2010-10-20T00:10:36 Top 10 players: dmj111(3697), cfaftw(3654), rsergio(3623), deepblue(3596), savchenko(3576), chotting(3575), EBraun(3542), hutabhu(3525), DocBot(3507), Polsky(3504) 2010-10-20T00:14:26 evening 2010-10-20T00:21:51 interesting game to watch http://72.44.46.68/canvas?game_id=580706 2010-10-20T00:42:51 *** codefun has joined #aichallenge 2010-10-20T00:43:07 I come bearing questions... 2010-10-20T00:43:14 *** Utkarsh has quit IRC (Ping timeout: 272 seconds) 2010-10-20T00:43:38 ask away 2010-10-20T00:44:03 Can anyone explain or suggest as to why my bot deletes it's entire empire (turns it's planets into neutral and all fleets in flight dissapear) after a certain amount of turns? It's usually 14 turns, then POOF I turn into neutral and the enemy wins. 2010-10-20T00:44:14 I can't find any errors in my code, I'm using perl 2010-10-20T00:45:42 I think I just fixed it.... nevermind XD 2010-10-20T00:45:57 ./failmodeoff 2010-10-20T00:47:21 *** Itkovian_ has joined #aichallenge 2010-10-20T00:49:19 *** Utkarsh has joined #aichallenge 2010-10-20T00:50:42 *** Itkovian has quit IRC (Ping timeout: 250 seconds) 2010-10-20T00:50:44 *** Itkovian_ is now known as Itkovian 2010-10-20T00:51:37 *** rebelxt_ has joined #aichallenge 2010-10-20T00:52:57 Agh no the problem wasn't solved... it just chose a larger number of turns until crash. 2010-10-20T00:53:07 Why do the coding gods hate me so! 2010-10-20T00:53:54 codefun: how the heck does a bot sacrifice itself anyways 2010-10-20T00:54:08 I have no idea 2010-10-20T00:54:18 It's an odd glitch. 2010-10-20T00:54:35 It's happened before but I knew the cause; it tried sending ships to a planet that didn't exist 2010-10-20T00:54:52 And that caused some error that wiped me out. This time that's not the issue though... 2010-10-20T00:56:10 *** rebelxt_ has quit IRC (Client Quit) 2010-10-20T00:58:01 OK now I found the issue.... 2010-10-20T00:58:26 I think... -_- 2010-10-20T00:59:03 *** Epona666 has quit IRC (Read error: Connection reset by peer) 2010-10-20T00:59:45 Good golly miss molly, firefox is using over 115 megabytes of ram; gonna close it down... it's been eating all my memory lately. Wtf? 2010-10-20T01:00:28 *** tty3 has joined #aichallenge 2010-10-20T01:00:29 *** krokkrok has quit IRC (Quit: Page closed) 2010-10-20T01:01:35 *** dmj111` has joined #aichallenge 2010-10-20T01:02:34 *** dmj111 has quit IRC (Ping timeout: 252 seconds) 2010-10-20T01:02:39 *** Epona666 has joined #aichallenge 2010-10-20T01:03:07 *** tty2 has quit IRC (Ping timeout: 252 seconds) 2010-10-20T01:04:16 *** codefun has quit IRC (Ping timeout: 265 seconds) 2010-10-20T01:08:07 *** hornairs has quit IRC (Remote host closed the connection) 2010-10-20T01:09:20 *** Azrathud has quit IRC (Ping timeout: 250 seconds) 2010-10-20T01:10:39 Top 10 players: dmj111(3671), cfaftw(3625), rsergio(3594), deepblue(3565), savchenko(3548), chotting(3546), EBraun(3513), hutabhu(3498), asavis(3486), Polsky(3479) 2010-10-20T01:12:32 *** Arthur_ has quit IRC (Ping timeout: 276 seconds) 2010-10-20T01:15:13 *** Arthur has joined #aichallenge 2010-10-20T01:16:03 *** JensT1 has joined #aichallenge 2010-10-20T01:16:06 *** JensT1 has left #aichallenge 2010-10-20T01:24:24 *** noid05826 has joined #aichallenge 2010-10-20T01:24:37 *** Mathnerd314 has quit IRC (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.8/20100722155716]) 2010-10-20T01:24:45 *** Accoun has quit IRC (Ping timeout: 265 seconds) 2010-10-20T01:26:06 *** Epona_ has joined #aichallenge 2010-10-20T01:29:02 *** Epona666 has quit IRC (Ping timeout: 255 seconds) 2010-10-20T01:29:02 *** iFire has quit IRC (Ping timeout: 255 seconds) 2010-10-20T01:33:06 *** JensT2 has joined #aichallenge 2010-10-20T01:35:23 *** Epona_ has quit IRC (Ping timeout: 265 seconds) 2010-10-20T01:35:28 *** iFire has joined #aichallenge 2010-10-20T01:48:43 *** wvdschel has joined #aichallenge 2010-10-20T02:10:38 Top 10 players: dmj111(3580), cfaftw(3565), rsergio(3525), deepblue(3488), EBraun(3484), Polsky(3472), savchenko(3464), asavis(3435), DocBot(3416), Rexxar(3413) 2010-10-20T02:17:28 *** hellman has left #aichallenge ("buffer overflow") 2010-10-20T02:30:42 *** sigh has quit IRC () 2010-10-20T02:31:13 *** perestrelka has joined #aichallenge 2010-10-20T02:33:55 *** Palmik has joined #aichallenge 2010-10-20T02:37:24 *** antimatroid1 has joined #aichallenge 2010-10-20T02:37:50 *** antimatroid has quit IRC (Ping timeout: 264 seconds) 2010-10-20T02:47:54 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-10-20T02:50:51 *** Appleman1234 has joined #aichallenge 2010-10-20T02:58:07 *** antimatroid has joined #aichallenge 2010-10-20T02:58:08 *** antimatroid1 has quit IRC (Read error: Connection reset by peer) 2010-10-20T03:03:48 *** vizier has quit IRC (Quit: Page closed) 2010-10-20T03:05:58 *** aerique has joined #aichallenge 2010-10-20T03:08:49 *** tapwater has quit IRC (Quit: tapwater) 2010-10-20T03:10:39 Top 10 players: dmj111(3490), cfaftw(3471), rsergio(3431), EBraun(3413), deepblue(3397), asavis(3385), Polsky(3383), savchenko(3383), Rexxar(3369), DracoAter(3364) 2010-10-20T03:11:20 *** mega1 has joined #aichallenge 2010-10-20T03:11:42 *** Itkovian has quit IRC (Quit: Itkovian) 2010-10-20T03:18:28 *** jaspervdj has joined #aichallenge 2010-10-20T03:18:59 *** TioLuiso has joined #aichallenge 2010-10-20T03:19:20 Hello people, good morning from spain 2010-10-20T03:19:41 I'm having some problem 2010-10-20T03:19:52 my bot si losing streaks of games 2010-10-20T03:20:08 since a few days ago 2010-10-20T03:20:12 in 1st turn 2010-10-20T03:20:26 I've looked in forums 2010-10-20T03:20:43 and looks like I'm not the only one having that problem 2010-10-20T03:21:11 I'm quite confident that it's not a bug in my code 2010-10-20T03:21:39 and I don't get the message that i have timed out or had an esception 2010-10-20T03:21:42 exceptin 2010-10-20T03:21:47 simply, y lose 2010-10-20T03:22:00 Is there any way I can get a log? 2010-10-20T03:23:38 *** smloh1 has joined #aichallenge 2010-10-20T03:24:28 no, there isn't 2010-10-20T03:28:10 the thing is I haven't updated my bot in some time 2010-10-20T03:28:26 and it didn't lose that many times before 2010-10-20T03:28:38 so I suspect there must be some problem in the server 2010-10-20T03:28:43 outside my bot 2010-10-20T03:28:51 however, I'll check my bot again 2010-10-20T03:28:54 thanks anyway 2010-10-20T03:28:59 I used to forfeit some games. 2010-10-20T03:29:08 Now I forfeit all. 2010-10-20T03:29:27 how's that? 2010-10-20T03:29:34 In retrospect it must have been better because the other servers are stronger. 2010-10-20T03:29:47 But now there is only a single, puny server. 2010-10-20T03:30:22 do you think it has any connection? 2010-10-20T03:30:45 that lack of resources can lead to forfeit games? 2010-10-20T03:32:49 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-20T03:38:59 possibly, I'm testing things by upload dummy and dummier players and checking what they do. 2010-10-20T03:39:04 *uploading 2010-10-20T03:39:26 not a very pleasant way of debugging 2010-10-20T03:42:15 *** Amanieu has quit IRC (Remote host closed the connection) 2010-10-20T03:42:25 *** Amanieu has joined #aichallenge 2010-10-20T03:47:41 *** narnach1 has joined #aichallenge 2010-10-20T03:52:37 *** narnach1 is now known as narnach_mb 2010-10-20T04:03:13 *** smloh1 has quit IRC (Quit: Page closed) 2010-10-20T04:03:48 *** boegel has joined #aichallenge 2010-10-20T04:09:54 *** Olathe has quit IRC (Ping timeout: 240 seconds) 2010-10-20T04:10:39 Top 10 players: cfaftw(3436), dmj111(3418), rsergio(3396), Polsky(3338), DracoAter(3331), EBraun(3331), deepblue(3311), DocBot(3292), smatcher(3288), Rexxar(3286) 2010-10-20T04:12:51 *** Alexander has joined #aichallenge 2010-10-20T04:25:58 *** gerald5122k has joined #aichallenge 2010-10-20T04:28:54 *** p4p4 has joined #aichallenge 2010-10-20T04:31:03 *** smellyhippy has joined #aichallenge 2010-10-20T04:31:50 *** phreeza has joined #aichallenge 2010-10-20T04:32:44 *** sigh has joined #aichallenge 2010-10-20T04:34:44 morning! lost my first game on the tcp server, but can't make much out of the konsole output, any description somewhere ? 2010-10-20T04:38:21 *** Florian95 has joined #aichallenge 2010-10-20T04:43:25 *** Naktibalda has joined #aichallenge 2010-10-20T04:44:41 *** Titankiller has joined #aichallenge 2010-10-20T04:44:55 ok, found out myself, just go http://72.44.46.68/canvas?game_id=584860 2010-10-20T04:49:30 *** Alexander has quit IRC (Ping timeout: 265 seconds) 2010-10-20T04:52:49 *** delt0r has joined #aichallenge 2010-10-20T04:56:20 *** wvdschel has quit IRC (Changing host) 2010-10-20T04:56:21 *** wvdschel has joined #aichallenge 2010-10-20T04:57:12 *** Lithosphere has quit IRC (Changing host) 2010-10-20T04:57:12 *** Lithosphere has joined #aichallenge 2010-10-20T04:58:08 *** mietek_ is now known as mietek 2010-10-20T05:10:42 Top 10 players: cfaftw(3502), dmj111(3389), rsergio(3371), smatcher(3341), DracoAter(3327), EBraun(3325), Polsky(3310), deepblue(3297), savchenko(3262), asavis(3257) 2010-10-20T05:11:00 *** Titankiller has quit IRC (Read error: Connection reset by peer) 2010-10-20T05:11:26 *** jesionaj has quit IRC (Ping timeout: 264 seconds) 2010-10-20T05:19:31 *** Olathe has joined #aichallenge 2010-10-20T05:25:35 so some guy i beat that was in the top 200 resubmitted his code 2010-10-20T05:25:53 now my rank dropped too 2010-10-20T05:26:19 when will the cloud be up again 2010-10-20T05:27:05 *** yasith has quit IRC (Ping timeout: 265 seconds) 2010-10-20T05:30:25 *** gerald5122k has quit IRC (Ping timeout: 240 seconds) 2010-10-20T05:35:13 probably nobody knows 2010-10-20T05:35:45 *** hellman has joined #aichallenge 2010-10-20T05:39:00 *** yasith has joined #aichallenge 2010-10-20T05:57:11 *** gerald5122k has joined #aichallenge 2010-10-20T05:59:14 *** smellyhippy has quit IRC (Ping timeout: 250 seconds) 2010-10-20T05:59:44 *** krokkrok has joined #aichallenge 2010-10-20T06:04:32 still testing why I forfeit games. Replaced all the logic in my code with a loop that allocates 15M garbage per turn and then just loops until 0.8s have passed. 2010-10-20T06:04:51 and it still manages to die: http://ai-contest.com/visualizer.php?game_id=5965366 2010-10-20T06:05:13 *** Palmik has joined #aichallenge 2010-10-20T06:05:42 mega1: :( i feel your pain 2010-10-20T06:05:52 i had that issue, it was incredibly frustrating 2010-10-20T06:06:05 worked fine locally 2010-10-20T06:06:06 but you use less than 0.1s now? 2010-10-20T06:06:18 i am using as much as i tell it to now 2010-10-20T06:06:19 but i was there 2010-10-20T06:06:21 then* 2010-10-20T06:06:37 language? 2010-10-20T06:07:07 c++ 2010-10-20T06:07:23 no gc then 2010-10-20T06:07:56 I suspect severe memory pressure 2010-10-20T06:08:13 uploaded a bot that does not allocate garbage just spins for 0.8s 2010-10-20T06:10:40 Top 10 players: cfaftw(3510), dmj111(3404), rsergio(3382), Polsky(3380), smatcher(3345), DracoAter(3338), EBraun(3321), savchenko(3303), deepblue(3303), asavis(3274) 2010-10-20T06:11:24 mega1: what lang? 2010-10-20T06:11:31 lisp 2010-10-20T06:11:50 is it quite a old GC? 2010-10-20T06:12:04 it has a good gc 2010-10-20T06:12:18 there has been a *lot* of R&D on gc and its bloody fast in java 2010-10-20T06:12:26 15M won't even show up... 2010-10-20T06:12:27 I know, java is ahead. 2010-10-20T06:12:36 neither in sbcl 2010-10-20T06:12:54 but if there is no physical memory ... 2010-10-20T06:13:00 is it a copy gc..or mark and sweep? 2010-10-20T06:13:09 copy 2010-10-20T06:13:52 well to me it sounds like something else is wrong? I mean on anything but crap, is pretty easy to *not* copy 15M 2010-10-20T06:14:07 but i haven't done lisp/scheme for a while 2010-10-20T06:14:24 well i have, but within the jvm aka kava 2010-10-20T06:14:25 I don't understand. 2010-10-20T06:14:34 easy to not copy? 2010-10-20T06:14:54 well a copy gc just copys live object to the new location... 2010-10-20T06:15:03 and? 2010-10-20T06:15:16 its performance is not dependant on the amount of garbarge 2010-10-20T06:15:21 mega1: are you timing out on the official server? sorry didn't read all the history 2010-10-20T06:15:26 other than more collections are needed 2010-10-20T06:15:35 are we trying to optimize the gc here? 2010-10-20T06:15:44 phreeza: yes 2010-10-20T06:16:08 I thought you said you are timing out without doing anything but producing 15M of garbarge 2010-10-20T06:16:18 yes 2010-10-20T06:16:34 ah and the theory is its GC... so i guess youre not using c++ 2010-10-20T06:16:52 *** asdd has joined #aichallenge 2010-10-20T06:16:59 I'm using lisp (sbcl) and we are discussing gc with delt0r. 2010-10-20T06:17:01 and it doesn't sound like that should be the problem.... unless you are allocating 15M every ms 2010-10-20T06:17:09 right 2010-10-20T06:17:24 unless the memory pressure on the system is high 2010-10-20T06:17:35 and gc is slowed down by swapping 2010-10-20T06:17:44 *** hellman has quit IRC (Remote host closed the connection) 2010-10-20T06:17:52 well even no gc is slowed down by swapping 2010-10-20T06:17:58 *** seisatsu has joined #aichallenge 2010-10-20T06:18:11 a copying gc touches more pages 2010-10-20T06:18:14 can't gc be suspended? 2010-10-20T06:18:19 mega1: true 2010-10-20T06:18:36 Has anyone here ever touched on or heard of Non-stationary Entropic Reduction Mapping? 2010-10-20T06:18:51 it can be suspended and I can optimize my code too, but if it's suspended it will need even more memory the next time 2010-10-20T06:19:00 swaping imo is dumb. its just too slow to be usefull on modern machines... it should be honest and just give a out of mem error 2010-10-20T06:19:07 one idea is to invoke it explicitly often 2010-10-20T06:19:19 It was a technology invented in 1996 and then abandoned. I became interested, and thought I would ask here. 2010-10-20T06:19:28 It's an AI method, of course. 2010-10-20T06:19:40 seisatsu: no but it sounds interesting 2010-10-20T06:19:55 It had the ability to be dropped into any system that it had no knowlege of, and learn on its own. 2010-10-20T06:20:21 By simply telling it when it made a mistake, it would reorganize its own internal workings to avoid it. 2010-10-20T06:20:33 But perhaps it was replaced by neural nets. 2010-10-20T06:20:33 then it probably sucked otherwise it wouldn't have been discarded? 2010-10-20T06:20:46 Mm, it was actually made for an interesting game. 2010-10-20T06:20:53 neural nets were around way before 1996 2010-10-20T06:20:53 The game development team invented it. 2010-10-20T06:21:04 I know, but you hear a lot about them now. 2010-10-20T06:21:25 But perhaps that's just my own view since I personally hear a lot more about them now. 2010-10-20T06:22:40 I'm looking into Case Based reasoning, have you heard about that? 2010-10-20T06:22:44 The game was praised as interesting. Basically, there is a robot that moves on its own. The only interaction allowed is prodding it on its right or left side to change direction. You can also manipulate the environment itself. You try to keep it from dying. 2010-10-20T06:22:50 From this it learns. 2010-10-20T06:22:54 Hmm, I have not. 2010-10-20T06:23:23 mm 2010-10-20T06:23:45 Cascading decisions upon previous decisions, basically? 2010-10-20T06:24:35 *** Olathe has quit IRC (Ping timeout: 240 seconds) 2010-10-20T06:24:39 Am I about right? 2010-10-20T06:27:47 depends on the implementation, some are cascading, some are just a single layer 2010-10-20T06:27:54 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-10-20T06:28:00 the basic idea is having a library of previous cases 2010-10-20T06:28:24 and matching the current situation to those known cases by some metric 2010-10-20T06:28:31 *** Olathe has joined #aichallenge 2010-10-20T06:28:46 the metric is tricky of course, how do you determine whether 2 scenarios are similar... 2010-10-20T06:29:03 so you probably need pattern matching 2010-10-20T06:29:32 try to make some kind of definitions that splits the game up into equivalence classes? 2010-10-20T06:29:34 *** TioLuiso has quit IRC (Ping timeout: 265 seconds) 2010-10-20T06:29:49 then look up tables or whatever? 2010-10-20T06:30:34 yea exactly 2010-10-20T06:31:07 and its neat because in theory you could look at replays and say, hey, this should have had a different reaction 2010-10-20T06:31:12 and add that to the library 2010-10-20T06:31:22 interesting concept, then you could try and train for different equivalence classes 2010-10-20T06:31:42 you could even try to "solve" some situations and include that 2010-10-20T06:31:50 or you can datamine replays to teach it 2010-10-20T06:33:07 its probably hard to really solve stuff in that level because these classes are abstracted. 2010-10-20T06:33:13 i'm doing a tree based approach where i make one move at a time, make a tree, then "play" each terminal node and then pick the best stream of moves at the end 2010-10-20T06:34:22 *** krokkrok has quit IRC (Quit: Page closed) 2010-10-20T06:34:23 i would be doing classes based off of structure of the geometry somehow, like do ones for situations of isosceles triangles, then move to general triangles 2010-10-20T06:34:24 yea i'm trying something similar. modelling the opponent is not working well for me though 2010-10-20T06:34:28 then start adding in more planets etc. 2010-10-20T06:34:36 maybe try partitioning the graph or something 2010-10-20T06:35:38 yea interesting idea... i was thinking of some kind of similarity metrics on the distance matrices... 2010-10-20T06:35:39 *** TioLuiso has joined #aichallenge 2010-10-20T06:37:01 try to make them isomorphic for some characteristic? 2010-10-20T06:37:29 yes 2010-10-20T06:38:52 *** Olathe has quit IRC (Ping timeout: 276 seconds) 2010-10-20T06:40:26 *** Rommel has joined #aichallenge 2010-10-20T06:40:35 g'day all :) 2010-10-20T06:41:28 Anyone here? 2010-10-20T06:48:05 I have actually just sent a formal request to Anark Corporation for details on their 1996 NERM AI technology. :D 2010-10-20T06:48:30 They haven't so much as touched it in 14 years, and neither has anyone else. 2010-10-20T06:48:55 ha, good luck with that 2010-10-20T06:48:58 Haha yeah 2010-10-20T06:49:17 you will get an email in like 3 years from someone saying wtf 2010-10-20T06:49:21 haha 2010-10-20T06:49:40 With NERM, an AI can be inserted into any system, and begin adapting to its environment and requirements automatically through trial and error with minimal outside interference. 2010-10-20T06:50:11 All you basically do is tell it when it's made a mistake. 2010-10-20T06:50:26 And give it a gentle nudge in the correct direction. 2010-10-20T06:50:43 It actually then reorganizes its internal structure to adapt 2010-10-20T06:50:57 It sounds very complicated. 2010-10-20T06:51:32 An article on it said that NERM is more complex of a system than any environment it might encounter. 2010-10-20T06:51:36 seisatsu: Too it sounds more like marketing tripe.... sorry 2010-10-20T06:51:39 http://www.domesro.com/2009/07/once-short-lived-robotic-intelligence.html 2010-10-20T06:51:46 nn 2010-10-20T06:52:00 *here's a 2009 article looking at it 2010-10-20T06:52:09 it was used in a single video game, once 2010-10-20T06:52:13 that was all. 2010-10-20T06:52:17 I'm going mad with trying to get the starter package for C++ ball rolling, I ran the following... and using just the default starter package, and MinGW, it just says that player 2 crashed :(... http://www.pastie.org/1235129 2010-10-20T06:52:20 It worked extraordinarily well. 2010-10-20T06:52:51 The only way you could interact directly with it was by touching the robot on either side; a nudge. 2010-10-20T06:52:59 It learned very quickly. 2010-10-20T06:53:05 I played that game once. 2010-10-20T06:53:13 yea i saw that... however there are no details at all. There is nothing published. Just wild assertions that it works... and the game got pretty bad reviews... 2010-10-20T06:53:32 Still, I will enquire. 2010-10-20T06:54:14 When something gets my interest I will move heaven and earth to grasp it. Until I get bored a few days later. 2010-10-20T06:54:24 *inquire 2010-10-20T06:54:27 lol--don't we all 2010-10-20T06:54:35 Any ideas anyone? 2010-10-20T06:54:39 i think there's needs to be an ai comp for a full on 3d game sooner or later 2010-10-20T06:54:46 there*** 2010-10-20T06:54:47 mm 2010-10-20T06:55:14 Neural nets now could probably do something similar to this "alleged" NERM, though. 2010-10-20T06:55:16 antimatroid: you mean 3d as in spacially... ie galcon 3d ? 2010-10-20T06:55:24 that would be more like the starcraft comp, i'm happy with these simple games 2010-10-20T06:55:25 Can anyone hear me? lol 2010-10-20T06:55:33 Rommel, we can 2010-10-20T06:55:34 Rommel: yes 2010-10-20T06:55:39 don't know 2010-10-20T06:55:39 nah, like first person shooter 2010-10-20T06:55:40 But nobody cares. :< 2010-10-20T06:55:51 right 2010-10-20T06:55:53 If it just crashes right away 2010-10-20T06:55:54 or 3d fighter plane fights or something 2010-10-20T06:56:04 it's either your fault for something you added 2010-10-20T06:56:09 antimatroid: that could be interesting 2010-10-20T06:56:14 i don't even play computer games (excepting flash and minesweeper) but writing a bot, sign me up 2010-10-20T06:56:17 in that the game and rules can be simple 2010-10-20T06:56:18 or the original writer's fault for not discovering a bug. 2010-10-20T06:56:26 lol cheers seisa... I had just finished my C++ backprop... was looking forward to giving it a shot... but if the starter package won't run lol 2010-10-20T06:56:27 the complexity is just too high... 2010-10-20T06:56:30 3d fighting would be solved by large numerical pde solutions 2010-10-20T06:56:43 antimatroid: not easily 2010-10-20T06:56:44 hello differential game theory 2010-10-20T06:56:56 cough multiplayer optimal control theory cough 2010-10-20T06:56:57 heh like these chasing games? 2010-10-20T06:56:59 if the turning is non holo... whatever 2010-10-20T06:57:11 then it gets real tricky real quick 2010-10-20T06:57:17 phreeza: i haven't thought about looking at planetwars as a chasing game 2010-10-20T06:57:22 i wonder if that would be useful 2010-10-20T06:57:40 you are sort of the pursuer and the evader in a way 2010-10-20T06:58:08 It looks like Anark Corporation hich designed NERM and the game, is exclusively a 3D technology company now. 2010-10-20T06:58:08 so the simple rules could be acceleration, turning drag, turning radius for speed... and "kill" condtions. 2010-10-20T06:58:14 They don't do games anymore. 2010-10-20T06:58:19 *which 2010-10-20T06:58:32 they replied already? 2010-10-20T06:58:33 Or AI. 2010-10-20T06:58:34 No 2010-10-20T06:58:38 delt0r: yep, i expect something like if they are in a certain distance and field of vision, you lose 2010-10-20T06:58:39 Aight I figured it out :P 2010-10-20T06:58:39 But I looked at the website 2010-10-20T06:58:46 ah 2010-10-20T06:58:49 you could start from various different positions 2010-10-20T06:58:55 that'd actually be a really awesome comp 2010-10-20T06:58:59 required libstdc++-6.dll :S 2010-10-20T06:58:59 well if you could track down the actual people involved 2010-10-20T06:59:04 They cannot have any possible benefit from NERM anymore 2010-10-20T06:59:06 ah, yes 2010-10-20T06:59:10 that would be fantastic 2010-10-20T06:59:13 the game visulizer would not be hard to do either 2010-10-20T06:59:23 amstan: i have the next game for you 2010-10-20T06:59:28 what was the game name? 2010-10-20T06:59:39 or the game engine 2010-10-20T06:59:59 Galapagos: Mendel's Escape 2010-10-20T07:00:01 >.> 2010-10-20T07:00:03 they need someone to make a kickass visualiser for such a game 2010-10-20T07:00:06 It was interesting 2010-10-20T07:00:12 game engine would be tricky if you want to simulate conrinuous time 2010-10-20T07:00:15 and space 2010-10-20T07:00:16 but the robot moved so damn slow that everyone got bored 2010-10-20T07:00:27 phreeza: you don't --just close to it 2010-10-20T07:00:31 That's probably why it rated poorly 2010-10-20T07:00:40 phreeza: yeah, you could define distance of "forward movement" each turn, and then players could set rotation 2010-10-20T07:00:42 lol what a name... 2010-10-20T07:00:45 And backtracking was a nightmare 2010-10-20T07:00:46 and that could be limited based on speed 2010-10-20T07:00:49 and a time consumer. 2010-10-20T07:00:57 discretise the problem? 2010-10-20T07:01:10 If I could get ahold of the original development team's documents 2010-10-20T07:01:20 I had a air to air combat board game once 2010-10-20T07:01:23 (assuming they kept documents, they're software developers :P) 2010-10-20T07:01:24 or remove time limits and allow players to make moves based on time? 2010-10-20T07:01:33 and they can change their move whenever? 2010-10-20T07:01:34 That would be fantastic. 2010-10-20T07:01:36 and keep it continuous 2010-10-20T07:01:51 brb 2010-10-20T07:01:56 *** TioLuiso has quit IRC (Ping timeout: 265 seconds) 2010-10-20T07:02:19 antimatroid: if its "true" continuous I would think doing something like the simulation football would be a good idea. 2010-10-20T07:02:35 ie tcp/ip messages 2010-10-20T07:02:49 time jitter is a problem for the bot... 2010-10-20T07:03:04 but a strict trun based system could be done as well 2010-10-20T07:03:13 *** HakanD has joined #aichallenge 2010-10-20T07:03:45 though many would find it hard. Calculus always trips people up... eps 1st year comp sci folk ;) 2010-10-20T07:03:59 i kind of like the idea of no concept of discrete moves, then players can just set rotation for a given time 2010-10-20T07:04:09 and it's up to them to estate where they will be in space 2010-10-20T07:04:23 i think 2d would be too easy to solve, so one would have to do 3d anyway 2010-10-20T07:04:55 antimatroid: simulation gets tricky however... there really still are "turns" --its just they are too short compared to "response" time 2010-10-20T07:05:26 how would you differentiate between chaser and chasee? :D 2010-10-20T07:05:32 antimatroid: 2d is not so easy... my planes in the RTS i am doing have all this... getting a "ai" to follow a path for optimal time is hard 2010-10-20T07:05:47 phreeza: you don't --its a dog fight 2010-10-20T07:05:48 neither, first to get their field of vision on the oponent wins, you can have draws, buti think it works 2010-10-20T07:05:57 oh so they can shoot? 2010-10-20T07:06:02 ahead? 2010-10-20T07:06:19 more like getting a lock like in Topgun 2010-10-20T07:06:30 but yea 2010-10-20T07:06:30 you could have a gun with bullet speed and health, but i think games would take too long 2010-10-20T07:06:53 antimatroid: there is a body of work on that... but guess who by? 2010-10-20T07:07:05 not so many will be able to get the papers... 2010-10-20T07:07:27 not sure? i have a couple of dover books on differential game theory 2010-10-20T07:07:37 <3 amazon 2010-10-20T07:08:00 if the speed affect the turn radius it really isn't such an easy problem 2010-10-20T07:08:19 if turn radius affects drag and changes speed its really hard 2010-10-20T07:08:25 *** hello has joined #aichallenge 2010-10-20T07:08:32 really hard is good 2010-10-20T07:08:44 i'm not sure one would get very many competitors though 2010-10-20T07:08:53 the learning curve to get started would be quite high 2010-10-20T07:09:04 but then this is higher than that of tron by a significant amount 2010-10-20T07:09:15 *** gerald5122k has quit IRC () 2010-10-20T07:09:18 there are some results on this... basicall showing a route is optimal (fastest, shortest) is not easy (IIRC np-hard) 2010-10-20T07:09:19 *** hello has quit IRC (Client Quit) 2010-10-20T07:09:52 would you allow people to "reverse"? 2010-10-20T07:10:23 well if we are going for "aircraft" then no... but since its a game... 2010-10-20T07:10:27 for ai 2010-10-20T07:10:33 or, car racing ai game would be pretty cool 2010-10-20T07:10:40 Top 10 players: cfaftw(3517), dmj111(3408), Polsky(3385), rsergio(3384), smatcher(3356), DracoAter(3341), EBraun(3325), savchenko(3311), deepblue(3308), asavis(3278) 2010-10-20T07:10:51 complex physics would put some people off.. 2010-10-20T07:10:51 it would probably need to be mixed up some more... i think straight up chasing has been dealt with extensively in literature, no? 2010-10-20T07:10:52 hard to make symmetric tracks though 2010-10-20T07:10:53 but 2010-10-20T07:11:03 motion isn't really hard 2010-10-20T07:11:11 *** Titankiller has joined #aichallenge 2010-10-20T07:11:13 like maybe a limited field of vision 2010-10-20T07:11:17 if you can write a bot that gets around the track easy... perhaps its not so bad 2010-10-20T07:11:33 antimatroid: just have "fastest time" 2010-10-20T07:11:59 not full race... or do you like "positional" play 2010-10-20T07:12:03 nah, for it to be a game you need both players on a track 2010-10-20T07:12:12 ie a car can block someone from passing 2010-10-20T07:12:13 otherwise it's more of a decision problem 2010-10-20T07:12:16 antimatroid: could use TORCS for that 2010-10-20T07:12:24 racing 2010-10-20T07:12:33 well, a game is a decision problem, but a decision problem isn't necessarily a game 2010-10-20T07:13:00 like mario ai wasn't a game 2010-10-20T07:13:14 not formally speaking anyway 2010-10-20T07:14:17 nplayer ai comps would be cool too, but the analysis gets sooo much harder 2010-10-20T07:14:37 and thus the game soo much simpler -,- 2010-10-20T07:14:38 cause you'd start wanting to form coalitions with the players near you at the start of the game 2010-10-20T07:14:40 :P 2010-10-20T07:14:51 well the solving mechanics 2010-10-20T07:14:54 hehe 2010-10-20T07:16:15 for example, in nplayer planetwars ai, for those of you who expand, you would want to take those who didn't out before fighting each other, otherwise they're going to take whoever wins out of you after you fight 2010-10-20T07:16:16 i'd like a game with a simulated stock market, with unlimited players... 2010-10-20T07:16:25 phreeza: you mean the stock market? 2010-10-20T07:16:34 hehehe 2010-10-20T07:16:55 well don't know if i would want to put money on my bot :) 2010-10-20T07:17:13 that's why you use other peoples finances :P 2010-10-20T07:17:26 so a real market then 2010-10-20T07:17:26 so i should join GS? :) 2010-10-20T07:18:28 ha, i am very tempted to run off and do algo trading rather than a phd 2010-10-20T07:18:52 i think i overestimate how much it'd be like a big ai contest though 2010-10-20T07:19:11 did you read about the guys who got arested for writing a trading algo that gamed the bigger ones? 2010-10-20T07:19:13 antimatroid: its day is nearly over... i wouldn't bother... ;) 2010-10-20T07:19:14 ridiculous 2010-10-20T07:19:27 why aren't they allowed to do that? 2010-10-20T07:19:34 i don't know 2010-10-20T07:19:37 not big engough 2010-10-20T07:19:39 really 2010-10-20T07:19:48 try investing in a hedge fund 2010-10-20T07:19:57 http://www.securitiestechnologymonitor.com/news/norwegian-day-traders-arrested-25971-1.html 2010-10-20T07:20:02 there are rules about how "rich" you have to be 2010-10-20T07:20:07 yea... 2010-10-20T07:20:16 i'm about 30k in debt from uni aha 2010-10-20T07:20:18 same with short selling and naked short selling 2010-10-20T07:20:32 but it's australia, so i get to pay it back when i can and don't have interest above inflation :) 2010-10-20T07:20:33 they whole thing is a crok of very smelly brown stuff 2010-10-20T07:20:44 antimatroid: 75k EU 2010-10-20T07:20:51 ouch 2010-10-20T07:21:00 changed courses a few times 2010-10-20T07:21:15 i get 2 degrees with 3 majors in a month and a bit 2010-10-20T07:21:26 mine dept is for NZ... 2010-10-20T07:21:39 and the interest rate is more than my CC here in Austria 2010-10-20T07:21:51 yeouch 2010-10-20T07:22:02 i don't have a credit card, debit cards save my life 2010-10-20T07:22:03 hehe i'm in germany we get our education virtually for free 2010-10-20T07:22:13 well i pay only the min... 3k EU per year 2010-10-20T07:22:20 not even half of the intreset 2010-10-20T07:22:22 i can easily pay for it with some freelancing 2010-10-20T07:22:54 phreeza: yea... thats why my daughter wants to pass her matura.. 2010-10-20T07:23:06 she is then alowed to go to uni here in Austria 2010-10-20T07:23:14 but she keeps failing 2010-10-20T07:23:17 does anyone else sometimes resort to the dreaded ctrl z method to fixing their bot? 2010-10-20T07:23:20 too lazy 2010-10-20T07:23:26 oh 2010-10-20T07:23:32 and lose all the wonderful changes they did make to somehow solve the one you can't find 2010-10-20T07:23:33 but shes already done with high school? 2010-10-20T07:23:45 phreeza: no 2010-10-20T07:24:05 year before the matura...or whatever its called... for the 2nd time 2010-10-20T07:24:14 hooray for HECS help animatroid :P 2010-10-20T07:24:30 :), where abouts are you? 2010-10-20T07:24:31 in germany its kind of inflationary with the Abitur, which is the german matura 2010-10-20T07:24:34 used to be hard 2010-10-20T07:24:35 tasmania <3 2010-10-20T07:24:38 Melbourne :) 2010-10-20T07:24:42 now you need it for almost any kind of job 2010-10-20T07:24:57 so its not as hard any more 2010-10-20T07:25:27 haha so we have middle europeans and oceanians here? 2010-10-20T07:25:32 day people and night people 2010-10-20T07:25:36 seems to be the case :D 2010-10-20T07:25:54 Make us sound like the above surface fairies and below surface trolls phreeza haha 2010-10-20T07:26:32 this ai comp is killing my productivity, it's past halfway through swot vac and i haven't started studying for exams yet 2010-10-20T07:26:49 well i came from down under... i am not their anymore...so then what am i? trolls can't roam around in daylight ;) 2010-10-20T07:27:21 antimatroid: thanks for reminding me... back to the day job 2010-10-20T07:27:57 haha well i am a troll then too, usually... today is an exception 2010-10-20T07:28:14 they see me trollin' 2010-10-20T07:31:57 http://ai-contest.com/visualizer.php?game_id=5964605 what a lame way to win? lol 2010-10-20T07:33:15 programs should over move change trees rather than just backwards and forwards 2010-10-20T07:33:24 cover* 2010-10-20T07:34:46 antimatroid: don't understand 2010-10-20T07:34:52 whats a move change tree? 2010-10-20T07:34:59 like a transposition table? 2010-10-20T07:36:45 say i go past to a past state of my "document" say, then make a change, it then drops the previous "future moves" you had before 2010-10-20T07:36:58 back to a past*, i need more sleep 2010-10-20T07:37:24 *** Sylph has quit IRC (Ping timeout: 265 seconds) 2010-10-20T07:37:58 oh 2010-10-20T07:38:10 text editors? 2010-10-20T07:38:19 yeah, or for programs 2010-10-20T07:38:32 which are essentially text editors on acid 2010-10-20T07:38:40 *** asdd has quit IRC (Ping timeout: 265 seconds) 2010-10-20T07:38:41 (colour) 2010-10-20T07:39:47 saw this for vim the other day: 2010-10-20T07:39:48 http://bitbucket.org/sjl/gundo.vim/src 2010-10-20T07:39:49 http://ai-contest.com/game_stats.php I think its down 2010-10-20T07:40:16 oh no its not 2010-10-20T07:40:29 0.2 gpm now \o/ 2010-10-20T07:40:30 it's just crawling 2010-10-20T07:42:03 How can there be 3.9 games per hour? 2010-10-20T07:42:08 *** Appleman1234 has joined #aichallenge 2010-10-20T07:42:14 Oh nvm... fail :D 2010-10-20T07:42:22 Another Australian :P 2010-10-20T07:42:28 Pfff we should be sleeping haha 2010-10-20T07:42:40 ha, i was up to 7am last night :| 2010-10-20T07:42:45 benefits of swot vac 2010-10-20T07:43:07 I work at a uni..... 2010-10-20T07:43:09 I'm curious as to whether a Neural net will take too much processing time 2010-10-20T07:43:15 and have a home office ;D 2010-10-20T07:43:23 depends how big the net is :P 2010-10-20T07:43:29 so its all vacation time 2010-10-20T07:44:05 Well, it takes my PC 0.1s for about 60,000 iterations... I hardly think that is possible to do since it won't get that many examples anyway... unless we're working on a 100mhz or something haha 2010-10-20T07:46:54 Rommel: wouldn't you send in a pre-trained network anyway? what do you want to do with online training? 2010-10-20T07:47:35 true that phreeza, though it might be useful to do some extended epoch learning considering I may want it to eventually learn some of the other opponents 2010-10-20T07:47:56 Not just against the example bots? 2010-10-20T07:48:10 tcp 2010-10-20T07:48:36 although i expect people already hate me for using 1 second already 2010-10-20T07:48:47 they'd hate you if you started using like 5 seconds a turn that it allows 2010-10-20T07:49:03 It really shouldn't... how are you using 1 second lol 2010-10-20T07:49:10 tree 2010-10-20T07:49:14 *** Sylph has joined #aichallenge 2010-10-20T07:49:24 i use as long as i tell it too 2010-10-20T07:49:56 and i'm not getting anywhere near the number of node generations as other people, but i think those are my functions for generating/updating data and picking moves etc. 2010-10-20T07:50:13 define tree? Also, do people modify PlanetWars.cc? I don't see how that would work... 2010-10-20T07:50:23 i do 2010-10-20T07:51:10 *** perestrelka has quit IRC (Read error: Connection reset by peer) 2010-10-20T07:51:26 my planetwars is now gamestate and only has a couple of things the original file did 2010-10-20T07:51:43 *** perestrelka has joined #aichallenge 2010-10-20T07:51:46 tree as in players pick moves at each node 2010-10-20T07:51:50 then run minimax on it 2010-10-20T07:52:20 although current version is no longer choosing moves for player 2 untill terminal nodes where it "runs games" and does player 1's turns individually 2010-10-20T07:52:49 oh... so your actually controlling the moves... ie playing, not AI? 2010-10-20T07:52:57 Or a mix 2010-10-20T07:52:58 ? 2010-10-20T07:53:15 what do you mean? 2010-10-20T07:53:24 any bot is going to need to be told what moves are possible 2010-10-20T07:53:30 "tree as in players pick moves at each node" 2010-10-20T07:53:37 Im guessing you don't mean real time 2010-10-20T07:53:37 yeah 2010-10-20T07:53:57 i have functions to generate moves for each player given the current game state 2010-10-20T07:54:01 each node has it's own gamestate 2010-10-20T07:54:26 it's not working very well yet 2010-10-20T07:55:05 but i think the winning bot is going to need to have a smart approach and one of the best heuristic appraoches backing it 2010-10-20T07:55:25 it's not like tron where the payoff function was pretty easy to make and move splits were tiny 2010-10-20T07:55:33 *** Olathe has joined #aichallenge 2010-10-20T07:55:34 *** zasdf has joined #aichallenge 2010-10-20T07:56:28 I haven't got my head around the system yet to understand what you mean by the gamestate, nor the nodes, other than the obvious/english definitions.. 2010-10-20T07:57:44 i think it helped a lot doing the tron contest before this one 2010-10-20T07:58:59 Nor how if you modified your planetWars.cc, it wouldnt be classed as cheating? ... although on second look... it appears all the planetWars.cc and .h files provide is some basic functions for yourself... all the other variables etc are setup and given elsewhere? 2010-10-20T07:59:21 Yay for not having used non-isolated programs before... :) 2010-10-20T07:59:46 *** tty2 has joined #aichallenge 2010-10-20T08:00:28 Rommel: yea thats definitely not cheating 2010-10-20T08:00:37 all that matters is what comes out of stdout 2010-10-20T08:00:39 Rommel: the starter packages are just there as a guide 2010-10-20T08:00:46 you can write you own bot from scratch if you want 2010-10-20T08:01:05 it's really just dealing with the engine by inputting and receving a string 2010-10-20T08:01:25 i read a rule somewhere about no threading but that will be hard to enforce i think, whit haskell and the like especially 2010-10-20T08:02:06 you're allowed green threads whatever they are 2010-10-20T08:02:17 otherwise that is meant to get you kicked out 2010-10-20T08:02:18 *** tty3 has quit IRC (Ping timeout: 240 seconds) 2010-10-20T08:02:20 Ah ok, the parseGameState is basically the big one that receives everything, then its just up to you how to handle it 2010-10-20T08:02:32 *** genericb1b has quit IRC (Ping timeout: 265 seconds) 2010-10-20T08:02:41 yeah, i have two of those now, one for setting up the game and another for updating it 2010-10-20T08:02:53 and they're buried right at the bottom 2010-10-20T08:03:25 ok, it all makes a lot more sense now :) 2010-10-20T08:03:58 *** genericbob has joined #aichallenge 2010-10-20T08:04:03 and is more comforting too... because the way it was setup at the moment just had to go :P 2010-10-20T08:04:04 i found out about the comp a few days before it opened, so i've had lots of time to wrap my head around things 2010-10-20T08:04:35 yeah, mine is pretty much completely rewritten 2010-10-20T08:04:42 everything became structs and public 2010-10-20T08:04:55 Why not OOP? 2010-10-20T08:05:11 i follow the communist approach to programming 2010-10-20T08:05:14 everything is public 2010-10-20T08:05:21 haha, way to be 2010-10-20T08:05:25 ;) 2010-10-20T08:05:33 but why structs... objects not doing it for you? 2010-10-20T08:05:47 structs are objects except everything is public by default :P 2010-10-20T08:06:14 ... and you are the only user, anyway. 2010-10-20T08:06:29 p4p4: that wouldn't ever stop me :P 2010-10-20T08:08:06 I haven't used structs in C++ yet haha, just been doing straight objects all this time... I did them in C, not sure if they are any different? 2010-10-20T08:08:35 same deal pretty sure 2010-10-20T08:09:08 Then why not just Objects anyway? Same thing, but possibility for constructors etc? Easier to manager surely? Esp. with the ability to have functions etc? 2010-10-20T08:09:17 *** jorge has joined #aichallenge 2010-10-20T08:09:20 i wouldn't personally take advice from me about technicalities with programming though 2010-10-20T08:09:32 I'm not, just curious if theres a reason :P 2010-10-20T08:10:42 Top 10 players: cfaftw(3521), dmj111(3426), rsergio(3399), Polsky(3390), smatcher(3363), DracoAter(3342), EBraun(3335), savchenko(3324), deepblue(3312), brunswick(3277) 2010-10-20T08:10:50 structs can have functions etc. 2010-10-20T08:11:04 there is literally no difference other than I don't have to write public: for a struct 2010-10-20T08:11:23 Ah... something that appears to be new in C++ for structs 2010-10-20T08:11:30 AFAIK in C you couldnt put functions in structs 2010-10-20T08:11:43 ah, well i only ever use c++ :P 2010-10-20T08:11:48 Any yeah upon Google searching around... "A struct in C++ is exactly the same as a class, except by default all of the members of a struct are public." 2010-10-20T08:11:55 but i plan to learn haskell and python at some point 2010-10-20T08:11:57 And* 2010-10-20T08:12:02 and i very much irrationally hate java 2010-10-20T08:12:35 it has its uses 2010-10-20T08:12:55 i don't understand how people can call c++ complicated while using java, imo c++ is way easier to learn 2010-10-20T08:15:15 What level of education you @ antima? 2010-10-20T08:15:26 final year undergrad 2010-10-20T08:15:29 what about you? 2010-10-20T08:15:29 antimatroid: thats far from true :P 2010-10-20T08:15:34 java is way easier 2010-10-20T08:15:53 no it's not, in c++ if i want to access elements of a string, myString[elementNo] bang 2010-10-20T08:15:58 lies error... I hate Java; myself: 1st year undergrad 2010-10-20T08:16:16 i want a matrix or graph, vector > myStuff; 2010-10-20T08:17:06 Rommel: what are you studying? 2010-10-20T08:18:35 Professional Software Development 2010-10-20T08:19:06 IMO this 1st year would have pushed me to suicide if I had not started doing 3rd subjects 2nd semester... 2010-10-20T08:19:16 3rd year subjects* 2010-10-20T08:19:23 why not just do cs? 2010-10-20T08:19:31 cs? 2010-10-20T08:19:36 computer science 2010-10-20T08:19:38 comp sci 2010-10-20T08:19:38 maths is fun :P 2010-10-20T08:19:46 *** zasdf has quit IRC (Ping timeout: 265 seconds) 2010-10-20T08:22:22 well, my course was previously named software engineering 2010-10-20T08:22:50 http://courses.swinburne.edu.au/courses/Bachelor-of-Science-(Professional-Software-Development)-I053/local 2010-10-20T08:23:16 In terms of what I would learn, its identical to most of the CS courses in my area 2010-10-20T08:23:20 *** Gorovkhaia2 has quit IRC (Remote host closed the connection) 2010-10-20T08:26:33 And the area I wanted to touch on most, AI, is barely covered by some of CS courses... atleast @ Swin they have a wider selection for it 2010-10-20T08:26:36 *** Mathnerd314 has joined #aichallenge 2010-10-20T08:28:07 What you doing antima? 2010-10-20T08:29:25 *** Gorovkhaia2 has joined #aichallenge 2010-10-20T08:29:58 cs, economics and maths 2010-10-20T08:30:14 i'm a sucker for being punished 2010-10-20T08:30:33 at the moment pure maths has me the most hooked 2010-10-20T08:30:42 along with game theory, but that side of it 2010-10-20T08:31:38 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-20T08:33:39 yeah 2010-10-20T08:34:10 microeconomics is awesome too 2010-10-20T08:34:20 *** Frontier has joined #aichallenge 2010-10-20T08:34:25 Well I made the mistake of not understanding maths in year 10... then just doing the general maths there on... was very frustrating because most of it made sense to me in late yr 11 after a year of physics :P... 2010-10-20T08:34:37 *** p4p4 has joined #aichallenge 2010-10-20T08:34:38 yeah can't stand commerce subjects :P 2010-10-20T08:34:48 i didn't do the top maths in year 12, then picked it up in my second year 2010-10-20T08:35:05 it's taken me 4 years to complete my undergrad (with lots of overloading) 2010-10-20T08:36:18 hmmm 2010-10-20T08:37:03 take some algebra and analysis classes, they'll turn you into a man 2010-10-20T08:37:49 *** novastar has joined #aichallenge 2010-10-20T08:38:08 hi 2010-10-20T08:40:41 i think you'll find microeconomic theory is more maths than anything 2010-10-20T08:40:52 particularly game theory 2010-10-20T08:44:29 novastar: hi ;-) 2010-10-20T08:45:49 gday :) 2010-10-20T08:46:26 hey error323, rommel 2010-10-20T08:46:59 is there any news on the server situation? 2010-10-20T08:48:22 *** jorge has quit IRC (Remote host closed the connection) 2010-10-20T08:49:26 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-10-20T08:50:23 Anyone here with experience with BP networks? 2010-10-20T08:50:30 Backprop AANs 2010-10-20T08:50:32 ANNs* 2010-10-20T08:50:32 *** tty2 has quit IRC (Changing host) 2010-10-20T08:50:32 *** tty2 has joined #aichallenge 2010-10-20T08:50:37 back propogation? 2010-10-20T08:50:42 ya 2010-10-20T08:50:44 not really 2010-10-20T08:50:45 :P 2010-10-20T08:51:45 I made one for fun, and it works great... but I'd say 50% of the time it just doesn't reach a solution, sits on the halfway mark and doesn't move... depends completely on the random initialisation of the weights... 2010-10-20T08:51:48 *** tty2 is now known as tty1 2010-10-20T08:52:13 *** perror has joined #aichallenge 2010-10-20T08:52:38 try a tree :P 2010-10-20T08:52:48 haha... not the point :P... 2010-10-20T08:52:54 still no idea what that is btw, wiki? 2010-10-20T08:53:13 you can't do a degree with programming and not know what a tree is :P 2010-10-20T08:53:40 look up minimax tree 2010-10-20T08:53:44 If your referring to like a massive if else set... I'm going to kill you :P 2010-10-20T08:53:46 k 2010-10-20T08:53:48 and alpha-beta pruning 2010-10-20T08:53:57 no massive if else set 2010-10-20T08:54:16 it's like a linked list, expect each node has a number of children 2010-10-20T08:54:23 except* 2010-10-20T08:54:28 aha 2010-10-20T08:54:41 i come up with some of the weirdest typos 2010-10-20T08:55:10 fyi minimax was the winning strategy for the tron contest 2010-10-20T08:55:15 but it's a lot harder and messier this time 2010-10-20T08:55:25 and you can only hope to approximate part of the actual gametree 2010-10-20T08:55:51 well, you get part of it, you're approximating the good parts of it 2010-10-20T08:56:48 regardless of any of those macro kind of ides, unless you handle the micro decisions of what moves to make well in this, your bot isn't going to do very well 2010-10-20T08:56:52 *** jarfhy has quit IRC (Read error: Operation timed out) 2010-10-20T08:56:58 i'm currently rethinking my choices of possible moves 2010-10-20T08:57:30 what did you use for depth? 2010-10-20T08:57:49 i have a queue of nodes to extend, and i just extend the tree for as long as i can 2010-10-20T09:00:14 *** iFire has quit IRC (Ping timeout: 265 seconds) 2010-10-20T09:00:14 hmmm... how would you define a plie in this case? 2010-10-20T09:00:31 *** seisatsu has quit IRC (Quit: Leaving) 2010-10-20T09:04:18 *** oPless has joined #aichallenge 2010-10-20T09:04:45 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-20T09:05:55 *** p4p4 has joined #aichallenge 2010-10-20T09:08:22 *** delt0r___ has joined #aichallenge 2010-10-20T09:09:19 *** novastar has left #aichallenge 2010-10-20T09:09:25 *** delt0r has quit IRC (Ping timeout: 240 seconds) 2010-10-20T09:10:42 Top 10 players: cfaftw(3528), dmj111(3429), rsergio(3407), Polsky(3401), smatcher(3368), DracoAter(3339), EBraun(3336), savchenko(3332), deepblue(3330), asavis(3314) 2010-10-20T09:11:41 *** iFire has joined #aichallenge 2010-10-20T09:13:17 *** jarfhy has joined #aichallenge 2010-10-20T09:16:19 morning 2010-10-20T09:18:08 night all 2010-10-20T09:18:21 night antima :) 2010-10-20T09:18:30 night 2010-10-20T09:18:36 I'd stay up... alas a OOP test tomor... today 2010-10-20T09:18:38 i'm going to be up a while yet :| 2010-10-20T09:18:45 this contest is very addictive, be warned 2010-10-20T09:18:53 I've noticied 2010-10-20T09:18:54 how is your treebot coming along antimatroid 2010-10-20T09:19:03 nn 2010-10-20T09:19:07 bhasker: terribly, i'm rethinking the underlying heuristics 2010-10-20T09:19:12 *** Rommel has quit IRC (Quit: Page closed) 2010-10-20T09:19:18 hmm 2010-10-20T09:19:28 it works well, but fails sometimes 2010-10-20T09:19:34 but i'm pretty sure that's the heuristics 2010-10-20T09:19:46 i can't win consistently, and i wont be happy till i'm in first place 2010-10-20T09:19:48 is testBot123 yours? 2010-10-20T09:19:52 ah k 2010-10-20T09:19:53 nope 2010-10-20T09:19:57 is it running on tcp 2010-10-20T09:19:58 ? 2010-10-20T09:19:59 i have used it before though 2010-10-20T09:20:00 nope 2010-10-20T09:20:06 ah k 2010-10-20T09:20:10 i haven't run it since this morning when i used antimatroid 2010-10-20T09:20:20 i use either antimatroid or youwillfail usually 2010-10-20T09:20:35 or something else if i don't want people to know who i am (used too many usernames that day already) 2010-10-20T09:21:54 do we know who CCF is yet? 2010-10-20T09:22:15 i wasn't even that intrigued until people kept asking, now i'm curious 2010-10-20T09:22:22 nope no idea 2010-10-20T09:22:27 its a good bot though 2010-10-20T09:22:28 running different bots under the same name seems kinda bad for the rating system, assuming it's within a 24h period that is 2010-10-20T09:22:34 i beat that earlier 2010-10-20T09:23:04 I think I read here once that felixcoto(?) is cccf 2010-10-20T09:23:42 flufffluff i guess fluff_21_0.9s is yours? 2010-10-20T09:24:08 yeah 2010-10-20T09:24:14 ah k 2010-10-20T09:24:21 so youwillfail is yours antimatroid 2010-10-20T09:24:30 yep 2010-10-20T09:24:34 it was a good bot though haven't seen it in quite sometime 2010-10-20T09:24:44 my last submission on official server is upto rank #47 2010-10-20T09:24:47 yeah, it's not amazing atm 2010-10-20T09:24:54 i'm in the thirties somewhere 2010-10-20T09:24:55 but the number of games is so low though:-\ 2010-10-20T09:24:59 i want top 10 at least 2010-10-20T09:25:59 don't we all 2010-10-20T09:26:03 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-20T09:26:13 yea but i doubt i will make it there 2010-10-20T09:26:18 i think i can make it to top 30 maybe 2010-10-20T09:27:03 I keep having timeout issues on the offical server :| I sent in a version the other day with a time cap of 0.01s. that one got to play 9 games without problems. 0.4s on the other hand died on the first game after 9 turns. lowering that to 0.1s made it last 21 turns on the first game 2010-10-20T09:27:53 *** p4p4 has joined #aichallenge 2010-10-20T09:28:18 lol http://72.44.46.68/canvas?game_id=588702 2010-10-20T09:28:27 don't know why the other bot never reacted 2010-10-20T09:28:32 i got disqualified last tournament with a bot that should have finished top 15 most likely 10, so i have a point to prove :P 2010-10-20T09:28:40 disqualified? 2010-10-20T09:28:42 why? 2010-10-20T09:28:53 exceeding time limit on one turn in one game 2010-10-20T09:29:03 stupidest rule ever, after much petitioning it's been changed this time 2010-10-20T09:29:24 heh 2010-10-20T09:29:40 that is a stupid rule 2010-10-20T09:29:49 i didn't even know that was the rule till it happened 2010-10-20T09:29:52 was incredibly annoying 2010-10-20T09:30:18 is anyone running time intensive bots without problems on the official server? 2010-10-20T09:30:35 biggest let down, and then when people who have no idea wtf you did ask you how you went and you say you got disqualified, they grill you so bad 2010-10-20T09:30:38 my bot barely takes anytime 2010-10-20T09:30:49 i haven't uploaded my tree bot there yet 2010-10-20T09:30:56 it's still too shit 2010-10-20T09:31:11 *** Arthur has quit IRC (Disconnected by services) 2010-10-20T09:31:13 i think my bot takes max of 200ms with full debug logging 2010-10-20T09:31:17 since i don't do any tree searches 2010-10-20T09:31:22 *** Arthur has joined #aichallenge 2010-10-20T09:31:26 and i abuse priority queues 2010-10-20T09:31:33 you would hope to be using almost nothing if you plan to use a tree later 2010-10-20T09:31:42 priority queues and overloading < are my best friend 2010-10-20T09:31:54 bhask: what's your name on tcp? 2010-10-20T09:31:57 malazan 2010-10-20T09:32:07 was playing my original versions with my name 2010-10-20T09:32:12 but when i rewrote it switched the name 2010-10-20T09:33:20 i wish more people would run their bots on tcp 2010-10-20T09:33:22 all the time 2010-10-20T09:33:29 i can never figure how good my bot is doing 2010-10-20T09:33:44 yeah, tcp rocks 2010-10-20T09:33:56 sorry, i should do that, next time i get a decent bot i will run a bat file indefinitely 2010-10-20T09:34:22 i often just run it as i'm fixing things 2010-10-20T09:34:39 as how it does against my past revisions isn't always a great indication of how it will do against other bots 2010-10-20T09:34:51 yea my current bot destroys my older bots 2010-10-20T09:34:57 but doesn't do that much better against other bots 2010-10-20T09:34:57 unlike tron where you could literally debug why you lost, it's not so easy to work out in this why you lost 2010-10-20T09:35:30 *** Titankiller has quit IRC (Read error: Connection reset by peer) 2010-10-20T09:35:39 actually figuring out reason for loss isnt that hard 2010-10-20T09:35:46 its usually 2-3 things 2010-10-20T09:35:49 a) the start was bad 2010-10-20T09:35:55 b) you picked the wrong neutral 2010-10-20T09:36:09 c) you picked the right neutral at the wrong time 2010-10-20T09:36:20 d) you forgot to defend something ^ 2010-10-20T09:37:13 so one rule that i have added that seems to do wonders 2010-10-20T09:37:23 when even or ahead don't attack any neutrals 2010-10-20T09:37:30 just sit 2010-10-20T09:37:35 you win by inaction 2010-10-20T09:37:41 or if enemy attacsk you snipe 2010-10-20T09:38:11 my bot now doesn't stupidly keep trying to expand 2010-10-20T09:38:30 yeah, i've added in variables for the expected final growth counts of each player 2010-10-20T09:38:34 bhasker: i think some bots also look ahead and see if they can be sniped, and don't attack neutrals if that is the case 2010-10-20T09:38:39 leading to things like this 2010-10-20T09:38:41 http://ai-contest.com/visualizer.php?game_id=5952104 2010-10-20T09:38:41 and make the decision to expand based off of that 2010-10-20T09:38:55 seems like I could use some of that... I'm expanding pretty much nonstop 2010-10-20T09:39:03 yup happens with my bot too phreeza 2010-10-20T09:39:09 fluff another rule 2010-10-20T09:39:15 if ahead in production but low on ships 2010-10-20T09:39:18 dmj taught me not to expand too much 2010-10-20T09:39:18 never attack anything 2010-10-20T09:39:25 by repeatedly crushing me in under 100 turns 2010-10-20T09:39:28 if you have less ships than enemy 2010-10-20T09:39:31 never attack anything 2010-10-20T09:39:40 case in point http://72.44.46.68/canvas?game_id=588825 2010-10-20T09:39:44 i win this game by inaction 2010-10-20T09:39:48 i was even but ahead in ships 2010-10-20T09:40:02 won the game by 14 ships ^ 2010-10-20T09:41:04 hehe, sweet 2010-10-20T09:41:11 i find my bot chooses to do nothing too often when i allow it the option :| 2010-10-20T09:41:29 i guess it sees too many awesome future moves that no longer remain possible 2010-10-20T09:41:30 well i added a few things 2010-10-20T09:41:34 i expand if enemy expands 2010-10-20T09:41:40 to maintain my advantage 2010-10-20T09:42:11 and i snipe enemy planets if he moves his ships away from a planet close to me 2010-10-20T09:42:26 http://72.44.46.68/canvas?game_id=588857 2010-10-20T09:42:30 another game draw this time 2010-10-20T09:43:42 *** sigh has quit IRC (Remote host closed the connection) 2010-10-20T09:44:46 bhasker: i'ma put current bot up even though it's not very good 2010-10-20T09:44:54 cause i'm going to make some large changes again 2010-10-20T09:44:58 bhasker: and do you think this is ideal? i can't see any obvious move to win that 2010-10-20T09:45:07 so maybe it is 2010-10-20T09:45:19 which might mean all the top bots will end up drawing? :D 2010-10-20T09:45:32 great, it plays me against bocsimacko right away 2010-10-20T09:45:39 that's going to hurt 2010-10-20T09:46:00 yea i think its ideal 2010-10-20T09:46:04 not to do anything 2010-10-20T09:46:12 shifting 60-70 ships from front 2010-10-20T09:46:16 could cause an immediate attack 2010-10-20T09:46:25 so its best not to do anything 2010-10-20T09:46:35 *** Azrathud has joined #aichallenge 2010-10-20T09:46:41 I just restarted because dmj appeared on tcp. 2010-10-20T09:46:42 actually 2010-10-20T09:47:04 hazard should have acquired the planet with 29 ships on it 2010-10-20T09:47:09 its a relatively safe planet to acquire 2010-10-20T09:47:24 *** Azrathud has quit IRC (Client Quit) 2010-10-20T09:48:13 mega1 your bot is interesting to watch 2010-10-20T09:48:15 *** noid05826 has left #aichallenge 2010-10-20T09:48:34 *** Accoun has joined #aichallenge 2010-10-20T09:48:34 though whats the reason behind the name? 2010-10-20T09:49:13 it's a character from a stupid child's tale for 4 year olds 2010-10-20T09:49:24 no special reason 2010-10-20T09:49:24 ooh k 2010-10-20T09:49:37 i tried to do no staging 2010-10-20T09:49:41 but my bot does horrible lol 2010-10-20T09:49:45 if i remove my staging code 2010-10-20T09:49:47 ha, defeated dmj 2010-10-20T09:50:03 bhasker: I think staging is crucial 2010-10-20T09:50:09 do you watch the games live using the viewer 2010-10-20T09:50:13 no 2010-10-20T09:50:18 ah just following logs? 2010-10-20T09:50:39 yea staging is needed, except i can't figure out when to stop sending to front sometimes 2010-10-20T09:50:50 yeah and checking out interesting games once they appear on the server 2010-10-20T09:50:54 enemy will be taking a planet near the back or i will lose some back planets 2010-10-20T09:51:09 and it takes a long time for my bot to realize that it has a threat at its back 2010-10-20T09:51:15 and it loses a bunch of planets in a short time 2010-10-20T09:51:34 trying to balance for that 2010-10-20T09:51:36 is a pain 2010-10-20T09:51:37 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-20T09:52:01 lost to antimatroid 2010-10-20T09:52:05 ha :P 2010-10-20T09:52:16 and i thought this bot sucked 2010-10-20T09:53:28 *** wvdschel has quit IRC (Quit: Leaving) 2010-10-20T09:53:34 fairly boring game though 2010-10-20T09:53:47 ah mega1 was slow to acquire the 2 growth planet 2010-10-20T09:53:56 thats what put him behind 2010-10-20T09:54:10 just realized that I had too many processes running 2010-10-20T09:54:23 looks like we have a match antimatroid 2010-10-20T09:54:26 and my bots weren't able to finish thinking 2010-10-20T09:54:44 ah 2010-10-20T09:55:38 if this bot keeps it up its going to get a nice rank boost once the old games expire aha 2010-10-20T09:55:58 upon review I don't think that was the reason it lost 2010-10-20T09:56:07 sorry anti broke your bots bubble:( 2010-10-20T09:56:12 i'm not using the full 1 second atm i don't think 2010-10-20T09:56:21 :P 2010-10-20T09:56:26 i'll get you this time 2010-10-20T09:56:33 *** p4p4 has joined #aichallenge 2010-10-20T09:57:00 :) 2010-10-20T09:57:06 i should head to gym 2010-10-20T09:57:10 will leave the bot running 2010-10-20T09:57:49 i had a punctured tyre on my bike today, was very annoying 2010-10-20T09:58:23 i also took a trip over my handlebars yesterday aha 2010-10-20T09:58:34 i will win this one too antimatroid unless my bot goes stupid 2010-10-20T09:58:38 i went riding down some dirt tracks nearby and decided to stop when i shouldn't have 2010-10-20T09:58:53 antimatroid: ouch, you puncture the tire at the same time? 2010-10-20T09:58:59 ouch trip over the handlebar 2010-10-20T09:59:03 that must have hurt 2010-10-20T09:59:06 foucist: i think it was just riding around on dirt tracks 2010-10-20T09:59:23 it's more of a cross between a road bike and a mountain bike and really shouldn't have been taken where it was 2010-10-20T09:59:53 didn't hurt so was all good, last time i went over the handlebars of a bike i broke my shoulder and didn't go to the doctor for a week, family gave me a fair bit of shit about that one 2010-10-20T10:00:00 antimatroid: one time i was riding in the rain, my foot slipped off the pedal, and i hit an embankment and went flying over, and the bike actually ended up twisted up a bit, i took it to a shop and they thought a car must've hit me 2010-10-20T10:00:14 i was shifting gears in my car with my right arm (you drive on the right side of the car in australia) 2010-10-20T10:00:28 bhasker :P 2010-10-20T10:00:38 :D 2010-10-20T10:01:17 weird that my bot went for those corner planets 2010-10-20T10:01:20 instead of your main base 2010-10-20T10:01:29 oh well 2010-10-20T10:01:33 http://72.44.46.68/canvas?game_id=589081 2010-10-20T10:01:33 *** brunswick has joined #aichallenge 2010-10-20T10:07:42 so, building clusters is a good thing ? 2010-10-20T10:10:43 Top 10 players: cfaftw(3509), dmj111(3423), rsergio(3413), Polsky(3392), smatcher(3365), DracoAter(3337), savchenko(3320), EBraun(3317), asavis(3312), deepblue(3307) 2010-10-20T10:11:38 *** JamesMG has joined #aichallenge 2010-10-20T10:12:14 *** Itkovian has joined #aichallenge 2010-10-20T10:12:36 p4p4: i don't have any code to do that in myine but i'd think so 2010-10-20T10:12:51 mutual defense is pretty easy that way 2010-10-20T10:14:18 *** Sarin has joined #aichallenge 2010-10-20T10:31:04 *** Florian95 has quit IRC (Read error: Connection reset by peer) 2010-10-20T10:34:09 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-20T10:35:20 *** Florian95 has joined #aichallenge 2010-10-20T10:35:33 *** Itkovian has quit IRC (Remote host closed the connection) 2010-10-20T10:35:37 dmj111`: how is dmj1.8's strength compared to what you have on the official server? 2010-10-20T10:35:41 *** Itkovian has joined #aichallenge 2010-10-20T10:35:55 *** p4p4 has joined #aichallenge 2010-10-20T10:37:43 contestbot: bye 2010-10-20T10:37:43 bye brunswick 2010-10-20T10:37:58 *** brunswick has quit IRC (Quit: Page closed) 2010-10-20T10:38:09 i like that ^^ 2010-10-20T10:39:19 *** Frontier has left #aichallenge 2010-10-20T10:44:24 *** rebelxt has joined #aichallenge 2010-10-20T10:44:34 *** Titankiller has joined #aichallenge 2010-10-20T10:44:35 *** Arthur has quit IRC (Disconnected by services) 2010-10-20T10:44:46 *** Arthur_ has joined #aichallenge 2010-10-20T10:46:54 *** aerique has quit IRC (Quit: ...) 2010-10-20T10:48:18 *** hellman has joined #aichallenge 2010-10-20T10:49:00 *** rebelxt has left #aichallenge 2010-10-20T10:49:09 *** Frontier has joined #aichallenge 2010-10-20T10:50:43 *** Frontier has joined #aichallenge 2010-10-20T10:53:00 *** rebelxt has joined #aichallenge 2010-10-20T10:55:53 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-20T10:57:13 thats a bit sad :) 2010-10-20T10:57:17 saying bye to a bot 2010-10-20T10:57:56 i hate bots, except contest bot 2010-10-20T10:57:57 *** p4p4 has joined #aichallenge 2010-10-20T10:58:03 !rainbow rainbows are awesome 2010-10-20T10:58:03 antimatroid: rainbows are awesome 2010-10-20T10:58:34 *** Frontier has left #aichallenge 2010-10-20T11:00:25 *** Itkovian has quit IRC (Ping timeout: 252 seconds) 2010-10-20T11:01:09 *** rebelxt has left #aichallenge 2010-10-20T11:02:41 case based reasoning for game AI, google tech talk 2010-10-20T11:02:41 http://www.youtube.com/watch?v=s9G7DRTuB5s 2010-10-20T11:02:57 there goes my productivity for the next 53 minutes 2010-10-20T11:03:39 *** Frontier has joined #aichallenge 2010-10-20T11:10:43 Top 10 players: smatcher(3574), cfaftw(3542), DracoAter(3473), dmj111(3441), rsergio(3431), Polsky(3421), asavis(3351), EBraun(3346), savchenko(3335), deepblue(3326) 2010-10-20T11:16:04 *** bhasker_ has joined #aichallenge 2010-10-20T11:17:51 *** bhasker has quit IRC (Ping timeout: 272 seconds) 2010-10-20T11:22:49 *** boegel has quit IRC (Ping timeout: 265 seconds) 2010-10-20T11:25:32 *** rebelxt has joined #aichallenge 2010-10-20T11:27:10 *** gerald512k has joined #aichallenge 2010-10-20T11:27:14 hi all 2010-10-20T11:27:22 finally i got a positive helo on TCP serv :DD 2010-10-20T11:28:31 \o/ 2010-10-20T11:31:10 cost a lot of work lol :) 2010-10-20T11:31:55 *** Sylph has quit IRC (Ping timeout: 240 seconds) 2010-10-20T11:34:18 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-10-20T11:37:00 *** Itkovian has joined #aichallenge 2010-10-20T11:40:51 *** Frontier has quit IRC (Read error: Connection reset by peer) 2010-10-20T11:40:57 *** Frontier has joined #aichallenge 2010-10-20T11:41:04 *** cfaftw has joined #aichallenge 2010-10-20T11:42:06 i think elo calculations broken http://www.ai-contest.com/profile_games.php?user_id=6749? 2010-10-20T11:42:09 http://www.ai-contest.com/profile_games.php?user_id=6749 *** 2010-10-20T11:42:16 that is first? 2010-10-20T11:42:16 lol @ current official server rankings 2010-10-20T11:42:41 cfaftw: congrats on your current deserved place :P 2010-10-20T11:42:45 antimatroid: I think its related to the very low number of games 2010-10-20T11:43:11 surely elo scores aren't a function of tme since game was played 2010-10-20T11:43:14 antimatroid, yes. "deserved". O:) 2010-10-20T11:43:16 and the quite high server load causing the better bots to loose to poor bots every now and then with timeouts 2010-10-20T11:43:30 toulouse 1er !!! 2010-10-20T11:43:37 comme au rugby :D 2010-10-20T11:45:00 1st bot are the most efficient now :) 2010-10-20T11:46:55 i think it goes back to the problem of game history being tied to user accounts and not specific submissions 2010-10-20T11:47:03 when a top 10 player resubmits everything gets crazy 2010-10-20T11:47:12 (Hazard just resubmitted yesterday) 2010-10-20T11:48:46 *** p4p4_ has joined #aichallenge 2010-10-20T11:51:48 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-10-20T11:53:09 *** p4p4_ is now known as p4p4 2010-10-20T12:06:04 i though game history was reset while calculating new ELO for new bots 2010-10-20T12:06:25 *** Bobng has joined #aichallenge 2010-10-20T12:06:52 it is so sad that we only got 1 fight every 6 hours on official ranking :( 2010-10-20T12:07:08 *** perestrelka has quit IRC (Read error: Connection reset by peer) 2010-10-20T12:07:16 me gets 1 game every 24 hours 2010-10-20T12:07:30 http://www.ai-contest.com/profile.php?user_id=3902 2010-10-20T12:07:41 *** perestrelka has joined #aichallenge 2010-10-20T12:07:43 mega1: the bot on tcp is my Sept 23rd submission. The latest submission is better, I think, but still would get beaten badly by yours. 2010-10-20T12:07:45 Your bot is doing the things I am trying to do, but cannot get working :( 2010-10-20T12:08:10 *** perestrelka has quit IRC (Read error: Connection reset by peer) 2010-10-20T12:08:16 *** Bobng has quit IRC (Client Quit) 2010-10-20T12:08:16 dmj111`: thanks. What is it that you cannot get working? 2010-10-20T12:08:47 *** perestrelka has joined #aichallenge 2010-10-20T12:08:48 Safely taking more neutrals "in the back". I either am too safe, like dmj1.8, or too agressive. 2010-10-20T12:08:52 *** Bobng has joined #aichallenge 2010-10-20T12:09:00 my current bot is pretty crappy 2010-10-20T12:09:11 *** dmj111` is now known as dmj111 2010-10-20T12:09:14 i'm current rewriting a lot of my heuristics stuff to make it a lot better 2010-10-20T12:09:27 not sure why my nickname had a ` in it... that is not the first time it has happened. 2010-10-20T12:09:43 roll with it? :P 2010-10-20T12:09:57 sql injection? 2010-10-20T12:10:02 emacs. 2010-10-20T12:10:07 have you spent a lot of time coding your bot dmj111? 2010-10-20T12:10:44 too much. a lot of it is in dead-end branches in hg. 2010-10-20T12:10:44 Top 10 players: smatcher(3573), cfaftw(3558), DracoAter(3474), dmj111(3445), rsergio(3434), Polsky(3426), asavis(3356), EBraun(3352), savchenko(3338), deepblue(3330) 2010-10-20T12:10:47 probably connection dropping and getting reconnected while the old nick is still in use and your client automatically uses ` to create a new nick 2010-10-20T12:11:39 Ahh... I have done more irc'ing in the last 5 weeks than I have for the rest of my life. 2010-10-20T12:12:03 thats cool hardcore geek coder player :D 2010-10-20T12:12:29 antimatroid: if I cannot get my new bot to do what I want, I may go back and try to add more heuristics to the official version. 2010-10-20T12:12:55 i've actually just rethought a lot of my decisions on moves in the last hour or two 2010-10-20T12:13:00 currently working on implementing it 2010-10-20T12:13:20 coding my bot learns me java language effectively and in a funny way very that is why i spend a lot of time n it. 2010-10-20T12:13:27 it's more common for clients to use _, but emacs? yeah that's not surprising :P 2010-10-20T12:13:49 gerald512k: you're up 2010-10-20T12:13:53 dmj i just beat you :P 2010-10-20T12:14:47 cool. 2010-10-20T12:15:10 it would be so nice if there was a server with 3 players :D 2010-10-20T12:15:41 *** lavalamp_ has joined #aichallenge 2010-10-20T12:16:49 completely changes the game 2010-10-20T12:17:02 yes that would just be for fun 2010-10-20T12:17:32 build it yourself :) 2010-10-20T12:17:47 I've been thinking about building a server for stuff like that after the competition 2010-10-20T12:17:54 ie. if one doesn't expand, the other two want to take them out before fighting, but who takes them out first? 2010-10-20T12:17:56 we'll see if there is still interest at that point though 2010-10-20T12:18:07 if you do it too fast the other player will take you both out 2010-10-20T12:18:07 Naktibalda i am very noob on java so i am not able to do this :) 2010-10-20T12:19:58 *** Frontier has left #aichallenge 2010-10-20T12:28:06 *** Spike_007 has joined #aichallenge 2010-10-20T12:28:13 hello all 2010-10-20T12:28:33 just checking irc, this is my first time ... :S 2010-10-20T12:28:52 *** bhasker_ has quit IRC (Quit: bhasker_) 2010-10-20T12:29:13 Hi 2010-10-20T12:29:28 hi 2010-10-20T12:29:45 gerald512k I think the engine already supports >2 players, no? 2010-10-20T12:31:19 unlikelly :) 2010-10-20T12:33:06 *** hehe has joined #aichallenge 2010-10-20T12:33:11 hello 2010-10-20T12:33:24 I wrote my bot with the potential for more in mind, but it just treats them all as one enemy blob 2010-10-20T12:33:28 if there are more than one 2010-10-20T12:33:48 *** Spike_007 has quit IRC () 2010-10-20T12:34:40 does everyone spend their lunch breaks playing minecraft? the server's been down for like 30 minutes! 2010-10-20T12:34:52 i guess i'll just work on my bot... :( 2010-10-20T12:35:25 I should get back on minecraft, he has to have fixed some stuff in multi by now... 2010-10-20T12:35:44 jamesmg, best to wait till oct 31... there's a big patch coming 2010-10-20T12:35:55 ah, ok 2010-10-20T12:36:13 last time I played minecarts, buckets and furnaces were all broken 2010-10-20T12:36:31 minecarts are still broken, as well as red stone 2010-10-20T12:36:37 and you can't place lava or water 2010-10-20T12:37:35 furnaces work 2010-10-20T12:37:42 bummer, that supposed to be in the oct 31 patch? 2010-10-20T12:38:02 eh... that's what everyone is hoping 2010-10-20T12:38:08 phreeza i think yes, but no sure that bot support it :D 2010-10-20T12:38:34 i'm so happy i haven't gotten into minecraft 2010-10-20T12:38:35 I was hoping he'd pseudo open source it at some point 2010-10-20T12:38:41 i wish he'd focus on getting the existing content working in every mode before adding next content 2010-10-20T12:38:47 next = new 2010-10-20T12:38:50 i wasted so much on dwarf fortress 2010-10-20T12:39:02 i dont know minecraft but is seems to be a geek game : 2010-10-20T12:39:04 I never did get into dwarf fortress thankfully 2010-10-20T12:39:06 lol, phreeza i tried dwarf fortress but couldn't get into it 2010-10-20T12:39:16 once he's bought his island in the carribean he might 2010-10-20T12:39:33 haha, he ever sort out his paypal problem? 2010-10-20T12:39:42 last I heard they had a hold on about 600k euros 2010-10-20T12:39:42 multiplayer minecraft has a bunch of sync issues in general, especially with chests 2010-10-20T12:39:43 yeah, he actually hired a few software developers 2010-10-20T12:39:47 they have a legit company now 2010-10-20T12:39:57 not sure, i think he did 2010-10-20T12:40:02 all my iron disappeared and i was pissed so i figure out the mechanism and now i can dupe items ad infinitum 2010-10-20T12:40:20 can someone give a one line "why is minecfact cool?" answer? i might try it after this contest is over ;) 2010-10-20T12:40:26 a1k0n, but that makes it not fun! 2010-10-20T12:40:38 rebelxt: it's virtual legos, for me anyways 2010-10-20T12:40:48 well, all the other brokenness, esp re: monsters, makes it not fun anyway 2010-10-20T12:40:53 although... i'm guilty of using the "repair" glitch. drop busted diamond pick on ground, pick it up, brand new diamond pick 2010-10-20T12:41:03 yeah cfaftw, I can't not do that 2010-10-20T12:41:05 a1k0n, too true 2010-10-20T12:41:15 i generally play single player, which is fun. multiplayer, well, i just dick around and build things 2010-10-20T12:41:25 scripted monsters? is this world of warcraft with lego mixed in? 2010-10-20T12:41:31 yeah rebelxt 2010-10-20T12:41:40 except no RPG elements really 2010-10-20T12:41:41 the only reason i bought the game is because me and a few friends got it together... one of them hosts a server and we all create worlds together 2010-10-20T12:41:43 it's awesome 2010-10-20T12:41:55 *** tty1 has quit IRC (Ping timeout: 240 seconds) 2010-10-20T12:41:57 i bet rpg elements will be forthcoming 2010-10-20T12:41:59 omg, i hate world of warcraft. so many days wasted ;) 2010-10-20T12:42:08 all I do in it is create buildings and caves and explode TNT deposits to create craters etc 2010-10-20T12:42:19 haha 2010-10-20T12:42:36 yes we duped a lot of TNT and blasted the shit out of our multiplayer map 2010-10-20T12:42:36 you should build an underwater walkway 2010-10-20T12:42:46 I just host a server for me and my friends to play on and do that kinda crap 2010-10-20T12:42:52 cfaftw: your bot climbed to #2. cool! 2010-10-20T12:43:03 cfaftw: when I discovered glass I made an underwater lab 2010-10-20T12:43:07 rebelxt, yeah... but it's not legit. the rankings are all jacked up 2010-10-20T12:43:20 too few games 2010-10-20T12:43:37 jamesmg, nice... my next project is going to be an underwater biocube 2010-10-20T12:43:40 i hope they will find new servers soon :( 2010-10-20T12:44:04 cfaftw: what would you grow in the biocube? 2010-10-20T12:44:08 is this like a hydroponics thing? 2010-10-20T12:44:19 foucist, everything 2010-10-20T12:44:22 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-20T12:44:25 I tried to get grass to grow underwater to grow some trees in my lab 2010-10-20T12:44:36 but I haven't spent enough time in game for it to grow down the path I set 2010-10-20T12:44:52 oh wait, we'er not talkinga bout the real world? :P 2010-10-20T12:45:05 *** hellman has quit IRC (Remote host closed the connection) 2010-10-20T12:45:12 what's the difference? 2010-10-20T12:45:32 jamesmg, what i'm going to do is build a crater in the water, build the cube, let grass grow, then build the ceiling and replace the water 2010-10-20T12:45:41 I can't make a waterfall out of a bucket of water in the real world :( 2010-10-20T12:45:52 this contest is probably going on in someone's virtual world anyway.. created in minecraft of course 2010-10-20T12:46:01 haha 2010-10-20T12:46:32 cfaftw: I never thought of making a deeper ocean by making a crater, good idea 2010-10-20T12:48:59 *** tapwater has joined #aichallenge 2010-10-20T12:49:04 *** bhasker has joined #aichallenge 2010-10-20T12:50:45 antimatroid: what game? 2010-10-20T12:51:46 amstan, galcon 2010-10-20T12:52:00 we're doing galcon now.. 2010-10-20T12:52:26 talking about the possibility of someone setting up a server that does 3 or more players in a match 2010-10-20T12:53:21 unless you're replying to something even further back in the logs :P 2010-10-20T12:54:32 *** bhasker_ has joined #aichallenge 2010-10-20T12:54:43 *** Frontier has joined #aichallenge 2010-10-20T12:56:14 how much time is allowed for the first move? 2010-10-20T12:56:18 1s or 3s? 2010-10-20T12:57:21 *** bhasker has quit IRC (Ping timeout: 265 seconds) 2010-10-20T12:57:33 3s 2010-10-20T12:58:26 didn't someone say that the engine waits 2s before sending the first game state? 2010-10-20T12:58:34 isn't it 2s? we looked at the code a few days ago... 2010-10-20T12:59:09 engine.py? 2010-10-20T13:00:05 1s for each turn, bots are given a startup time of 2s at start, but game state is given after that, than you have 1s to do your move 2010-10-20T13:01:00 HakanD: thanks, I can see it now in the code. 2010-10-20T13:01:33 http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/engine.py - is it what you are looking at? 2010-10-20T13:02:24 branch has different logic hmm: http://code.google.com/p/ai-contest/source/browse/branches/20100929-games-in-the-cloud/planet_wars/backend/engine.py 2010-10-20T13:03:08 *** Itkovian has quit IRC (Quit: Itkovian) 2010-10-20T13:03:54 maybe someone knows what branch they are running on the server that's up 2010-10-20T13:04:51 * mega1 uploads a fixed version, crosses fingers 2010-10-20T13:07:54 mega1, good luck. can't wait to get pwned by your bot ;) 2010-10-20T13:08:02 it's just lost 2010-10-20T13:08:14 wth? first turn? 2010-10-20T13:09:07 no, 17th 2010-10-20T13:10:44 Top 10 players: smatcher(3575), cfaftw(3557), DracoAter(3476), dmj111(3450), rsergio(3436), Polsky(3360), asavis(3346), savchenko(3342), EBraun(3341), deepblue(3330) 2010-10-20T13:11:08 *** bhasker has joined #aichallenge 2010-10-20T13:11:22 mega1, server or TCP? 2010-10-20T13:11:37 server, but I think I got a handle on it ... 2010-10-20T13:11:52 the ancient sbcl they are running is playing tricks 2010-10-20T13:13:02 *** boegel has joined #aichallenge 2010-10-20T13:13:28 *** HakanD has quit IRC (Ping timeout: 245 seconds) 2010-10-20T13:13:48 *** bhasker__ has joined #aichallenge 2010-10-20T13:14:43 *** bhasker_ has quit IRC (Ping timeout: 245 seconds) 2010-10-20T13:16:55 *** boegel has quit IRC (Client Quit) 2010-10-20T13:17:28 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2010-10-20T13:18:16 *** p4p4 has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2010-10-20T13:18:57 *** p4p4 has joined #aichallenge 2010-10-20T13:20:29 *** p4p4 is now known as Les_Shadoks 2010-10-20T13:20:58 * Les_Shadoks "To reduce the numbers of unhappy people, always beat up the same individuals." 2010-10-20T13:21:08 *** Les_Shadoks is now known as p4p4 2010-10-20T13:21:50 * p4p4 finally found out, what all this reminds me of! 2010-10-20T13:23:16 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-10-20T13:26:08 *** Frontier has left #aichallenge 2010-10-20T13:28:57 *** Frontier has joined #aichallenge 2010-10-20T13:32:09 *** jamenbhalek| has joined #aichallenge 2010-10-20T13:34:05 *** gerald512k has quit IRC (Ping timeout: 265 seconds) 2010-10-20T13:34:23 hmm.. has anyone tried to compile own tcp server? 2010-10-20T13:35:19 *** McLeopold has joined #aichallenge 2010-10-20T13:36:54 my dev code is 100 ELO worse than my old code apparently 2010-10-20T13:36:56 oops. 2010-10-20T13:37:00 rabidus_: I did, but it was a while back 2010-10-20T13:37:57 i'm just getting compile errors of some basic functions... tried to install libbsd-dev packages and such.. 2010-10-20T13:38:23 currently expire.cpp hangs with function 'time' and function 'strlcpy' 2010-10-20T13:38:34 hangs? 2010-10-20T13:38:42 umm... makes an error :) 2010-10-20T13:38:59 'was not declared in this scope' 2010-10-20T13:39:04 'time' is surely just some different headers between bsd and linux 2010-10-20T13:39:41 strlcpy isn't in normal glibc because drepper is a dork 2010-10-20T13:40:15 you can convert to use strncpy, or snap the implementation of strlcpy from openbsd or elsewhere 2010-10-20T13:40:30 s/snap/snag/ 2010-10-20T13:40:36 yeah, including time.h fixed that time issue 2010-10-20T13:40:41 hmm.. ok, i'll try 2010-10-20T13:40:54 iirc, source for strlcpy is about 12 lines long :) 2010-10-20T13:41:07 *** Itkovian has joined #aichallenge 2010-10-20T13:41:36 oh! :D 2010-10-20T13:41:49 *** Itkovian has quit IRC (Remote host closed the connection) 2010-10-20T13:41:59 *** Itkovian has joined #aichallenge 2010-10-20T13:42:21 oh, that helped a lot 2010-10-20T13:42:35 thx, now its of strlcat but i can carry on by my own 2010-10-20T13:42:37 thx 2010-10-20T13:43:07 *** perestrelka has quit IRC (Read error: Connection reset by peer) 2010-10-20T13:43:30 *** ebrahim has joined #aichallenge 2010-10-20T13:43:44 *** perestrelka has joined #aichallenge 2010-10-20T13:45:04 rabidus_: http://www.openbsd.org/cgi-bin/cvsweb/~checkout~/src/lib/libc/string/strlcpy.c?rev=1.11;content-type=text%2Fplain 2010-10-20T13:45:15 i found those too 2010-10-20T13:45:21 :) 2010-10-20T13:45:27 thnx for help, it works :) 2010-10-20T13:47:27 is it only me, or are there more ppl out there spending more time on hacking servers, analyzers, war drivers, etc, than improving their bots ? 2010-10-20T13:49:19 those of us who waste time on our bots are just less vocal... ;) 2010-10-20T13:51:44 *** Titankiller has quit IRC (Ping timeout: 272 seconds) 2010-10-20T13:53:57 hi all 2010-10-20T13:54:02 *** jamenbhalek| is now known as gerald512k 2010-10-20T13:54:32 *** Titankiller has joined #aichallenge 2010-10-20T13:55:46 about ai-contect.com it is very weird, i had no game today, and some other players had 10+ 2010-10-20T13:56:59 i've seen quite a few quite a few requests about profile changes in the Introduction bit on the forum, i've posted mine there - I hope thats an alright place to do it 2010-10-20T13:57:18 *** Frontier has left #aichallenge 2010-10-20T14:00:36 yay my new bot actually works now. let's see how it does. 2010-10-20T14:01:38 yay it's about to lose 2010-10-20T14:01:49 my new bot has fighten yours many times today : but it was 0.2 2010-10-20T14:02:13 yeah that's the one i wrote before the contest and i have yet to actually beat it objectively 2010-10-20T14:02:14 my new bot is not running yet but at least I can see the light at the end of the tunnel 2010-10-20T14:02:26 *** Frontier has joined #aichallenge 2010-10-20T14:02:29 :) 2010-10-20T14:02:55 hmm, accoun just disconnected against me and he was winning 2010-10-20T14:03:00 i am actually searching some java command to write a log file, but i dont really understand how i can do that. 2010-10-20T14:03:03 ok this new bot is terrible 2010-10-20T14:03:04 *** mceier has joined #aichallenge 2010-10-20T14:03:10 terrible in the good sense? 2010-10-20T14:03:13 no. 2010-10-20T14:03:17 lol:) 2010-10-20T14:04:01 :-D 2010-10-20T14:04:12 any time you spend making good logs will pay for itself multiple times over 2010-10-20T14:04:18 gerald512k: to write to a file you mean? 2010-10-20T14:04:34 *** choas has joined #aichallenge 2010-10-20T14:04:55 yes, i want to write a log file for searching bug, because i had to look 20 matches to see if my new method works ok ... 2010-10-20T14:04:59 i loose a lot of time 2010-10-20T14:05:01 for debuging 2010-10-20T14:05:18 because what i write never works on the first time lol^^ 2010-10-20T14:05:18 still, debug > log for me :) 2010-10-20T14:05:45 look into the File class, and FileOutputStream 2010-10-20T14:05:52 you said you were learning Java with this right? 2010-10-20T14:06:08 yes i never used java before, neither other language 2010-10-20T14:06:20 yeah, have a look at those then, should get you started 2010-10-20T14:06:28 so i understand class system, clonable, comparable private etc ... but i never used file i/o 2010-10-20T14:06:47 yes thanks you i will take a look as soon as i have implemented by infrontline() method :) 2010-10-20T14:07:10 Here's the logging command line flags from my bot: 2010-10-20T14:07:20 var LoggingEnabled *bool = flag.Bool("e", false, "set to true to enable logging to stderr") 2010-10-20T14:07:29 var Errfile *string = flag.String("f", "", "specify a file to append the log to (automatically enables logging)") 2010-10-20T14:07:36 var Recvfile *string = flag.String("r", "", "specify a file to record received data to") 2010-10-20T14:07:43 var Logturn *int = flag.Int("t", -1, "specify a turn to log exclusively") 2010-10-20T14:07:51 var VP *int = flag.Int("vp", -1, "specify a planet for extra verbosity") 2010-10-20T14:07:52 *** Amanieu has left #aichallenge 2010-10-20T14:07:57 var LogOnlyPanics *bool = flag.Bool("op", false, "Only use the error logging mechanisms for panics") 2010-10-20T14:08:06 does not look like java :-D 2010-10-20T14:08:11 this is python? 2010-10-20T14:08:13 no that is golang 2010-10-20T14:08:16 ok 2010-10-20T14:08:18 golang.org 2010-10-20T14:08:20 you dont have to use ";" ? 2010-10-20T14:08:22 hahaha! 2010-10-20T14:08:26 nope 2010-10-20T14:08:36 never had to read golang before 2010-10-20T14:08:38 this is kewl :) 2010-10-20T14:08:40 what's "flag"? 2010-10-20T14:08:46 newline triggers ; in the compiler. 2010-10-20T14:08:55 flag is a package for parsing the command line 2010-10-20T14:08:59 ah 2010-10-20T14:09:10 ok i dont understand anything in your command line ^^ 2010-10-20T14:09:32 basically I declare those variables globally like that and call flag.Parse() somewhere, and everything is set 2010-10-20T14:09:42 no wonder, since you have little exp. in programming, this might seem pretty hardocre to you :-P 2010-10-20T14:09:51 lavalamp_, you think, they'll get go support running, before the compo's over ? 2010-10-20T14:10:00 I SURE HOPE SO 2010-10-20T14:10:41 I don't even want to add up all the hours I've spent on this... 2010-10-20T14:10:45 *** Spike_007 has joined #aichallenge 2010-10-20T14:10:46 Top 10 players: smatcher(3585), cfaftw(3567), DracoAter(3487), dmj111(3462), rsergio(3444), Polsky(3369), asavis(3353), savchenko(3353), EBraun(3352), deepblue(3342) 2010-10-20T14:11:25 *** hehe has quit IRC (Quit: Page closed) 2010-10-20T14:12:00 ahhh fuck, i hadn't stumbled across that in a while, handling the fleets incoming for neutral planets is a pain in the ass 2010-10-20T14:12:37 *** jesionaj has joined #aichallenge 2010-10-20T14:12:49 gerald512k, basically I have the option to log to stderr, or to a file, and I can turn on logging for only a particular turn or planet. And, if playing on the tcp server, I can record all the incoming data to a file so I can replay it through my bot later to see what it did 2010-10-20T14:12:51 *** perror has quit IRC (Quit: Bye all !) 2010-10-20T14:14:25 yes that is very cool to use 2010-10-20T14:14:41 but log file may be big for one match 2010-10-20T14:14:48 oh, and I have the option to only log crashes (panics) 2010-10-20T14:15:09 that's why I have the turn option, it doesn't print anything until it gets to that turn 2010-10-20T14:15:27 a heavy gain of time 2010-10-20T14:15:40 (only makes sense if you're replaying a game through the bot) 2010-10-20T14:15:49 yes of course i understood that :) 2010-10-20T14:15:59 wll guys gonna work on the bot a little, see u later 2010-10-20T14:16:01 :) 2010-10-20T14:16:04 good luck 2010-10-20T14:16:48 thanks :) i have not a bad ranking withouth have implemented defense, si i have good hope about this :) cya 2010-10-20T14:16:54 *** gerald512k is now known as ger512kAFK 2010-10-20T14:21:51 are there more, public tcp servers ? 2010-10-20T14:22:37 one is not enough to contain your bot's awesomeness? 2010-10-20T14:23:40 lavalamp_: I gave my bot an explain command line option that logs detailed messages for a particular turn to stderr. Looks like a lot of people have the same ideas here. 2010-10-20T14:23:43 problem is, it's not that awesome, let me win at least once in a while.. 2010-10-20T14:24:12 haha 2010-10-20T14:24:17 lol 2010-10-20T14:24:27 ;) 2010-10-20T14:24:52 *** BtbN has joined #aichallenge 2010-10-20T14:25:03 p4p4: Sometimes you get a match against RageBot on the TCP server. 2010-10-20T14:25:21 yeah, won that one! 2010-10-20T14:25:24 RageBot is actually a good test case 2010-10-20T14:25:28 Hopefully on a map with a large gap between the starting planets ^^ 2010-10-20T14:25:56 sometimes i get draw againt ragebot :( 2010-10-20T14:26:00 but ragebot got a surprisingly good elo 2010-10-20T14:27:04 it's because people don't do the "how many ships can I send and not lose my planet" math 2010-10-20T14:27:38 which is ridiculously easy to do 2010-10-20T14:27:41 damn, i just got completely ruined by the way neutral clashes are handled 2010-10-20T14:28:00 I think I need to collect some 'character' info on my opponent. RageBot would be 100% aggressive for example which in turn means I can snipe neutrals wherever I want. 2010-10-20T14:29:14 my firt bot was like ragebot. (if i had mode ships then enemy, i attack with all my planets all his weakest planet, else expand) Was 20 line and go rank 400 very fast 2010-10-20T14:29:15 antimatroid: It is easy - the difference of the two largest fleets is the remaining fleet on the planet. If it is other than 0 the planet changes ownership. 2010-10-20T14:29:17 if I get it working, my new bot will eventually compare what they did with what I thought they would do, and use that to know how well I'm predicting its moves 2010-10-20T14:30:01 *** Florian95 has quit IRC (Remote host closed the connection) 2010-10-20T14:30:06 Frontier: no, not if the largest fleet is less than the number of neutral planets 2010-10-20T14:30:07 damn... anyone else being brainwashed by steve jobs? 2010-10-20T14:30:09 and resolution is fine 2010-10-20T14:30:16 it's when i want to do other things that it complicates thing 2010-10-20T14:31:15 I had departures leaving planets at the wrong time which forced me to rewrite some things this morning 2010-10-20T14:31:18 antimatroid: I don't understand you. What does the number of neutral planets have to do with a single battle on a planet? 2010-10-20T14:31:33 neutral ships** 2010-10-20T14:31:53 oh wait, that might work 2010-10-20T14:32:16 so?... largest fleet is neutral - incoming fleet = x remaining neutral, owner changes to ... tada: neutral 2010-10-20T14:32:48 still doesn't fix me having to store more data for what i want to do though 2010-10-20T14:33:21 hm... if player a and b both send 20 ships to a 10 ship neutral, what happens? It's left with 0 ships and still neutral? 2010-10-20T14:33:33 yep 2010-10-20T14:33:42 I'm pretty sure that's what my code does 2010-10-20T14:33:56 there are basically 3 ways it could have been done, none were obviously the best choice, so a decision was made 2010-10-20T14:34:04 maybe... I just handle the present ships on the planet as an addition to the owners 'incoming' fleets on that turn and do the math with just the total ship counts for each player 2010-10-20T14:34:31 Frontier: me too 2010-10-20T14:34:48 *** Titankiller has quit IRC (Ping timeout: 272 seconds) 2010-10-20T14:34:49 * lavalamp_ checks my source... yes it gets that case right 2010-10-20T14:34:50 i worked out a solution, nevermind 2010-10-20T14:34:53 it's a bit messy but works 2010-10-20T14:35:26 *** yasith has quit IRC (Ping timeout: 272 seconds) 2010-10-20T14:36:09 *** Itkovian has quit IRC (Quit: Itkovian) 2010-10-20T14:37:36 I lost my drive a while back. Calculating how many enemy ships might potentially attack a single planet over the course of x turns is really not trivial. Not only is there growth and fleets in transit, but also neutrals beeing captured mean time or worst scenario of all: 2010-10-20T14:38:54 I might capture an enemy planet if he doesn't defend it which would change the amount of ships the enemy could send from turn n on. Then again, if he defends that will reduce the amount of ships at first (for the defense) but give him more ships after the attack. 2010-10-20T14:39:44 my first bot did that in two steps 2010-10-20T14:39:52 1. handle fleets in transit 2010-10-20T14:40:06 2. do all that complicated stuff 2010-10-20T14:40:14 So you end up either ignoring these cases, calculating an absolute maximum for each round or try to predict if the enemy defends or not and go with that. 2010-10-20T14:40:35 I missed the case where he captures a neutral first in that bot 2010-10-20T14:41:01 the other cases I think I handled pretty well 2010-10-20T14:41:05 lavalamp_: I do that too. The point is, I cannot be sure I will actually CAPTURE a certain planet. The enemy could send reinforcements in time. 2010-10-20T14:41:28 *** Itkovian has joined #aichallenge 2010-10-20T14:42:40 Frontier: Yeah, I didn't do it quite right for planets I was attacking 2010-10-20T14:42:44 Imagine my fleet is 5 turns away and he has a planet with reinfocements ready in a distance of 3 turns. Your bot would believe that it captures the enemy planet while it is 95% certain it wont due to the nearby reinforcements 2010-10-20T14:43:07 oh no I take into account nearby reinforcements 2010-10-20T14:43:22 *** ebrahim has quit IRC (Remote host closed the connection) 2010-10-20T14:43:27 oh that's cool :) 2010-10-20T14:44:12 yeah... unless those reinforcements are coming from a neutral he hasn't captured yet 2010-10-20T14:45:12 Ok, but that is the same story. I have seen that often: An advanced bot sends enough ships to capture 3 planets to just one neutral and goes from there. 2010-10-20T14:45:15 wait, neutrals aren't sending stuff, are they ? 2010-10-20T14:45:39 p4p4: Hell no. We are talking about advanced predictions. 2010-10-20T14:45:47 k. 2010-10-20T14:45:56 *** RainCT has joined #aichallenge 2010-10-20T14:46:04 p4p4: An enemy could take a neutral which then becomes a threat. 2010-10-20T14:47:04 so, its steps +2 2010-10-20T14:47:26 I hadn't even considered the possibility in my code 2010-10-20T14:47:44 a1k0n: is 0.5 your new bot? 2010-10-20T14:48:29 yeah 2010-10-20T14:48:38 looks like i finally broke into positive ELO again 2010-10-20T14:48:44 but it isn't a lot better 2010-10-20T14:48:51 it also isn't a lot different 2010-10-20T14:49:08 lavalamp_: If we do the obvious stuff in phase 1 (i.e. fleets in transit) and the complicated stuff in phase 2 we still have a bot which doesn't behave right in some situations that a human would easily realize. 2010-10-20T14:49:25 yes 2010-10-20T14:49:37 Which is why my current bot is different 2010-10-20T14:49:37 a1k0n: it just beat my dev code 2010-10-20T14:49:58 i saw, heh 2010-10-20T14:50:48 To do the heuristics right you sorta have to consider everything all at once 2010-10-20T14:51:44 lavalamp_: What do you do differently? I think about some iterative approach, but I have to be aware of possible deadlocks if I say: "I don't know if the enemy can hold planet 1, let's predict planet 2 first which could send reinforcements. Ok planet 2 depends on 1 again... deadlock" ^^ 2010-10-20T14:52:16 I'm trying to write a minimaxish bot 2010-10-20T14:53:04 *** rebelxt has left #aichallenge 2010-10-20T14:53:07 Rank 177 :( - I need to do something soon 2010-10-20T14:54:10 It's going to do the math to detect rock-paper-scissors situations and choose randomly (weighted appropriately) 2010-10-20T14:54:24 i clearly need to implement the knapsack first turn 2010-10-20T14:54:28 my bot can tell on turn 2 it's going to lose 2010-10-20T14:54:35 lol 2010-10-20T14:54:39 hahaha 2010-10-20T14:54:43 rofl 2010-10-20T14:55:01 how good is it's prediction? 2010-10-20T14:55:16 they used to say that one day the smartest ai will watch you make the first move in chess, calculate furiously, then concede 2010-10-20T14:55:40 pretty good: http://72.44.46.68/canvas?game_id=592914 2010-10-20T14:56:01 it knew it was going to lose because it did something unimaginably stupid 2010-10-20T14:56:06 ....quite 2010-10-20T14:56:21 because it thought its opponent was going to do something just as dumb. i guess it plays better with that turned off 2010-10-20T14:56:34 wow, that's quite the first move 2010-10-20T14:56:44 er... so you wrote a fail bot and wonder how it can predict its loss? ^^ 2010-10-20T14:57:10 well, it seems to have failed to hit the global maximum on the first turn, put it that way 2010-10-20T14:57:11 the problem with minimax/alphabeta is you generally have to assume the opponent will do what you would do 2010-10-20T14:58:04 i'm actually getting quite a lot of success with my current tree approach 2010-10-20T14:58:28 I have some plans to take care of that, including simulating e.g. ragebot as an opponent 2010-10-20T14:58:38 So far I choose planets only factored by enemy distance, my distance, growth, ships... stuff like that. No knapsack, no maximin. 2010-10-20T14:58:46 *** Migi32 has joined #aichallenge 2010-10-20T14:59:53 has anyone actually found a real example of rock-paper-scissors mechanics in PlanetWars? 2010-10-20T15:00:24 If I get it working, I will use information on how well I'm predicting their behavior in my prediction 2010-10-20T15:00:25 Like an example map with 3 or 4 planets that demonstrates it? 2010-10-20T15:00:25 I know I always said I was convinced they were there, just subtle and hard to find 2010-10-20T15:00:41 no, voidptr shot down my example 2010-10-20T15:00:50 but the more I think about it, the more I get convinced that there actually are no RPS mechanics 2010-10-20T15:00:53 Frontier, yes 2010-10-20T15:01:13 I can't find one example 2010-10-20T15:01:56 once I get my bot working I will have it log any it finds ^^ 2010-10-20T15:02:33 * raedwulf attacks his bot with a spoon 2010-10-20T15:03:06 raedwulf, but does your bot realise the truth? 2010-10-20T15:03:30 if so your attacks will be pretty ineffective ;) 2010-10-20T15:03:32 in planetwars, there is no truth, only estimation 2010-10-20T15:03:55 * p4p4 tries to fork his mind, ... but fails ... 2010-10-20T15:04:04 *** fluffluff has quit IRC (Quit: Leaving.) 2010-10-20T15:04:05 Migi32: i think mcleopold had one 2010-10-20T15:04:26 McLeopold, ping! 2010-10-20T15:04:29 well, it was a map where the optimal strategy was always doing nothing 2010-10-20T15:04:36 so maybe not quite rp 2010-10-20T15:04:38 * rps 2010-10-20T15:05:07 rps needs 3 possibilities I guess :) 2010-10-20T15:05:08 *** perestrelka has quit IRC (Read error: Connection reset by peer) 2010-10-20T15:05:40 no, rps needs a circular chain of options, each better than the next 2010-10-20T15:05:45 *** perestrelka has joined #aichallenge 2010-10-20T15:05:45 nontransitive dice, etc. 2010-10-20T15:05:46 *** zerd has joined #aichallenge 2010-10-20T15:06:08 my old bot just butchered my new one, definite progress here 2010-10-20T15:08:58 *** Azrathud has joined #aichallenge 2010-10-20T15:09:34 *** narnach1 has joined #aichallenge 2010-10-20T15:10:03 *** narnach1 is now known as narnach_mb 2010-10-20T15:10:46 Top 10 players: smatcher(3581), cfaftw(3562), DracoAter(3487), dmj111(3459), rsergio(3438), Polsky(3366), asavis(3356), savchenko(3350), EBraun(3345), deepblue(3339) 2010-10-20T15:13:02 Hi people i have a question for java coders, may you help me 2 minutes please? 2010-10-20T15:13:14 *** ger512kAFK is now known as gerald512kk 2010-10-20T15:13:32 shoot 2010-10-20T15:14:28 *** Accoun has quit IRC (Ping timeout: 276 seconds) 2010-10-20T15:14:41 Supposing i have 2+ planet. For each planet, i want to create a list ordered by nearest to farest. so i have pw.MyPlanets().size() list that lenght pw.MyPlanets().size()-1. 2010-10-20T15:15:08 may i use a different type then List to do this ? for instance HashMap would not be beter? 2010-10-20T15:15:24 ( ordered by nearest to farest friend planet i mean ) 2010-10-20T15:15:50 hashmap is not ordered 2010-10-20T15:15:56 I don't know java (well, anyway) but list sounds appropriate for that 2010-10-20T15:16:32 use a SortedList 2010-10-20T15:16:33 nested lists is probably better 2010-10-20T15:17:03 so you have eg, closest[0] is an order list of planets by closeness 2010-10-20T15:17:08 for the 0th planet 2010-10-20T15:17:11 Migi32: :P 2010-10-20T15:17:27 on turn 1 I make a 2d array listing every planet in order from every planet 2010-10-20T15:17:43 *** fluffluff has joined #aichallenge 2010-10-20T15:17:52 then to get friendly planets I just go through the list asking if each one is friendly 2010-10-20T15:18:01 as do i 2010-10-20T15:18:18 it's much more flexible because you can ask if it's friendly at any particular time 2010-10-20T15:18:31 yes i see. The probleme i have is i dont know how to name the list linked to one planet. i need to declare pw.MyPlanets().size() list si i am stuck 2010-10-20T15:18:47 ok i see lavalamp this semm very good 2010-10-20T15:18:49 you are on your 3rd minute! 2010-10-20T15:18:53 ;) 2010-10-20T15:19:05 haha sory so i shut up :( 2010-10-20T15:19:12 haha no, go on 2010-10-20T15:19:17 lol:) 2010-10-20T15:19:23 i was just joking 2010-10-20T15:19:47 do you know, why (after copying starting java package) into NetBeans as new project, there are missing package statements at the beggining of each *.java file ??? 2010-10-20T15:19:59 i dont wanna flood the chanel but i am really stuck. I dont know how to create pw.Myplanets().size() lists :( 2010-10-20T15:20:10 sorry I was distracted. So let me rephrase my question as "has anyone found a situation in PlanetWars where the best move was a "mixed strategy", which involves probabilities (like in rock-paper-scissors, where the best move is pick any with 33% probability each)?" 2010-10-20T15:20:12 it's not exactly NB problem, but still, that statement is required as far as I know 2010-10-20T15:21:09 so RPS or RPS-like things 2010-10-20T15:21:24 Spike_007: it's only required for actual packages, i think 2010-10-20T15:21:31 which is what netbeans is assuming 2010-10-20T15:21:40 might be rock paper scissors lizard spock :P 2010-10-20T15:22:01 Sorry, gerald, don't know java well enough to comment... if you google "java list" or something like that I bet you'd find something 2010-10-20T15:22:31 gerald512kk: ArrayList will let you do adjustable size 2010-10-20T15:22:36 oh, thx :) I've never worked with java via cmd line (shame on me, I know) so it took my by surprise ... 2010-10-20T15:22:50 ok yes i gonna use it tyvm! 2010-10-20T15:23:08 i forget if you can do arrays like that. String[] = new String[pw.MyPlanets().size()]; 2010-10-20T15:23:14 Migi32: if/when my bot finds one I will be posting it somewhere :) 2010-10-20T15:24:12 *** OrgChangePllss has joined #aichallenge 2010-10-20T15:24:13 *** cfaftw has quit IRC (Quit: Leaving) 2010-10-20T15:24:25 any word on when organizations will be updated? 2010-10-20T15:26:45 patience is a virtue and all, 2+ weeks is starting to be noticable though :( 2010-10-20T15:27:41 *** Accoun has joined #aichallenge 2010-10-20T15:29:38 *** Azrathud has quit IRC (Ping timeout: 255 seconds) 2010-10-20T15:31:30 hmmm, anyone can give me a hint how to integrate sources into NetBeans ... ? 2010-10-20T15:32:08 Migi32: I think McLeopold had found one a while back 2010-10-20T15:32:54 http://pastebin.com/2CSkuKne 2010-10-20T15:33:14 *** fry_ has quit IRC (Ping timeout: 255 seconds) 2010-10-20T15:34:09 Migi32: http://contestbot.hypertriangle.com/channellogs/%23aichallenge/%23aichallenge.2010-10-15.log 2010-10-20T15:34:33 the converstation starts around 12:19 2010-10-20T15:35:14 thanks 2010-10-20T15:35:19 np 2010-10-20T15:36:41 sweet, I will use that as a test case for my bot 2010-10-20T15:37:14 heh, good luck 2010-10-20T15:38:58 preeza: ah ok but it wasn't perfect 2010-10-20T15:38:58 attack beats neutral, neutral beats hold, hold sucks... 2010-10-20T15:39:03 That's why I need a neutral, so that holding can beat an attack. 2010-10-20T15:39:33 hold will never beat attack 2010-10-20T15:39:37 only draw 2010-10-20T15:39:57 which makes the perfect strategy a RageBot 2010-10-20T15:44:05 can anyone help me woth integration sources into NetBeans please ? 2010-10-20T15:44:05 perhaps there is a RPS like this: attack beats expand fast, expand fast beats expand slow, expand slow beats attack? 2010-10-20T15:44:06 but that'd require at least 6 planets and is probably too complex for humans to analyse 2010-10-20T15:44:06 or 5, but still 2010-10-20T15:44:06 any more news on the almighty cloud? 2010-10-20T15:44:07 or rather its lack of being almighty :P 2010-10-20T15:44:07 Spike_007, what programming language? 2010-10-20T15:44:07 Migi32: yea... probably 2010-10-20T15:44:07 *** boegel has joined #aichallenge 2010-10-20T15:44:08 Ming32: He uses Java with a 98% probability, and C++ with 1% and PHP with 1% 2010-10-20T15:44:19 Frontier, I use NetBeans + C++ 2010-10-20T15:44:30 Migi32 -> Java 2010-10-20T15:44:35 ok 2010-10-20T15:45:19 Migi32: see, for you I would have calculated different probabilities of course. It is all about estimating from other peoples forst moves. 2010-10-20T15:46:03 well for c++ I just imported the sources as a regular project because it just requires command-line input and output (std::cin and std::cout) 2010-10-20T15:46:08 I've some error alerts in NetBeans, probably because 'package ...' statement is missing in the beggining of each file (or I've somehow screwed integration) 2010-10-20T15:46:12 If he was familiar with C++ in NetBeans that would raise the probability that he ever had to import sources into the IDE thus reducing the probability of his question. It's like RageBot or not RageBot. 2010-10-20T15:46:41 Spike_007, java needs no "package ..." afaik 2010-10-20T15:46:52 Spike_007: That is ok. The usage of the default package is stongly discouraged, you can disable the warnings though 2010-10-20T15:47:01 agree, still, no idea where errors are from .. 2010-10-20T15:47:17 what's the error? 2010-10-20T15:47:33 Migi32: I have another scenario you might be interested in. 2010-10-20T15:48:01 pfff, screw good practice if you're the only one who's ever going to be using the code 2010-10-20T15:48:06 oh hi McLeopold :) 2010-10-20T15:48:13 shoot 2010-10-20T15:48:29 let me dig it up... 2010-10-20T15:48:31 with a 33% probability, shoot the guy next to me otherwise 2010-10-20T15:48:55 i.e: DualBot.java yells cannot find symbol method Prodction(int) location PlanetWarrs; ... I've checked, it is there, still the error occures 2010-10-20T15:49:03 !dice 1d3 2010-10-20T15:49:03 Frontier: 2 2010-10-20T15:49:35 phew 2010-10-20T15:49:41 P 10.0 10.0 2 90 5 2010-10-20T15:49:41 P 5.67 7.5 1 34 2 2010-10-20T15:49:42 P 4.0 12.0 1 66 3 2010-10-20T15:49:53 1= left, 2= you, 3= right? 2010-10-20T15:50:03 uh oh... 2010-10-20T15:50:10 !roulette 2010-10-20T15:50:10 Migi32: *click* 2010-10-20T15:50:14 naah 2010-10-20T15:50:15 DEBUG - Distance Table: 2010-10-20T15:50:16 DEBUG - 0 1 2 2010-10-20T15:50:16 DEBUG - 0 0 5 7 2010-10-20T15:50:16 DEBUG - 1 0 5 2010-10-20T15:50:16 DEBUG - 2 0 2010-10-20T15:50:57 !roulette 2010-10-20T15:50:58 *** rabidus_ was kicked by contestbot (BANG!) 2010-10-20T15:50:59 * contestbot reloads and spins the chambers. 2010-10-20T15:51:00 there are holes in your table 2010-10-20T15:51:07 *** rabidus_ has joined #aichallenge 2010-10-20T15:51:09 :< 2010-10-20T15:51:15 McLeopold, that's a brilliant idea for displacing distances, I never considered a table 2010-10-20T15:51:36 McLeopold: starting planets are 0 and 1? 2010-10-20T15:51:48 It's not a normal start 2010-10-20T15:52:01 It's just a scenario 2010-10-20T15:52:18 ok but 2 is neutral? 2010-10-20T15:52:23 McLeopold, a triangular matrix would make sense but your distance table doesn't 2010-10-20T15:52:23 ah no ok 2010-10-20T15:52:30 no neutrals 2010-10-20T15:52:43 i missed the specs before 2010-10-20T15:52:43 Player 1 has two planets, and technically enough ships and growth to hold, but only if he can guess player 2's first turn. 2010-10-20T15:52:53 very cool 2010-10-20T15:53:01 McLeopold, what if he attacks player 2? 2010-10-20T15:53:19 *** bhasker has joined #aichallenge 2010-10-20T15:53:36 *** ratatata has joined #aichallenge 2010-10-20T15:53:42 Player 2 has enough to hold as well, in fact, if player 2 holds, he would win with more ships. 2010-10-20T15:54:18 then how can player 1 win? 2010-10-20T15:54:49 Migi32: where are the holes? 2010-10-20T15:54:56 Migi32: I agree, a triangular matrix makes sense :) 2010-10-20T15:55:10 McLeopold, I see this: 2010-10-20T15:55:14 x 0 1 2 2010-10-20T15:55:17 0 0 5 7 2010-10-20T15:55:19 1 x 0 5 2010-10-20T15:55:21 2 x x 0 2010-10-20T15:55:23 x is a "hole" 2010-10-20T15:55:40 the x's are redundant info 2010-10-20T15:55:59 oh wait the 0 1 2 is "planet 0", "planet 1", "planet 2" XD 2010-10-20T15:56:02 I am dumb :P 2010-10-20T15:56:10 ^^ 2010-10-20T15:56:30 *** bhasker__ has quit IRC (Ping timeout: 272 seconds) 2010-10-20T15:56:31 Migi32: me too, now i see it is a triangular matrix 2010-10-20T15:57:11 Thanks for the case McLeopold... I am very excited to run that through my bot when I get it working 2010-10-20T15:57:32 but my question stands, McLeopold, how can player 1 win? 2010-10-20T15:57:38 McLeopold: is attacking planet 2 an option for player 1 though? 2010-10-20T15:57:53 player 2, sorry 2010-10-20T15:58:00 Image that scenario on the map twice with the places reversed. If you solve your side before the opponent does, you capture that side and send ships to the other. 2010-10-20T15:58:15 phreeze: everything is an option 2010-10-20T15:58:45 but can he capture it? i think not? 2010-10-20T15:58:52 no 2010-10-20T15:59:02 unless player 2 sends everything 2010-10-20T16:00:33 McLeopold, you can also wait until player 2 attacks, then counter-attack his main 2010-10-20T16:00:47 actually...I think I messed up the ship counts... 2010-10-20T16:02:55 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-20T16:03:02 P 10.0 10.0 2 100 5 2010-10-20T16:03:02 P 5.67 7.5 1 34 2 2010-10-20T16:03:02 P 4.0 12.0 1 56 3 2010-10-20T16:03:16 I had been messing with the scenario... 2010-10-20T16:03:25 do you use a map editor for this? 2010-10-20T16:03:37 I use a text editor 2010-10-20T16:03:40 And a whiteboard 2010-10-20T16:03:45 5.67 is a weird number 2010-10-20T16:03:49 * javagamer needs a whiteboard 2010-10-20T16:03:55 2.5 * sqrt(3) 2010-10-20T16:03:59 ish 2010-10-20T16:04:03 :P 2010-10-20T16:04:11 or something like that 2010-10-20T16:05:27 Migi32: okay, with the update numbers, you can't wait because it will be too late. 2010-10-20T16:05:28 Whiteboard: http://admin.home.nerdiii.operaunite.com/whiteboard/ 2010-10-20T16:05:45 sorry, http://home.nerdiii.operaunite.com/whiteboard/ 2010-10-20T16:05:53 *** p4p4 has joined #aichallenge 2010-10-20T16:06:17 same link 2010-10-20T16:06:25 Frontier: is this your whiteboard? 2010-10-20T16:06:44 Yes it is, but 'sigh' used it some time ago 2010-10-20T16:07:43 awesome web app 2010-10-20T16:07:43 oooh cool 2010-10-20T16:08:11 the diagram helps a lot 2010-10-20T16:08:30 indeed. McLeopold do you use a drawing pad? If not, do you play FPS games? :P 2010-10-20T16:08:53 No drawing pad. 2010-10-20T16:09:04 Is that a complement to my drawing skill? 2010-10-20T16:09:15 ...except for the random scribbling... 2010-10-20T16:09:27 I'm impressed by your mouse drawing skills 2010-10-20T16:09:38 McLeopold, yes it is. You draw fast and nice 2010-10-20T16:09:44 for a mouse anyway 2010-10-20T16:09:50 Hmm.. I turn of mouse acceleration. Maybe that helps. 2010-10-20T16:09:57 off 2010-10-20T16:10:28 well the drawing has been f'd up... 2010-10-20T16:10:46 Top 10 players: smatcher(3570), cfaftw(3550), dmj111(3471), rsergio(3470), DracoAter(3463), Polsky(3353), asavis(3342), savchenko(3336), DocBot(3331), deepblue(3328) 2010-10-20T16:11:53 In C++ if I want a function to return two ints is the only way to make a struct/class w/ two ints? 2010-10-20T16:12:21 In python you return a tuple. Switch languages... 2010-10-20T16:12:29 :) 2010-10-20T16:12:47 McLeopold, I am missing that about python very much atm 2010-10-20T16:12:48 javagamer: You can place the return values in two int& parameters 2010-10-20T16:13:03 yeah, pass pointers, mess with stuff directly 2010-10-20T16:13:04 *** p4p4 has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2010-10-20T16:13:07 javagamer, std::pair f(...) { ... } 2010-10-20T16:13:26 std::pair 2010-10-20T16:13:28 Migi32, just what I needed, thanks! 2010-10-20T16:13:30 ah, beat me to it 2010-10-20T16:13:38 :P 2010-10-20T16:13:55 but if the pair has a real meaning I'd use a struct anyway 2010-10-20T16:13:59 makes nicer code 2010-10-20T16:14:14 golang lets you return multiple values 2010-10-20T16:14:15 ...but the overhead of creating an object of type pair! 2010-10-20T16:14:33 Just want to return the owner and number of ships a planet will have from my prediction code 2010-10-20T16:14:46 *** JensTi has joined #aichallenge 2010-10-20T16:14:47 Frontier, it's no more than allocating two ints 2010-10-20T16:14:58 you might as well make a struct for that, you'll be doing it all over the place 2010-10-20T16:15:16 don't worry about speed in c++ (but don't go around copying std::vectors all the time like the starter packages do) 2010-10-20T16:15:48 honestly I use a struct for that purpose, too. even worse i have a struct that consists of a turn number AND a struct of owner and ship count 2010-10-20T16:15:51 typedef std::pair YourPair 2010-10-20T16:16:20 or FutureInt or whatever it means 2010-10-20T16:16:26 yeah, it's just that "first" and "second" are a lot more confusing than "ships" and "owner" 2010-10-20T16:16:45 so make a struct :P 2010-10-20T16:17:25 Migi32: That still happens to me a lot. I often write functions that return a vector& but I store it in a new vector object copying all the elements just because I forgot that ampers-and 2010-10-20T16:18:21 in C the usual thing to do would be to save return values for status codes and pass actual results via pointer parameters. not saying it's pretty, but it's really the "best" way to go about it in C most of the time. C++ has a few more options, but whether they are better, i don't know 2010-10-20T16:18:50 wait, it copies the elements by default? ew 2010-10-20T16:19:03 Frontier, I think you're using getters and setters too much ;) 2010-10-20T16:19:17 i guess vectors aren't immutable so it makes sense but... yuck :) 2010-10-20T16:19:45 yeah don't misuse that... 2010-10-20T16:21:01 i haven't worked on my bot in like a week 2010-10-20T16:21:24 Considering I probably will use this all over I guess I'll make a struct, but good to know std::pair exists 2010-10-20T16:21:50 Migi32: I like how you can return a const vector&. That allows you to get right at the source of an objects data without the danger of changing anything without the object knowing 2010-10-20T16:21:50 jmcarthur, sorry I keep defending c++ but the & that Frontier talked about changes that default behavior 2010-10-20T16:21:56 Migi32: i know 2010-10-20T16:22:08 but only because it's pass by reference 2010-10-20T16:22:37 my intuition would have been that the underlying array shouldn't be copied anyway, but that's because my mind makes the incorrect assumption that std::vector is immutable 2010-10-20T16:22:45 but I definitely agree that in c++ you can end up with hidden, sneaky copies 2010-10-20T16:22:46 c++ makes it easy to shoot yourself in the foot in a lot of ways I appreciate more since starting to write things in go 2010-10-20T16:23:30 python lets you shoot yourself in the foot, but only with rubber bullets 2010-10-20T16:23:30 I really want to check out go 2010-10-20T16:23:37 McLeopold: :D 2010-10-20T16:23:41 I read about go's features and I liked almost everything 2010-10-20T16:23:42 * jmcarthur goes back to his haskell projects 2010-10-20T16:23:56 * boegel joins jmc 2010-10-20T16:24:02 in C you can shoot yourself in the foot, but you have to disable all the safeties 2010-10-20T16:24:04 * boegel joins jmcarthur 2010-10-20T16:24:14 I love it, but it needs some sort of generics before it will be mainstream 2010-10-20T16:24:27 that's basically my only complaint 2010-10-20T16:24:27 by which i mean, it's only a little harder to do than in c++ 2010-10-20T16:24:48 i kind of disagree with the direction of that metaphor 2010-10-20T16:24:49 yeah you can't accidentally copy a vector in C ;) 2010-10-20T16:25:00 really? most people i talk to about c++ or java, templates/generics are the most depised things 2010-10-20T16:25:20 in C you can take aim at whatever you want and fire directly at the target, including your foot. in C++ you don't necessarily know what you're aiming at 2010-10-20T16:25:23 lavalamp_, no generics? :( 2010-10-20T16:25:23 you can't accidentally copy an array in C 2010-10-20T16:25:35 if you pass it as an argument, it returns a pointer 2010-10-20T16:25:36 Well, I have the same 4 lines of code copied for a bunch of different structs, it gets old 2010-10-20T16:25:43 er, it passes a pointer 2010-10-20T16:25:45 Migi32, not yet, it's on their lits 2010-10-20T16:25:48 *list 2010-10-20T16:26:04 of course, this means you can do silly things like declare an array in a function and attempt to return it 2010-10-20T16:26:19 this is one way to shoot yourself in the foot 2010-10-20T16:26:29 is anyone here using Java + NetBeans ? 2010-10-20T16:27:32 Zannick: you can copy an array in C if it's a struct member, but that's presumably what you want to happen anyway if it's a struct member 2010-10-20T16:27:46 Zannick, I don't know about Java but templates in C++ are obscenely complicated, the go authors will not do it that way 2010-10-20T16:27:52 that is, if you pass the struct as a parameter 2010-10-20T16:28:06 oh right, structs will be passed in full 2010-10-20T16:28:06 also garbage collection. I'm still a big fan of RAII and manually deleting things 2010-10-20T16:28:19 but that's only if the entire array is in the struct, and not a pointer 2010-10-20T16:28:26 and in fact, you usually want to pass a large struct by reference, not by value 2010-10-20T16:28:32 i use pointers because my arrays aren't constant-length 2010-10-20T16:28:33 whether it has arrays or not 2010-10-20T16:28:33 Zannick, yes only fixed size arrays 2010-10-20T16:28:41 yes 2010-10-20T16:28:42 Migi32: I used to agree, but garbage collection makes a lot of stuff really convienient 2010-10-20T16:28:59 lavalamp_: java is pretty similar 2010-10-20T16:29:13 garbage collection can actually have less overhead than manual allocations/frees sometimes, especially if the allocates in regions 2010-10-20T16:29:24 there are some good examples of terrible generics in java 2010-10-20T16:29:27 *if the gc allocates in regions 2010-10-20T16:29:28 in terms of how you have to specify things 2010-10-20T16:29:39 jmcarthur, I know, it's not about the speed. It's about predictability. 2010-10-20T16:29:48 Accidently copying a C++ vector is even worse considering the copy-cascade you set of for every object contained in it and every object member of these objects and so on. If the copy constructors allocate file handles or lots of memory it's becomming really creepy when you see where you end up in a debugger ^^. 2010-10-20T16:29:55 Yeah I think java didn't have generics when I learned it, or I never got to them 2010-10-20T16:29:59 public static T ... 2010-10-20T16:30:39 Migi32: explicit or inferred regions are predictable gc, but it's not really in common language runtimes 2010-10-20T16:30:46 jmcarthur: that's why i like to use the stack. moving the stack pointer up and down is pretty low overhead 2010-10-20T16:30:54 Migi32: it's only the mark-and-sweep style gc that's less predictable 2010-10-20T16:31:38 Best thing is: If you haven't defined a copy-constructor for an object containing dynamically allocated memory in a pointer you have the same pointer in two objects and when you attempt to free the pointer in both objects you get an exception for a duplicate 'delete'. 2010-10-20T16:31:46 Zannick: sharing statically allocated data is even faster :) 2010-10-20T16:31:49 jmcarthur, and the ways it can screw up your profiling. I was profiling something once that ended up having a garbage collection cycle right in it 2010-10-20T16:32:06 Migi32: now that depends on the language and vm 2010-10-20T16:32:11 true 2010-10-20T16:32:19 *** sigh has joined #aichallenge 2010-10-20T16:32:22 *** OrgChangePllss has quit IRC (Quit: Page closed) 2010-10-20T16:32:42 In go you can request the GC run, if you want it to be more predictable 2010-10-20T16:32:49 oooh, statically allocated :) 2010-10-20T16:32:50 yeah most gc languages allow that 2010-10-20T16:32:55 well, that's never gc'ed, so 2010-10-20T16:32:57 :P 2010-10-20T16:33:01 ok that makes sense 2010-10-20T16:33:05 but yes. Stack > gc and manually managed memory. If only we had infinite stack space 2010-10-20T16:33:34 Zannick: okay then, heap allocated at init time. you can still continue to share that and beat the stack for speed overall 2010-10-20T16:33:41 Zannick: for certain kinds of programming, of course 2010-10-20T16:33:52 well, with no gc'ing the memory, certainly 2010-10-20T16:34:12 GC is poorly defined in this conversation 2010-10-20T16:34:15 the cool thing about C++ is you can actually override "new" and write any memory management scheme you want 2010-10-20T16:34:17 the only caveat with using the stack is the shoot-yourself-in-the-foot logic i mentioned earlier 2010-10-20T16:34:21 GC doesn't mean that the memory has to be scanned all the time 2010-10-20T16:34:40 i mean it won't be deleted or removed until the program ends 2010-10-20T16:34:50 Migi32: Have you done that? 2010-10-20T16:35:17 I have done that (custom memory management) 2010-10-20T16:35:23 it's almost always been a terrible idea 2010-10-20T16:35:24 Migi32: you can do it in haskell too :P 2010-10-20T16:35:41 one of my systems projects from college was an implementation of malloc 2010-10-20T16:35:42 at least if you use the FFI like C 2010-10-20T16:35:44 Frontier, umm no I am not 1337 enough :P 2010-10-20T16:36:57 Migi32: of course in haskell "new" isn't really a first class thing, so i kinda cheated 2010-10-20T16:37:01 err 2010-10-20T16:37:06 it's not language defined, that is 2010-10-20T16:37:13 it's actually more first class then in C++ in that sense 2010-10-20T16:37:25 jmcarthur, do you even have variables to allocate in haskell? I thought it was a functional language 2010-10-20T16:37:25 *than 2010-10-20T16:37:26 btw... can I have RTTI in C++ for some objects? I'd like to scan objects and classes for their members and do some magic with the information. 2010-10-20T16:37:48 Migi32: you can basically write C in haskell. it's usually used for binding to C libraries, but it can also be used to write general low level code 2010-10-20T16:38:01 Frontier: you mean as opposed to all objects? 2010-10-20T16:38:09 I don't think so 2010-10-20T16:38:30 lavalamp_: yeah, I don't want to bloat the code for just one class of ...er .. classes 2010-10-20T16:38:35 Migi32: your "C" code in haskell will still have a bit more overhead than raw C would though, of course 2010-10-20T16:38:55 I think it's on or off, and who knows how their server is set up 2010-10-20T16:40:04 lavalamp_: It is set up to compile with -O3 and -funroll-loops and links to the math library 2010-10-20T16:40:29 Migi32: but for example, i've used the haskell FFI to manage resources on the GPU in a very low level way and then expose purely functional objects for the rest of the library that have impure implementations 2010-10-20T16:40:31 Frontier: good luck then 2010-10-20T16:40:56 At least in MSVC, you have to turn on RTTI, it's not on by default 2010-10-20T16:41:35 Same goes for gcc. 2010-10-20T16:41:41 thought so 2010-10-20T16:41:52 Don't recommend depending on it, then :) 2010-10-20T16:42:28 *** Bobng has quit IRC (Quit: Leaving) 2010-10-20T16:43:02 *** Bobng has joined #aichallenge 2010-10-20T16:43:30 *** JensTi1 has joined #aichallenge 2010-10-20T16:45:11 I have a few parameters in my bot which I'd like to tune (i.e. via command-line) and right now I have to add a lot of boiler plate code to my bot object. how many tunable parameters are there, what are their names, what are their addresses inside a particular bot object (in case i have two bots competing), ... some RTTI would reduce the overhead. 2010-10-20T16:45:58 *** JensTi has quit IRC (Ping timeout: 276 seconds) 2010-10-20T16:51:51 this is something I've really been wondering about: we can basically execute arbitrary code on the server. Can't that be exploited? 2010-10-20T16:52:50 I thought they used a sandbox 2010-10-20T16:52:57 and if not why do most web servers provide only hosting of php files and no executables? 2010-10-20T16:53:22 supposedly they have safeguards against things 2010-10-20T16:53:38 most web servers don't know how to sandbox things appropriately 2010-10-20T16:54:17 Also, the rules show they are aware and concerned of the possibility of somebody exploiting something. 2010-10-20T16:54:26 but by most I meant all those that I know of. Well for the "no executables" part at least (not the "only php") 2010-10-20T16:55:22 anything that lets you host cgi is letting you run executables 2010-10-20T16:56:09 also, if my bot is allowed to read files can't it browse to other people's bots and read their code? 2010-10-20T16:56:32 what good will that do it? 2010-10-20T16:56:34 only if you could find it 2010-10-20T16:57:46 I wouldn't mind seeing dmj's code :P 2010-10-20T16:58:02 oh wait it can't send me that code after it read it 2010-10-20T16:58:34 have it write the source code out in ASCII by fleet counts :) 2010-10-20T16:58:47 haha, nice! :P 2010-10-20T16:58:50 Migi32: on the cloud servers, UDP 'connections' are not sandboxed 2010-10-20T16:59:12 so just read every file you find and send it to your pc 2010-10-20T16:59:29 exactly 2010-10-20T17:00:53 from the rules: Any attempt to disrupt the normal operation of the contest software or the contest servers will result in the immediate involvement of law enforcement officials. Our policy is to always prosecute. 2010-10-20T17:01:06 I wouldn't risk it ^^ 2010-10-20T17:01:53 It wouldn't disrupt the server, but it was a bad idea to ... wake sleeping dogs 2010-10-20T17:02:35 Frontier, like any script kiddie will read this long boring conversation ^^ 2010-10-20T17:02:54 at least, I'm assuming the guys who do these kind of evil things are script kiddies 2010-10-20T17:03:05 otherwise they'd be smart enough to write their own bot, wouldn't they :P 2010-10-20T17:03:55 I'd be interested in seeing a diff of each bot with every other bot on final submission day 2010-10-20T17:04:01 just for statistic purposes 2010-10-20T17:05:13 Well when I was on a vocational school they had some guys write a soccer wm web-app. And a cracker in my class found out it wasn't safe against SQL injections. When I heard that I asked him what that is and the day after I wrote an app that reconstructs the database structure and content from the generated error messages by the web server. 2010-10-20T17:05:56 The next day I used the unused account of a secretary and changed my bets after the matches had been played. 2010-10-20T17:07:00 Frontier, you didn't dare ;) 2010-10-20T17:07:04 MySQL allowed for multiple INSERTS in a row, but I needed one dummy insert first so I used the unused account for that. 2010-10-20T17:08:03 *** delt0r___ has quit IRC (Read error: Operation timed out) 2010-10-20T17:08:21 Migi32: Hell, I felt like the god of the known universe. It's not the script kiddies I'm worried about, its ppl like me. 2010-10-20T17:08:22 *** delt0r___ has joined #aichallenge 2010-10-20T17:08:38 the kid next to me in our cisco networking class put a copy of our highschool's enrollment data online 2010-10-20T17:08:42 never did find out how he got it 2010-10-20T17:09:19 my high school had a web server and I ran a Nessus scan against it. Turned out they had a "test" MySQL database with account-name "test" and password "" 2010-10-20T17:09:20 I did this: 2010-10-20T17:09:26 CREATE TABLE test 2010-10-20T17:09:50 (TEXT 'message') (or something) 2010-10-20T17:09:56 INSERT INTO test VALUES "hi" 2010-10-20T17:09:57 And in the unix classes when it comes to 'ssh into other ppls computers' it always ended up with some jerk altering the target run level to 6 2010-10-20T17:10:05 INSERT INTO test (SELECT * FROM test) 2010-10-20T17:10:06 INSERT INTO test (SELECT * FROM test) 2010-10-20T17:10:06 INSERT INTO test (SELECT * FROM test) 2010-10-20T17:10:09 and so on :P 2010-10-20T17:10:18 ...20 computers constantly rebooting ^^ 2010-10-20T17:10:47 Migi32: ...lol... 2010-10-20T17:10:48 Top 10 players: cfaftw(3607), smatcher(3579), DracoAter(3483), dmj111(3462), rsergio(3442), Polsky(3368), asavis(3360), EBraun(3339), deepblue(3337), savchenko(3335) 2010-10-20T17:11:41 I could have continued and caused the whole system to run out of space but I only continued until like 5 GB of "hi" messages, then deleted all but one message saying something funny 2010-10-20T17:11:47 don't remember what it was though :( 2010-10-20T17:12:17 Also cool: Play with hard links under linux and change ownership to someone else so their contingent of disk space gets used up without them actually having any of their files in their home directory 2010-10-20T17:13:11 maybe I could write a C++ bot that downloaded and installed go on the server... 2010-10-20T17:13:44 lavalamp_, use python and write "import go" 2010-10-20T17:13:46 :P 2010-10-20T17:14:05 lavalamp_: Hey! You can install it into the sandbox environment and go from there. your c++ bot would act as a starter. but wait... downloading and installing takes longer than a sec right? 2010-10-20T17:14:23 python >= go? 2010-10-20T17:14:27 I'd have to have a bot good enough to last long enough to accomplish all this 2010-10-20T17:14:39 just let it try until it plays ragebot 2010-10-20T17:14:40 Yeah, I'm not sure how that would work 2010-10-20T17:14:40 then sit 2010-10-20T17:14:53 (importing go from python) 2010-10-20T17:15:31 Frontier, no I was refering xkcd comics like this: http://xkcd.com/353/ 2010-10-20T17:15:36 Or... write a C++ bot that CONTAINS go as a data section, like a self-extracting zip file. 2010-10-20T17:15:36 write a c++ UDP server that connects to your local machine and executes your go bot when it's run... 2010-10-20T17:15:54 JamesMG: best idea ever! 2010-10-20T17:16:25 Awesome! 2010-10-20T17:17:12 The amount of "What...?" if you open sourced that after the comp would be hilarious 2010-10-20T17:18:16 maybe I could disassemble a compiled version of my bot and include that as inline assembly in a c bot 2010-10-20T17:19:02 lavalamp_, or you could just compile your bot and execute it from inside C 2010-10-20T17:19:10 I think that's allowed actually 2010-10-20T17:19:36 Migi32: thanks, maybe I will do that if they never get go running 2010-10-20T17:19:57 !seen j3camero 2010-10-20T17:19:57 Migi32: j3camero was last seen in #aichallenge 2 weeks, 1 day, 19 hours, and 11 seconds ago: amstan: yes, but no names, haha. 2010-10-20T17:20:02 hell, I will do that when I get home and see how my last bot does 2010-10-20T17:20:14 lavalamp_ google for "obfuscation" 2010-10-20T17:20:18 lavalamp_, that's rather likely actually 2010-10-20T17:20:20 yay! space pirate ship: http://torrentfreak.com/pirate-parties-plan-to-shoot-torrent-site-into-orbit-101020/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed:+Torrentfreak+(Torrentfreak)&utm_content=Google+Feedfetcher 2010-10-20T17:20:48 there is lots of code that can make your program non understandable 2010-10-20T17:20:52 Migi32: la la la I'm not listening 2010-10-20T17:20:55 amstan, lol that is awesome 2010-10-20T17:21:33 :P 2010-10-20T17:21:50 amstan, is that allowed, if I write a c++ "bot" that just executes my precompiled go bot? 2010-10-20T17:22:03 lavalamp_: idk if you can 2010-10-20T17:22:16 you might be blocked by the sandbox 2010-10-20T17:22:18 can you unban him if he gets banned for trying? :) 2010-10-20T17:22:24 I'm wondering if they'd actually shoot that ship out of space :P 2010-10-20T17:22:41 Migi32, on what grounds? 2010-10-20T17:22:42 it's probably allowed, especially since you're just trying it as a workaround for nonexistant go support 2010-10-20T17:22:50 the rules say reading from files you supply is okay, but not writing 2010-10-20T17:22:55 shooting a spacecraft equals war 2010-10-20T17:23:00 davidd, copyright infringement 2010-10-20T17:23:10 there's no rules that i'm aware of that prevent you from doing this 2010-10-20T17:23:23 but i'm wondering if the sandbox allows you to execute other processes like that 2010-10-20T17:23:24 I'd include go source, too 2010-10-20T17:23:26 if it works I may do it to sidestep the C# compiler issue actually 2010-10-20T17:23:27 copyright infringment isnt a crime, its civil disobedience 2010-10-20T17:23:29 *** sigh has quit IRC (Remote host closed the connection) 2010-10-20T17:23:34 cool, I'll try it after I get this bot working 2010-10-20T17:23:34 we are aware of one person who has been submitting a lua interpreter and his bot in lua 2010-10-20T17:23:42 civil court in most countries 2010-10-20T17:23:57 Zannick: yes, but it's not an exernal executable that had to be executed 2010-10-20T17:24:13 lavalamp_: how are you going to launch it? fork won't work 2010-10-20T17:24:14 true 2010-10-20T17:24:23 *** nullkuhl has joined #aichallenge 2010-10-20T17:24:30 amstan: system? 2010-10-20T17:24:37 execvp or something 2010-10-20T17:24:40 a lot of the syscalls are not allowed 2010-10-20T17:24:54 hmm 2010-10-20T17:25:03 i'm not sure how that's prevented 2010-10-20T17:25:05 lavalamp_: look up the sandbox stuff 2010-10-20T17:25:12 lavalamp_: you might not be able to 2010-10-20T17:25:22 Where is the sandbox stuff? 2010-10-20T17:25:28 @where repo 2010-10-20T17:25:28 amstan: repo = http://code.google.com/p/ai-contest/ 2010-10-20T17:25:34 in there somewhere 2010-10-20T17:25:35 thx 2010-10-20T17:25:36 lol 2010-10-20T17:25:42 i would think exec would work/be allowable 2010-10-20T17:25:48 since it replaces the running process 2010-10-20T17:26:00 but then you have be careful about exploits 2010-10-20T17:26:50 if all fails you can read the file to memory then in assembly write "call (address of main function inside that memory block" 2010-10-20T17:27:06 hahaha ew 2010-10-20T17:27:09 but yeah 2010-10-20T17:27:20 good luck 2010-10-20T17:27:25 I might need it 2010-10-20T17:27:30 i' 2010-10-20T17:27:31 really not that hard I think 2010-10-20T17:27:32 ow i made a mistake 2010-10-20T17:27:35 link not correct 2010-10-20T17:27:39 it would be so much easier for you guys to just install go >.< 2010-10-20T17:27:46 i'll cover you as long as you don't do anything malicious, i don't think jeff will have any problem with this 2010-10-20T17:27:50 *** Appleman1234 has joined #aichallenge 2010-10-20T17:27:53 especially since it's his fault 2010-10-20T17:27:57 i don't know what all the fcntl stuff does, that might prevent forking 2010-10-20T17:27:58 ok I appreciate it :) 2010-10-20T17:28:11 lavalamp_: i would, but idk how to access the cloud comps 2010-10-20T17:28:13 I want to see the gobot compete at the end of this 2010-10-20T17:28:21 i blame jeff for starting the contest and not sticking with it 2010-10-20T17:28:23 lavalamp_: i would do it if it was just the main server, i have access to that 2010-10-20T17:28:46 amstan: Yeah, I understand, I don't blame you 2010-10-20T17:28:53 http://www.semdesigns.com/products/obfuscators/JavaObfuscator.html 2010-10-20T17:28:56 even jeff should know about N+2 2010-10-20T17:29:15 actually it's the only way I can think of that would make this work, because calling the compiled program as a process can be considered a "thread" 2010-10-20T17:30:18 davidd, obfuscating is old-fashioned, you should write underhanded code 2010-10-20T17:31:18 underhanded code? Gonna need a sample :) 2010-10-20T17:31:31 ah, no, the fcntl stuff is just file descriptors 2010-10-20T17:32:23 haha so far my 1.5 months of development have yielded 6 elo points 2010-10-20T17:32:52 beats teh -120 elo points i had previously anyway 2010-10-20T17:33:03 *** elijahbal has quit IRC (Remote host closed the connection) 2010-10-20T17:33:06 don't feel too bad, the last week has yielded -100 for me 2010-10-20T17:33:26 i'm not on the charts yet :) 2010-10-20T17:33:36 i feel pretty good about the idea though. there's a lot of room for improvement here 2010-10-20T17:33:39 the absolute elo score doesn't matter anyway 2010-10-20T17:33:54 i really wish they were normalized against ragebot rather than just adding 2000 to them 2010-10-20T17:33:58 yeah but i'm talking relative elo points here 2010-10-20T17:34:06 relative to what? 2010-10-20T17:34:30 to... everybody. elo1-elo0 = 6 2010-10-20T17:34:47 JamesMG, the contest: http://underhanded.xcott.com/?p=8 the winners: http://underhanded.xcott.com/?p=12 2010-10-20T17:34:51 relative to my old version, but like, if you just directly compare the two it beats the old version 99% of the time 2010-10-20T17:34:59 "everybody" has a single elo score you're comparing to? 2010-10-20T17:35:08 oh okay 2010-10-20T17:35:28 *** boegel has quit IRC (Quit: Leaving) 2010-10-20T17:35:40 so it makes sense to say "i gained 6 elo points relative to everybody" 2010-10-20T17:35:41 although that may not be the best measure since it doesn't factor "everybody" into the equation 2010-10-20T17:35:44 as long as you have a baseline 2010-10-20T17:35:47 overfitting is a possibility 2010-10-20T17:36:01 very much so but it's a lot less overfit than just comparing two bots to one another. 2010-10-20T17:36:34 oh i thought that's what you were implying that you did 2010-10-20T17:36:45 no i'm still running all my bots on tcp server 2010-10-20T17:36:49 oh! 2010-10-20T17:36:53 okay now i see 2010-10-20T17:37:35 Migi32: this contest is awesome 2010-10-20T17:37:46 JamesMG, I know 2010-10-20T17:37:47 haha wo 2010-10-20T17:37:49 wow * 2010-10-20T17:38:04 have you read the winner's code? It's just pure genius 2010-10-20T17:38:18 so in your face that it hurts :P 2010-10-20T17:38:37 has the code of the tron-contest winner been opened? 2010-10-20T17:38:51 davidd, it's open-source 2010-10-20T17:39:05 a1k0n made it 2010-10-20T17:39:20 yeah it's pretty terrible 2010-10-20T17:39:30 http://github.com/a1k0n/tronbot 2010-10-20T17:40:01 much oblidged 2010-10-20T17:41:42 sudo aptitude install git 2010-10-20T17:42:48 and click on downloads on the site :) 2010-10-20T17:43:13 or just read the source here as this is pretty much all there is: http://github.com/a1k0n/tronbot/blob/master/cpp/MyTronBot.cc 2010-10-20T17:43:35 a1k0n: didn't you also write a post mortem 2010-10-20T17:43:42 i see you packed the example bots and tools too 2010-10-20T17:43:48 yes, http://a1k0n.net/blah/archives/2010/03/index.html#e2010-03-04T14_00_21.txt 2010-10-20T17:43:55 what are the // {{{ // }}} for? 2010-10-20T17:44:02 just visual inspection of components? 2010-10-20T17:44:02 vim foldmethod=marker 2010-10-20T17:44:09 ah 2010-10-20T17:44:16 i just use multiple files :) 2010-10-20T17:44:25 instead of, like.. yeah, modularizing my code i just folded it. not sure why. 2010-10-20T17:44:37 i didn't really trust the compiler script on the site i think 2010-10-20T17:44:38 my code right now is so modularized 2010-10-20T17:44:41 haha textfile based blog? awesome! 2010-10-20T17:44:42 i think i have 15 source files now 2010-10-20T17:44:52 phreeza: it's nanoblogger 2010-10-20T17:44:55 Migi32: The bitter thing about these comics on xkcd is when you know some of the references like the faulty Pentium FPU in this comic: http://xkcd.com/394/ 2010-10-20T17:44:58 7 c files, 8 h files, or some such 2010-10-20T17:45:04 which is also sorta terrible, but written entirely in bash 2010-10-20T17:45:24 *** gerald512kk has quit IRC () 2010-10-20T17:45:27 haha hadn't heard of that 2010-10-20T17:50:09 hm 2010-10-20T17:50:14 #if 0 2010-10-20T17:50:15 Zannick: I'm sorry Dave, err Zannick; I cannot 'if'. 2010-10-20T17:50:23 ^ pro 2010-10-20T17:51:30 there are three types of comments in C: single-line //, multi-line /* */, and these lines don't exist #if 0 2010-10-20T17:51:45 from the nanoblogger website 2010-10-20T17:52:00 cons: comes with a user manual 2010-10-20T17:52:10 lol 2010-10-20T17:52:22 *** Sarin has quit IRC (Remote host closed the connection) 2010-10-20T17:53:20 *** Itkovian has quit IRC (Remote host closed the connection) 2010-10-20T17:53:42 *** Itkovian has joined #aichallenge 2010-10-20T18:01:53 *** Azrathud has joined #aichallenge 2010-10-20T18:04:23 i have no idea what my commandline arguments were supposed to do 2010-10-20T18:04:49 also the if(this) {} else { is pretty unlike me 2010-10-20T18:04:57 *** Spike_007 has quit IRC () 2010-10-20T18:05:25 a1k0n, maybe you planned to later add to it? 2010-10-20T18:05:44 i am unlikely to have been planning much of anything 2010-10-20T18:06:02 i think i was just extremely lazy when doing anything unrelated to tuning the bot 2010-10-20T18:06:02 *** mega1 has quit IRC (Ping timeout: 264 seconds) 2010-10-20T18:07:37 *** JensTi has joined #aichallenge 2010-10-20T18:09:18 *** JensTi1 has quit IRC (Ping timeout: 245 seconds) 2010-10-20T18:09:54 *** JamesMG has quit IRC (Quit: Leaving.) 2010-10-20T18:10:04 *** Appleman1234 has quit IRC (Ping timeout: 265 seconds) 2010-10-20T18:10:49 Top 10 players: cfaftw(3622), smatcher(3581), DracoAter(3486), dmj111(3454), rsergio(3444), EBraun(3357), asavis(3348), deepblue(3346), savchenko(3329), Polsky(3323) 2010-10-20T18:16:16 2,5 hours since code upload, still no first game.. should I re-upload? or is it a server issue? 2010-10-20T18:16:43 I never ever coded anything above 10 lines outside of an IDE. Except when I had this old 64 MB RAM notebook and wrote Java stuff for school. Man I even unzipped all the runtime jars to make things go faster. 2010-10-20T18:17:24 constesbot: Should ratatata re-upload? 2010-10-20T18:17:35 contesbot: Should ratatata re-upload? 2010-10-20T18:17:35 ratatata: successful compile? 2010-10-20T18:17:45 contestbot: Should ratatata re-upload? 2010-10-20T18:17:46 Frontier: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2010-10-20T18:17:47 yep, even received e-mail within the minute 2010-10-20T18:18:44 well, the main server is going at 5 gpm 2010-10-20T18:18:48 :/ 2010-10-20T18:24:19 *** Bobng has quit IRC (Ping timeout: 245 seconds) 2010-10-20T18:30:56 *** RainCT has quit IRC (Remote host closed the connection) 2010-10-20T18:31:53 *** choas has quit IRC (Quit: leaving) 2010-10-20T18:41:10 *** BtbN has quit IRC (Quit: Verlassend) 2010-10-20T18:43:15 gpm ? 2010-10-20T18:43:47 games per minute 2010-10-20T18:44:28 buy moar ram :P 2010-10-20T18:50:26 *** davidjliu has joined #aichallenge 2010-10-20T18:53:38 *** ratatata has quit IRC (Quit: Page closed) 2010-10-20T19:00:25 *** borovaka has joined #aichallenge 2010-10-20T19:01:52 hello can you tell me how the game calculate distance between the planets 2010-10-20T19:02:47 sqrt((x1-x2)^2 + (y1-y2)^2) 2010-10-20T19:02:57 ceiling of the eucilidean distance 2010-10-20T19:03:19 *** jesionaj` has joined #aichallenge 2010-10-20T19:03:25 http://ai-contest.com/specification.php 2010-10-20T19:03:35 *** Itkovian has quit IRC (Quit: Itkovian) 2010-10-20T19:03:40 *** jesionaj has quit IRC (Ping timeout: 252 seconds) 2010-10-20T19:03:42 yeah ceil not round... important 2010-10-20T19:04:12 thanks a lot 2010-10-20T19:04:19 sorry for my bad english :) 2010-10-20T19:04:20 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-20T19:04:23 np 2010-10-20T19:05:52 *** mceier has quit IRC (Quit: leaving) 2010-10-20T19:06:35 *** McLeopold has quit IRC (Quit: Leaving) 2010-10-20T19:10:50 Top 10 players: cfaftw(3587), smatcher(3550), dmj111(3474), DracoAter(3451), rsergio(3424), deepblue(3325), EBraun(3325), ruilov(3316), savchenko(3307), asavis(3303) 2010-10-20T19:15:06 *** FlagCapper has joined #aichallenge 2010-10-20T19:18:02 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-10-20T19:19:00 *** Itkovian has joined #aichallenge 2010-10-20T19:20:08 *** borovaka has quit IRC (Quit: Leaving) 2010-10-20T19:35:55 *** nann has joined #aichallenge 2010-10-20T19:38:52 *** Apophis has quit IRC (Ping timeout: 240 seconds) 2010-10-20T19:40:33 *** lavalamp_ has quit IRC (Quit: bye) 2010-10-20T19:47:11 something seems really fishy with the rankings... 2010-10-20T19:49:24 *** Codefun has joined #aichallenge 2010-10-20T19:49:53 All real games on the server end after 200 turns, right? 2010-10-20T19:50:07 yeah 2010-10-20T19:50:48 Good, because my bot seems to always take more than 1 second on turn 334... 2010-10-20T19:50:56 But if it ends at turn 200 I guess it doesn't matter. 2010-10-20T19:50:57 heh 2010-10-20T19:51:17 but that's on your machine, so it might matter 2010-10-20T19:51:20 Curse you optimization protocols! :D 2010-10-20T19:53:30 *** wh1teside has quit IRC (Ping timeout: 265 seconds) 2010-10-20T20:01:37 *** narnach has quit IRC (Remote host closed the connection) 2010-10-20T20:03:55 *** Codefun has quit IRC (Quit: Page closed) 2010-10-20T20:04:13 *** AlliedEnvy has quit IRC (Ping timeout: 276 seconds) 2010-10-20T20:05:35 *** yam has quit IRC (Ping timeout: 240 seconds) 2010-10-20T20:07:44 *** Azrathud has quit IRC (Ping timeout: 265 seconds) 2010-10-20T20:10:51 Top 10 players: smatcher(3552), cfaftw(3491), dmj111(3477), rsergio(3444), DracoAter(3381), chotting(3349), ruilov(3319), EBraun(3313), deepblue(3312), user42(3310) 2010-10-20T20:10:58 *** Migi32 has quit IRC (Remote host closed the connection) 2010-10-20T20:19:17 *** AlliedEnvy has joined #aichallenge 2010-10-20T20:25:10 *** Frontier has quit IRC (Ping timeout: 276 seconds) 2010-10-20T20:25:40 *** sigh has joined #aichallenge 2010-10-20T20:33:21 *** sigh has quit IRC (Quit: Leaving...) 2010-10-20T20:35:57 *** phreeza has joined #aichallenge 2010-10-20T20:43:03 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-20T20:48:22 *** Apophis has joined #aichallenge 2010-10-20T20:50:35 *** nann has quit IRC (Ping timeout: 240 seconds) 2010-10-20T20:53:51 *** Arthur_ has quit IRC (Remote host closed the connection) 2010-10-20T21:01:19 *** Smloh1 has joined #aichallenge 2010-10-20T21:06:17 *** yam has joined #aichallenge 2010-10-20T21:10:51 Top 10 players: smatcher(3553), cfaftw(3503), dmj111(3483), rsergio(3447), DracoAter(3382), chotting(3355), Polsky(3353), ruilov(3334), user42(3318), EBraun(3318) 2010-10-20T21:13:31 *** Itkovian has quit IRC (Quit: Itkovian) 2010-10-20T21:27:07 *** rommel has joined #aichallenge 2010-10-20T21:27:27 antima still here? haha 2010-10-20T21:28:09 and hello all :) 2010-10-20T21:36:49 *** amram99 has joined #aichallenge 2010-10-20T21:39:00 I think theres a problem with my bot... when I played it against some of example bots on the forums... it would get to around turn 12 then the other bot would crash... I just have a slightly modified starter package bot atm... just tried http://72.44.46.68/canvas?game_id=599960 , and it appears that the game crashed aswell? 2010-10-20T21:39:11 Note that my bot seems fine AFAIK? 2010-10-20T21:40:47 Damnit everytime I don't get more than a few turns in... :| 2010-10-20T21:40:59 *** FlagCapper has quit IRC (Quit: Page closed) 2010-10-20T21:42:39 *** amram99 has quit IRC (Quit: Page closed) 2010-10-20T21:46:55 *** Smloh1 has quit IRC (Ping timeout: 265 seconds) 2010-10-20T21:47:46 *** bhasker has quit IRC (Ping timeout: 252 seconds) 2010-10-20T21:55:18 *** JCS^ has joined #aichallenge 2010-10-20T22:00:28 *** JensTi has quit IRC (Quit: Leaving.) 2010-10-20T22:01:59 *** qacek has joined #aichallenge 2010-10-20T22:09:51 *** davidjliu has quit IRC (Quit: Page closed) 2010-10-20T22:10:19 *** davidjliu has joined #aichallenge 2010-10-20T22:10:51 Top 10 players: smatcher(3584), cfaftw(3499), dmj111(3479), rsergio(3445), DracoAter(3381), chotting(3354), Polsky(3347), ruilov(3328), user42(3312), EBraun(3309) 2010-10-20T22:11:01 *** amstan has quit IRC (Read error: Operation timed out) 2010-10-20T22:11:03 *** amstan_ has joined #aichallenge 2010-10-20T22:11:03 *** ChanServ sets mode: +o amstan_ 2010-10-20T22:14:00 it seems to me as if the server dumps all games older than around... 3 days-ish, and doesn't consider them in rankings... 2010-10-20T22:17:43 lol, literally, like, nobody is here 2010-10-20T22:21:03 that's a feature, not a bug 2010-10-20T22:22:11 ah... I guess it's just a little messed up right now with the 5 games per minute 2010-10-20T22:22:22 I assume that feature will be changed for the final contest? heh, hopefully, at least 2010-10-20T22:23:33 the final contest is a month away 2010-10-20T22:23:40 we might even have admins by then :P 2010-10-20T22:27:46 it be nice if they would at least acknowledge the situation 2010-10-20T22:30:14 there are very few available to do so 2010-10-20T22:30:45 oh yeah, is there any knkown word on bogusbrainpower's server offer? 2010-10-20T22:56:55 *** hellman has joined #aichallenge 2010-10-20T23:05:06 *** bhasker has joined #aichallenge 2010-10-20T23:06:01 *** Azrathud has joined #aichallenge 2010-10-20T23:10:45 oh god, i forgot how much is sucks programming the backend crap 2010-10-20T23:10:51 Top 10 players: smatcher(3568), cfaftw(3503), dmj111(3477), rsergio(3437), Polsky(3356), chotting(3347), ruilov(3343), deepblue(3319), EBraun(3317), user42(3316) 2010-10-20T23:10:52 i'm rewriting all my heuristic stuff 2010-10-20T23:11:07 what is going on with these scores? 2010-10-20T23:11:13 up and down 2010-10-20T23:11:16 down then up 2010-10-20T23:11:23 *** Itkovian has joined #aichallenge 2010-10-20T23:11:36 is this because people are changing their code? 2010-10-20T23:11:38 or just randomness? 2010-10-20T23:11:55 dunno, first place seems wrong to me 2010-10-20T23:12:09 http://www.ai-contest.com/profile_games.php?user_id=6749 2010-10-20T23:12:13 look at the number of losses they have 2010-10-20T23:12:16 i agree 2010-10-20T23:12:18 wtf? 2010-10-20T23:12:29 if it only uses scores from the past 3 days there probably aren't many games to use 2010-10-20T23:12:41 it shouldn't just be using scores from the last 3 days surely 2010-10-20T23:12:56 that's what davidjliu said earlier 2010-10-20T23:13:09 he won against second though.. 2010-10-20T23:13:21 shouldn't put you into first 2010-10-20T23:13:30 also won against chotting 2010-10-20T23:13:41 http://www.ai-contest.com/profile_games.php?user_id=4147 2010-10-20T23:13:41 against ebraun too 2010-10-20T23:14:06 well smatcher has won every game the past 3 days so it makes sense he would be in first place 2010-10-20T23:14:23 maybe he looses against noobs, but wins against good bots 2010-10-20T23:14:27 like.. elo will do this 2010-10-20T23:15:05 yeah, elo is not very good for rps 2010-10-20T23:15:13 rps? 2010-10-20T23:15:25 rock paper scissors 2010-10-20T23:15:38 not to be confused with rps matches, which would work 2010-10-20T23:16:40 i don't think this is rock paper scissors 2010-10-20T23:17:00 people keep throwing that around a lot 2010-10-20T23:17:52 my point is actually that elo won't work well for a nontransitive graph 2010-10-20T23:18:05 who is the better player, when a > b > c > a? 2010-10-20T23:19:45 i don't know how elo works but it seems to be used to rank plenty of other nontransitive graphs 2010-10-20T23:23:18 well 2010-10-20T23:23:22 sent an email rant 2010-10-20T23:23:31 hopefully jeff will notice 2010-10-20T23:25:49 i was actually thinking last night about a possible different way of "sorting" bots in a final contest 2010-10-20T23:25:55 *** seisatsu has joined #aichallenge 2010-10-20T23:26:14 i was thinking maybe players doing individual tournaments in other tournaments that sort of do a link of mergesort? 2010-10-20T23:26:35 you can never have perfect rankings for the point Zannick made, but it'd work quite well i think 2010-10-20T23:32:18 whee, just got to kick two instances back into running games :) 2010-10-20T23:33:50 *** Accoun has quit IRC (Ping timeout: 255 seconds) 2010-10-20T23:34:03 now back to trying to sync code from the server to svn 2010-10-20T23:35:31 Woot I won my first game on the TCP :P 2010-10-20T23:37:06 amstan_: want to do me a favour? :P 2010-10-20T23:37:49 remove (Australia) off the end of the organization name for University of Tasmania? (I originally added that on the end of my forum post request for it to be added under the right country, not to be added in the name, but at the time i was just happy for it to have been added) 2010-10-20T23:39:00 WOAAH 2010-10-20T23:39:04 so many servers 2010-10-20T23:39:04 heh 2010-10-20T23:39:48 *** qacek has quit IRC (Ping timeout: 240 seconds) 2010-10-20T23:53:08 gah, i'm almost tempted to completely rewrite aha