpserver chokes, but i'm afraid i won't be able to do much with this... 2011-12-15T04:42:38 0.0% 42 182.7 183.2 179.1 203.2 4.6 2011-12-15T04:42:42 from west coast us 2011-12-15T04:42:51 <_flag> The pings seem to be fine, I'll test it more tommorow but for now I need to get some sleep 2011-12-15T04:43:41 night 2011-12-15T04:44:14 there's a reason any sort of real time game uses udp networking instead of tcp :/ 2011-12-15T04:44:47 *** mviel_ has quit IRC (Ping timeout: 252 seconds) 2011-12-15T04:45:26 <_flag> janzert: I was thinking that might have something to do with it too, waiting for acks and such, especially in games with large numbers of ants, may contribute to it 2011-12-15T04:48:13 http://ants.fluxid.pl/replay.34503 2011-12-15T04:48:17 thestinger: what a weird game 2011-12-15T04:48:23 your strcats walled themselves in 2011-12-15T04:54:06 *** kaemo has quit IRC (Ping timeout: 252 seconds) 2011-12-15T04:57:33 *** Palmik_ has quit IRC (Ping timeout: 245 seconds) 2011-12-15T04:58:09 ah, two good games on main server 2011-12-15T04:58:36 but only +0.31 to skill, lol 2011-12-15T05:00:44 *** Surya has joined #aichallenge 2011-12-15T05:00:58 http://aichallenge.org/visualizer.php?game=240095&user=757 bah, instead of spawnkilling my ants get onto enemy hill and die one by one 2011-12-15T05:03:25 *** Palmik_ has joined #aichallenge 2011-12-15T05:07:08 *** nha has joined #aichallenge 2011-12-15T05:11:46 i don't think this bot in thing is working for me :\ 2011-12-15T05:11:53 can someone paste bin it form me? 2011-12-15T05:12:04 http://ants.fluxid.pl/replay.33663 i need turns 1-860 2011-12-15T05:12:54 *** Samvara has joined #aichallenge 2011-12-15T05:13:38 Question: how do I download the map if I know the game number? I want to play my test games in the same map 2011-12-15T05:13:56 *** Cyndre_ has quit IRC (Read error: Connection reset by peer) 2011-12-15T05:14:25 http://aichallenge.org/visualizer.php?game=235320&user=31 i see starterbots timeouting... 2011-12-15T05:14:50 timing out* 2011-12-15T05:15:12 antimatroid: which one are you? 2011-12-15T05:15:26 SkyCaptain? 2011-12-15T05:15:27 SkyCaptain 2011-12-15T05:15:28 yeah 2011-12-15T05:16:09 something really weird happened in that game 2011-12-15T05:16:31 almost like my map information is wrong :\ 2011-12-15T05:16:35 pasting 2011-12-15T05:16:48 http://paste.pocoo.org/show/521178/ 2011-12-15T05:16:55 I think the tournament is more popular than anyone expected and the servers are getting hammered. Fortunately I write in C so clock cycles are something for other people to worry about :-) 2011-12-15T05:17:05 Yeah, you keep killing yourself around 45-50 2011-12-15T05:17:31 antimatroid: oh boy, this looks bad! 2011-12-15T05:17:49 yeah it's not good :P 2011-12-15T05:17:53 but hopefully i can just fix it 2011-12-15T05:17:56 cheers 2011-12-15T05:18:00 Samvara, I'm also looking for downloading maps directly from the viewer, I don't know if this feature exists 2011-12-15T05:19:17 I'd hack it up myself but I'm running out of time for genuine coding. I have to work tomorrow 2011-12-15T05:19:54 Surya: the more wtf thing is why are they even congregating on the dead end side of a corridor? 2011-12-15T05:20:00 i think they think there's an enemy or something 2011-12-15T05:21:31 even then they shouldn't kill themselves at all... 2011-12-15T05:21:36 Antimatroid: You don't have any protection so they don't go on each other? I mean, even with false informations, my ants wouldn't kill themselves 2011-12-15T05:21:49 fluxid: i know, there's 2 wtfs in one 2011-12-15T05:22:03 hope you'll find it 2011-12-15T05:22:10 i don't expect it to be hard 2011-12-15T05:22:16 oh, i know that it could be 2011-12-15T05:22:21 BETA PARTICLES 2011-12-15T05:22:33 once i get the debug info (i'm being lazy), it'll tell me exactly what I moved them for 2011-12-15T05:23:04 i don't have the 2am problem 2011-12-15T05:23:12 with debug info i can work out exactly what my bot is doing 2011-12-15T05:23:15 btw those logs almost never worked for me 2011-12-15T05:23:28 somehow seeds are often screwed 2011-12-15T05:23:42 i don't read the seed anyway 2011-12-15T05:23:47 ah, ok 2011-12-15T05:23:56 i have small randomness builtin 2011-12-15T05:23:57 You can get debug info back from the game? How does that work? 2011-12-15T05:24:16 Samvara: last button ont the right 2011-12-15T05:24:19 BOT IN 2011-12-15T05:24:19 i hate randomness, i like to be able to work out exactly what it did and why 2011-12-15T05:25:26 well, i use it only to choose one direction of movement if there one than one with highest weight 2011-12-15T05:25:53 i've considered randomly ordering each squares land neighbours in the first turn 2011-12-15T05:25:55 i may still do that 2011-12-15T05:26:02 so path finding isn't always favouring north etc. 2011-12-15T05:26:38 Antimatroid: Why did you insert randomness, i mean, for debugging purpose, I comment the random value and pick a constant. I doesn't change the behavior of my bot but make it possible to debug. 2011-12-15T05:26:57 i don't have any randomness 2011-12-15T05:27:05 Enemy have I guess 2011-12-15T05:27:07 ok 2011-12-15T05:28:56 Oh yeah, I know about that button... Hmmm, I get it that's what player_seed is for, so you can be deterministic and random and thus reconstruct the game. 2011-12-15T05:29:29 Thanks for the wake-up, I really have lost serious sleep and time trying to bang in a last minute entry 2011-12-15T05:29:37 *** mviel_ has joined #aichallenge 2011-12-15T05:32:28 *** mviel__ has quit IRC (Ping timeout: 276 seconds) 2011-12-15T05:32:35 *** mviel_ is now known as mviel 2011-12-15T05:33:30 *** nek has quit IRC (Ping timeout: 255 seconds) 2011-12-15T05:36:37 * antimatroid wonders if we should have distributed food less uniformly? 2011-12-15T05:36:50 i think my bot would favour the fact that it's fairly uniform 2011-12-15T05:36:58 and it would require a smarter bot to exploit it not being so 2011-12-15T05:37:21 ie. so ants could discover more patches of food 2011-12-15T05:38:03 *** greghaynes has quit IRC (Read error: Connection reset by peer) 2011-12-15T05:41:06 It would have added a big amount of luck 2011-12-15T05:42:12 *** greghaynes has joined #aichallenge 2011-12-15T05:45:16 aichallenge: janzert epsilon * r8fa9673 / manager/random_pairing.py : Script to pair rounds of randomly paired matches - http://git.io/rDw4Kw 2011-12-15T05:45:18 aichallenge: janzert epsilon * rc1f3d25 / (9 files in 4 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/ytvozQ 2011-12-15T05:47:15 *** greghaynes has quit IRC (Read error: Connection reset by peer) 2011-12-15T05:51:02 *** Akranis has joined #aichallenge 2011-12-15T05:54:28 *** Palmik_ has quit IRC (Remote host closed the connection) 2011-12-15T05:54:36 *** greghaynes has joined #aichallenge 2011-12-15T05:54:48 *** Palmik_ has joined #aichallenge 2011-12-15T05:58:38 grrr for starters my unseen border is incorrect 2011-12-15T06:01:13 *** greghaynes has quit IRC (Read error: Connection reset by peer) 2011-12-15T06:13:00 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-15T06:25:35 *** greghaynes has joined #aichallenge 2011-12-15T06:28:19 *** Areks has joined #aichallenge 2011-12-15T06:33:41 *** ikaros has quit IRC (Read error: Operation timed out) 2011-12-15T06:50:16 well i found one bug 2011-12-15T06:50:23 but now i have to revert it to find the collision bug :P 2011-12-15T06:55:24 *** pairofdice has joined #aichallenge 2011-12-15T06:57:59 *** mviel_ has joined #aichallenge 2011-12-15T06:58:32 *** olexs has joined #aichallenge 2011-12-15T07:01:26 *** mviel has quit IRC (Ping timeout: 248 seconds) 2011-12-15T07:10:08 *** mviel_ is now known as mviel 2011-12-15T07:10:12 *** Anilm3 has joined #aichallenge 2011-12-15T07:14:44 Hi 2011-12-15T07:15:18 iH 2011-12-15T07:18:39 Hi pairofdice, how's you bot doing? 2011-12-15T07:20:00 Dropped almost 100 places since I uploaded it 2011-12-15T07:20:09 Over a month ago 2011-12-15T07:22:54 Learning Java so haven't touched it 2011-12-15T07:24:54 ewww java :P 2011-12-15T07:27:38 *** QuirionPT has joined #aichallenge 2011-12-15T07:28:27 yeah ewww haha 2011-12-15T07:29:02 I don't understand why my bot is still gaining skill 2011-12-15T07:31:03 i think i fixed all my bugs from that game before 2011-12-15T07:31:37 *** ikaros has joined #aichallenge 2011-12-15T07:35:14 he better do well in this tcp game 2011-12-15T07:38:17 ....and i timed out 2011-12-15T07:38:18 typical 2011-12-15T07:41:08 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-15T07:41:39 *** TheLinker has joined #aichallenge 2011-12-15T07:44:49 *** liberforce has joined #aichallenge 2011-12-15T07:46:31 *** u_ has joined #aichallenge 2011-12-15T07:54:25 *** raemde_ has quit IRC (Read error: Connection reset by peer) 2011-12-15T07:54:33 http://ants.fluxid.pl/replay.34780 things were going so well :( 2011-12-15T07:54:46 *** raemde_ has joined #aichallenge 2011-12-15T08:13:26 wow xathis is getting near 100 2011-12-15T08:14:05 kill it with ants! 2011-12-15T08:14:58 i;m waiting for (i think) thestinger to finish me 2011-12-15T08:17:53 *** Anilm3_ has joined #aichallenge 2011-12-15T08:18:16 Anilm3: leave, dammit 2011-12-15T08:18:52 *** Anilm3 has quit IRC (Ping timeout: 244 seconds) 2011-12-15T08:18:56 thanks 2011-12-15T08:18:58 *** Anilm3_ is now known as Anilm3 2011-12-15T08:21:06 *** sigh has quit IRC (Remote host closed the connection) 2011-12-15T08:23:27 *** BEASTMODE has quit IRC (Remote host closed the connection) 2011-12-15T08:23:27 *** denysonique__ has quit IRC (Read error: Connection reset by peer) 2011-12-15T08:23:27 *** nickjohnson has quit IRC (Remote host closed the connection) 2011-12-15T08:23:48 *** BEASTMODE has joined #aichallenge 2011-12-15T08:31:29 *** nickjohnson has joined #aichallenge 2011-12-15T08:35:34 if your ant has seen an enemy hill before and it loses it out of sight (maybe your ant got killed or moved away) can your bot still remember the location of the enemy hill? or you just rely on the hill given in the input? 2011-12-15T08:35:54 TL_: you have to track that information yourself 2011-12-15T08:36:00 similarly for food 2011-12-15T08:37:38 ok. how about previous food locations? do you keep track of it so that when the next food spawn occurs your ants are near it? 2011-12-15T08:38:31 what? 2011-12-15T08:40:53 the food are being spawned in the same place right? so if you could remember the location of previous foods you could make your ants wait in the location for the next re-spawn. i think i have read it somewhere. not sure if i remembered it correctly though. 2011-12-15T08:41:53 each location on the grid will spawn food before a location spawns it twice 2011-12-15T08:42:02 excepting when a food item is waiting to spawn where a static ant sits 2011-12-15T08:42:30 you might want to track the last time food was seen at a square, that way for example you can go back to not visible squares that you haven't seen food at yet 2011-12-15T08:43:41 i'm not sure that'd be very useful though 2011-12-15T08:45:30 yeah. there's not much a bot could do for that data. thanks antimatroid. hope you fix your timeout bugs. back to working on my bot now. 2011-12-15T08:45:30 *** AlliedEnvy has quit IRC (Remote host closed the connection) 2011-12-15T08:45:44 i don't think the timeout bug is my problem 2011-12-15T08:45:55 it sounds like flag was having the same issues 2011-12-15T08:47:35 *** g0llum has joined #aichallenge 2011-12-15T08:47:38 might be your internet connection. i sometimes timeout consecutively for so many games, so i would stop testing on tcp. you could confirm if it is not a timeout bug if you upload it in the main server 2011-12-15T08:48:01 *** AlliedEnvy has joined #aichallenge 2011-12-15T08:48:37 but then i can't have my submission count as 1 at the end :P 2011-12-15T08:48:46 and come in and surprise people 2011-12-15T08:48:48 *** Kurnevsky has joined #aichallenge 2011-12-15T08:48:57 it's either going to be a wild success of a complete failure 2011-12-15T08:49:01 either way it's kind of fun 2011-12-15T08:51:11 create a dummy account to test if it's working there. you will be disqualified if you get caught though. :P 2011-12-15T08:51:37 actually i probably wouldn't, but nah 2011-12-15T08:51:53 it's only when someone has an active submission on two accounts that we care 2011-12-15T08:51:58 but it would be against the spirit 2011-12-15T08:54:07 do you remember a game, where you control a spider, and you need to walk on screen in such way so you can "take" as much as possible, but you can't allow your just being bult "net" to be broken by *something*... 2011-12-15T08:54:17 for example second player 2011-12-15T08:55:06 when you get from one edge to the other, your net "solidifies" and so you've taken a part of area 2011-12-15T08:55:29 i wonder if this is good for ai challenge 2011-12-15T08:56:20 sounds cool 2011-12-15T08:56:26 no idea what you're talking about 2011-12-15T08:56:48 i'm looking for a working example 2011-12-15T08:57:12 *** avdg has joined #aichallenge 2011-12-15T08:57:25 what about something like tron, except you can walk over coloured squares and turn them into your own 2011-12-15T08:57:42 there's a maximum number of turns and the person who has the most coloured at the end wins 2011-12-15T08:58:01 sounds like reversi 2011-12-15T08:58:30 i'm waiting for a tcp game to finish so i can start like 3 more and watch a mythbusters ep 2011-12-15T08:59:32 *** sigh has joined #aichallenge 2011-12-15T09:05:58 *** mviel_ has joined #aichallenge 2011-12-15T09:06:17 *** Antimony has joined #aichallenge 2011-12-15T09:09:26 *** mviel has quit IRC (Ping timeout: 248 seconds) 2011-12-15T09:24:53 antimatroid: http://www.youtube.com/watch?v=-1r7d4mAaIM similar to this 2011-12-15T09:25:17 you take areas by drawing from edge to edge 2011-12-15T09:25:26 this could work for 2 players or even more 2011-12-15T09:25:34 but it may not be too advanced... 2011-12-15T09:26:19 *** lhb__ has joined #aichallenge 2011-12-15T09:26:23 i mean, challenging 2011-12-15T09:27:36 Was there anyway I could have won this game? Or was it just bad luck? http://aichallenge.org/visualizer.php?game=240183&user=3845 2011-12-15T09:27:46 *** bergmark has joined #aichallenge 2011-12-15T09:29:11 unlucky, red and pink piled up on you 2011-12-15T09:29:15 Antimony: you're parasprites, right? 2011-12-15T09:29:18 *** raemde_ has quit IRC (Ping timeout: 255 seconds) 2011-12-15T09:30:12 i was sad from Y-mas timing out, i wanted to see the end of game :( 2011-12-15T09:31:45 the funny thing is the bot never saw y-mas hill, so he was going blind 2011-12-15T09:37:07 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-15T09:37:58 es I am 2011-12-15T09:38:12 I don't like those big blobs 2011-12-15T09:38:22 then, sorry.. :) 2011-12-15T09:38:27 From an efficiency standpoint 2011-12-15T09:40:33 but i have common target for all ants so it's a bit hard not to make blobs 2011-12-15T09:40:54 the huge blob at the end gets split in search for Y-mas hill tho 2011-12-15T09:44:21 *** emilk has left #aichallenge 2011-12-15T09:48:27 *** olexs has quit IRC (Quit: Leaving.) 2011-12-15T09:50:36 *** raemde_ has joined #aichallenge 2011-12-15T09:52:41 *** raemde has joined #aichallenge 2011-12-15T09:53:46 *** lhb__ has quit IRC (Ping timeout: 276 seconds) 2011-12-15T09:55:09 *** raemde_ has quit IRC (Ping timeout: 252 seconds) 2011-12-15T09:56:24 *** Blkt has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-12-15T10:02:27 *** Darhuuk has quit IRC (Quit: leaving) 2011-12-15T10:02:54 *** Darhuuk has joined #aichallenge 2011-12-15T10:03:03 @rankings 2011-12-15T10:03:05 pairofdice: Top 10 players: xathis(97.4), pguillory(92.6), a1k0n(91.0), delineate(89.5), ThisIsNotABug(88.5), protocolocon(88.4), cumbuz(88.2), bix0r4ever(87.9), bucash(87.8), itzkow(87.6) 2011-12-15T10:03:10 A bit of a lead 2011-12-15T10:03:13 *** denysonique__ has joined #aichallenge 2011-12-15T10:03:59 >97.4 2011-12-15T10:04:14 i wonder if xathis gets over 100 before 18th 2011-12-15T10:04:42 *** Blkt has joined #aichallenge 2011-12-15T10:06:15 *** raemde has quit IRC (Ping timeout: 248 seconds) 2011-12-15T10:10:20 *** peerai has joined #aichallenge 2011-12-15T10:10:59 *** jstrong has joined #aichallenge 2011-12-15T10:12:11 *** sigh has quit IRC (Remote host closed the connection) 2011-12-15T10:15:54 *** TheLinker has quit IRC (Remote host closed the connection) 2011-12-15T10:17:28 *** Anilm3 has quit IRC (Quit: leaving) 2011-12-15T10:17:36 *** UncleVasya has joined #aichallenge 2011-12-15T10:21:05 *** QuirionPT has quit IRC (Remote host closed the connection) 2011-12-15T10:26:27 *** QuirionPt has joined #aichallenge 2011-12-15T10:27:31 *** Anilm3 has joined #aichallenge 2011-12-15T10:31:08 *** jstrong has quit IRC (Quit: Leaving.) 2011-12-15T10:32:35 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-12-15T10:36:33 *** amstan has joined #aichallenge 2011-12-15T10:36:33 *** ChanServ sets mode: +o amstan 2011-12-15T10:40:06 *** peerai has quit IRC (Quit: Page closed) 2011-12-15T10:45:51 http://aichallenge.org/visualizer.php?game=242948&user=757 protocolocon behaves funny at the end 2011-12-15T10:46:28 *** Anilm3 has quit IRC (Read error: Connection reset by peer) 2011-12-15T10:48:22 amstan: http://www.youtube.com/watch?v=-1r7d4mAaIM how about making ai challenge similar to this? i mean, two players trying to dominate as much area as possible and trying to prevent enemy to do the same (somehow) 2011-12-15T10:48:52 Fluxid: oh man, i love that commentator 2011-12-15T10:49:01 Fluxid: see the one for ecobear 2011-12-15T10:49:38 *** pairofdice has quit IRC (Ping timeout: 245 seconds) 2011-12-15T10:51:22 *** pairofdice has joined #aichallenge 2011-12-15T10:52:49 amstan: i like him too, but i wonder if idea of the game is good for ai challenge 2011-12-15T10:55:00 Fluxid: that is basically a twist on jezzball 2011-12-15T10:57:18 *** Boscop has joined #aichallenge 2011-12-15T10:58:13 rwest: ya 2011-12-15T10:59:42 maybe 2011-12-15T10:59:51 I think another game like ants with different types of ants would be cool 2011-12-15T11:00:34 non-discrete grid 2011-12-15T11:00:47 makes planning ahead more..interesting 2011-12-15T11:01:26 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-12-15T11:03:01 I want projectiles! 2011-12-15T11:04:44 I was thinking flying ants 2011-12-15T11:04:50 that could go over water 2011-12-15T11:05:03 projectiles would be interesting though 2011-12-15T11:05:33 carpenter ants that could build bridges over water or destroy them (corny I know, haha) 2011-12-15T11:06:06 *** BEASTMODE is now known as AFKMODE 2011-12-15T11:08:25 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-15T11:09:08 *** aerique has quit IRC (Quit: ...) 2011-12-15T11:09:30 I think more than 500ms would be needed for that though 2011-12-15T11:11:47 Well, tron had one agent. Planet wars had hundreds of agents. Ants has hundreds of agents. How about 1 < x < 20 agents. Also, stochasticity would be something new 2011-12-15T11:11:50 ;) 2011-12-15T11:13:16 *** QuirionPt has quit IRC (Ping timeout: 240 seconds) 2011-12-15T11:16:13 if you fire projectiles, you can add noise to aiming 2011-12-15T11:16:18 computational pool does this 2011-12-15T11:17:53 *** Anilm3 has joined #aichallenge 2011-12-15T11:18:20 Hey 2011-12-15T11:19:58 o/ 2011-12-15T11:20:17 haha 2011-12-15T11:20:23 \o 2011-12-15T11:20:43 *** Anilm3 has quit IRC (Client Quit) 2011-12-15T11:20:49 3 days left! And I got my gf at home, whoouhouu 2011-12-15T11:22:48 *** Anilm3 has joined #aichallenge 2011-12-15T11:30:47 *** codetiger has joined #aichallenge 2011-12-15T11:31:44 *** spicydog has joined #aichallenge 2011-12-15T11:34:02 *** GeorgeSebastian has joined #aichallenge 2011-12-15T11:36:19 *** spicydog has quit IRC (Client Quit) 2011-12-15T11:39:48 *** dvladim has joined #aichallenge 2011-12-15T11:40:15 *** GeorgeSebastian has quit IRC (Ping timeout: 255 seconds) 2011-12-15T11:45:56 *** kapoc has joined #aichallenge 2011-12-15T11:46:11 *** SMJ has joined #aichallenge 2011-12-15T11:50:14 hey, look: http://aichallenge.org/visualizer.php?game=243103&user=11648 this bot build circles... 2011-12-15T11:50:40 but those circles does now warp, maybe it's a spiral 2011-12-15T11:51:46 http://aichallenge.org/visualizer.php?game=242137&user=11648 this is awesome! 2011-12-15T11:52:32 he certainly build ripples across the map! 2011-12-15T11:52:42 *** GeorgeSebastian has joined #aichallenge 2011-12-15T11:54:57 *** treeform has joined #aichallenge 2011-12-15T11:57:05 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-12-15T12:00:05 Fluxid: that's kinda cool lol 2011-12-15T12:00:27 would be pretty ineffective against good bots though 2011-12-15T12:01:17 *** bergmark has quit IRC (Ping timeout: 252 seconds) 2011-12-15T12:02:07 Like saying building sand castles is ineffective against tanks 2011-12-15T12:02:39 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-15T12:03:30 *** raemde has joined #aichallenge 2011-12-15T12:08:51 protocolocon, a new challenger appears 2011-12-15T12:12:49 He ended up 11th during the planet war contest 2011-12-15T12:14:02 I really wish I had time to work on my bot :-/ 2011-12-15T12:15:16 ah, so not new at all 2011-12-15T12:16:07 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-15T12:16:47 *** nha has quit IRC (Ping timeout: 240 seconds) 2011-12-15T12:18:04 *** Blkt has quit IRC (Remote host closed the connection) 2011-12-15T12:19:08 *** Bosc0p has joined #aichallenge 2011-12-15T12:20:41 *** grwip has joined #aichallenge 2011-12-15T12:21:10 *** Bo5cop has joined #aichallenge 2011-12-15T12:21:12 do we already know what the turn time limit will be for the contest? 2011-12-15T12:21:25 100 2011-12-15T12:21:48 100ms ? 2011-12-15T12:22:13 j/k 500 they said it is staying I believe 2011-12-15T12:22:17 500ms 2011-12-15T12:22:27 oh ok, 2011-12-15T12:22:28 thx 2011-12-15T12:22:32 *** Boscop has quit IRC (Ping timeout: 252 seconds) 2011-12-15T12:23:05 *** mcstar has joined #aichallenge 2011-12-15T12:23:13 If it was 100ms, my bot would cry 2011-12-15T12:23:20 mine too! 2011-12-15T12:23:34 he's already a bit afraid of the 500... 2011-12-15T12:23:41 ahaha, same here :P 2011-12-15T12:24:08 *** pguillory has joined #aichallenge 2011-12-15T12:24:13 *** Surya has quit IRC (Quit: Page closed) 2011-12-15T12:25:28 *** Bosc0p has quit IRC (Ping timeout: 268 seconds) 2011-12-15T12:28:21 *** treeform has quit IRC (Remote host closed the connection) 2011-12-15T12:31:02 antimatroid: hey 2011-12-15T12:31:19 *** avdg1 has joined #aichallenge 2011-12-15T12:31:49 hmm i thought it was a second? 2011-12-15T12:31:54 500ms hmm 2011-12-15T12:35:00 *** avdg has quit IRC (Ping timeout: 252 seconds) 2011-12-15T12:39:34 @later tell antimatroid how come you havent submitted yet? 2011-12-15T12:39:34 mcstar: OK 2011-12-15T12:39:49 *** bergmark has joined #aichallenge 2011-12-15T12:41:01 so, is anyone else using java having trouble with first turn timeouts? 2011-12-15T12:41:02 :( 2011-12-15T12:41:41 nope 2011-12-15T12:42:19 I believe many people had a problem with that using the java starter bot, apparently the parser in it is very inefficient 2011-12-15T12:42:32 I used the java start bot 2011-12-15T12:42:36 ah, that may be it, i'm definitely using that parser... 2011-12-15T12:42:43 I am using that parser too 2011-12-15T12:42:50 it's weird though, because my turn time on turn 2+ is like 5-30 ms 2011-12-15T12:42:58 so I think it's mostly the JIT hosing me... 2011-12-15T12:43:08 I just remember seeing some forum threads about it 2011-12-15T12:43:19 *** grwip has quit IRC (Remote host closed the connection) 2011-12-15T12:43:32 in fact i didn't touch the parser at all 2011-12-15T12:43:33 I believe it was particularly a problem on some multi-hill maps that sent a lot of data on the first turn 2011-12-15T12:44:11 ah, i have this problem on a lot of maps, seemingly randomly 2011-12-15T12:44:18 and I haven't seen any new threads on it in a while so someone may have fixed it as well 2011-12-15T12:44:38 if it's just turn 1 timeouts though it wouldn't be so bad to bump the setup time 2011-12-15T12:44:44 well 2011-12-15T12:44:50 it's not *setup* timeouts 2011-12-15T12:44:59 (I don't do any setup work at all now, trying to diagnose this) 2011-12-15T12:45:04 it's first actual turn timeouts 2011-12-15T12:45:32 yoden: do you do any initialization on first turn that you don't on others? 2011-12-15T12:45:46 *** TheLinker has joined #aichallenge 2011-12-15T12:45:54 though... maybe i can try faking a turn during setup, using that longer amount of time to let the JIT do it's work 2011-12-15T12:46:42 sort of; i have work I do for each newly visible tile 2011-12-15T12:47:14 but it shouldn't be 10x slower than when i have a hundred ants exploring (should probably be faster) 2011-12-15T12:47:31 I also think the timeouts are "clusted" 2011-12-15T12:47:42 e.g, if you look at two of my timeouts from last night: 2011-12-15T12:47:49 http://aichallenge.org/visualizer.php?game=238161&user=2374 2011-12-15T12:47:50 http://aichallenge.org/visualizer.php?game=238195&user=2374 2011-12-15T12:47:58 everyone else in those games timed out too, lol 2011-12-15T12:53:30 visualizer doesn't really handle 1 turn games that well 2011-12-15T12:53:59 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-15T12:54:03 it'd be nice if we could actually see how long things were taking on the server 2011-12-15T12:54:37 the other 2 players that timeout there timed out 7 and 4 of there last 10 games each so not too surprising that they timed out 2011-12-15T12:57:46 why did the game rate get very fast at ~5:20 AM server time? I played like 10+ games in 20 minutes 2011-12-15T12:58:17 hmm, i guess it's bursty in general, seems like I usually play games in pairs... 2011-12-15T12:58:27 yes it is 2011-12-15T12:59:34 janzert: you change languages often 2011-12-15T13:00:25 programing languages? ;) Not for regular use much, although I'll play around with others 2011-12-15T13:02:55 i checked your submissions page 2011-12-15T13:02:56 yeah, i know you were just testing things out... 2011-12-15T13:02:57 *** liberforce has left #aichallenge 2011-12-15T13:02:57 *** Murashka has joined #aichallenge 2011-12-15T13:03:08 ahh, yes 2011-12-15T13:03:16 i feel like the wolverine in swordfish 2011-12-15T13:03:23 programming under pressure... 2011-12-15T13:03:27 its almost the end 2011-12-15T13:03:55 we're getting close :) 2011-12-15T13:04:12 I've still got a bit to get in place to be able to run the finals too 2011-12-15T13:04:20 although only one major part left 2011-12-15T13:04:26 how long will that take? 2011-12-15T13:04:30 pairing cutoff 2011-12-15T13:04:35 http://tcpants.com/replay.961 2011-12-15T13:04:37 the finals that is 2011-12-15T13:04:40 if only I had some actual attack code 2011-12-15T13:04:49 rather than just stay close and out of range 2011-12-15T13:05:08 less than a week although I need to discuss it with amstan and McLeopold to decide on a firm length 2011-12-15T13:05:40 o.O 2011-12-15T13:05:44 thats long 2011-12-15T13:06:14 planetwars was 4 days 2011-12-15T13:06:15 i realize that less than a week could mean 1 sec, but i guess you meant close to that right? 2011-12-15T13:06:34 I mean somewhere around the same as planetwars most likely 2011-12-15T13:07:41 *** yoden-cloud has joined #aichallenge 2011-12-15T13:08:05 how many games do you think it will take most bots to get a stable trueskill? 2011-12-15T13:09:07 yoden-cloud: depends on your definition of stable ;) but you should be pretty close with 25 2011-12-15T13:09:26 usually less 2011-12-15T13:09:36 well, i mean, you can define stable in terms of the sigma 2011-12-15T13:09:38 yes? 2011-12-15T13:10:07 I think most people would define in terms of distance to "true" rank 2011-12-15T13:10:19 isn't that what the trueskill sigma represents? 2011-12-15T13:10:29 possible error relative to 'true' rank? 2011-12-15T13:10:57 mostly 2011-12-15T13:11:53 it also measures variance in game play, so could theoretically never narrow beyond a certain amount 2011-12-15T13:12:01 right 2011-12-15T13:12:42 so once it ~stops decreasing, you're at approximate the correct amount of games 2011-12-15T13:13:16 20 games will almost always get your sigma under 3 if that gives you any extra feel for it 2011-12-15T13:14:38 right 2011-12-15T13:14:39 maybe almost always is too strong, but certainly usually in that I've never observed a counter example 2011-12-15T13:14:57 i guess, it seems weird to stop playing games, when playing more games would cause a bot's trueskill to go up 2011-12-15T13:15:49 ? 2011-12-15T13:16:33 in games 25-29 of my last submission, trueskill has gone up by about 2 2011-12-15T13:16:49 which isn't *that* much, but does push this submission to the highest skill of them =P 2011-12-15T13:17:27 mine gained like 30 points around games 40-50 2011-12-15T13:17:36 *** dvladim has quit IRC (Ping timeout: 240 seconds) 2011-12-15T13:17:40 then lost 50 over the next 15 2011-12-15T13:17:44 lol 2011-12-15T13:18:01 I think a lot of people resubmitted 2011-12-15T13:18:05 new uploads change the ranking much 2011-12-15T13:18:10 I am getting crushed by bots climbing back up 2011-12-15T13:18:35 yeah, there's definitely a lot of noise in the current rankings because of how resubmissions work 2011-12-15T13:18:49 last night i went as high as 95th, but now im at 122th 2011-12-15T13:19:01 i even took screenshots 2011-12-15T13:19:05 well, overall ranking will fluctuate a ton 2011-12-15T13:19:06 sure 2011-12-15T13:19:14 but shouldn't trueskill fluctuate less? 2011-12-15T13:19:19 oh segfaults 2011-12-15T13:19:23 how i love thee 2011-12-15T13:19:24 amstan: hey 2011-12-15T13:19:40 and i don't even know where to start, this used to work! 2011-12-15T13:20:36 *** Jacob_Strauss has joined #aichallenge 2011-12-15T13:20:48 you start with gdb 2011-12-15T13:21:07 *** Antimony has joined #aichallenge 2011-12-15T13:21:10 @seen protocolocon 2011-12-15T13:21:10 Jacob_Strauss: protocolocon was last seen in #aichallenge 1 year, 1 week, 6 days, 16 hours, 50 minutes, and 31 seconds ago: meneky, agreed 2011-12-15T13:21:34 rwest: now what? http://pastebin.com/4gXVsNRg 2011-12-15T13:21:57 i'm thinking more and more to just do this code in python 2011-12-15T13:22:04 drivers in python is much cooler 2011-12-15T13:22:34 isnt jack a bit outdated? 2011-12-15T13:22:44 outdated by what? 2011-12-15T13:23:06 pulseaudio or similar 2011-12-15T13:23:09 just asking 2011-12-15T13:23:26 jack can give you like 1ms latency and can also transport midi 2011-12-15T13:23:38 pulseaudio is more for general purpose sound outputting 2011-12-15T13:24:19 *** arscan_ has joined #aichallenge 2011-12-15T13:24:28 what are you doing with it? 2011-12-15T13:24:47 with this guitarseq program of yours 2011-12-15T13:24:59 it's meant to be a driver for a midi guitar 2011-12-15T13:25:24 where does the notes come from? 2011-12-15T13:25:33 events from the device 2011-12-15T13:25:34 *** Redgis has joined #aichallenge 2011-12-15T13:25:46 fret press/release, and strum press/release 2011-12-15T13:27:00 amstan: put a breakpoint just before the call and check to make sure all the variables are what they should be 2011-12-15T13:27:21 i think the problem is that you left out C# 2011-12-15T13:27:26 *** kapoc has left #aichallenge 2011-12-15T13:28:08 amstan: try running it with valgrind it might give you a hint 2011-12-15T13:28:15 i usually do that 2011-12-15T13:28:18 valgrind is a good tool 2011-12-15T13:28:22 i never used gdb 2011-12-15T13:28:31 i should really learn how to use those tools 2011-12-15T13:28:32 *** treeform has joined #aichallenge 2011-12-15T13:28:39 usually i just go with prints 2011-12-15T13:28:45 yep 2011-12-15T13:28:46 it's rare that i need to debug segfaults 2011-12-15T13:28:47 I was forced to learn them hah 2011-12-15T13:29:04 oh 2011-12-15T13:29:08 in school I had a class where I had to "difuse a binary bomb" 2011-12-15T13:29:11 on MCUs usually it's hard to debug stuff with them 2011-12-15T13:29:13 jabbotts debug vis. is excellent 2011-12-15T13:29:22 it helped me quite much 2011-12-15T13:29:34 (not for debugging, but for realtime visualizing) 2011-12-15T13:29:43 and usually not needed, since segfaults don't exist when you don't have an os, it's just that you might end up with random values where you don't expect 2011-12-15T13:30:07 <3 gdb 2011-12-15T13:30:29 amstan: you could try programming an emulator 2011-12-15T13:30:50 mcstar: problem with emulators is that they can't emulate everything from hardware 2011-12-15T13:31:11 when i code on mcus it's either i do some algorithm that i could compile on the comp the same way 2011-12-15T13:31:25 or.. it has to do with using the chip's features, like a communication bus with another device 2011-12-15T13:32:45 amstan: a debugger after my own heart 2011-12-15T13:32:53 *** treeform has quit IRC (Remote host closed the connection) 2011-12-15T13:32:55 usually if something segfaults it's more of an "oops, I know what I did" 2011-12-15T13:33:06 if it's my code 2011-12-15T13:33:15 *** choas has joined #aichallenge 2011-12-15T13:33:25 anyone have a crash course on valgrind? 2011-12-15T13:33:25 *** treeform has joined #aichallenge 2011-12-15T13:33:38 it doesn't like the program as its argument 2011-12-15T13:33:45 ah 2011-12-15T13:33:49 happened to me too 2011-12-15T13:33:50 I get worse problems, I do a lot of work in daemons. I don't notice I crash them since they get restarted instantly 2011-12-15T13:33:57 when i tried to run my bot through valgrind 2011-12-15T13:34:03 I reboot and get a warning that I have 500 crash reports 2011-12-15T13:34:03 lol 2011-12-15T13:34:07 amstan: command not found? 2011-12-15T13:34:18 mcstar: yes 2011-12-15T13:34:28 ./ seems to work better 2011-12-15T13:34:31 when i try to run it from the shell it always worked 2011-12-15T13:34:33 *** thagomizr has joined #aichallenge 2011-12-15T13:34:33 yeah 2011-12-15T13:34:42 that's a lot of output 2011-12-15T13:34:58 Syscall param write(buf) points to uninitialised byte(s) 2011-12-15T13:35:02 getting closed i guess 2011-12-15T13:35:05 when i developed my qt app, well there was a hell lot of output too 2011-12-15T13:36:12 worst thing ever is getting stuck between 2 bots that kill your and take your hill then both run in opposite directions 2011-12-15T13:36:17 colluding bastards! 2011-12-15T13:36:18 i don't even have a variable named buf 2011-12-15T13:36:20 lol 2011-12-15T13:37:13 *** ikaros has joined #aichallenge 2011-12-15T13:37:14 amstan: when you ran gdb did you compile with -g? 2011-12-15T13:37:15 amstan: you are using the jack lib 2011-12-15T13:37:18 yes 2011-12-15T13:37:21 to both 2011-12-15T13:37:26 ok making sure 2011-12-15T13:37:31 i mean those errors probably come from the lib 2011-12-15T13:37:41 or warnings 2011-12-15T13:37:45 mcstar: yes, but why? 2011-12-15T13:37:51 its not clean 2011-12-15T13:37:54 possibly being passed an empty pointer? 2011-12-15T13:38:50 people dont use valgrind enough, and c/c++ is nothing more than high-level assembly so you got memory errors, uninitilized variables .... 2011-12-15T13:38:55 well, here's the guilty code: https://github.com/amstan/guitar/blob/master/guitarseq/jack.c 2011-12-15T13:39:22 segfault happens because of the jack_ringbuffer_peek on line 15 2011-12-15T13:40:02 *** thagomizr has quit IRC (Quit: thagomizr) 2011-12-15T13:40:52 ringbuffer is a global or something? 2011-12-15T13:40:59 mcstar: was just gonna ask that 2011-12-15T13:40:59 lol 2011-12-15T13:41:09 yeah, declared in jack.h 2011-12-15T13:41:15 :S 2011-12-15T13:41:22 initialized in jack_init 2011-12-15T13:43:01 is it possible that the callback is being called on an empty ringbuffer? 2011-12-15T13:43:21 or uninitialized 2011-12-15T13:43:34 (I don't know when the callback is called, only where it's set) 2011-12-15T13:43:37 *** arscan_ has quit IRC (Quit: Page closed) 2011-12-15T13:43:45 rwest: hmm 2011-12-15T13:43:48 possibly 2011-12-15T13:43:55 rwest: let me try that 2011-12-15T13:44:04 jack_ringbuffer_reset(ringbuffer); 2011-12-15T13:44:10 can reset NULL it? 2011-12-15T13:44:19 rwest: hurray 2011-12-15T13:44:27 *** Surya has joined #aichallenge 2011-12-15T13:45:23 apparently the new version of jack is very eager and executes the callback while i'm still initing 2011-12-15T13:45:50 heh 2011-12-15T13:46:48 *** kire has joined #aichallenge 2011-12-15T13:47:09 *** choas has quit IRC (Ping timeout: 255 seconds) 2011-12-15T13:51:00 how do the finals work? 2011-12-15T13:52:32 1st everyone submits their sekrit killer bots 2011-12-15T13:52:33 basically, you put a lot of hard work in your bot, and it will turn out useless for most of the people 2011-12-15T13:53:45 something wrong eith the workers ? i click around random profiles & the last game played was ~5:18 wherever i look 2011-12-15T13:54:47 there's a few less running at the moment, but they're still churning out games 2011-12-15T13:54:58 workers are working, cache trouble? 2011-12-15T13:55:37 well 3 or four less than normal shouldn't make a dent 2011-12-15T13:55:37 g0llum: check the latest games page http://aichallenge.org/games.php 2011-12-15T13:55:38 pairofdice: Nov 11th 6:24pm 2011-12-15T13:55:58 k. convinced. 2011-12-15T13:56:15 is this a common strategy that everyone keeps his "sekrit" bot for the finals? 2011-12-15T13:56:34 sry, "sekrit killer" 2011-12-15T13:56:41 there are generally a lot of last minute submissions 2011-12-15T13:56:57 pairofdice: why didnt you upload recently? 2011-12-15T13:57:13 I haven't worked on the bot 2011-12-15T13:57:23 i mean, i took this week off work to work on ants (well, and play starcraft xD) 2011-12-15T13:57:25 but you are here all the time :) 2011-12-15T13:57:54 turned out that im more effective when im not around irc 2011-12-15T14:00:00 I wasn't very well prepared programming-skill wise for the challenge. I started preparing for the next one ;) 2011-12-15T14:01:04 i should help with the next one :P 2011-12-15T14:01:12 i was really bad about that for this one 2011-12-15T14:01:16 id love to too 2011-12-15T14:01:40 i spent soo much time here, i should have contributed in some way... 2011-12-15T14:02:24 *** Gnome has quit IRC (Quit: Gnome) 2011-12-15T14:02:32 any ideas for the next one? 2011-12-15T14:02:54 were you discussing it at all? 2011-12-15T14:02:56 i haven't been checking the forum 2011-12-15T14:03:19 and i haven't been designing any games myself ;) 2011-12-15T14:04:19 *** Bo5cop is now known as Boscop 2011-12-15T14:04:21 *** Boscop has joined #aichallenge 2011-12-15T14:04:30 *** bhasker has quit IRC (Ping timeout: 252 seconds) 2011-12-15T14:05:31 lol, the server gives player seeds as longs... but the test script does not on my laptop 2011-12-15T14:05:42 Hmmm, it looks like generalizing a piece of code actually increased the total code size 2011-12-15T14:06:00 so much for removign duplication 2011-12-15T14:06:44 :o 2011-12-15T14:07:34 ö 2011-12-15T14:09:05 how long is the competition expected to take? 2011-12-15T14:09:19 a day? 2011-12-15T14:09:38 *** Frib has joined #aichallenge 2011-12-15T14:09:44 submissions end on dec 18 2011-12-15T14:09:59 historically it's run for at least a few days after that 2011-12-15T14:10:07 cool 2011-12-15T14:10:09 thanks 2011-12-15T14:10:44 so little time 2011-12-15T14:11:21 whats your Antimony 2011-12-15T14:11:25 yours* 2011-12-15T14:12:50 xathis has a huge lead 2011-12-15T14:13:18 its somewhat annoying 2011-12-15T14:13:35 cause i watched a couple of his games, and he doesnt seem to do anything special 2011-12-15T14:13:39 not that i recognize 2011-12-15T14:14:16 I'm currently rank 103, mainly because my ants are too scared to ever attack 2011-12-15T14:14:47 parasprites is unguessable from Antimony 2011-12-15T14:15:13 I had this name on IRC before I joined hte ai contest 2011-12-15T14:15:53 *** Akranis has quit IRC (Ping timeout: 245 seconds) 2011-12-15T14:16:34 the code that plays now doest defend its own hills 2011-12-15T14:16:40 i mean mine 2011-12-15T14:17:42 *** TL_ has quit IRC (Ping timeout: 258 seconds) 2011-12-15T14:18:00 im very much looking forward to the end otherwise 2011-12-15T14:18:15 im really interested in knowing how did other ppl solve the problems 2011-12-15T14:18:18 my ants used to defend, but then they got smarter at combat 2011-12-15T14:18:27 screw this, I'm not dieing for my hill 2011-12-15T14:18:35 :) 2011-12-15T14:18:49 this was a big lesson for me 2011-12-15T14:18:58 turned out, the simplest things are the bestr 2011-12-15T14:19:11 or at least you have to try them 2011-12-15T14:19:26 i spent much time just thinking about the "optimal" thing 2011-12-15T14:19:28 which is bs 2011-12-15T14:19:30 hows your bot doing? 2011-12-15T14:19:37 me too 2011-12-15T14:19:50 i climbed up to 95th yesterday night 2011-12-15T14:19:55 but now im 122nd 2011-12-15T14:20:11 oh 2011-12-15T14:20:22 janzert: are we on a light diet of workers now? 2011-12-15T14:20:30 only 2 rows, we had 3.5 rows before 2011-12-15T14:21:31 2 rows of totaal workers, 0.1 rows of amazon workers 2011-12-15T14:23:06 *** Accoun has quit IRC () 2011-12-15T14:24:15 accoun is not participating this time? 2011-12-15T14:24:24 *** Scryer has joined #aichallenge 2011-12-15T14:24:44 wait, 13k users? Does that mean that almost 5k have yet to submit? 2011-12-15T14:26:49 i dont expect them to 2011-12-15T14:27:20 neither those with only 1 version 2011-12-15T14:28:23 i think there are <3k active users 2011-12-15T14:28:37 woah I had an Official timeout 2011-12-15T14:29:01 im glad 2011-12-15T14:29:17 crap, I was doing pretty well in that game 2011-12-15T14:29:19 BenJackson: I have had 2 now 2011-12-15T14:29:33 I got started late in the comp 2011-12-15T14:29:36 :/ 2011-12-15T14:29:43 I am around 1300 with a pretty dumb bot though 2011-12-15T14:29:49 there is nothing that should time out on those maps though 2011-12-15T14:30:00 they are tiny and my bot only cares about map size 2011-12-15T14:30:08 in regards to computation tim 2011-12-15T14:30:09 e 2011-12-15T14:30:54 last time I had an Official timeout I measured at home and that turn should have taken about 100ms 2011-12-15T14:30:58 and earlier turns had taken 5x lnoger 2011-12-15T14:31:39 the workers seems somewhat uneven in terms of performance 2011-12-15T14:32:36 *** ChrisH_ has joined #aichallenge 2011-12-15T14:33:22 BenJackson: paradoxe did have trouble controlling a 1000 ants 2011-12-15T14:34:55 janzert: Maybe it's just dumb luck, but my last few submissions seem to get a lot of 1v1 matchups against bots with much lower mu while they are climbing the ladder. My last game was 81.29 mu vs. 71.00 mu. 2011-12-15T14:35:05 my ants get increasingly suicidal as I go over 200 and 250 ants 2011-12-15T14:35:18 *** foRei has joined #aichallenge 2011-12-15T14:35:20 plus in most games without multi-hill it's hard to spawn too many ants 2011-12-15T14:35:31 by the time you dominate enough to get that much food there aren't enough turns left 2011-12-15T14:35:38 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-15T14:35:43 ChrisH_: same happened to me 2011-12-15T14:35:51 I think what's happening is that we're not the seeds in those games 2011-12-15T14:36:12 but when you get pulled into a game it uses up your queue position (afaict) 2011-12-15T14:36:47 well that's bad because those games always happen when I am at the front of the queue (as far as I can tell). 2011-12-15T14:37:01 so it's not like they are bonus games 2011-12-15T14:37:08 someone explained that by the fact that bots near your skill are also at the front of the queue at the same time 2011-12-15T14:37:19 it took me a lot more games to rank up this time 2011-12-15T14:37:30 normally I win like 4-5 in a row and hit some competition 2011-12-15T14:37:35 this time I won like 11 or 12 in a row 2011-12-15T14:37:54 I think it has to do with some changes to the matchup in which I think they loosened the requirements in some way, but that's just a suspicion 2011-12-15T14:38:09 *** Accoun has joined #aichallenge 2011-12-15T14:38:33 ChrisH_: the one that really gets me is the one where I beat GreenTea, but he was also climbing (and below me) so it didn't count for anything 2011-12-15T14:40:05 as usual though I've got a new version before my old one is even stabilized 2011-12-15T14:40:12 I'll probably resubmit tonight after reviewing fluxid results 2011-12-15T14:40:25 I suspect it has to do with how the queue goes in waves with sections of the queue dominated by similar mu's, but if you're toward one end of the wave perhaps all the good matchups have played too recently to get another game so you get stuck with poorer matchup 2011-12-15T14:40:42 (didnt, i meant to write didnt jesus h christ) 2011-12-15T14:41:43 ChrisH_: then you get a good matchup and it's a multi-hill maze 2011-12-15T14:42:27 *** u_ has quit IRC (Ping timeout: 240 seconds) 2011-12-15T14:42:33 a lot of my recent features have been focused on better multi-hill performance, so I want to see those games the most 2011-12-15T14:42:35 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-12-15T14:44:07 but it's frustrating to have a mu ~80 and get three consectutive matches against opponents ~70 1v1 2011-12-15T14:44:57 http://tcpants.com/replay.1029 2011-12-15T14:45:00 that is a comical game 2011-12-15T14:45:20 is tcpants back? 2011-12-15T14:45:21 *** QuirionPt has joined #aichallenge 2011-12-15T14:45:40 BenJackson: yea 2011-12-15T14:46:58 rwest: why is it comical? 2011-12-15T14:47:03 btw that is a hard map 2011-12-15T14:47:18 the walls near water 2011-12-15T14:47:19 haha 2011-12-15T14:47:32 we are having standoffs over water 2011-12-15T14:47:52 walls near water? All 'walls' are water :) 2011-12-15T14:48:02 walls of ants 2011-12-15T14:48:05 he means 2011-12-15T14:48:07 I think he means that due to thin walls two bots are thikning they're fighting 2011-12-15T14:48:07 yea 2011-12-15T14:48:07 oh, right 2011-12-15T14:48:18 they are, kind of 2011-12-15T14:48:27 but thats easy ti fix 2011-12-15T14:48:40 mcstar: not we my combat 2011-12-15T14:48:46 we = with 2011-12-15T14:48:52 well lack or combat 2011-12-15T14:48:56 I gotta fix that first 2011-12-15T14:49:00 even easier 2011-12-15T14:49:16 if you have no comabt code then its super easy to fix 2011-12-15T14:49:24 I have combat code kind of 2011-12-15T14:49:32 I just don't move into kill range of enemy 2011-12-15T14:49:35 lol 2011-12-15T14:49:40 thats not combat code 2011-12-15T14:49:46 but I surround them 2011-12-15T14:50:00 do you play music to them? i hope so 2011-12-15T14:50:04 yes! 2011-12-15T14:50:09 I hold the boombox over my head 2011-12-15T14:50:26 yellow likes it much 2011-12-15T14:51:02 that wall thing reminds me of one of my ugliest bugs 2011-12-15T14:51:17 I had this idea of projecting the "attacking surface" of the enemies through time 2011-12-15T14:51:26 *** Surya has quit IRC (Ping timeout: 258 seconds) 2011-12-15T14:51:27 except it managed to go trhough walls like that 2011-12-15T14:51:31 producing some very odd behavior 2011-12-15T14:51:50 if I had attack code, I would fight them over the watter 2011-12-15T14:51:52 water 2011-12-15T14:51:56 wow I fail at typing today 2011-12-15T14:52:41 mine fought through water too, but i fixed it 2011-12-15T14:52:47 obvioulsy with an obvious change 2011-12-15T14:52:58 i dont know what can be said about battle safely 2011-12-15T14:53:10 i dont think i want to share my new idea that im working on 2011-12-15T14:53:33 I have a similar approach to memetix but it's not ironed out 2011-12-15T14:53:35 actually its not battle-code per say 2011-12-15T14:53:41 its pre-battle-code 2011-12-15T14:53:42 I should just use his and call it a day 2011-12-15T14:54:44 rwest: how do you know what he has? 2011-12-15T14:54:50 post on the forum 2011-12-15T14:54:51 heh 2011-12-15T14:54:56 what?? 2011-12-15T14:55:04 i dont read forums 2011-12-15T14:55:10 only in case of emergencies 2011-12-15T14:55:18 memetix and a1k0n have posted their approaches to combat 2011-12-15T14:55:24 yea 2011-12-15T14:55:26 pfff 2011-12-15T14:55:29 why? 2011-12-15T14:55:33 whats insane 2011-12-15T14:55:36 w->t 2011-12-15T14:55:46 they posted how they do evaluation 2011-12-15T14:55:53 not how they choose moves based on the evaluation 2011-12-15T14:56:20 i removed that code from my bot 2011-12-15T14:56:27 that did enumeration + evaluation 2011-12-15T14:56:36 and this increased its skill by 15 2011-12-15T14:58:22 probably i should read it 2011-12-15T14:58:33 i mean i dont get what do you mean by evaluation 2011-12-15T14:58:53 but not how they choose upon that 2011-12-15T15:00:57 enumeration + evaluation => information needed to make good decisions 2011-12-15T15:01:08 choosing moves => makind decisions based on that information 2011-12-15T15:01:16 s/makind/making/ 2011-12-15T15:01:56 you mean mankind? 2011-12-15T15:02:32 *** Jacob_Strauss_ has joined #aichallenge 2011-12-15T15:02:57 from my perspective after you know evaluation, acting upon that is easy 2011-12-15T15:03:03 at least not that hard 2011-12-15T15:03:49 true, but it's still an important piece of the puzzle with room for a lot of creativity 2011-12-15T15:04:05 *** Murashka has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:04:06 have you read it? 2011-12-15T15:04:24 memetix' yes 2011-12-15T15:04:28 *** Jacob_Strauss has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:04:28 *** rebelxt has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:04:33 did it help you? 2011-12-15T15:04:48 did you do something differently beacuse you read it? 2011-12-15T15:04:51 *** ChrisH_ has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:04:51 *** pguillory has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:04:51 *** dragonfyre13 has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:04:55 a little, but I haven't implemented it yet.. 2011-12-15T15:05:09 where are you now? 2011-12-15T15:05:10 I have some bugs I need to sort out and I don't really have time to do it 2011-12-15T15:05:14 *** arscan has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:05:37 *** Scryer has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:06:00 *** yoden-cloud has quit IRC (Ping timeout: 258 seconds) 2011-12-15T15:06:27 i cant type i into the search field :( 2011-12-15T15:06:41 the visualizer cathches the keystroke 2011-12-15T15:06:59 my bot has some pretty advanced logic that 'almost and mostly' works.. 2011-12-15T15:07:16 I can't build anything good on it until it works close to 100% 2011-12-15T15:07:25 it is eliminated pretty often 2011-12-15T15:07:44 well yes, that's an old version 2011-12-15T15:08:23 it works much better than my current bot 2011-12-15T15:08:49 and the version before it worked even better 2011-12-15T15:09:03 *** bhasker has joined #aichallenge 2011-12-15T15:11:15 *** arscan has joined #aichallenge 2011-12-15T15:14:08 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-15T15:22:47 *** Antimony has joined #aichallenge 2011-12-15T15:32:46 *** yoden-cloud has joined #aichallenge 2011-12-15T15:32:47 *** ChrisH_ has joined #aichallenge 2011-12-15T15:34:55 BenJackson: are you running on fluxid atm? 2011-12-15T15:37:07 *** cyphase has quit IRC (Ping timeout: 240 seconds) 2011-12-15T15:41:31 *** TheLinker has quit IRC (Read error: Connection reset by peer) 2011-12-15T15:43:34 <_flag> Bah, everytime I upload to the main server there's something I want to add or change the very next day :( 2011-12-15T15:46:09 yeah, especially now that it's crunch time 2011-12-15T15:46:31 your latest versions have been doing very well though, so nice job 2011-12-15T15:46:49 _flag: very nice considering c 2011-12-15T15:47:24 i don't know if my bot will catch up to you and xathis, but i'm hoping to press for a top 10 or 20 slot 2011-12-15T15:47:41 "16 year old high school student. 2011-12-15T15:47:43 " 2011-12-15T15:47:47 _flag: really? 2011-12-15T15:47:52 <_flag> mcstar: yup 2011-12-15T15:48:06 i suddenly feel stupid 2011-12-15T15:48:12 very nice 2011-12-15T15:48:42 <_flag> Thanks 2011-12-15T15:49:12 When I was that age we were still hacking BBSes with 300 baud modems. ;) 2011-12-15T15:49:55 *** cyphase has joined #aichallenge 2011-12-15T15:50:13 And my computer's CPU did not have a native multiply or divide operation. :P 2011-12-15T15:50:49 oh, and it had 3 8 bit registers. 2011-12-15T15:51:08 glad I don't have to work within those constraints any more 2011-12-15T15:51:32 i had my first computer when i was ~17 2011-12-15T15:52:10 The first computer I ever programmed on had 1024 bytes of RAM. 2011-12-15T15:52:18 yes, BYTES 2011-12-15T15:52:31 is that a lot? 2011-12-15T15:52:36 LOL 2011-12-15T15:52:38 :D 2011-12-15T15:52:39 That's loads! 2011-12-15T15:52:46 it was cool back then 2011-12-15T15:53:04 *** Migi32 has joined #aichallenge 2011-12-15T15:53:11 i think there are some oscilloscopes today with similar amount 2011-12-15T15:55:20 ChrisH_: what is this area42? 2011-12-15T15:55:40 Of course this conversation is headed this direction, so I'll stop now: http://dilbert.com/strips/comic/1992-09-08/ 2011-12-15T15:56:34 made me laugh 2011-12-15T15:56:38 XD 2011-12-15T15:56:40 mcstar: It's a game store for strategy games that my brother operates: http://www.area42games.com/ 2011-12-15T15:57:02 O O OOO O O O OOO OO OO O OO 2011-12-15T15:57:46 @translate O O OOO O O O OOO OO OO O OO 2011-12-15T15:57:46 mcstar: Error: The command "translate" is available in the Format and Google plugins. Please specify the plugin whose command you wish to call by using its name as a command before "translate". 2011-12-15T15:58:31 don't try to decode it, it's just pseudorandom 2011-12-15T15:59:51 hm, i just piped it into Enterprise E's universal translator and it means Good Day in Binar 2011-12-15T16:00:09 but yeah, you didn't really have ones and zeroes when you worked with punch cards, you only had holes 2011-12-15T16:00:27 now come on 2011-12-15T16:00:39 that strip is hilarious 2011-12-15T16:00:59 *** TL_ has joined #aichallenge 2011-12-15T16:01:12 *** kire has quit IRC (Remote host closed the connection) 2011-12-15T16:01:21 it is mind boggling how did people put up with systems "back in the day" 2011-12-15T16:01:55 i cant stand c anymore for example, since i started to pay attention to programming im looking for really high level languages 2011-12-15T16:02:03 only in retrospect 2011-12-15T16:02:07 mcstar: we weren't "putting up" with them, they were cooler than anything else available 2011-12-15T16:02:26 ChrisH_: you could have chosen another profession 2011-12-15T16:02:40 now they know how many holes it takes to fill the albertr haaall.. 2011-12-15T16:02:47 probably its stupid hearing it from me 2011-12-15T16:02:50 early computers could do things that had seemed impossible only a few years earlier 2011-12-15T16:02:57 no I couldn't, my brain was prewired to do this stuff 2011-12-15T16:03:07 in retrospect they seem primitive, but at the time they were amazing 2011-12-15T16:03:20 ok 2011-12-15T16:03:24 i dont disagree 2011-12-15T16:03:40 but nowadays you arent "uncomfortable" with the system 2011-12-15T16:03:50 its pretty easy to do anything 2011-12-15T16:03:51 yeah, imagine when I moved from 1K of memory to owning a computer with 64K, that was a 64 times improvement in only a couple years. 2011-12-15T16:03:51 yes I am 2011-12-15T16:04:07 I just started in a job where I have to use windows in my daily work 2011-12-15T16:04:11 *shudder* 2011-12-15T16:04:24 and back than, that increased amount of hardware capacity really meant something 2011-12-15T16:04:31 but today, it doesn mean that much 2011-12-15T16:04:50 Yeah a measly doubling every 18 months... pff! 2011-12-15T16:04:52 now you can focus on ai+theoretical questions 2011-12-15T16:05:05 in a nice and comfortable architecture/software environment 2011-12-15T16:05:08 *** Kurnevsky has left #aichallenge 2011-12-15T16:05:35 but still computers are slower, smaller and dumber than the human brain 2011-12-15T16:05:45 pairofdice: the hardware may double, but my ability to use that hardware does not 2011-12-15T16:05:46 if that isn't primitive I don't know what is 2011-12-15T16:06:36 *** Gnome has joined #aichallenge 2011-12-15T16:06:42 computers are bigger and faster than the human brain 2011-12-15T16:06:50 and not that much dumber 2011-12-15T16:06:58 i think human intelligence is overrated 2011-12-15T16:07:02 highly 2011-12-15T16:07:02 depends what you measure 2011-12-15T16:07:06 12:34 < ChrisH_> BenJackson: are you running on fluxid atm? 2011-12-15T16:07:13 ChrisH_: should be 2011-12-15T16:07:18 also fired up on tcpants 2011-12-15T16:07:42 BenJackson: what's your name on fluxid? 2011-12-15T16:07:44 visual processing, extracting information and making decisions based on it 2011-12-15T16:08:10 computers are barely able to drive cars and hit moving targets with weapons 2011-12-15T16:08:26 http://dilbert.com/strips/comic/1992-09-08/ 2011-12-15T16:08:29 but they improve with incredible speed 2011-12-15T16:08:37 much faster than human evolution 2011-12-15T16:08:40 ChrisH_: every time I see that dilbert I think of writing basic programs on a manual typewriter 2011-12-15T16:08:45 which really didn't have 1 2011-12-15T16:08:47 it only had l 2011-12-15T16:09:09 they're improving quickly yes, but they still have much catching up to do 2011-12-15T16:09:11 ChrisH_: bj_vXX 2011-12-15T16:09:12 Imagine when AIs will do the development. 2011-12-15T16:09:17 v12 at the moment 2011-12-15T16:09:20 LOL, I never had that experience, but I did learn to type on an electric typewriter. 2011-12-15T16:09:22 they were improving just as quickly 50 years ago 2011-12-15T16:09:38 that was before I had a computer 2011-12-15T16:09:42 so I'd write programs on the typewriter 2011-12-15T16:09:46 then type them in later 2011-12-15T16:10:41 I was lucky, I came of age right when the C64 came down to $200. 2011-12-15T16:10:57 I got my first C64 for $210 on Dec 23, 1983 2011-12-15T16:11:01 *** _Sander_ <_Sander_!~sandre@ip54565dd6.speed.planet.nl> has joined #aichallenge 2011-12-15T16:11:03 after saving all summer 2011-12-15T16:11:14 BenJackson: Four days before me. 2011-12-15T16:11:15 that has to be the best investment ever 2011-12-15T16:11:22 well i hate computers, they are the cause of my unsuccesful career 2011-12-15T16:12:04 BenJackson: Same here, sounds like we have very similar experiences. :) 2011-12-15T16:12:38 that was about a year before I was born 2011-12-15T16:12:47 and I'm 27 now 2011-12-15T16:12:53 damn, time passes quickly 2011-12-15T16:12:54 Minthos: you mean you were still in two places at once then. ;) 2011-12-15T16:13:06 yeah 2011-12-15T16:13:15 well, more 2011-12-15T16:13:20 my version 11 still seems to be on track to be about 9th 2011-12-15T16:13:20 and I still am 2011-12-15T16:13:24 coming up on 26, yep too quickly 2011-12-15T16:13:25 but I get what you mean 2011-12-15T16:13:41 ChrisH_: btw V12 is where I got in the mode you described: TRY ANYTHING 2011-12-15T16:13:56 so it has some improvements but the new code is even crazier than the old code 2011-12-15T16:14:31 also I think the version running on fluxid has some kind of west-going bias due to my ultra hackish last-minute diffusion of one of my goal maps 2011-12-15T16:14:36 damn, i timed out on flux.id 2011-12-15T16:15:17 Well, I hope my V8 on the main server does well, because I like how it should work in theory, and what I've got on Fluxid is a slight generalization of that. I think I've got a couple mor things to try and maybe some speed profiling to do before finals. 2011-12-15T16:15:18 V11 = make state global, I don't have time to keep passing it correctly to all these inner functions 2011-12-15T16:15:26 V12 = make bot global, I don't have ... 2011-12-15T16:15:30 mcstar: happens to me all the time 2011-12-15T16:15:50 cat bot.0.01.py | sort -R > bot.0.02.py 2011-12-15T16:16:22 V11 shaved a lot of speed off 2011-12-15T16:16:24 did flux.id reset again? 2011-12-15T16:16:28 I haven't checked V12 2011-12-15T16:16:47 xathis's mu is nearly 102 now 2011-12-15T16:17:04 I can't believe what a huge lead he's opened up 2011-12-15T16:17:50 *** flowenol has joined #aichallenge 2011-12-15T16:18:07 thestinger: I just passed you on aichallenge. remind me what you're offering me not to switch over to c++11 and win that category? ;-) 2011-12-15T16:19:31 < BenJackson> xathis's mu is nearly 102 now ↠it was over 102 earlier today 2011-12-15T16:20:11 I assume this is trueskill's way of making enough headroom for him and eventually it will push everyone's mu down 2011-12-15T16:21:00 *** Palmik_ has quit IRC (Remote host closed the connection) 2011-12-15T16:22:17 *** ChrisH_ has quit IRC (Ping timeout: 258 seconds) 2011-12-15T16:22:19 <_flag> I'm pretty sure he just lost the mu over his last two games, probably nothing to do with trueskill 2011-12-15T16:22:30 *** skunx has joined #aichallenge 2011-12-15T16:23:42 *** ChrisH_ has joined #aichallenge 2011-12-15T16:25:37 *** Jacob_Strauss_ has quit IRC (Quit: Page closed) 2011-12-15T16:26:34 *** skunx has quit IRC (Quit: leaving) 2011-12-15T16:29:10 i think the games on fluxid are taking way longer now that everyone is trying to do combat logic~~ 2011-12-15T16:31:50 I think some people are lagging badly 2011-12-15T16:31:54 or simply ignoring the time limit 2011-12-15T16:32:22 I watched a game with M24 take nearly 30 minutes until he timed out (which meant his bot took >5s for a move!) and then the game zipped by 2011-12-15T16:32:38 >ignoring time limit 2011-12-15T16:32:41 aw yeah, way to go 2011-12-15T16:33:42 Live fast, die young 2011-12-15T16:34:01 Or rather, Live slow, die old 2011-12-15T16:34:08 Fluxid: why is that i cant see the listing of the games i played in? 2011-12-15T16:34:21 I like lefant's name, too bad I just figured it out 2011-12-15T16:35:07 mcstar: what, again 2011-12-15T16:35:15 what, again 2011-12-15T16:35:28 yeah, there is something broken in the code 2011-12-15T16:35:34 Fluxid has a data integrity problem 2011-12-15T16:35:44 i cant see anyone's listing 2011-12-15T16:35:48 mcstar, there was some reset yesterday, games got deleted 2011-12-15T16:35:49 oh no 2011-12-15T16:35:58 I actually NEED those fluxid games this time 2011-12-15T16:36:09 but it says i played 71 games 2011-12-15T16:36:19 so the counter isnt reset? 2011-12-15T16:36:24 mcstar: games were definitely dropped without resetting the summary 2011-12-15T16:36:33 but that was yesterday 2011-12-15T16:36:34 damn 2011-12-15T16:36:42 i wanted to take a look at them 2011-12-15T16:36:59 http://ants.fluxid.pl/player/bj_v12 2011-12-15T16:37:03 *** TL__ has joined #aichallenge 2011-12-15T16:37:07 I played those games overnight 2011-12-15T16:37:33 http://ants.fluxid.pl/replay.35421 2011-12-15T16:37:44 haha I couldn't figure out how I didn't do well in that game based on the graph 2011-12-15T16:37:50 *** yoden-cloud_ has joined #aichallenge 2011-12-15T16:37:52 but GreenTea and I both timed out on the same turn 2011-12-15T16:37:55 handing the win to red 2011-12-15T16:38:11 ffffuck 2011-12-15T16:38:23 *** yoden-cloud has quit IRC (Ping timeout: 258 seconds) 2011-12-15T16:38:23 *** arscan has quit IRC (Ping timeout: 258 seconds) 2011-12-15T16:38:46 *** ChrisH_ has quit IRC (Ping timeout: 258 seconds) 2011-12-15T16:39:11 good news for my V12 is that my losses on fluxid are timeouts 2011-12-15T16:39:52 lemme try to work it out 2011-12-15T16:39:55 *** TL_ has quit IRC (Ping timeout: 258 seconds) 2011-12-15T16:41:49 BenJackson: lolol 2011-12-15T16:42:02 http://ants.fluxid.pl/replay.35506 2011-12-15T16:42:05 39th turn 2011-12-15T16:42:16 or 40th 2011-12-15T16:42:27 mcstar: which one are you? 2011-12-15T16:42:36 theres a hill at 33.55 2011-12-15T16:42:40 why didnt you raze it? 2011-12-15T16:42:58 Fluxid: what do you mean? 2011-12-15T16:43:01 nick 2011-12-15T16:43:06 mcstar 2011-12-15T16:44:14 aaaaah 2011-12-15T16:44:19 last seen 2011-12-15T16:44:20 13.12.2011 15:47:47 2011-12-15T16:44:24 *** kevlar has quit IRC (Read error: Connection reset by peer) 2011-12-15T16:44:40 ok, so, i purged db in such way that ony players were left with skill etc 2011-12-15T16:44:54 so in this case it's not db inconsistency 2011-12-15T16:44:56 *** kevlar has joined #aichallenge 2011-12-15T16:45:01 sorry about that 2011-12-15T16:45:17 i got purged cause my bot was stupid? 2011-12-15T16:45:33 that is discrimination 2011-12-15T16:46:17 the weird thing is that i just played there ~30min ago, i had a timeout, but i still cant see the game 2011-12-15T16:46:21 you got purged, because you were away. 2011-12-15T16:46:42 13.12 hasnt passed yet 2011-12-15T16:46:48 its only 12.15 2011-12-15T16:46:58 weird 2011-12-15T16:47:04 mcstar: no, everything was purged 2011-12-15T16:47:18 k, doesnt really matter 2011-12-15T16:47:19 also, 13.12.2011 ↠day first, second month 2011-12-15T16:47:24 come on 2011-12-15T16:47:27 im kidding 2011-12-15T16:47:30 you're trolling me 2011-12-15T16:47:33 ;( 2011-12-15T16:47:35 lol 2011-12-15T16:49:35 *** u_ has joined #aichallenge 2011-12-15T16:50:24 Fluxid, BenJackson, I'm curious what your guys background is in programming? Did you go to school for it? Or just work in it? 2011-12-15T16:53:39 they had a pneumonia shot infected with a retorvirus when they were young, a common side effect was that those children exhibited enhanced abilities with digital systems 2011-12-15T16:54:07 Aww many, my bot was at 87.7mu, and now it's down in the 85s 2011-12-15T16:55:00 mcstar: lol 2011-12-15T16:55:33 i will treasure this game always: http://aichallenge.org/visualizer.php?game=243931&user=432 2011-12-15T16:56:01 but i guess staying alive while not always being in 1st place is a better strategy 2011-12-15T16:56:20 Gnome: i've started programming in visualbasic in elementary school, in middle school learned it more, in highschool played with pascal and php and in uni java, c, prolog, python, and now working with python full-time: banking systems and financial stuff 2011-12-15T16:57:02 I started with visual basic as a kid, and then taught myself C++ in highschool 2011-12-15T16:57:07 and later Python 2011-12-15T16:57:13 visual basic. pah, basic wasn't visual in my day 2011-12-15T16:57:27 why is it visual??? 2011-12-15T16:57:40 cause you can make a gui with it? 2011-12-15T16:57:44 That's cool, Fluxid, how is your bot doing in this challenge? 2011-12-15T16:57:52 a1k0n: actually i started with qbasic in msdos 6 2011-12-15T16:57:54 plus it doesn't have line numbers 2011-12-15T16:58:02 msdos 6?! 2011-12-15T16:58:09 but then i've got visual basic 5 for CHRISTMAS 2011-12-15T16:58:11 nice. 2011-12-15T16:58:17 best christmas ever, i'm telling you 2011-12-15T16:58:22 mcstar: https://upload.wikimedia.org/wikipedia/en/7/75/Microsoft_Visual_Studio_6_screenshot.png 2011-12-15T16:58:31 hey thestinger 2011-12-15T16:58:36 hi 2011-12-15T16:58:43 i had the ROM basic on a TRS-80 CoCo. until we got a 286 when i was in middle school where i could use GW-BASIC for DOS 3.3. 2011-12-15T16:58:50 thestinger: i know, i used it 2011-12-15T16:59:00 i just dont get the "visual" adjective 2011-12-15T16:59:03 it's probably just their crappy visual studio branding 2011-12-15T16:59:03 mcstar: yeah, guis and stuff, and OO 2011-12-15T16:59:09 visual C++, etc. 2011-12-15T16:59:22 well, visual c++ also had gui editor.. 2011-12-15T16:59:25 k, but its just stupid advertising idiom 2011-12-15T16:59:27 but then they went full retard 2011-12-15T16:59:31 my best christmas ever was when i got a cassette recorder for my TRS-80 so i could actually save/load my programs 2011-12-15T16:59:31 https://www.microsoft.com/visualstudio/en-us/products/2010-editions/visual-cpp-express 2011-12-15T16:59:35 yeah, they call it "Visual C++" 2011-12-15T16:59:41 *** Murashka has joined #aichallenge 2011-12-15T16:59:45 Gnome: around 22nd place i think 2011-12-15T16:59:45 i even wrote a program on my computer to wake me up when santa came 2011-12-15T16:59:52 you know, there are realy VISUAL programming environment 2011-12-15T17:00:07 (i was 6) 2011-12-15T17:00:15 where you can build up your program interactively, (not gui, program) 2011-12-15T17:00:17 Have you done any previous AI stuff or is this your first go? I'm finding it hard to just start coding mine, haha, I think I'm over analyzing my first attempt. 2011-12-15T17:00:18 a1k0n: oh boy 2011-12-15T17:00:25 *** ChrisH_ has joined #aichallenge 2011-12-15T17:00:33 mcstar: lego mindstorms, scratch, etc.? 2011-12-15T17:00:37 a1k0n: you attached a motion sensor to it? 2011-12-15T17:00:38 ok, i've started at around 12 2011-12-15T17:00:46 Gnome: neural networks on uni 2011-12-15T17:00:48 heh no, it required his cooperation. 2011-12-15T17:00:59 but, uh, not really applied in ants 2011-12-15T17:01:24 thestinger: dunno about their names, some lisp or prolog variants probably 2011-12-15T17:01:33 logo was derived of a lisp 2011-12-15T17:01:47 oh lego 2011-12-15T17:01:54 dunno about that either 2011-12-15T17:02:03 @later tell janzert A lot of my matches after my last couple submissions seem to have pretty poor match quality. Maybe it's just bad luck, but it seems too frequent. 2011-12-15T17:02:03 ChrisH_: Ready to serve. 2011-12-15T17:03:26 *** amstan has quit IRC (Remote host closed the connection) 2011-12-15T17:04:19 ag yeah, quartz composer and simulink are good examples 2011-12-15T17:05:35 lol 2011-12-15T17:05:42 Note: Microsoft Visual Studio and the languages it encompasses (Visual Basic, Visual C#, Visual J#, etc.) are commonly confused to be but are not visual programming languages. All of these languages are textual and not graphical. MS Visual Studio is a visual programming environment, but not a visual programming language, hence the confusion. 2011-12-15T17:05:59 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-12-15T17:06:10 BenJackson: Ugh, I just got my 5th consecutive 1v1 low mu matchup. Eight games, eight straight wins, and I'm still only #693. My V5 got to #53 in fewer games. 2011-12-15T17:07:41 *** jstrong has joined #aichallenge 2011-12-15T17:08:00 ChrisH_: I won 12 in a row on this submit I think 2011-12-15T17:08:47 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Ping timeout: 252 seconds) 2011-12-15T17:09:06 Gnome: I started programming on my own at a young age and eventually majored in CS 2011-12-15T17:09:16 but most of the important things I learned were from practical experience and not the classroom 2011-12-15T17:09:46 werent you amazed by 6.001 ? 2011-12-15T17:10:33 mit decided to teach 6.01, and switched from scheme to python 2011-12-15T17:10:41 BenJackson, yeah I get that. I've been coding since around middle school, got my degree in cs (software emphasis). Been trying to break into the AI community and get good at it, but it's proven a challenging area. 2011-12-15T17:11:20 I had an ai book when I was a kid 2011-12-15T17:11:30 but I didn't really grok it beyond writing a tic-tac-toe game 2011-12-15T17:11:52 I did a lot of graphics hacking on my c64 2011-12-15T17:11:56 and wrote/ran a bbs for years 2011-12-15T17:12:15 like a character from Hackers 2011-12-15T17:12:19 I still think people should learn to program on something like a C64 2011-12-15T17:13:56 Gnome: it was actually lego mindstorms (which I got as part of a LEGO revival when I realized I could afford all the LEGO I wanted) 2011-12-15T17:14:08 that got me interested in "real life" AI 2011-12-15T17:14:17 I'd written so many programs where the inputs were "known" 2011-12-15T17:14:32 trying to write a program with incomplete, incorrect, misleading information was interesting 2011-12-15T17:14:49 then I entered the 1998 (?) ICFP pousse contest 2011-12-15T17:14:53 which was only 3 days long 2011-12-15T17:15:17 Do you enter a lot of contests now? 2011-12-15T17:15:22 that was interesting because I had no idea how to play myself (it's vaguely like tic-tac-toe meets connect-4) 2011-12-15T17:15:34 no, not really 2011-12-15T17:15:38 I didn't notice tron 2011-12-15T17:15:50 and when I saw PW it didn't appeal 2011-12-15T17:15:57 *** amstan has joined #aichallenge 2011-12-15T17:15:57 *** ChanServ sets mode: +o amstan 2011-12-15T17:16:47 yeah, ants is super interesting 2011-12-15T17:16:52 how do you guys handle defense? 2011-12-15T17:16:59 i've been wanting to try for so long, so i had to do ants 2011-12-15T17:17:06 the best defense is a good offense. :) 2011-12-15T17:17:40 I've been trying to find a community that does these types of puzzles frequently. I think ants is an awesome challenge. 2011-12-15T17:17:49 I found the web page for the winning pousse entry 2011-12-15T17:17:53 keep in mind, 3 days 2011-12-15T17:17:56 I was working alone 2011-12-15T17:18:09 I used to put up a ring of ants around my hill that expanded depending on how many ants I had and how close the enemy was, but now I just go right at them and attack the enemy ants to keep them out of my territory. 2011-12-15T17:18:09 Yeah that would be tough man, how did you end up doing? 2011-12-15T17:18:18 the winners were a HD sophomore, a MIT sophomore, 2 grad students, a research scientist, two professors and a system administrator 2011-12-15T17:18:22 as a team 2011-12-15T17:18:39 *** treeform has quit IRC (Ping timeout: 255 seconds) 2011-12-15T17:18:40 I don't remember my exact place, something like 6th 2011-12-15T17:18:57 Wow, not bad at all. 2011-12-15T17:18:57 ChrisH_: that's largely what V11 does that's new for me too 2011-12-15T17:19:00 yeah, but for this to work you have to make a call how many ants to leave behind 2011-12-15T17:19:04 V12 is about doing that more proactively 2011-12-15T17:19:20 thats very tricky. too few and you get killed. too little and you can't eat food fast enough and get strategically lost 2011-12-15T17:19:21 which it does so well that none of my personal test bots usually even find my hill 2011-12-15T17:19:24 That was what put my V5 into the top 5, now a lot more people are doing it too. 2011-12-15T17:19:29 ChrisH_ I was thinking of your original strategy as well. Did it not work as well as just straight up offense? 2011-12-15T17:19:50 defending is a waste of ants 2011-12-15T17:19:54 ChrisH_: yeah, I would say that after combat defined the top tier it is being redefined by strategy again 2011-12-15T17:20:13 OH and V12 solves (I think!) the rotating combat problem 2011-12-15T17:20:24 Gnome: the problem was it was just too inefficient, it let the enemy get too close and the ants on defense were frequently defending against nothing in one or more directions. 2011-12-15T17:20:44 *** treeform has joined #aichallenge 2011-12-15T17:20:50 The defending ants aren't doing anything useful most of the time 2011-12-15T17:21:13 I think the defending in my old version wored decently 2011-12-15T17:21:30 now they're too scared to attack enemies until it's too late 2011-12-15T17:21:33 Think of this: You shouldn't defend your hills, you should defend your feeding grounds around your hill to preserve exclusive access to the biggest food supply possible. 2011-12-15T17:21:54 Yeah, a continuous perimeter you have visuals on 2011-12-15T17:22:10 And if there is no enemy in sight, then move farther out. 2011-12-15T17:22:23 When you make contact with the enemy you need good combat. 2011-12-15T17:22:54 Then it becomes a matter of who can resupply their lines the fastest with the most ants, or who can out flank the other. 2011-12-15T17:23:20 http://web.archive.org/web/20060831061035/http://www.ai.mit.edu/extra/icfp-contest/phase1.html 2011-12-15T17:23:38 that's the closest thing I can find to a full result for that contest 2011-12-15T17:23:43 the final announce only goes to 3rd place 2011-12-15T17:24:05 ChrisH_: I would argue that it's who can resupply their lines most efficiently 2011-12-15T17:24:08 that's what xathis is crazy good at 2011-12-15T17:24:33 one thing I'm wondering about. Do you intentionally create a grid of ants to watch every square? 2011-12-15T17:24:35 I think that's just two words for the same thing. 2011-12-15T17:24:36 xathis appears to 2011-12-15T17:25:01 I do this, but I'm having problems routing ants to it (timeouts) 2011-12-15T17:25:23 if it works, it produces setups similar to xathis 2011-12-15T17:25:49 my V11 sometimes gets surprised by ants and has to push them back 2011-12-15T17:25:55 there's also issues in maps with lots of water 2011-12-15T17:25:59 also it sometimes values slipping combat ants past the combat 2011-12-15T17:26:02 when they should be defending 2011-12-15T17:26:05 was it gets tricky to determine where to put the grid 2011-12-15T17:26:18 V12 is better at being in the right place at the right time 2011-12-15T17:26:21 and holding the line 2011-12-15T17:26:39 Antimony: I intentionally maximize visibility as one factor in moving ants 2011-12-15T17:27:04 although you have to turn that off early 2011-12-15T17:27:35 when do you turn it off? i only try to expand vis if nothing better to do :P 2011-12-15T17:27:49 which will probably be much less effective once I have real combat logic~~ 2011-12-15T17:28:02 I have a sliding scale of bonus for exploring that's so high initially that it swamps maintaining visibility 2011-12-15T17:28:08 because you really need your first ants to get out fast 2011-12-15T17:28:20 *** Murashka has quit IRC (Quit: Page closed) 2011-12-15T17:28:24 you need the initial chokepoints to form as far away from you as possible 2011-12-15T17:28:25 ah 2011-12-15T17:28:28 yeah 2011-12-15T17:28:36 a few steps can be the difference between needing to hold a 20 ant wide wall and a 4 ant wide wall 2011-12-15T17:28:56 and like chrish said I've scrapped all "defense" 2011-12-15T17:29:04 defense is not letting you anywhere near my hill 2011-12-15T17:30:01 Here is a fun game to watch in which my bot and pguillory's form a huge line of battle between our territories before his bigger army eventually overwhelms me. http://aichallenge.org/visualizer.php?game=238002&user=10 2011-12-15T17:32:01 there's one of my recent games including pguillory where at least 3 times I have him pinned with a defensive line and then we twist sideways and I let him by 2011-12-15T17:32:05 and then pick him up with a new line 2011-12-15T17:32:07 and repeat 2011-12-15T17:32:11 hopefully V12 does not do that 2011-12-15T17:33:21 i love when a bot breaks a defensive line 2011-12-15T17:33:24 and tons of ants die for free 2011-12-15T17:33:29 feels like real strategy then :D 2011-12-15T17:33:58 I experimented with some changes to my pathing to combat in V6 and it was worse, so V7 restored my V5 behavior, but with a few bug fixes. Like you said earlier BenJackson I've been improving (hopefully) my strategic planning in the last couple versions. 2011-12-15T17:34:51 i only have 1 version 2011-12-15T17:35:48 ah 2011-12-15T17:35:52 i wanted to ask 2011-12-15T17:36:09 do you guys write your bots in an object oriented fashion? 2011-12-15T17:36:21 most of you program in java/c++ 2011-12-15T17:36:57 what constitutes object oriented? 2011-12-15T17:37:20 inheritance, polymorphism 2011-12-15T17:37:21 mcstar, mine is in Go, which is type oriented and has methods but isn't strictly OO 2011-12-15T17:37:22 I use Go, which is only partially OO, but I do not use those features much because most of my logic is just manipulating arrays of integers. 2011-12-15T17:37:22 somesuch 2011-12-15T17:37:28 ChrisH_, Hi! 2011-12-15T17:37:56 kevlar_work: hi 2011-12-15T17:37:59 mcstar: no inheritance, and the polymorphism comes from templates (so basically, just like type classes in haskell) 2011-12-15T17:38:03 i cant imagine doing my thing in a pure functional fashion 2011-12-15T17:38:14 I have lots of random objects but I'm mostly just using them as records 2011-12-15T17:38:22 there's not really a lot of need for polymorphism... 2011-12-15T17:38:32 im heavily using stl 2011-12-15T17:38:43 i do have a few interfaces for strategies and such 2011-12-15T17:39:12 for me an ant is usually just an integer (it's position on the map), and sometimes two integers (it's position and the turn that it is there). 2011-12-15T17:39:18 and all my algorithsm operate heavily on mutable data structures 2011-12-15T17:39:22 *** mj41 has joined #aichallenge 2011-12-15T17:39:32 i really want to have a look at a top haskell player's code 2011-12-15T17:39:38 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-15T17:39:40 everything is about ants and turns, so OO isn't really applicable at that level 2011-12-15T17:40:05 I do have a "tile" object, but that's mostly for my sanity 2011-12-15T17:40:28 I copy the map state into int[]s for most calculations, so the algos aren't coupled to the map repr 2011-12-15T17:40:38 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-15T17:41:09 i have my ants as a class, and i store pointers to them in lists and on a 2d array 2011-12-15T17:42:03 and i have 22 whole-map sized 2d arrays 2011-12-15T17:42:18 more really 2011-12-15T17:42:29 thats not couting the ones for pathfinding 2011-12-15T17:43:06 meh, the tcp server is so slow now 2011-12-15T17:43:10 *** yoden-cloud_ has quit IRC (Ping timeout: 258 seconds) 2011-12-15T17:43:10 mcstar: you're alive! 2011-12-15T17:43:21 people are using up way more than 500ms again 2011-12-15T17:43:21 mcstar: i'm just waiting :) 2011-12-15T17:43:23 i was alive the whole time 2011-12-15T17:43:26 hello 2011-12-15T17:43:29 hey 2011-12-15T17:43:37 *** Darhuuk_ has joined #aichallenge 2011-12-15T17:43:55 oh, so you gonna submit only for the finals? 2011-12-15T17:44:02 yeah 2011-12-15T17:44:20 you must have some top secret strategy 2011-12-15T17:44:20 gutsy 2011-12-15T17:44:28 that's how i roll :P) 2011-12-15T17:44:30 antimatroid: have you played on tcp? 2011-12-15T17:44:35 yeah, skycaptain 2011-12-15T17:44:43 i'm working on fixing bugs and getting rid of stupidness 2011-12-15T17:45:04 and it liked to timeout yesterday, but i'm fairly confident that was network issues 2011-12-15T17:45:46 I had to add an explicit garbage collect to my game loop to smooth out the response times 2011-12-15T17:46:01 antimatroid: which server? i dont see you on flux.id 2011-12-15T17:46:03 so I could cut down on my leeway time 2011-12-15T17:46:12 SkyCaptain sorry 2011-12-15T17:46:16 it's case sensitive 2011-12-15T17:46:34 i didnt know that 2011-12-15T17:46:41 *** Darhuuk has quit IRC (Ping timeout: 244 seconds) 2011-12-15T17:47:18 sky captain and the world of tomorrow? 2011-12-15T17:47:22 ive seen such a movie 2011-12-15T17:47:33 SkyCaptain from future cop lapd preccint assault 2011-12-15T17:47:38 oh 2011-12-15T17:47:46 ok, you might have told me that b4 2011-12-15T17:48:40 yeah my combat isn't as good as it should be 2011-12-15T17:48:52 i'm going to work on picking our better sample moves to test 2011-12-15T17:50:47 I'm just going to merge everything and reupload :P 2011-12-15T17:52:07 *** Elderwolf has quit IRC (Ping timeout: 240 seconds) 2011-12-15T17:52:41 http://ants.fluxid.pl/replay.35629 making my bot really aggressive/reckless works well 2011-12-15T17:52:58 i hope my new macro strategy merged with a tiny bit of "not leaving my hill's sight" will result in a decent place 2011-12-15T17:54:04 I don't try to enumerate a lot of possible enemy moves now 2011-12-15T17:54:12 *** mj41 has quit IRC (Ping timeout: 255 seconds) 2011-12-15T17:54:27 just need a "good" enemy offensive move to test against 2011-12-15T17:54:34 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-15T17:54:46 who? 2011-12-15T17:54:51 and if they do something different, I'll just make 1 for 1 trades and hold ground/gain ground 2011-12-15T17:55:12 delete that 2011-12-15T17:55:19 i didnt see the 'I' 2011-12-15T17:55:29 well, I was trying to get ~3-10 good sets of enemy moves (for each battle), but it's better if I just do 1 2011-12-15T17:55:34 *** tomsax has joined #aichallenge 2011-12-15T17:56:07 bots with really good combat are cautious and they don't hold ground well 2011-12-15T17:56:16 I can make some stupid moves that would let them trade with me, but they won't 2011-12-15T17:56:41 You can push bots with "good" combat around 2011-12-15T17:56:48 worse is better :) 2011-12-15T17:58:02 I use something like Memetix posted on the forums to stop obviously stupid things from happening 2011-12-15T17:58:09 *** tomsax has quit IRC (Client Quit) 2011-12-15T17:58:29 *** kire has joined #aichallenge 2011-12-15T18:00:22 i feel like im in a disadvantage that i havent read it 2011-12-15T18:00:32 but i would hate myself if i read it 2011-12-15T18:00:36 it didn't really work for me as a full combat implementation, at all 2011-12-15T18:00:44 it was way worse than what I was already doing 2011-12-15T18:00:55 but it works well as an extra sanity check 2011-12-15T18:01:05 *** Elderwolf has joined #aichallenge 2011-12-15T18:01:18 so now I can be more reckless with my old method, and it won't be a bad thing 2011-12-15T18:01:53 like in a 2v1 I'll assume they will move towards one of my ants 2011-12-15T18:02:23 a cautious/good bot will just run away anyway 2011-12-15T18:02:50 *** choas has joined #aichallenge 2011-12-15T18:02:59 I used to come up with all moves they could make, and do the least worst thing - and since I don't trade 1v1 I wouldn't push them around 2011-12-15T18:04:21 *** Jak_o_Shadows has joined #aichallenge 2011-12-15T18:04:25 :S 2011-12-15T18:04:53 if i have have 2 vs 1 and if the other one makes a mistake i kill it 2011-12-15T18:05:18 and in other situations as well 2011-12-15T18:05:28 and cheaply 2011-12-15T18:05:32 nuff said 2011-12-15T18:08:52 *** _Sander_ <_Sander_!~sandre@ip54565dd6.speed.planet.nl> has quit IRC (Quit: Ik ga weg) 2011-12-15T18:10:31 thestinger: are you picking their moves before or after you've "made yours"? 2011-12-15T18:10:47 i changed to picking their moves for each set of my moves 2011-12-15T18:10:57 so i attempt to find the actual worst outcome for each of my sets 2011-12-15T18:11:42 *** meduza has joined #aichallenge 2011-12-15T18:12:59 antimatroid: have you tried running your new bot against your beta bot? 2011-12-15T18:13:04 *** pairofdice has quit IRC (Quit: When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl.) 2011-12-15T18:13:17 mcstar: my beta bot didn't know about hills 2011-12-15T18:13:22 lol 2011-12-15T18:13:24 i know 2011-12-15T18:13:28 have you? 2011-12-15T18:13:31 no 2011-12-15T18:13:39 my "stupid algorithm" bot has done better on live than my "smart" bot 2011-12-15T18:13:45 I love AI sometimes 2011-12-15T18:15:05 that is AAI 2011-12-15T18:16:45 I wonder if you could employ machine learning via the TCP game server to develop a good strategy 2011-12-15T18:16:46 *** antimatroid has quit IRC (Read error: Connection reset by peer) 2011-12-15T18:16:57 @rankings 2011-12-15T18:16:58 BenJackson: Top 10 players: xathis(97.3), pguillory(92.4), a1k0n(91.8), FlagCapper(90.9), protocolocon(89.7), delineate(89.1), ThisIsNotABug(88.7), bix0r4ever(88.7), cumbuz(87.4), BenJackson(87.4) 2011-12-15T18:17:02 ok! now I can resubmit 2011-12-15T18:17:04 *** antimatroid has joined #aichallenge 2011-12-15T18:17:24 pretty over ubuntu freezing up on me 2011-12-15T18:19:09 *** kire has quit IRC (Quit: Leaving) 2011-12-15T18:19:28 @later tell ChrisH that it took 48 games to get to #10 2011-12-15T18:19:28 BenJackson: I think that worked... 2011-12-15T18:19:50 hi, does anybody knows how to add new value to affiliations list? right now i can only pick existing one on my profile page 2011-12-15T18:20:07 meduza: used to be a forum post to request in 2011-12-15T18:20:11 but I thought they automated it 2011-12-15T18:20:22 maybe they only automated changing it and not adding new ones 2011-12-15T18:20:29 janzert: see above 2011-12-15T18:20:36 related forum thread is closed for new posts 2011-12-15T18:21:12 haha 2011-12-15T18:21:25 i still have a function that produces random coordinates for output 2011-12-15T18:21:33 i probably can safely delete that 2011-12-15T18:24:12 BenJackson: from the forum: Users are able to submit organization changes(even add new ones) on the site now. Please look around in your profile. 2011-12-15T18:24:36 BenJackson: but it seems there is no such feature on the site 2011-12-15T18:27:19 *** Antimony has joined #aichallenge 2011-12-15T18:27:30 *** Alexer has quit IRC (Ping timeout: 244 seconds) 2011-12-15T18:27:38 *** Alexer has joined #aichallenge 2011-12-15T18:27:40 meduza: when logged in click "change" next to your current affiliation 2011-12-15T18:28:04 BenJackson: and you're pretty certainly overrated at that 2011-12-15T18:29:29 i.e. it took 3 wins in a row to take you from #20 up to #10 2011-12-15T18:29:56 janzert: yep. i can pick value from the list then. but i can't insert value, that is not currently in the list 2011-12-15T18:30:15 yeah, I think you can only add a new organization when signing up 2011-12-15T18:30:36 janzert: my bad... 2011-12-15T18:31:07 it's just that no one has written the feature to add a new one after the fact 2011-12-15T18:31:20 not that it's not desired :} 2011-12-15T18:32:05 janzert: but is there any workaround? 2011-12-15T18:33:50 is there anyway to automatically download replays and generate the input files ourselves? 2011-12-15T18:34:03 Antimony: bot in button 2011-12-15T18:34:13 ? 2011-12-15T18:34:25 on the rhs of the visualiser is a bot in button 2011-12-15T18:34:26 use that 2011-12-15T18:34:46 "automatically" 2011-12-15T18:34:51 oh 2011-12-15T18:34:56 sorry :P 2011-12-15T18:35:14 i think you'd have to automate that 2011-12-15T18:35:15 it only generates 1 input and it requires you to click through several dialogs 2011-12-15T18:35:24 antimatroid: using c++ or c++11? 2011-12-15T18:35:37 c++11 2011-12-15T18:35:37 anything. I guess Python would probably be easies 2011-12-15T18:35:42 :) 2011-12-15T18:35:53 gimme unordered_Set 2011-12-15T18:36:08 I should try switching to unordered set sometime and see if my bot gets any faster 2011-12-15T18:36:33 you have to write an index function which is a bit of a pita 2011-12-15T18:36:45 goes from like set to unordered_set 2011-12-15T18:37:01 i should really typedef locset 2011-12-15T18:37:01 no, you can override the hash function in teh std namespace 2011-12-15T18:37:04 or something 2011-12-15T18:37:15 admittedly, doing so is also a PITA if you aren't in the global namespace 2011-12-15T18:37:31 what is? 2011-12-15T18:37:39 extending std::hash 2011-12-15T18:37:40 can typedefs include template parameters? 2011-12-15T18:37:47 thats not a PITA 2011-12-15T18:37:51 thats easy as pie 2011-12-15T18:37:57 *** Paradoxial has joined #aichallenge 2011-12-15T18:37:59 it is inside a nonglobal namespace 2011-12-15T18:38:04 becasue you have to create a hole in your namespace 2011-12-15T18:38:28 namespace std{} 2011-12-15T18:38:34 there's a hole in my namespace, my namespace... 2011-12-15T18:38:41 that only works in the global namepsace! 2011-12-15T18:38:54 whats your point? 2011-12-15T18:39:01 everything is global in c++ 2011-12-15T18:39:05 ... 2011-12-15T18:39:07 has a fed up namespce 2011-12-15T18:40:34 how long do people wait before revisiting a not visible square? 2011-12-15T18:41:38 I never explicitly revist them 2011-12-15T18:41:45 I used to, but it was hurting my bot 2011-12-15T18:41:54 same here 2011-12-15T18:42:06 technically, my bot doesn't even explore in the first place 2011-12-15T18:42:26 antimatroid: 5+ steps 2011-12-15T18:42:47 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-15T18:43:18 main server is down to 18 workers? 2011-12-15T18:43:21 i have it at 10 atm 2011-12-15T18:43:41 although i'm collecting the notvisible and potential enemy boarders at the same time 2011-12-15T18:43:51 i might do potential enemy border then not visible border 2011-12-15T18:47:51 Zannick, is there any easy way to get actual replay data, not just one input at a time? 2011-12-15T18:48:50 i think there's a json link 2011-12-15T18:48:51 not sure 2011-12-15T18:49:03 do you know what the link is/ 2011-12-15T18:50:16 Oh wait, I figured it out. http://aichallenge.org/game/246017 2011-12-15T18:50:29 you need to put a slash between game and number 2011-12-15T18:50:49 ah 2011-12-15T18:50:53 yes, just got that 2011-12-15T18:50:57 the russians have the most people in top 100 2011-12-15T18:51:01 game_info.php no longer works 2011-12-15T18:51:59 *** QuirionPt has quit IRC (Quit: Leaving) 2011-12-15T18:56:30 turn 1220, not sure if winning or losing :\ 2011-12-15T18:58:06 when I open it in the browser, I get json, but urllib seems to be returning binary 2011-12-15T18:59:27 *** choas has quit IRC (Ping timeout: 255 seconds) 2011-12-15T19:02:00 *** Frib has quit IRC (Read error: Connection reset by peer) 2011-12-15T19:02:49 http://ants.fluxid.pl/replay.35751 2011-12-15T19:02:50 draw aha 2011-12-15T19:02:58 i went from doing well to not so well though 2011-12-15T19:03:39 *** Frib has joined #aichallenge 2011-12-15T19:07:25 *** SMJ has joined #aichallenge 2011-12-15T19:11:32 who is M? 2011-12-15T19:14:10 *** honkblargio has joined #aichallenge 2011-12-15T19:20:49 *** Redgis has quit IRC (Ping timeout: 252 seconds) 2011-12-15T19:27:08 damn 2011-12-15T19:27:15 my bot just ate all my ram 2011-12-15T19:28:59 *** replore has joined #aichallenge 2011-12-15T19:34:33 mcstar, I almost hung my computer last night when I ran 4 copies of my bot and had a flash app open in Chrome and my computer decided to start doing a backup 2011-12-15T19:35:15 Ctrl+Alt+F1 2011-12-15T19:36:11 nah, command-space "acmo" enter *kill flash* 2011-12-15T19:36:41 then all was good, the bots ate memory and the backup had the disk all to itself. 2011-12-15T19:36:51 s/memory/cpu and memory/ 2011-12-15T19:37:07 *** Boscop has quit IRC (Ping timeout: 240 seconds) 2011-12-15T19:37:58 when you run out of memory on linux it can be pretty severe 2011-12-15T19:38:14 this is mac os 2011-12-15T19:38:15 sometimes you dont have the time to switch to terminal and kill the app 2011-12-15T19:38:19 yeah 2011-12-15T19:38:30 i guess on that you have a bit more time 2011-12-15T19:38:54 when something eats up memory, i can no longer use X 2011-12-15T19:38:57 I think it's only because it takes apps longer to get ahold of the memory the system is keeping around as "inactive" instead of "free" 2011-12-15T19:39:39 presumably in case the original owner of the memory comes back for it or something, not sure. 2011-12-15T19:39:51 but yeah, X11 makes matters difficult 2011-12-15T19:39:53 i think theres some process priority differences between the two systems 2011-12-15T19:40:01 likely 2011-12-15T19:40:09 osx is very much graphical 2011-12-15T19:40:16 lol, yeah 2011-12-15T19:40:34 i mean Darwin is behind that, which is just a bsd, but still, aqua has high priority 2011-12-15T19:40:42 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-15T19:41:45 bsd in general has some different philosophies about memory 2011-12-15T19:41:49 not that I have any idea what they are, lol 2011-12-15T19:42:00 yes, its very much different 2011-12-15T19:42:26 different kernels, different philosophies, and different virtual memory systems 2011-12-15T19:42:34 im not familiar with the details either 2011-12-15T19:42:57 but i read up on it sometime ago, and maybe bsd felt more "thought out" 2011-12-15T19:45:13 i made my BSF faster 2011-12-15T19:45:24 at the cost of this memory error 2011-12-15T19:45:44 so most of the time its faster, but sometimes it never terminated 2011-12-15T19:45:46 s 2011-12-15T19:45:55 but that happends only once per run 2011-12-15T19:46:02 so this is an improvement right? 2011-12-15T19:47:40 there are just way too many typos in the prev. lines to correct.... 2011-12-15T19:49:06 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-15T19:49:47 after reading tweets from my mother for years, my eyes just autocorrect for spelling errors. 2011-12-15T19:49:50 er, texts* 2011-12-15T19:49:58 she doesn't tweet (yet?)... 2011-12-15T19:50:55 *** Jak_o_Shadows has quit IRC (Read error: Connection reset by peer) 2011-12-15T19:53:17 *** Jak_o_Shadows has joined #aichallenge 2011-12-15T19:56:11 *** meduza has quit IRC (Ping timeout: 258 seconds) 2011-12-15T19:58:22 does anyone know how to generate input files from the json? 2011-12-15T19:59:10 you can generate bot input files from the visualizer directly 2011-12-15T19:59:20 other than that I know of no easy way. 2011-12-15T19:59:32 read source for the visualizer and extrapolate? 2011-12-15T20:00:39 aww man, it's in javascript 2011-12-15T20:01:11 at least it's not obfuscated like the web version 2011-12-15T20:06:19 *** Paradoxial has quit IRC (Read error: Connection reset by peer) 2011-12-15T20:06:39 *** Paradoxial has joined #aichallenge 2011-12-15T20:09:16 weird 2011-12-15T20:09:39 i finish consecutive turns in 5ms and 200ms 2011-12-15T20:09:55 thats huge variance 2011-12-15T20:10:04 I've seen some things like that 2011-12-15T20:10:08 I never did figure out why 2011-12-15T20:12:26 i think the bot im measuring doesnt get even time-shares of the CPU 2011-12-15T20:13:34 theres some option to the engine whether to run the bots parallel or in serial 2011-12-15T20:13:54 I'm really tempted to take my new strategy to avoid timeouts from my smart but unskilled bot and graft it onto my dumb but apparently skilled bot 2011-12-15T20:14:19 *** kincody has joined #aichallenge 2011-12-15T20:14:21 its main problem gaining rank was when it timed out >250 ants or so 2011-12-15T20:15:06 whats your language? 2011-12-15T20:15:44 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-15T20:16:08 *** SMJ has joined #aichallenge 2011-12-15T20:16:11 Go 2011-12-15T20:16:39 oh, nice, i get down to 2ms if i run in serial 2011-12-15T20:16:48 *** honkblargio has left #aichallenge 2011-12-15T20:16:56 ~30 ants in 4 ms 2011-12-15T20:17:16 much better 2011-12-15T20:17:27 with how much visible area? 2011-12-15T20:18:01 why? 2011-12-15T20:18:16 i mean why does that matter? 2011-12-15T20:18:28 is your algorithm's complexity not based on how much of the board you have to consider? 2011-12-15T20:18:39 not really 2011-12-15T20:19:59 when I started comparing multiple goals for each ant (instead of blindly going toward the "closest shiny object") I found that my timing became somewhat more dependant on how much visible area there was to check 2011-12-15T20:21:26 my bot is stuck around rank 11 to 15 so I guess I'll reupload 2011-12-15T20:22:03 nothing really improves it anymore though 2011-12-15T20:22:18 thestinger: not even you? 2011-12-15T20:22:30 mcstar: :) 2011-12-15T20:22:34 no, I just make it worse 2011-12-15T20:22:57 randomly deleting lines of code would probably work pretty well 2011-12-15T20:24:17 you are too humble 2011-12-15T20:24:29 thestinger: my goal is to beat you :P 2011-12-15T20:24:32 cause our combat is similar 2011-12-15T20:25:36 my goal is to get to the first page again and stay there this time 2011-12-15T20:25:42 *** kincody has quit IRC (Ping timeout: 258 seconds) 2011-12-15T20:25:47 I want to get in the top 25 2011-12-15T20:25:58 so, I'm not going to risk merging my fancy new stuff 2011-12-15T20:26:03 i wont be happy until i place top 10 in an ai comp 2011-12-15T20:26:05 not enough time left to test it 2011-12-15T20:26:07 but i doubt this will be my comp 2011-12-15T20:26:18 my comp? 2011-12-15T20:26:22 to get top 10 in 2011-12-15T20:26:25 i haven't had enough time 2011-12-15T20:26:31 oh 2011-12-15T20:26:50 i still have things to work on with my bot though 2011-12-15T20:26:58 so i'm hoping the sucky things will improve yet 2011-12-15T20:27:20 i also want first in australia 2011-12-15T20:27:28 solifiquid might be an issue 2011-12-15T20:27:32 who is that? smiley? 2011-12-15T20:27:53 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-12-15T20:28:04 1983? 2011-12-15T20:28:11 i cant remember 2011-12-15T20:29:01 meh, _flag is canadian :P 2011-12-15T20:29:07 yep sure 2011-12-15T20:29:11 ocaml, thats him 2011-12-15T20:29:13 if he wasn't, 1st in canada would be easy 2011-12-15T20:29:48 what about forumidable? 2011-12-15T20:29:57 fourmidable? 2011-12-15T20:30:00 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-15T20:30:14 oh, true - didn't notice he reuploaded 2011-12-15T20:30:23 http://aichallenge.org/profile.php?user=4622 2011-12-15T20:30:34 *** yoden-cloud has joined #aichallenge 2011-12-15T20:30:45 3rd in canada it is then :P 2011-12-15T20:30:50 wooo, combat logic that works for HoldBot 2011-12-15T20:30:51 :) 2011-12-15T20:30:58 progress~~ 2011-12-15T20:31:04 my goal is to get a bot to skill 70 :) 2011-12-15T20:31:14 15:28 < janzert> BenJackson: and you're pretty certainly overrated at that 2011-12-15T20:31:18 but I started two days ago, so. 2011-12-15T20:31:27 janzert: I don't follow 2011-12-15T20:31:48 BenJackson: you should make a ruby script that execs a compiled version of your bot or something :P 2011-12-15T20:32:14 *** SMJ has joined #aichallenge 2011-12-15T20:32:31 thestinger: a transparent ploy to keep c++11 to yourself 2011-12-15T20:32:44 at least use a token lambda if you switch to C++11 xD 2011-12-15T20:32:50 LOLOLOLOL damn 2011-12-15T20:32:59 no the delicious irony woudl be absolutely 0 c++11 features 2011-12-15T20:33:12 i noticed a 5 turn periodic increase in turntime 2011-12-15T20:33:12 I think it will use C++11 features anyway, it will be faster 2011-12-15T20:33:27 due to the new move semantics, etc. 2011-12-15T20:33:45 (its something that does something turn%5==0, damn) 2011-12-15T20:34:00 maybe _flag will switch to C++11 but just keep using C 2011-12-15T20:34:27 then he'll have to add a bunch of casts to malloc() calls 2011-12-15T20:35:45 he can just #define that 2011-12-15T20:35:46 muahaha 2011-12-15T20:35:54 the only c++11 thing i'm using is unordered_set i think ehe 2011-12-15T20:36:03 antimatroid: no 'auto'? 2011-12-15T20:36:05 std::tr1 has unordered_set anyway 2011-12-15T20:36:11 yeah, tr1/unordered_set 2011-12-15T20:36:13 nope? 2011-12-15T20:36:20 i don't understand what auto does 2011-12-15T20:36:27 type inference 2011-12-15T20:36:33 like templates do 2011-12-15T20:36:35 but locally 2011-12-15T20:36:50 antimatroid: auto it = my_vector.begin() 2011-12-15T20:37:00 so you don't have to spell out vector::iterator 2011-12-15T20:37:29 * antimatroid goes to try that out! 2011-12-15T20:37:53 the first C++11 feature I used was the right-angle bracket thing 2011-12-15T20:37:55 xD 2011-12-15T20:38:03 i told him that weeks before, never mind 2011-12-15T20:38:06 std::vector> 2011-12-15T20:38:09 (auto) 2011-12-15T20:38:13 barely even a feature 2011-12-15T20:38:14 mcstar: i'm terrible at these things :) 2011-12-15T20:38:36 yeah, it really annoys me that most compilers can catch that >> now 2011-12-15T20:38:44 but of course can't accept it if you're not in c++11 mode 2011-12-15T20:39:01 I use unordered_map and unordered_set a lot 2011-12-15T20:39:07 omg, 60 ants battle+pathfinding in 12ms 2011-12-15T20:39:09 because I'm used to coding in languages with those 2011-12-15T20:39:32 and lambdas/closures/higher-order functions 2011-12-15T20:39:49 even using lambdas just as nested functions to limit scope is nice 2011-12-15T20:40:12 well I'm king of tcpants 2011-12-15T20:40:13 thestinger: whats wrong with {}? 2011-12-15T20:40:22 mcstar: I mean, the scope of functions 2011-12-15T20:40:30 instead of having 30 global functions 2011-12-15T20:40:48 hm 2011-12-15T20:40:58 i dont see what does "scope" has to do with it 2011-12-15T20:41:10 has->have 2011-12-15T20:41:32 well, I can define a little function nested inside another and use it 3 times - just makes the code easier to look over 2011-12-15T20:42:12 ok i get it 2011-12-15T20:42:16 i just dont see the scope 2011-12-15T20:42:21 you limit functionality 2011-12-15T20:42:23 and reuse that 2011-12-15T20:42:29 thats ok 2011-12-15T20:42:40 scope for me means "scope of a name" 2011-12-15T20:42:51 well, it does change the scope of the name 2011-12-15T20:42:52 but probably it can be applied in a broader sense 2011-12-15T20:42:57 im just not used to it 2011-12-15T20:43:05 there's something like 'get_nattackers' as a local variable, instead of a global function 2011-12-15T20:43:42 it's somewhat of an idiom in python to use lots of nested functions, even if you aren't using them as closures 2011-12-15T20:43:46 so I'm used to it 2011-12-15T20:44:08 I use closures a lot in my Go 2011-12-15T20:44:33 i just got to know recently that you can use continuations in c 2011-12-15T20:44:53 I also have lots of little anonymous functions (passing them as arguments to higher-order functions) 2011-12-15T20:45:02 the syntax is pretty ugly though 2011-12-15T20:46:29 anyone know who Flower is on tcpants? 2011-12-15T20:46:55 (I just can't help reading that as T C Pants, lol) 2011-12-15T20:47:21 kevlar_work: nobody can :) 2011-12-15T20:48:56 *** NightExcessive has joined #aichallenge 2011-12-15T20:50:36 strcat_next 2011-12-15T20:50:41 playing against it 2011-12-15T20:50:44 im gonna die 2011-12-15T20:50:54 :P 2011-12-15T20:51:07 that's actually the last game it will play 2011-12-15T20:51:11 whats the load-time on tcp? 2011-12-15T20:51:17 *** alc has joined #aichallenge 2011-12-15T20:51:18 it takes ages for the game to start 2011-12-15T20:51:29 I think some bots are "cheating" the turntime 2011-12-15T20:51:35 and loadtime too probably 2011-12-15T20:51:46 you're told that 500ms is the turntime, but you get up to 5s 2011-12-15T20:51:51 my games can take 30-45 mins now 2011-12-15T20:52:02 I'm sure there are bots using more than 500ms, it's annoying as hell 2011-12-15T20:52:11 probably not intentional 2011-12-15T20:52:14 *** Antimony has joined #aichallenge 2011-12-15T20:52:43 http://aichallenge.org/visualizer.php?game=246691&user=3473 2011-12-15T20:52:47 arrgh 3-way tie for 1st 2011-12-15T20:53:27 lots of names I don't recognize in the top 100 now 2011-12-15T20:54:11 :( timed out 2011-12-15T20:54:24 might be a network thing 2011-12-15T20:54:30 it is 2011-12-15T20:54:33 always is 2011-12-15T20:54:38 im just annopyed 2011-12-15T20:54:39 10, 11, 12 are separated by .03 2011-12-15T20:54:39 -p 2011-12-15T20:54:41 yeah i time out all the time 2011-12-15T20:54:45 on tcp 2011-12-15T20:54:49 "14-10 are separated by .25 2011-12-15T20:54:57 but my bot thinks it returns within like 100-200ms so i'm not too worried 2011-12-15T20:54:57 I almost never time out on TCP 2011-12-15T20:54:57 some days my bots timeout 50% of the time, and they drop in rank a lot 2011-12-15T20:55:26 rwest_: i'm in tasmania :P 2011-12-15T20:55:36 on wireless 2011-12-15T20:55:41 with a crappy isp 2011-12-15T20:55:42 antimatroid: is the devil there? 2011-12-15T20:55:46 ^^ 2011-12-15T20:55:50 lots of 'em :P 2011-12-15T20:55:55 janzert: can we get the top by mu query? 2011-12-15T20:56:00 I thikn there are a lot of resubs percolating up 2011-12-15T20:56:47 time to watch my bot play pointless games against starter bots for 5 hours 2011-12-15T20:56:53 So I am working on this puzzle for an inteview, I can't believe they give this shit for a web dev job hah 2011-12-15T20:57:07 has to compute in under a minute 2011-12-15T20:57:28 one minute? did you mean 1 second? 2011-12-15T20:57:34 I am raising a 36700x36700 matrix to the 10th power 2011-12-15T20:58:07 yikes 2011-12-15T20:58:09 and then summing the resulting values 2011-12-15T20:58:16 it supposed to fit in a 64-bit int 2011-12-15T20:58:18 arbitrary matrix? or one with tricks 2011-12-15T20:58:27 I have split it up 2011-12-15T20:58:30 into squares 2011-12-15T20:58:43 so I do A^2 x A^2 to get A^4 2011-12-15T20:58:45 and the like 2011-12-15T20:59:05 is it a dense matrix? 2011-12-15T20:59:07 hasn't even returned yet, gotta be like 10 min 2011-12-15T20:59:16 its an adjacentcy matrix 2011-12-15T20:59:24 oh 2011-12-15T20:59:26 is it diagonalizable? 2011-12-15T20:59:34 yep 2011-12-15T20:59:38 antimatroid: I highly doubt it 2011-12-15T20:59:38 i mean try that 2011-12-15T20:59:49 actuall let me look at my printed matrix 2011-12-15T20:59:52 there's got to be a trick. let me think about it 2011-12-15T20:59:59 rwest_: is the matrix online? 2011-12-15T21:00:02 now we all want to play 2011-12-15T21:00:14 there are several techniques for faster matrix mul 2011-12-15T21:00:16 haha the matrix is just the solution I came up with 2011-12-15T21:00:22 but the best ones only apply to certain kinds of matrices 2011-12-15T21:00:24 there's a way to do something like that with the eigenvectors i think, but pita to get those 2011-12-15T21:00:37 sec I can dropbox the pdf 2011-12-15T21:00:41 Even with repetead squaring, 37600 row multiplication is slow 2011-12-15T21:00:45 gauss does it in n^3 i think 2011-12-15T21:00:48 I think the trick invovles taking advantage of the fact 2011-12-15T21:00:52 I knew a very strange grad student who was obsessed with giant matrix mul 2011-12-15T21:00:54 that you only need the sum 2011-12-15T21:00:59 and used early 64-proc parallel sun boxes to work on them 2011-12-15T21:01:31 https://www.dropbox.com/s/89hd7283yx3546i/Block_Puzzle.pdf 2011-12-15T21:01:34 rwest_: you know about project euler? 2011-12-15T21:01:44 omg 1000 hits already 2011-12-15T21:01:46 it's actually a modified problem from project euler 2011-12-15T21:01:51 I did some of project euler when I first started programming 2011-12-15T21:02:03 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-15T21:02:07 I did the first 25 or so project euler problems in shell (using bc and tail and stuff) 2011-12-15T21:02:08 tbh I didn't really like it, the problems required a tiny bit of code to solve so it wasn't really a good way to learn 2011-12-15T21:02:11 like problem 215 2011-12-15T21:02:16 I googled it haha 2011-12-15T21:02:19 which worked because computers are so much more powerful than they scaled for when they made those up 2011-12-15T21:02:23 no real help beyond what I already have 2011-12-15T21:02:34 since I already had what people were hinting at 2011-12-15T21:02:51 I really think this is because I am using BigInteger in Java 2011-12-15T21:02:52 lol 2011-12-15T21:02:57 to sum it 2011-12-15T21:02:58 hm 2011-12-15T21:03:06 wtb uint64 2011-12-15T21:03:12 why 37600^2 matrix for that problem? 2011-12-15T21:03:30 because it's faster than brute forcing 2011-12-15T21:03:33 a lot faster 2011-12-15T21:03:33 I think you gave the wrong problem 2011-12-15T21:03:36 there's no matrix 2011-12-15T21:03:39 you gonna need some recursive alg or somesuch 2011-12-15T21:03:49 recursion is so slow on it 2011-12-15T21:03:58 BenJackson: http://projecteuler.net/profile/strcat.png that's how far I got when I was using it to learn python :) 2011-12-15T21:04:01 make it TCO 2011-12-15T21:04:12 Antimony: I use an adjacency matrix for the different rows of blocks 2011-12-15T21:04:23 I ended up getting stuck on some problems though due to speed issues and gave up 2011-12-15T21:04:23 so I only calculate whether they can stack once 2011-12-15T21:04:41 then if I raise the matrix to the power of height -1 2011-12-15T21:04:47 I get the number of walls I can build 2011-12-15T21:04:55 well you should have said that then 2011-12-15T21:05:06 I was building a graph 2011-12-15T21:05:09 there could be a shortcut that doesn't involve huge matrices 2011-12-15T21:05:18 this reeks of a dynamic programming problem 2011-12-15T21:05:21 and the shortcut to the graph was the matrix 2011-12-15T21:05:28 and from here I am stuck 2011-12-15T21:05:30 I'm really bad at spotting the recurrence relations in DP though 2011-12-15T21:05:43 I am about to switch to C so I can have 64 bit ints 2011-12-15T21:05:45 but I try to be alert to the possibility because if it is DP then it is usually really fast 2011-12-15T21:05:57 they say 0.1s in C 2011-12-15T21:06:02 I can't really see how 2011-12-15T21:06:21 you're probably doing it wrong 2011-12-15T21:06:30 Antimony: that would be my guess lol 2011-12-15T21:06:38 -floop-parallelize-all :P 2011-12-15T21:06:41 I get all the right answers for the example answers shown 2011-12-15T21:07:04 I should just run it on the cluster at work 2011-12-15T21:07:48 precompute the table 2011-12-15T21:07:53 I don't get the problem 2011-12-15T21:08:05 I can't find any interpertation that gives the numbers they have 2011-12-15T21:08:09 10 possible heights, 96 (or less) possible widths 2011-12-15T21:08:14 ok, I'm going to have to write something to figure out which bots are using like 4s a turn 2011-12-15T21:08:18 table of 960 int64 2011-12-15T21:08:19 done 2011-12-15T21:08:26 thestinger: M24 is at the top of my list 2011-12-15T21:08:28 BenJackson: hahah 2011-12-15T21:08:42 I'd hire someone who presented me with the right table 2011-12-15T21:08:48 960 is not too big for a lookup table 2011-12-15T21:08:53 probably smaller than the code to compute it 2011-12-15T21:09:22 BenJackson: the possible block combos for width 48 is 37500ish 2011-12-15T21:09:38 that's a number IN the table 2011-12-15T21:09:38 how the heck do you even get 2 for 7.5x1? 2011-12-15T21:09:48 Antimony: 3 first, 3 second 2011-12-15T21:10:01 I thik that's given to answer questions about symmetry 2011-12-15T21:10:01 interchange the rectangles 2011-12-15T21:10:03 4.5,3 and 3,4.5 2011-12-15T21:10:05 oops 2011-12-15T21:10:14 I was thinking of blocks that are 1x1 and 1.5x1 2011-12-15T21:10:16 oh yeah and step 1 I'd factor out the 1.5 and solve for 3, 2 2011-12-15T21:10:16 nvm 2011-12-15T21:10:23 me too 2011-12-15T21:10:31 I just accidently factored by .5 isntead of 1.5 2011-12-15T21:10:54 BenJackson: dabino_sym (9) bugnutsv6 (2) strcat (2) sevik_v25c (0) sevik_v25c_1 (0) SFX_BotV1 (0) rebelxt16 (0) 2011-12-15T21:10:59 that game took 2.7s per turn 2011-12-15T21:11:03 etc 2011-12-15T21:11:10 I'll just write a script to figure out which bots do it 2011-12-15T21:11:11 how do you get duration? 2011-12-15T21:11:23 well, there's a column with the # of turns 2011-12-15T21:11:28 I only notice M24 by seeing turns go by in my game until he timed out 2011-12-15T21:11:31 and then it took off 2011-12-15T21:11:33 I'm assuming I get a game again right away (I pretty much always do) 2011-12-15T21:11:35 *** AntDroid has joined #aichallenge 2011-12-15T21:11:42 also M24 is pretty good but I don't know who that is on aichallenge 2011-12-15T21:11:48 first thing I do is get the planks with certain numbers of blocks, then I get all permutations of each plank 2011-12-15T21:11:50 but with 2+ seconds of cpu time maybe he's not that good 2011-12-15T21:12:01 for the 48 width 2011-12-15T21:12:05 that is like .1s 2011-12-15T21:12:19 ohh that's your adjacency matrix 2011-12-15T21:12:20 meh, I'm too lazy to deal with html atm 2011-12-15T21:12:24 I'll just do this by hand xD 2011-12-15T21:12:24 you're finding all the possible rows 2011-12-15T21:12:28 yes 2011-12-15T21:12:28 and trying to find out how they could stack up 2011-12-15T21:12:41 so that info is only calculated once 2011-12-15T21:13:03 well the left and right edges have to alternate 2011-12-15T21:13:08 yes 2011-12-15T21:13:12 that is my first drop out 2011-12-15T21:13:40 calculating the first matrix takes around 1s 2011-12-15T21:13:44 under 2s 2011-12-15T21:13:47 under 1s 2011-12-15T21:13:48 err 2011-12-15T21:13:54 what's your representation of the matrix? 2011-12-15T21:13:59 array 2011-12-15T21:14:11 BenJackson: nvm, I think tcp just eats games 2011-12-15T21:14:20 game 35972: survived score: 21 turn: 562 2011-12-15T21:14:25 that was the last game strcat_next played 2011-12-15T21:14:34 http://ants.fluxid.pl/player/strcat_next nowhere to be seen 2011-12-15T21:14:47 * antimatroid has been thinking some games aren't showing up 2011-12-15T21:14:54 can't really figure out the duration of games if some don't show up 2011-12-15T21:15:00 I need to analyze my v12 games and resubmit before fluxid fully implodes 2011-12-15T21:15:10 I think it did 2011-12-15T21:15:16 I'm sure my last 10 or so games haven't shown up 2011-12-15T21:15:18 :\ 2011-12-15T21:15:33 Fluxid: we need you :P 2011-12-15T21:15:47 whoops its a 3329x3329 matrix 2011-12-15T21:16:14 http://ants.fluxid.pl/player/strcat_stupid for example, this bot has been playing games 2011-12-15T21:16:16 @calc 3329 * 3320 2011-12-15T21:16:16 rwest_: 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". 2011-12-15T21:16:20 @math calc 3329 * 3320 2011-12-15T21:16:20 rwest_: 11052280 2011-12-15T21:16:29 the last one on there was almost an hour ago 2011-12-15T21:16:36 but it has been playing nonstop 2011-12-15T21:16:39 BenJackson: http://pastebin.com/iGbXHzJr top 20 mu 2011-12-15T21:17:18 janzert: I think we can rule out xathis as a mathematical anomaly 2011-12-15T21:17:21 all in favor?? 2011-12-15T21:17:25 lol 2011-12-15T21:17:28 hehe 2011-12-15T21:17:44 I had 91.75 before I reuploaded - I think it's more packed at the top than we think 2011-12-15T21:17:48 because people keep reuploading 2011-12-15T21:17:49 toss out the outlier 2011-12-15T21:17:56 well good news is I'm solidly in the top 20 2011-12-15T21:18:01 now to try to get solidly in the top 10 2011-12-15T21:18:03 BenJackson: maybe :P 2011-12-15T21:18:14 i bet there's at least 30 people who feel that way 2011-12-15T21:18:28 http://aichallenge.org/profile.php?user=4622 2011-12-15T21:18:33 lots of bots like that are climbing back up 2011-12-15T21:18:34 11 million BigInteger calls 2011-12-15T21:18:35 lol 2011-12-15T21:18:38 that has to be it 2011-12-15T21:18:39 well I've been inching my way up, losing ground to other bots' improvements but gaining with my own 2011-12-15T21:18:41 fourmidable isn't listed there 2011-12-15T21:19:57 thestinger: a couple of games just turned up for me 2011-12-15T21:20:26 were you eliminated in them? those don't show up write away until they finish 2011-12-15T21:20:31 right away* 2011-12-15T21:20:43 fuck it, back to ants 2011-12-15T21:20:59 http://ants.fluxid.pl/replay.35902 bloody timeouts aha 2011-12-15T21:21:08 thestinger: yeah i timed out in both 2011-12-15T21:21:18 i know i have to wait, just noting that games did indeed turn up 2011-12-15T21:21:58 I'm keeping a list of games that get finished, and some never turn up (I'm just looking at the ones with 'survived' to make it simple) 2011-12-15T21:22:20 rwest_: how much time do you have to solve the problem? 2011-12-15T21:22:43 *are you given* 2011-12-15T21:22:52 mcstar: at least until monday, but I have like 3 other places I have to visit with next week anyway 2011-12-15T21:23:32 there is a similar PE problem, my friend couldnt solve it in 2 days 2011-12-15T21:23:46 I know there is 2011-12-15T21:23:47 he is a graduated mathematician 2011-12-15T21:23:49 I found it 2011-12-15T21:23:56 *** raemde has quit IRC (Read error: Connection reset by peer) 2011-12-15T21:24:00 I am gonna email what I got 2011-12-15T21:24:10 and whatever, it's not a big loss 2011-12-15T21:24:21 i'm hoping to avoid job hunting for the moment 2011-12-15T21:24:26 but that depends on getting phd funding 2011-12-15T21:24:42 the moment the contest is over, im gonna apply to ericsson 2011-12-15T21:24:42 jobs in the Boston area are so plentiful 2011-12-15T21:24:57 i have to move to get a job 2011-12-15T21:25:04 there's no decent jobs in tasmania 2011-12-15T21:25:12 fishing? 2011-12-15T21:25:15 well for Software Engineering 2011-12-15T21:25:15 lol 2011-12-15T21:25:25 antimatroid: just means you have to create your own 2011-12-15T21:25:47 janzert: :P, i'm going to try and weasel my way into the algo trading industry if i have to get a job 2011-12-15T21:25:47 there's this great thing called the internet will let you reach a global customer base too ;) 2011-12-15T21:25:55 and that means going to sydney 2011-12-15T21:26:14 ahh, yeah I imagine there's not a big industry in tasmania for that :) 2011-12-15T21:26:46 plus my best bet for a programming type job would be graduatish because my programming skills aren't very good 2011-12-15T21:26:58 i'm going to spend a bit of time in january learning how to use stuff like debuggers etc. properly 2011-12-15T21:27:06 and git 2011-12-15T21:27:39 but hopefully i can delay this for another 3-4 years :P 2011-12-15T21:27:55 the job part, still going to do the learning /rambling 2011-12-15T21:28:00 :) 2011-12-15T21:28:31 antimatroid: you can learn git in 15 mins :) 2011-12-15T21:28:55 yeah well, better 2011-12-15T21:28:58 if the debugger is gdb, then good luck xD 2011-12-15T21:29:01 and i want to learn how to use linux better 2011-12-15T21:29:23 install Arch or Gentoo, sure way to learn lots of stuff 2011-12-15T21:29:34 you'll break stuff and have to learn tons to fix it 2011-12-15T21:29:56 http://tcpants.com/replay.1343 2011-12-15T21:29:57 i suspect i'm still too much of a newb to be able to do that 2011-12-15T21:30:03 all I ever do on that map is build walls 2011-12-15T21:30:13 antimatroid: you can definitely install Arch, there's an awesome Beginners' Guide 2011-12-15T21:30:29 you can just install it in virtualbox and have it open in a browser 2011-12-15T21:31:13 https://wiki.archlinux.org/index.php/Beginners%27_Guide 2011-12-15T21:31:15 i shall do that in the new year 2011-12-15T21:32:07 and then stuff like compiling a kernel can be fun 2011-12-15T21:32:23 easier than learning C++ because there are actually nice docs :P 2011-12-15T21:32:51 i mostly just learn c++ on a need to use basis 2011-12-15T21:33:02 i need to do something and ask people how i do it 2011-12-15T21:33:06 yeah, that's pretty much how I learned Linux stuff 2011-12-15T21:33:12 hence i don't actually know how to properly describe stuff :P 2011-12-15T21:33:22 I just installed Arch as my first distro, and had to learn everything 2011-12-15T21:34:23 im going to give haskell a spin 2011-12-15T21:34:32 i was looking into functinal languages for some time 2011-12-15T21:34:46 i messed with ocaml, and f# i didnt like it 2011-12-15T21:34:49 only haskell remained 2011-12-15T21:35:02 and i wanna be better at lisp too 2011-12-15T21:35:29 antimatroid: you should try haskell 2011-12-15T21:35:41 yeah haskell still interests me, but it's way lower on the to learn list than it used to be 2011-12-15T21:36:03 i wish c++ had haskells type system 2011-12-15T21:36:05 you place stuff like git and linux above it? 2011-12-15T21:36:15 more practical for the moment 2011-12-15T21:36:38 algebraic data types are so nice 2011-12-15T21:36:55 haskell has type classes 2011-12-15T21:36:58 really ncie 2011-12-15T21:36:59 i know 2011-12-15T21:37:16 i wrote my new alg in it 2011-12-15T21:37:23 the one that will improve my battle 2011-12-15T21:37:38 (well, just prototyped it) 2011-12-15T21:38:46 if only ghc wasn't 780MB 2011-12-15T21:39:56 *** nek has joined #aichallenge 2011-12-15T21:40:21 isnt that the whole haskell-platform? 2011-12-15T21:40:31 no, just ghc and the libs it comes with 2011-12-15T21:40:57 *** NightExcessive has quit IRC (Quit: Leaving) 2011-12-15T21:40:58 Download Size : 63224.66 K 2011-12-15T21:41:00 Installed Size : 621383.00 K 2011-12-15T21:41:09 7.2.2 is in [testing] still 2011-12-15T21:41:29 Arch is dropped haskell-platform because they are way behind the ghc versions, and all the libs in it are old 2011-12-15T21:41:33 is dropping* 2011-12-15T21:41:54 well, im not using it actively 2011-12-15T21:42:04 well, you can still get all the libraries 2011-12-15T21:42:05 and the tutorial i sbased on 6.x that im reading 2011-12-15T21:42:18 Real World Haskell or something? 2011-12-15T21:42:30 the learn yourself a haskell for greater good 2011-12-15T21:42:38 the one with the phunny pictures 2011-12-15T21:42:48 yeah, I learned with that one 2011-12-15T21:42:55 Real World Haskell is a good follow up to that 2011-12-15T21:43:10 I haven't actually finished it yet 2011-12-15T21:43:10 i read the first chapter of that in the spring 2011-12-15T21:43:19 *** alc has quit IRC (Ping timeout: 244 seconds) 2011-12-15T21:43:26 thats when i chose common lisp instead of haskell to learn first 2011-12-15T21:43:29 real world haskell is so much better than learn you a haskell 2011-12-15T21:43:43 as cool as the presentation of learn you a haskell is 2011-12-15T21:43:47 learn you a haskell takes like a day though 2011-12-15T21:44:05 it covers the basics of everything really quickly which is nice, because you can just start making haskell programs 2011-12-15T21:44:29 im at recursion 2011-12-15T21:45:16 you read it in a day? the whole thing? 2011-12-15T21:45:37 mcstar: thanks for not making me feel so stupid :) 2011-12-15T21:45:51 mcstar: yeah 2011-12-15T21:45:56 but lots of the stuff I had learned from other languages 2011-12-15T21:46:13 well, if i dont do anything else, and only READ it, i can do it in a day 2011-12-15T21:46:22 I did it on a sunday or something :P 2011-12-15T21:46:23 but with messing with code and such, no 2011-12-15T21:47:11 thestinger: there are 14 chapters now 2011-12-15T21:47:18 when did you read it? :D 2011-12-15T21:47:31 maybe 6 months ago? 2011-12-15T21:47:42 *** Scryer has joined #aichallenge 2011-12-15T21:48:01 it looks the same 2011-12-15T21:48:12 sure, i didnt think otherwise 2011-12-15T21:48:16 the stuff near the end becomes almost like an appendix/tour of the standard library though 2011-12-15T21:48:27 you're almost done the real stuff 2011-12-15T21:48:43 well, nvm 2011-12-15T21:49:10 near the end it's more about functional programming in general and the standard lib though 2011-12-15T21:49:47 *** treeform has quit IRC (Remote host closed the connection) 2011-12-15T21:50:41 chapter 7 isn't really a chapter, and then 10, 13 and 14 are basically examples 2011-12-15T21:53:18 it would be nice if matchmaking matched you based on where you were on your last upload instead of starting over 2011-12-15T21:54:08 we just need moar servers 2011-12-15T21:54:45 damn tcp games are long :P 2011-12-15T21:54:49 98th yippppie 2011-12-15T21:54:54 some people take ages 2011-12-15T21:55:05 dang 2011-12-15T21:55:07 dropped 30 places 2011-12-15T21:55:08 to 62 :( 2011-12-15T21:55:12 i beat sir macelon 2011-12-15T21:55:42 http://aichallenge.org/visualizer.php?game=246723&user=12 2011-12-15T21:55:47 yeah 2011-12-15T21:55:48 pls take a look at it 2011-12-15T21:55:56 you wont see another one like this from me again 2011-12-15T21:56:15 that's some very nice movement 2011-12-15T21:56:49 thx 2011-12-15T21:57:08 *** bhasker has quit IRC (Ping timeout: 245 seconds) 2011-12-15T21:58:22 winrar 2011-12-15T21:58:33 meanwhile i friggin timeout again 2011-12-15T21:58:57 120ms to do the turn though :\ 2011-12-15T21:59:08 http://ants.fluxid.pl/player/strcat_stupid 2011-12-15T21:59:17 oh my god 2011-12-15T21:59:20 look at the game 3rd down from the top, compared to the 4th 2011-12-15T21:59:23 i just dropped 7 THOUSAND places 2011-12-15T21:59:30 took over an hour 2011-12-15T21:59:33 jstrong: resubmittied? :P 2011-12-15T21:59:39 yeah troid :) 2011-12-15T21:59:43 thestinger: weird 2011-12-15T22:00:19 http://ants.fluxid.pl/replay.36043 friggin timeouts aha 2011-12-15T22:00:22 yeah, so that bot is taking 2.5 to 3.0 seconds per turn 2011-12-15T22:00:43 http://ants.fluxid.pl/player/Truusje_v115 this one 2011-12-15T22:00:53 I narrowed it down 2011-12-15T22:01:38 I'm 60% sure :P 2011-12-15T22:02:27 *** flowenol has quit IRC (Ping timeout: 240 seconds) 2011-12-15T22:03:02 http://ants.fluxid.pl/player/Truusje_v115 2011-12-15T22:03:23 meh 2011-12-15T22:03:26 maybe I'm wrong 2011-12-15T22:07:23 i wonder if protocolocon worked alone or with his comrads 2011-12-15T22:09:12 rufes? 2011-12-15T22:09:20 he is the only one 2011-12-15T22:11:31 check planet wars 2011-12-15T22:11:49 http://planetwars.aichallenge.org/organization_profile.php?org_id=327 2011-12-15T22:12:12 *** raemde has joined #aichallenge 2011-12-15T22:12:23 yeah we talked about this 2011-12-15T22:12:26 in may 2011-12-15T22:12:34 thats why im surprised too 2011-12-15T22:12:47 yeah, so that's why i wondered out loud whether he worked alone 2011-12-15T22:14:38 teams can participate 2011-12-15T22:14:47 there were some who asked 2011-12-15T22:14:53 so there are teams out there 2011-12-15T22:15:04 how do people spread their ants out? 2011-12-15T22:15:16 i think it should be noted on the profile page though 2011-12-15T22:15:30 antimatroid: non-visible locations are used as targets to look for food (not just unseen ones) 2011-12-15T22:16:09 thestinger: but what about like when you have heaps of vision? 2011-12-15T22:16:20 presumably you want to spread your ants out to minimise food collection time 2011-12-15T22:16:40 well, I only send 1 ant to an area for that 2011-12-15T22:16:50 so they spread out and cover the map if possible 2011-12-15T22:16:50 http://aichallenge.org/visualizer.php?game=246776&user=4513 2011-12-15T22:16:54 watch that xathis game 2011-12-15T22:16:58 he maximises visibility 2011-12-15T22:17:22 http://aichallenge.org/visualizer.php?game=245863&user=2255 2011-12-15T22:17:29 my ants usually spread out and cover a lot of the map 2011-12-15T22:17:45 I don't directly control it, it's emergent due to how goals are weighted for finding food 2011-12-15T22:18:05 like, they won't turn around and go back to the hill at the beginning 2011-12-15T22:18:27 because unseen locations are weighted more than tiles they could see last turn, or a few turns ago 2011-12-15T22:18:33 rwest_: in case you are still online & looking for a solution to your programming problem, a friend & me once spend 3 days on it with little sleep, you can do it insanely quickly using graphs & pathfinding. If you want, I can look up the solution for you. Just sent me a private message (I should be on here 24/24, so I'll see it eventually and respond). 2011-12-15T22:18:49 rwest: in case you are still online & looking for a solution to your programming problem, a friend & me once spend 3 days on it with little sleep, you can do it insanely quickly using graphs & pathfinding. If you want, I can look up the solution for you. Just sent me a private message (I should be on here 24/24, so I'll see it eventually and respond). 2011-12-15T22:19:17 spreading out to look for food is one of the few things I have working well :P 2011-12-15T22:20:39 I can beat bots with better combat only because I find more food 2011-12-15T22:22:25 the lambda I pass into my BFS to make the gradient map for finding food (spreading out basically) records the distance to a target, and the location of that target 2011-12-15T22:22:27 *** Kingpin13 has quit IRC (Quit: quit) 2011-12-15T22:22:40 so I know which targets are taken (and mark off the whole area around them as taken too) 2011-12-15T22:22:49 i have the solution right here 2011-12-15T22:23:01 but i cant formalize it for the computer 2011-12-15T22:23:12 its 4:23 2011-12-15T22:24:03 anyway, it just took a lot of trial and error and random tuning to get spreading to work 2011-12-15T22:24:40 I've tried adding on territory control (ants are attracted to the borders I control to fight enemies), but my combat isn't good enough to make it work 2011-12-15T22:24:56 so my foragers just have to hope the enemy doesn't hunt them down atm 2011-12-15T22:25:53 works well enough when I'm not fighting one of the top 5 2011-12-15T22:26:03 *** Antimony has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-12-15T22:26:38 Darhuuk_: you make all the possible lists of 3 and 4.5 that adds up to N, after that you sum of those lists one by one, you get new, lists, with succesive sums, you drop the last element of each, you assign numbers to these lists, you cycle through them, you can get from list A to list B if there are no intersections between the two, you get these little graphs, and you just have to enumerate the 2011-12-15T22:26:40 possible paths 2011-12-15T22:34:16 *** hydrogenesis has joined #aichallenge 2011-12-15T22:37:09 ok, cleaning up V12 to check in 2011-12-15T22:37:32 BenJackson: last check in? 2011-12-15T22:38:54 How do you decide it's significantly better? Looks like it hasn't come up to v11 on Fluxid yet. 2011-12-15T22:39:11 Scryer: fluxid.pl is basically broken atm :P 2011-12-15T22:39:26 games go missing, and bots are using 2-3s a turn 2011-12-15T22:39:35 Ah. 2011-12-15T22:39:38 I don't think there's enough time to properly test anything with the game rate on there now 2011-12-15T22:39:39 antimatroid: I just do a bunch of stuff and then use git add -p to split it up into meaningful blocks 2011-12-15T22:39:45 and I take that opportunity to code review 2011-12-15T22:39:46 usually 2011-12-15T22:42:03 *** Migi32 has quit IRC (Remote host closed the connection) 2011-12-15T22:46:23 "So now there is total of 108 improvements and bug fixes available in Qt Commercial 4.8.0 that are not part of the LGPL release." 2011-12-15T22:46:27 and.... Qt is dead 2011-12-15T22:46:36 http://lists.qt-project.org/pipermail/development/2011-December/000908.html 2011-12-15T22:46:55 I was thinking about learning it but not anymore :P 2011-12-15T22:47:27 meh 2011-12-15T22:47:39 they say they don't intend to keep it that way but it seems like BS to me 2011-12-15T22:47:49 the commercial version just lets you use it without having to deal with the LGPL 2011-12-15T22:48:03 deal with LGPL? 2011-12-15T22:48:08 it's not that hard 2011-12-15T22:48:13 well 2011-12-15T22:48:22 companies pay them to avoid having to distribute the code for Qt basically... 2011-12-15T22:48:28 that's where all the money comes from :P 2011-12-15T22:48:39 you don't have to distribute the code 2011-12-15T22:48:51 you just need to dynamically link it 2011-12-15T22:49:00 and not do any modifications on it 2011-12-15T22:49:01 yeah, corporations are stupid though 2011-12-15T22:49:24 they would probably have to pay lawyers to make sure everything was alright 2011-12-15T22:50:21 maybe their current model isn't earning them any money, and that's why they didn't apply the fixes to the LGPL version 2011-12-15T22:50:37 holy cow 2011-12-15T22:50:45 762M of debug logs from fluxid and tcpants 2011-12-15T22:50:55 snapshots of those are going to clog my disk for months :) 2011-12-15T22:51:15 ZFS? 2011-12-15T22:51:20 yes 2011-12-15T22:51:31 is that why you use FreeBSD? :P 2011-12-15T22:51:41 I've been using freebsd since 1994 2011-12-15T22:52:04 I prefered it over linux due to my SunOS experience 2011-12-15T22:52:10 and other BSD variants 2011-12-15T22:52:52 ok this cell maze kind of sucked, but it was neat to see the visualization showing how my bot knew when it had sealed the edges and ensured there were no enemies inside 2011-12-15T22:53:26 I bet you can get into the top 5 2011-12-15T22:53:59 top 10 is my current goal 2011-12-15T22:54:02 for some reason your bot seems to have bad luck on the official site 2011-12-15T22:54:12 not sure it's luck 2011-12-15T22:54:17 my bot makes some silly mistakes 2011-12-15T22:54:20 V12 fixes a major one 2011-12-15T22:54:32 mcstar: I forgot how we solved it, but I don't think we even needed to calculate sums, the solution was pretty neat :). Anyway, going to sleep now, will read any msgs with my nick in them in the morning. 2011-12-15T22:57:37 tcpclient kept going when I tried to stop it on tcpants 2011-12-15T22:57:44 lost 3 quick timeout games and went to 3rd 2011-12-15T22:58:04 *** dvladim has joined #aichallenge 2011-12-15T22:59:40 oh fluxid isn't adding any games to my page 2011-12-15T22:59:57 BenJackson: yeah :) 2011-12-15T23:00:10 no point in a quick fluxid sanity test then 2011-12-15T23:00:17 we could surprise the hell out of bmh 2011-12-15T23:00:21 it still adds like 1/10 games 2011-12-15T23:00:27 sometimes. 2011-12-15T23:00:40 for _next it stopped adding them completely 2011-12-15T23:01:37 probably if any broken player is in it doesn't add or something 2011-12-15T23:01:47 http://ants.fluxid.pl/player/strcat_next 2 hours of games haven't been added 2011-12-15T23:02:10 they still finish, and I see the score printed :P 2011-12-15T23:03:08 ahh I should capture that I guess 2011-12-15T23:03:15 didn't win 2nd game on tcpants but it's not my bug 2011-12-15T23:03:19 winner just had a weak neighbor 2011-12-15T23:03:22 *** nek has quit IRC (Remote host closed the connection) 2011-12-15T23:03:43 actually I *was* going to win 2011-12-15T23:03:46 750 turn limit? bogus 2011-12-15T23:04:22 ok, resubmit 2011-12-15T23:04:46 *** avdg1 has quit IRC (Quit: Leaving.) 2011-12-15T23:04:52 arrgh run the tests before my queue time 2011-12-15T23:07:10 I guess I'll try tcpants 2011-12-15T23:07:33 I'm alone with weak bots on bhickey 2011-12-15T23:07:36 we could take that over 2011-12-15T23:08:20 running the ver I just reuploaded to the site on there now 2011-12-15T23:08:30 on tcpants? 2011-12-15T23:08:32 me too 2011-12-15T23:08:36 yeah 2011-12-15T23:11:03 my skill is -0 2011-12-15T23:11:11 to balance xathis's mu no doubt 2011-12-15T23:13:46 *** Jak_o_Shadows1 has joined #aichallenge 2011-12-15T23:14:08 wow, someone got to the top 100 with octave 2011-12-15T23:14:08 thestinger: it would be AWESOME if we played in every game on our way to the top 2011-12-15T23:14:14 nice! 2011-12-15T23:14:15 http://aichallenge.org/profile.php?user=11792 2011-12-15T23:14:26 *** treeform has joined #aichallenge 2011-12-15T23:14:35 I may want to win for some language but I'm not that hardcore 2011-12-15T23:15:01 http://aichallenge.org/profile.php?user=13690 lol, pascal 2011-12-15T23:15:35 *** dvladim has quit IRC (Ping timeout: 252 seconds) 2011-12-15T23:15:58 I want to see someone in the top 10 using tcl 2011-12-15T23:16:18 *** Jak_o_Shadows has quit IRC (Ping timeout: 252 seconds) 2011-12-15T23:16:49 *** treeform has quit IRC (Read error: Operation timed out) 2011-12-15T23:17:35 I wrote an approximately infinite amount of code in Tcl…maybe I should have done that 2011-12-15T23:21:18 http://ants.fluxid.pl/replay.36180 I hate these maps 2011-12-15T23:21:29 where you start really close to other hills 2011-12-15T23:22:20 *** ChrisH_ has joined #aichallenge 2011-12-15T23:22:59 I've come to the conclusion that they might be entertaining even if they aren't good for ranking 2011-12-15T23:23:09 ChrisH_: I @latered you as ChrisH (no _) 2011-12-15T23:23:15 something about how lnog it took to hit the top 10 2011-12-15T23:23:25 i'll check the logs 2011-12-15T23:23:32 about 45 games 2011-12-15T23:23:39 then...I resubmitted 2011-12-15T23:24:09 "that it took 48 games to get to #10" 2011-12-15T23:25:31 with randmo games like your last one you're on a similar path 2011-12-15T23:26:29 oh HOW DID I KNOW I'd get a nice 1v1 game right after I submitted 2011-12-15T23:26:34 yeah, at least I got a better matchup, but two of the other bots rolled over too easy for InsaneMalkavian, so he got early points and I came in 2nd, still acceptable. 2011-12-15T23:26:59 but when I get to #20 how about a 10-way circus with bots ranked from #5 to #150 2011-12-15T23:28:37 It seems like there is range of bots in the 100-200 range that are pretty aggressive but not so great at defense and they can really throw the "balance" of a game off. 2011-12-15T23:28:56 seems crazy that I can beat a bot at pos 3200 which is clearly a starter kit and still not outrank even a fraction of the starter kits 2011-12-15T23:29:06 tcpants is so much faster 2011-12-15T23:29:15 ChrisH_: yes, I'm hoping my new version is better against those 2011-12-15T23:29:38 ChisH_: http://tcpants.com/player/SkyCaptain my bot :) 2011-12-15T23:29:56 I think I'd rather take a few 2nd places in those games on my way past that group than try too hard to win them outright. 2011-12-15T23:30:14 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-15T23:31:15 ChrisH_: I just mean my defense has improved 2011-12-15T23:31:25 V11 was getting by making some big defensive mistakes 2011-12-15T23:31:34 if it happens on the fringe it gets to try again 2011-12-15T23:31:41 not if it happens close to home 2011-12-15T23:32:11 antimatroid: nice, I'll have to run a few games on tcpants 2011-12-15T23:32:33 BenJackson: oh, ok 2011-12-15T23:32:35 antimatroid: a pleasure stalemating with you: http://tcpants.com/replay.1439 2011-12-15T23:33:23 BenJackson: you keep telling yourself that :P with more time i would have had you 2011-12-15T23:34:32 there's a game where symmetry might have helped 2011-12-15T23:34:37 neither of us saw the other player's hill 2011-12-15T23:34:42 I just fired up a single game on tcpants to get a foot in the door, but I have some code to work on tonight, so I can't run too many. 2011-12-15T23:35:13 yeah i'm still working on my bot 2011-12-15T23:35:17 I've been running tons of bots without much regard for the CPU 2011-12-15T23:35:22 i want to get better samples of moves for battles 2011-12-15T23:35:22 on a quad core 2011-12-15T23:35:25 seems like a slow match though, the network light isn't blinking too fast 2011-12-15T23:35:33 yeah my bot is quite fast 2011-12-15T23:35:45 usually takes less than 100ms to pick all its moves, even with a couple of hundred ants 2011-12-15T23:36:04 i friggin timed out on tcpants too 2011-12-15T23:36:11 I have to do my development inside a VM, so I can't overload it too much even though the hardware is a Quad. 2011-12-15T23:36:19 This looks like a word problem. "If there are 164 players ahead of you, and the rate is 31 players per minute, how many minutes before your turn?" Answer: 29. 2011-12-15T23:36:28 one small way I can tell V12 is better is that V11 would mostly win or totally lose (0 point) whereas V12 is better at getting a few points and scoring in the middle in adverse circumstances 2011-12-15T23:36:29 wait no i didn't 2011-12-15T23:36:35 opened wrong terminal window aha 2011-12-15T23:37:19 BenJackson: if i can draw with you i'm going to extrapolate and say I'm top 50 at least 2011-12-15T23:37:29 assuming my bot doesn't have major crash/timeout issues 2011-12-15T23:38:11 probably true 2011-12-15T23:38:13 thestinger beat me :( 2011-12-15T23:38:54 I just got pwned on a similar maze map 2011-12-15T23:39:03 maybe i need to care more about enemy hills? 2011-12-15T23:39:09 i had the most ants at the end :( 2011-12-15T23:40:44 So I've been working on multi-hill strategy the last few days and it has turned out rather well. Watch my bot use spawn control ants nicely in this replay: http://ants.fluxid.pl/replay.35482 2011-12-15T23:41:38 which one is you? 2011-12-15T23:41:47 Hint: I win. 2011-12-15T23:42:56 And I don't lose any hills. 2011-12-15T23:43:47 it's pretty cool to watch 2011-12-15T23:43:56 http://tcpants.com/replay.1459 2011-12-15T23:44:00 I barely noticed the bug here 2011-12-15T23:44:17 as in the hunt for red october, my ant closed the distance BEFORE IT COULD ARM ITSELF 2011-12-15T23:44:34 at 8 ants I wasn't putting enemy hills into my attack function 2011-12-15T23:44:37 but I did WALK RIGHT BY 2011-12-15T23:44:39 yeah, what's with not stepping on that hill 2011-12-15T23:45:03 there was no specific "step on hill" rule in the explore code 2011-12-15T23:45:06 you must have a higher priority target for that ant for some reason 2011-12-15T23:45:10 and it wasn't a general goal because I had so few ants 2011-12-15T23:45:13 *** Gnome has quit IRC (Quit: Gnome) 2011-12-15T23:45:19 oh 2011-12-15T23:45:25 there's a timidity function 2011-12-15T23:45:27 *** Gnome has joined #aichallenge 2011-12-15T23:45:34 with 8 ants it will only attack hills if NOTHING is around 2011-12-15T23:45:36 it's not very bright 2011-12-15T23:45:44 what approach do you guys take for going after hills? 2011-12-15T23:45:51 that's an illegal map anyway 2011-12-15T23:45:53 One thing I like about my latest versions is that all the pathing is done for all targets at once, so any ant not in combat can go to any target without stuff like that happening 2011-12-15T23:45:57 atm i'm just zorg rushing them along with enemies after other stuff 2011-12-15T23:46:20 ChrisH_: mine too. that hill is just not a target (yet) 2011-12-15T23:46:40 my code has actually gotten simpler in many ways as I've made better use of the tools I've built. 2011-12-15T23:46:58 hills are always targets if my bot has seen them 2011-12-15T23:47:32 on the tcp servers sometimes I watch the wrong color 2011-12-15T23:47:36 my "explore code" is just configuring explore targets 2011-12-15T23:47:36 and wonder wtf my ants are doing 2011-12-15T23:47:44 lol, i've done that 2011-12-15T23:47:58 http://tcpants.com/replay.1461 :D 2011-12-15T23:48:06 or I'll watch someone elses game and forget it isn't my bot controlling the white ants 2011-12-15T23:48:15 ohh I'm wrong 2011-12-15T23:48:19 that hill was on the attack list 2011-12-15T23:48:32 that was the result of my new graduated explore bonus at start 2011-12-15T23:48:37 it did what I wanted: it expanded fast 2011-12-15T23:48:44 uh oh, someone just got beat by skycaptain. ;) 2011-12-15T23:50:28 no matter what happens I'm pretty damn pleased with games like this: http://tcpants.com/replay.1449 2011-12-15T23:51:47 my first game on tcpants: http://tcpants.com/replay.1457 2011-12-15T23:52:02 "Ants not razing hills" jeez, what more could I do. 2011-12-15T23:52:39 you couldn't have won that anyway 2011-12-15T23:52:42 er that last hill 2011-12-15T23:52:47 turn limit is 750 2011-12-15T23:53:04 maybe 2011-12-15T23:53:18 ChrisH: who lost to skycaptain? 2011-12-15T23:53:27 everyone in a 10 player game 2011-12-15T23:53:39 i thought he meant him or something 2011-12-15T23:53:47 i only just won 2011-12-15T23:53:48 yeah, the game you linked, which included BenJackson 2011-12-15T23:54:02 yep, I didn't break out initially 2011-12-15T23:54:10 xathis would ahve gotten out of that 2011-12-15T23:54:46 now i'm hoping for top 10 aha 2011-12-15T23:54:54 i'm no xathis 2011-12-15T23:55:26 antimatroid: you're bot is looking pretty good from what I've seen. 2011-12-15T23:55:37 It will be fun to see how you do. 2011-12-15T23:55:44 it's still pretty stupid with battle sometimes 2011-12-15T23:55:58 like moves away from the enemy when it could be more aggressive 2011-12-15T23:55:59 mine is really quite stupid with battle 2011-12-15T23:56:09 I could probably put in the "simple combat" from the forum and do better 2011-12-15T23:56:10 or makes stupid trades 2011-12-15T23:56:13 but I'm sticking with what brung me 2011-12-15T23:56:17 News flash: My bot just beat xathis on the main servers: http://aichallenge.org/visualizer.php?game=247684&user=10 2011-12-15T23:56:22 woot! 2011-12-15T23:56:24 winnar 2011-12-15T23:56:38 *** xar0l has joined #aichallenge 2011-12-15T23:57:26 ChrisH_: any ideas on how much spawn control improves the quality of your bot on multi hill maps? 2011-12-15T23:57:28 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-15T23:57:31 *** cyphase has quit IRC (Quit: http://www.cyphase.com/) 2011-12-15T23:57:32 dat hill control 2011-12-15T23:57:47 antimatroid: just follow the link 2011-12-15T23:57:57 yes i watched 2011-12-15T23:58:14 but from other testing he's done, it may or may not have shown significant improvement 2011-12-15T23:58:31 it's actually easier than I thought 2011-12-15T23:58:39 you know if you're eating 2011-12-15T23:58:44 so you don't have to suffer any lag or anything 2011-12-15T23:59:07 antimatroid: I'm still not sure how much it helps, I think combat is still king, but saving steps by putting ants closer to their targets at spawn time has to help. 2011-12-15T23:59:27 with the penalty of the ants on the hills initially 2011-12-15T23:59:39 it depends on the map, but I will take my victory over those bots on that map as strong evidence that it helps 2011-12-15T23:59:42 the trick would seem to be knowing when to cap them 2011-12-15T23:59:52 yeah 2011-12-15T23:59:55 yeah, i did a lot of experimentation with that part 2011-12-15T23:59:55 because early is when it's going to help most