2010-10-30T00:12:52 Top 10 players: felixcoto(3934), bocsimacko(3915), Accoun(3899), GreenTea(3853), rsergio(3833), rs5147(3820), joakim_mjardner(3777), dmj111(3767), shangas(3750), Hazard(3750) 2010-10-30T00:25:05 *** Mjothvitnir has joined #aichallenge 2010-10-30T00:29:21 *** Rodya has joined #aichallenge 2010-10-30T00:51:28 *** sigh has joined #aichallenge 2010-10-30T00:53:35 *** Utkarsh has quit IRC (Ping timeout: 265 seconds) 2010-10-30T00:54:24 *** McLeopold has joined #aichallenge 2010-10-30T01:00:00 *** Utkarsh has joined #aichallenge 2010-10-30T01:01:55 *** demerzel_ has left #aichallenge 2010-10-30T01:04:23 *** davidjliu has quit IRC (Quit: Page closed) 2010-10-30T01:09:56 not even halloween, and already a graveyard in here. ;-) 2010-10-30T01:12:43 Top 10 players: felixcoto(3935), bocsimacko(3934), Accoun(3920), rsergio(3869), GreenTea(3859), rs5147(3812), dmj111(3799), joakim_mjardner(3775), shangas(3734), deepblue(3734) 2010-10-30T01:16:03 *** delt0r___ has joined #aichallenge 2010-10-30T01:16:26 hi delt0r___ 2010-10-30T01:17:12 *** delt0r has quit IRC (Ping timeout: 245 seconds) 2010-10-30T01:22:33 * amstan raises from the dead, looks around 2010-10-30T01:23:04 davidd: so, you just discovered irc today? 2010-10-30T01:23:10 amstan: do you have any ability to make changes to the official site? 2010-10-30T01:23:20 McLeopold: yes, why? 2010-10-30T01:23:34 How hard would it be to allow people to change their profile info? 2010-10-30T01:23:49 it would involve writing code 2010-10-30T01:23:57 and i really don't want to do that, i have my own code too 2010-10-30T01:24:19 i will svn update if there's any appropriate code that people want 2010-10-30T01:24:44 own code? 2010-10-30T01:25:21 there basically isn't anything to enable profile edits 2010-10-30T01:25:38 no code paths edit the info 2010-10-30T01:25:59 so, isn't the site open source somewhere? 2010-10-30T01:26:19 yeah 2010-10-30T01:26:43 i think it is simply "someone has to get round to do it" 2010-10-30T01:26:51 d? 2010-10-30T01:27:06 could I get a link? 2010-10-30T01:27:17 @url repo 2010-10-30T01:27:17 sigh: repo = http://code.google.com/p/ai-contest/ 2010-10-30T01:27:18 @repo 2010-10-30T01:27:19 Zannick: repo = http://code.google.com/p/ai-contest/ 2010-10-30T01:28:08 I haven't been paying attention for liek 2 weeks 2010-10-30T01:28:19 have there been any major advancements in the bots? 2010-10-30T01:28:44 they kick-ass 2010-10-30T01:28:46 :P 2010-10-30T01:28:51 haha 2010-10-30T01:29:30 well, they can't vote yet, but we're trying 2010-10-30T01:29:46 :) 2010-10-30T01:29:50 ^^ 2010-10-30T01:29:57 Zannick: have you got a bot up and running yet 2010-10-30T01:30:36 nope :P 2010-10-30T01:30:58 i am finishing up neutral grabbing, then all i have left is attack 2010-10-30T01:31:11 hopefully i can get something running before monday 2010-10-30T01:31:21 since my november has a lot less free time for this 2010-10-30T01:32:01 cool, good luck :) 2010-10-30T01:32:14 hmm... McLeopold still hasn't uploaded a new bot o_O 2010-10-30T01:32:33 nope 2010-10-30T01:32:45 I have huge performance issues. 2010-10-30T01:32:56 ah :( 2010-10-30T01:32:56 And life... 2010-10-30T01:32:59 hehe, i have to finish off my bot and test it... wrote several hundred lines of code without testing 2010-10-30T01:33:04 lmao... debugging fun 2010-10-30T01:33:22 raedwulf: that's always dangerous 2010-10-30T01:33:54 yea, unfortunately the code needed other code to test >.< 2010-10-30T01:34:15 ok bbl run 2010-10-30T01:34:18 cya 2010-10-30T01:34:48 amstan: is the web site running on apache? 2010-10-30T01:34:55 tes 2010-10-30T01:35:06 sigh: you're back :P 2010-10-30T01:35:17 *** superflit has joined #aichallenge 2010-10-30T01:35:19 can you provide the apache config so I can run one local? 2010-10-30T01:35:25 or is it in the trunk? 2010-10-30T01:35:32 antimatroid: I think... I'm planning to start work on my bot again today 2010-10-30T01:35:42 I still haven't started though :P 2010-10-30T01:35:55 McLeopold: it's not a complicated site.. you just put them somewhere web accessible 2010-10-30T01:35:57 i've been working hard and it sucks :P 2010-10-30T01:35:59 nothing special done through apache 2010-10-30T01:36:03 i have a couple of ideas in the works though 2010-10-30T01:36:16 cool 2010-10-30T01:36:22 and i found quite a few bugs with my internals 2010-10-30T01:36:29 I need to remember my ideas again... I think I have them written down somewhere 2010-10-30T01:36:38 ah, that's good... easy wins 2010-10-30T01:37:07 is minimax working out? 2010-10-30T01:37:22 hmmm, not really, but i think it's cause it's not set up well 2010-10-30T01:37:38 i pretty much have things set up to try it in a zero sum simultaneous game 2010-10-30T01:37:46 do you know if anyone else is using it to any degree of success? 2010-10-30T01:38:01 no idea 2010-10-30T01:38:14 JamesMG has something working, but not to great success 2010-10-30T01:38:28 at the moment my tree is using a decision tree at each node for players to pick moves, and it just takes too long to get anywhere 2010-10-30T01:38:43 yeah, I can imagine 2010-10-30T01:39:24 i might take the tree back to single moves per node without decision trees 2010-10-30T01:39:34 but i need the decision trees to generate pure strategies for zero sum games 2010-10-30T01:39:45 i can't work out how else to do it, dp would be nice, but there's sooo much data 2010-10-30T01:40:16 not sure exactly what you mean by all that :) 2010-10-30T01:40:55 the idea then is to find pure strategy nash equilibria if they exist, otherwise use iterated strict dominance (and maybe have to throw out some other strategies) then use simplex for msne or some kind of numerical approximation 2010-10-30T01:41:22 msne? 2010-10-30T01:41:30 mixed strategy nash equilibria 2010-10-30T01:41:35 ah, ok 2010-10-30T01:41:36 so like rock paper scissors, you have to randomise 2010-10-30T01:41:44 yeah, I know 2010-10-30T01:41:45 otherwise you'll be pounced if they try to predict you 2010-10-30T01:41:49 :) 2010-10-30T01:41:50 yep :) 2010-10-30T01:42:15 even in my minimax tree, if two nodes are equal if(getNum(0,1)) addChild :) 2010-10-30T01:42:24 as best child** 2010-10-30T01:42:57 I don't know what that means 2010-10-30T01:43:05 i understand where minimax on tree's fails miserably now 2010-10-30T01:43:31 where? other than the enormous branching factor? 2010-10-30T01:44:03 no, if there's no saddle points then it doesn't work 2010-10-30T01:44:08 but still spits out an answer for you 2010-10-30T01:44:18 no saddle points means no pure strategy nash equilibria 2010-10-30T01:44:33 and the other bot can exploit the fact that you're going to do that 2010-10-30T01:45:33 ah ok 2010-10-30T01:46:14 is this an issue that has come up in practice for this game, or just theoretical? 2010-10-30T01:47:26 theoretical, but a1k0n kept saying it in the past and I couldn't work out where it happened 2010-10-30T01:47:40 and such situations will happen, but it's probably not easy to exploit 2010-10-30T01:47:46 i see 2010-10-30T01:47:57 but i try to approximate optimal, not suboptimal :P 2010-10-30T01:48:07 haha 2010-10-30T01:56:51 * Mjothvitnir crumbles to dust 2010-10-30T01:56:54 *** Mjothvitnir has quit IRC (Quit: Page closed) 2010-10-30T01:59:42 *** seisatsu has quit IRC (Quit: Leaving) 2010-10-30T02:02:20 * Zannick facepalm 2010-10-30T02:02:56 no wonder this section wasn't working 2010-10-30T02:03:02 player 1 wasn't the closest player 2010-10-30T02:03:08 player 134514384 was 2010-10-30T02:03:44 ah, that's right, you are using C 2010-10-30T02:04:02 :P 2010-10-30T02:07:59 *** Sarin has joined #aichallenge 2010-10-30T02:09:08 okay, great 2010-10-30T02:09:14 i've changed it to 23 now 2010-10-30T02:09:38 changed what? 2010-10-30T02:09:46 the closest player is now player 23 2010-10-30T02:10:12 oh, geez 2010-10-30T02:10:13 player 23? you mean planet? 2010-10-30T02:10:23 i forget to set that flag 2010-10-30T02:11:32 bahahaha 2010-10-30T02:11:35 it works 2010-10-30T02:11:52 except now it tries to conquer everything 2010-10-30T02:12:00 and loses spectacularly to ragebot in the process 2010-10-30T02:12:17 everyone 2010-10-30T02:12:23 everyon's been hrough that 2010-10-30T02:12:28 *through 2010-10-30T02:12:41 i give up, I suck at typing 2010-10-30T02:12:43 Top 10 players: bocsimacko(3941), felixcoto(3926), Accoun(3901), GreenTea(3846), rsergio(3827), rs5147(3804), dmj111(3778), joakim_mjardner(3751), deepblue(3723), shangas(3706) 2010-10-30T02:13:38 oh, i know why that's doing that 2010-10-30T02:13:44 i didn't adjust the right budget 2010-10-30T02:14:00 well, that added 20 turns of life :P 2010-10-30T02:17:41 *** hellman has joined #aichallenge 2010-10-30T02:19:54 *** virdo has quit IRC (Ping timeout: 240 seconds) 2010-10-30T02:30:13 why there is no reference in the waterloo page about this contest? 2010-10-30T02:30:38 superflit: we're slow to update that page 2010-10-30T02:31:18 it seems like fake sorry 2010-10-30T02:32:06 ok....? 2010-10-30T02:33:30 It fails on the credibility 2010-10-30T02:33:47 if at least on the events of waterlook there was a link 2010-10-30T02:33:57 that was ok 2010-10-30T02:34:10 no links, no citation on google blog 2010-10-30T02:34:30 you probably want to talk to jeff then 2010-10-30T02:35:13 he's the one that wrote that text 2010-10-30T02:35:43 as far as i see it, we host our own servers and past contestants are admins 2010-10-30T02:36:19 the contest seems nice and well developed 2010-10-30T02:36:45 but seriously, it lacks credibility. there is nothing on waterloo only the past contest 2010-10-30T02:36:58 even the past contest page do not say anything about this contest 2010-10-30T02:37:15 what does it mean to not be credible? 2010-10-30T02:37:18 credibility for what though? 2010-10-30T02:37:20 yeah.. 2010-10-30T02:37:32 if you use the name google 2010-10-30T02:37:32 it's not like we're making people give us their credit card info 2010-10-30T02:37:41 google seems to not be using your name... 2010-10-30T02:37:50 that's their problem.. 2010-10-30T02:37:53 with 3800+ players I don't think we can withstand any more credibility 2010-10-30T02:37:53 no you are just collecting cheap algo 2010-10-30T02:38:29 google blog is for important things 2010-10-30T02:38:31 and doing research papers with free data 2010-10-30T02:38:44 maybe next time 2010-10-30T02:39:22 * raedwulf smells a troll. 2010-10-30T02:39:23 superflit: research papers? i don't think any of us are in grad school 2010-10-30T02:39:47 i did part of a research paper as an undergrad 2010-10-30T02:39:58 http://csclub.uwaterloo.ca/ 2010-10-30T02:40:00 our bots are really going to be used to control the next US airial drone strike? 2010-10-30T02:40:02 w/e 2010-10-30T02:40:15 wtf is this, i don't even? 2010-10-30T02:40:23 honestly, though, it wouldn't be the first fun someone has put something fun on the internet in order to get info for papers 2010-10-30T02:40:39 see also: luis von ahn and the esp game 2010-10-30T02:40:54 McLeopold: no, interstellar drone strikes 2010-10-30T02:40:57 Analysis of randomized algorithms via the probabilistic method. ... .... ... 2010-10-30T02:41:11 heh, well whoever's writing that paper I wish they'd come help run this thing :P 2010-10-30T02:41:14 as we are clearly designing such simulators 2010-10-30T02:41:22 superflit: is that the paper you think was derived from the contest? 2010-10-30T02:41:41 amstan: no yet, but the second will.. 2010-10-30T02:42:15 seriously, stop trolling >.< 2010-10-30T02:42:43 Good think I'm programming in a backdoor... 2010-10-30T02:43:03 he's more entertaining that accoun 2010-10-30T02:43:21 superflit: are you going to join us and write a bot? 2010-10-30T02:43:57 McLeopold: nice trick.....at least trick or treat.. 2010-10-30T02:43:58 McLeopold: good, cause my front door isn't coming along great 2010-10-30T02:44:05 *** superflit has left #aichallenge 2010-10-30T02:44:18 O_o 2010-10-30T02:44:21 wow 2010-10-30T02:44:23 I don't understand... 2010-10-30T02:44:29 * Zannick just being silly 2010-10-30T02:44:47 McLeopold: no man.. you clearly wanted his intellectual property! 2010-10-30T02:44:54 how dare you profit off him? 2010-10-30T02:45:13 I would prefer bocsimako's IP 2010-10-30T02:45:31 @rankings 2010-10-30T02:45:33 amstan: Top 10 players: bocsimacko(3954), felixcoto(3941), Accoun(3914), GreenTea(3860), rsergio(3854), rs5147(3812), dmj111(3792), joakim_mjardner(3785), deepblue(3731), Hazard(3718) 2010-10-30T02:46:02 you though you were going to have a comeback like, "no you mean felixcoto's", didn't you 2010-10-30T02:46:14 sorry we don't log ip addresses of submitters :P 2010-10-30T02:46:31 janzert: he meant intellectual property 2010-10-30T02:46:40 not internet protocol address 2010-10-30T02:46:46 yeah, I know 2010-10-30T02:46:47 wait.. 2010-10-30T02:46:52 oh god 2010-10-30T02:46:56 * amstan is oblivious again 2010-10-30T02:46:57 ? 2010-10-30T02:47:26 McLeopold: you should team up with accoun, or something 2010-10-30T02:47:44 My beard isn't long enough 2010-10-30T02:47:49 lol 2010-10-30T02:47:50 i think we're done here 2010-10-30T02:48:29 Zannick: did you ever figure out my r/p/s map? 2010-10-30T02:48:42 oh, i haven't looked at it since i left work 2010-10-30T02:48:46 there's maps for that? 2010-10-30T02:48:53 *** tapwater has quit IRC (Quit: tapwater) 2010-10-30T02:49:15 yea mega1 open sourcing after the contest? :D 2010-10-30T02:49:28 sure, after I clean it up 2010-10-30T02:49:39 so.. who's mega1? in the rankings? 2010-10-30T02:49:40 provided that I'm motivated to do so 2010-10-30T02:49:48 bocsimacko 2010-10-30T02:50:03 http://pastebin.com/ZZK3TkzF 2010-10-30T02:50:11 amstan: I had a "did not start/timeout/crashed" yesterday 2010-10-30T02:50:23 mega1: I like your bot. I"m going to mimic it. 2010-10-30T02:50:37 McLeopold: i do think it isn't very interesting since the minimum distance is 7 and players start with 7 ships 2010-10-30T02:51:04 McLeopold: you need bug for bug compatibility, its strengths is in the bugs :-) 2010-10-30T02:51:15 oh wait 2010-10-30T02:51:21 yeah, I know what you mean... 2010-10-30T02:51:25 no, yeah 2010-10-30T02:51:27 mega1: hmm, not sure, you could ask janzert 2010-10-30T02:51:48 if player 1 launches toward his neutral and player 2 launches towards the same one 2010-10-30T02:51:55 hmm? 2010-10-30T02:51:57 p2 snipes 2010-10-30T02:52:00 yep 2010-10-30T02:52:15 Change the starts to 1 growth, and it is probably the same. 2010-10-30T02:52:33 oh, they're 0 growth 2010-10-30T02:52:36 how did i miss that 2010-10-30T02:52:46 I think you could make it more complecated until it became an official map. 2010-10-30T02:53:39 that...makes the game much more narrow 2010-10-30T02:53:47 since now there are only three moves 2010-10-30T02:54:12 go for 0, 3, or wait 2010-10-30T02:54:16 but at least I've got proof there is no dominant strategy 2010-10-30T02:54:21 except waiting always loses to the first two 2010-10-30T02:54:31 no, it wins against going for the far one 2010-10-30T02:54:42 oh, true 2010-10-30T02:55:02 yeah, that's rps 2010-10-30T02:55:04 janzert: our cool guy on top here is not happy that he timed out for unknown reasons yesterday 2010-10-30T02:55:22 amstan: I'm pretty mad about his loss too 2010-10-30T02:55:34 hey, he's still first... 2010-10-30T02:55:45 It cheapens my victory when I beat him. :P 2010-10-30T03:02:36 *** seisatsu has joined #aichallenge 2010-10-30T03:12:43 Top 10 players: bocsimacko(3952), felixcoto(3941), Accoun(3898), GreenTea(3888), rsergio(3882), rs5147(3803), dmj111(3793), joakim_mjardner(3782), shangas(3712), deepblue(3700) 2010-10-30T03:13:18 McLeopold: good luck with that ;) 2010-10-30T03:13:30 ? 2010-10-30T03:13:34 with what 2010-10-30T03:13:34 beating mega1 2010-10-30T03:13:39 thanks! 2010-10-30T03:14:22 my bot wont make any moves on the first turn now 2010-10-30T03:14:23 grrr 2010-10-30T03:14:25 McLeopold: umm. what about attacking the enemy planet? 2010-10-30T03:15:16 amstan: on the example map: 2010-10-30T03:15:23 it's a zero grwoth 2010-10-30T03:16:20 wait what? 2010-10-30T03:16:22 aww 2010-10-30T03:16:24 that's boring then 2010-10-30T03:16:53 It's not suppose to be exciting. It just proves there is no dominant strategy. 2010-10-30T03:17:11 for some maps 2010-10-30T03:17:51 what do you mean by not dominant? :P 2010-10-30T03:18:14 antimatroid: no winning strategy 2010-10-30T03:18:23 well, more generically, some maps favor aggressive attacks, some favor waiting more 2010-10-30T03:18:27 there's technically no winning strategy on any map 2010-10-30T03:18:29 this one just sits on the line 2010-10-30T03:18:42 antimatroid, I think on some there are. 2010-10-30T03:18:56 not if it's symmetric, there is a msne with equal payoffs to each player 2010-10-30T03:18:57 I've got map30 pretty much nailed down 2010-10-30T03:19:20 yes, most maps will be draw with perfect play 2010-10-30T03:19:39 but with my example, it shows that there can be situations where it is random 2010-10-30T03:20:08 there might be small locations in a larger map that resemble the r/p/s situation 2010-10-30T03:20:34 yeah, didn't we go through that already? 2010-10-30T03:20:54 my example with your correction does it 2010-10-30T03:20:56 not with a fool proof example 2010-10-30T03:22:00 was that the one with the 1 distance neutrals in the center? 2010-10-30T03:22:23 yeah, but you can change the distances 2010-10-30T03:22:28 they were picked arbitrarily 2010-10-30T03:22:48 oh wait, yeah, you guys tore that apart didn't you? :P 2010-10-30T03:22:53 it was close, but attacking the close neutral was dominate in that one 2010-10-30T03:23:12 if one moves the middle neutrals further apart would it switch? 2010-10-30T03:23:27 the change here is, the neutrals are far enough apart that you can't hop after picking the wrong planet 2010-10-30T03:23:49 yeah okay 2010-10-30T03:23:56 and you lose to many ships as well 2010-10-30T03:24:01 i was going to say, just move one up and the other down 2010-10-30T03:24:23 so, my example can fall apart with various changes 2010-10-30T03:24:36 on one side, waiting becomes dominant, on the other attacking 2010-10-30T03:24:49 A good bot will need to be able to see the difference 2010-10-30T03:25:14 and mega1 has it down pretty good 2010-10-30T03:26:25 my bot sucks at specifics, i need to work that out better 2010-10-30T03:26:32 every time i try my brain just goes to mush 2010-10-30T03:27:51 I use spreadsheets to help me think things through 2010-10-30T03:33:42 i use a piece of paper 2010-10-30T03:33:45 and a pen 2010-10-30T03:33:47 lol 2010-10-30T03:33:53 pacer* 2010-10-30T03:34:04 white board and dry erase marker 2010-10-30T03:34:16 don't have one handy :( 2010-10-30T03:34:23 oh man, everyone is ruining me at having a whiteboard these days 2010-10-30T03:34:40 like half my analysis class started talking about their whiteboards the other day 2010-10-30T03:34:47 and i was like wtf, i am so behind the times 2010-10-30T03:35:12 lol 2010-10-30T03:35:17 get a tablet pc and show them up 2010-10-30T03:35:22 ^^ 2010-10-30T03:35:30 yea wish that they weren't all so darn expensive 2010-10-30T03:36:07 meh, i'm a maths student, computers are largely useless 2010-10-30T03:36:09 sweet 2010-10-30T03:36:14 so, how do you brag about a whiteboard, other than you have one? 2010-10-30T03:36:14 pure maths** 2010-10-30T03:36:22 my bot can now take over every planet then sit on its thumbs 2010-10-30T03:36:25 that's bragging rights right there :P 2010-10-30T03:36:41 and defend while ragebot tries to chip away 2010-10-30T03:36:42 antimatroid: in my computer maths lectures i wrote up maths in latex in realtime 2010-10-30T03:36:52 Zannick: tcp! 2010-10-30T03:36:58 i want to see 2010-10-30T03:36:59 :P 2010-10-30T03:37:02 there's no attack code yet! 2010-10-30T03:37:06 meh, latex documents are pretty, but i'm not going to type notes in it 2010-10-30T03:37:11 how does it take a planet then? 2010-10-30T03:37:18 i did, it was fine :P 2010-10-30T03:37:20 ragebot doesn't go for neutrals 2010-10-30T03:37:27 you get pretty quick at it after a bit of practice 2010-10-30T03:37:38 my main problem was my old laptop keyboard really failed 2010-10-30T03:37:42 then when there's a diagram put on the board? 2010-10-30T03:37:48 i'm using "conquer" and "attack" to mean separate things wrt neutral or enemy 2010-10-30T03:37:50 all my maths lectures are entirely off the whiteboard 2010-10-30T03:37:57 antimatroid: then you scrabble for a piece of paper :P 2010-10-30T03:38:08 didn't have any diagrams 2010-10-30T03:38:10 so, attack would be pretty eash 2010-10-30T03:38:13 in that course 2010-10-30T03:38:21 ships + 1 + (distance * growth) 2010-10-30T03:38:25 my analysis lecturer loves diagrams 2010-10-30T03:38:26 module* 2010-10-30T03:38:30 ah right 2010-10-30T03:38:54 eash is the cool way to pronounce easy. it's like "yesh" 2010-10-30T03:39:00 well it does depend on the type of diagram 2010-10-30T03:39:32 all my lecturers now use slides 2010-10-30T03:39:37 which are available online 2010-10-30T03:39:46 there's rarely anything extra 2010-10-30T03:39:56 that conquer function is nearly 200 lines long 2010-10-30T03:40:10 copy paste? 2010-10-30T03:40:30 no, just too many separate components 2010-10-30T03:40:39 McLeopold: are you still all heuristics 2010-10-30T03:40:47 seriously? a planet is a planet is a planet? 2010-10-30T03:40:48 my quick heuristic bot last night did really well (t) 2010-10-30T03:40:51 (h**) 2010-10-30T03:41:04 for how fast i threw that together, compared to my tree bots 2010-10-30T03:41:06 yes, all heuristics 2010-10-30T03:41:12 this isn't pure heuristic 2010-10-30T03:41:19 it got you 2010-10-30T03:41:27 i am casing on things like which player is closest 2010-10-30T03:41:40 isn't that still heuristics? 2010-10-30T03:41:44 i need to grab a piece of paper for my heuristics i think 2010-10-30T03:41:59 in some manner of speaking 2010-10-30T03:42:02 Zannick: I still don't understand what you were saying about player 23? 2010-10-30T03:42:12 there is no player 23 2010-10-30T03:42:15 sigh, you behind by like, 2 hours 2010-10-30T03:42:31 somehow it thought player "23" was closest to the planet 2010-10-30T03:42:35 nah, what he said about closest player just reminded me 2010-10-30T03:42:43 oh 2010-10-30T03:42:55 i am amused at the errors 2010-10-30T03:43:19 anyway, I reserve the right to raise dead conversations 2010-10-30T03:43:24 Zannick: you must have a list of neutral planets? what if an enemy planet just happened to be on the list? would that be so bad? 2010-10-30T03:43:26 tree bot? kind of, it uses heuristics for each players moves, but also generates a minimax tree and picks moves based off that 2010-10-30T03:43:48 i just discovered earlier that once the bot was satisfied the planet was well defended for the next turn, it would happily send ships off that planet 2010-10-30T03:44:00 bug in my budgeting >.< 2010-10-30T03:44:03 quickly fixed that 2010-10-30T03:44:23 are people actually using 0-1 knapsack for first move? 2010-10-30T03:44:23 McLeopold: no, what is done is even more highly cased on what type of planet it is 2010-10-30T03:44:35 doesn't seem like it really takes into account what the other player is then going to do to you to me 2010-10-30T03:44:39 or your position strength etc. 2010-10-30T03:44:43 antimatroid: according to the forum like a month ago, yes 2010-10-30T03:45:02 position strength? this is first turn, remember 2010-10-30T03:45:23 yeah, isn't that when it's most important 2010-10-30T03:45:32 I think he mean the pos. strength of the planets you are aquiring 2010-10-30T03:45:35 oh, ok 2010-10-30T03:45:37 map7, turn 1, take the center 2010-10-30T03:45:43 really? 2010-10-30T03:45:44 well, you can factor that into your planet value 2010-10-30T03:46:12 i have that open right next to the window already, i think you might be right 2010-10-30T03:47:06 wait, if they take their 12 first move and you take middle, what then? 2010-10-30T03:47:56 um...profit? 2010-10-30T03:48:08 wouldn't you lose? 2010-10-30T03:48:57 well, you would, but not me 2010-10-30T03:49:00 :) 2010-10-30T03:49:06 haha 2010-10-30T03:49:10 but why not? 2010-10-30T03:49:18 what moves are you planning to make following that? 2010-10-30T03:49:42 it depends on the enemy 2010-10-30T03:49:54 I'm just saying good position is taking the center 2010-10-30T03:50:04 and you can determine that on turn 1 2010-10-30T03:50:10 wow, i think i might have a working bot tomorrow 2010-10-30T03:50:13 they send 13 first move and you send 38, you have no other information about the opoonent, what do you do? 2010-10-30T03:50:15 but i need to break now 2010-10-30T03:50:30 let me get my whiteboard... 2010-10-30T03:50:54 i'm download that better visualizer that lets me play my bot 2010-10-30T03:51:07 delt0r's? 2010-10-30T03:51:33 yeah 2010-10-30T03:51:39 i know i know, should have already 2010-10-30T03:51:42 I have trouble beating RageBot in human mode. 2010-10-30T03:52:36 i didn't think it would be that hard 2010-10-30T03:53:13 RageBot has a human mode! 2010-10-30T03:53:17 :P 2010-10-30T03:53:23 did you see that debug image yesterday? it was awesome 2010-10-30T03:53:29 what image? 2010-10-30T03:53:33 http://pastebin.com/L7Tzcy8Y 2010-10-30T03:53:54 that's lavalamps's debugging 2010-10-30T03:54:42 that looks very confusing 2010-10-30T03:54:45 wow 2010-10-30T03:55:24 *** Palmik has joined #aichallenge 2010-10-30T03:57:47 yes, but it's still awesome 2010-10-30T03:57:54 hmm, my computer seems to hate the visualiser 2010-10-30T04:01:19 lol 2010-10-30T04:01:21 okay, it works 2010-10-30T04:09:06 *** seisatsu has quit IRC (Quit: Leaving) 2010-10-30T04:12:44 Top 10 players: felixcoto(3951), bocsimacko(3942), Accoun(3894), GreenTea(3879), rsergio(3872), dmj111(3791), joakim_mjardner(3769), rs5147(3767), Hazard(3694), deepblue(3654) 2010-10-30T04:13:45 *** Appleman1234 has joined #aichallenge 2010-10-30T04:15:18 McLeopold: middle planet is a terrible move :P 2010-10-30T04:16:59 hop to the 4+2 2010-10-30T04:17:03 then decide 2010-10-30T04:18:34 i don't thnk that works either, i think you need to take your 12 2010-10-30T04:19:12 can i change the send ratio? 2010-10-30T04:19:41 yeah okay i see 2010-10-30T04:21:23 McLeopold: so.. hold on 2010-10-30T04:21:31 McLeopold: if you have any map with 0 growth 2010-10-30T04:21:36 you'll have same problem 2010-10-30T04:21:52 because there will never be any reason to expand 2010-10-30T04:22:02 makes the game uninteresting 2010-10-30T04:22:12 amstan: what if the growth is 1, it still works 2010-10-30T04:22:31 what if the middle is collectively worth much more than the starting position 2010-10-30T04:22:31 not if there's waiting involved between moves 2010-10-30T04:24:54 *** irchs has joined #aichallenge 2010-10-30T04:34:14 *** amstan_ has joined #aichallenge 2010-10-30T04:34:14 *** ChanServ sets mode: +o amstan_ 2010-10-30T04:35:37 *** irchs has quit IRC (Quit: irchs) 2010-10-30T04:38:55 *** perestrelka has quit IRC (Ping timeout: 255 seconds) 2010-10-30T04:41:39 *** amstan_ has quit IRC (Remote host closed the connection) 2010-10-30T04:45:58 *** perestrelka has joined #aichallenge 2010-10-30T05:10:32 *** perestrelka has quit IRC (Ping timeout: 240 seconds) 2010-10-30T05:12:45 Top 10 players: felixcoto(4007), bocsimacko(3965), GreenTea(3923), Accoun(3922), rsergio(3885), dmj111(3831), joakim_mjardner(3810), rs5147(3808), Hazard(3754), Mistmanovx(3681) 2010-10-30T05:17:39 *** Commandir has joined #aichallenge 2010-10-30T05:30:52 *** p4p4 has joined #aichallenge 2010-10-30T05:35:51 it would be cool if the home planets didn't always have the maximum growth rate 2010-10-30T05:36:52 why 2010-10-30T05:37:09 cause at present you pretty much always have the dominant strategy to defend it 2010-10-30T05:37:43 and perhaps several starting planets 2010-10-30T05:37:45 you could have close starting matches with smaller home growth rates where it's better to give up that planet and move to another one 2010-10-30T05:37:59 my bot could handle several starting planets 2010-10-30T05:38:01 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-10-30T05:38:05 i could start it at any point in a game and it'd be good to go 2010-10-30T05:38:38 yeah, just saying it would be interesting with more variation 2010-10-30T05:40:27 that is, I agree with your initial statement 2010-10-30T05:41:34 *** Bobng has joined #aichallenge 2010-10-30T05:42:12 *** Utkarsh has joined #aichallenge 2010-10-30T05:46:55 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-10-30T05:57:48 *** McLeopold has quit IRC (Ping timeout: 240 seconds) 2010-10-30T05:58:32 *** irchs has joined #aichallenge 2010-10-30T06:04:19 *** jaspervdj has joined #aichallenge 2010-10-30T06:06:35 *** irchs has quit IRC (Quit: irchs) 2010-10-30T06:10:29 *** Sarin has quit IRC (Quit: Verlassend) 2010-10-30T06:11:35 *** Appleman1234 has joined #aichallenge 2010-10-30T06:12:46 Top 10 players: bocsimacko(4022), felixcoto(4018), Accoun(3973), rsergio(3971), GreenTea(3958), rs5147(3872), dmj111(3859), joakim_mjardner(3827), Slin-.-(3764), Hazard(3750) 2010-10-30T06:13:08 *** perestrelka has joined #aichallenge 2010-10-30T06:15:56 *** ABC_ has joined #aichallenge 2010-10-30T06:17:12 contestbot: rankings 10 2010-10-30T06:17:13 ABC_: Top 10 players: bocsimacko(4020), felixcoto(4016), Accoun(3972), rsergio(3969), GreenTea(3956), rs5147(3871), dmj111(3857), joakim_mjardner(3826), Slin-.-(3762), Hazard(3749) 2010-10-30T06:18:31 *** ABC_ has quit IRC (Client Quit) 2010-10-30T06:20:16 *** Tobu has quit IRC (Ping timeout: 276 seconds) 2010-10-30T06:21:14 hmmm interesting, i have a map i made for which my bot crashes on using the visualiser with human play but not the original 2010-10-30T06:23:18 *** phreeza has joined #aichallenge 2010-10-30T06:25:30 *** boegel has joined #aichallenge 2010-10-30T06:25:42 *** Tobu has joined #aichallenge 2010-10-30T06:47:47 *** Cromulent has joined #aichallenge 2010-10-30T06:48:31 *** Cromulent has left #aichallenge 2010-10-30T06:58:56 *** RainCT has joined #aichallenge 2010-10-30T07:11:43 *** boegel has quit IRC (Quit: This computer has gone to sleep) 2010-10-30T07:12:46 Top 10 players: bocsimacko(4029), Accoun(4003), rsergio(3998), GreenTea(3961), felixcoto(3960), rs5147(3918), dmj111(3839), joakim_mjardner(3827), Slin-.-(3795), george(3752) 2010-10-30T07:28:07 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-10-30T07:29:17 morning 2010-10-30T07:30:16 *** n9thbit has joined #aichallenge 2010-10-30T07:34:10 *** Utkarsh has joined #aichallenge 2010-10-30T07:52:22 *** javagamer_cpp has quit IRC (Remote host closed the connection) 2010-10-30T07:58:03 *** javagamer has joined #aichallenge 2010-10-30T08:04:34 ahem, this is why i think the time limit should be extended 2010-10-30T08:04:35 http://72.44.46.68/canvas?game_id=776925 2010-10-30T08:05:07 the outcome of that game would be incorrect at present in the final tournmanet 2010-10-30T08:05:24 why antimatroid 2010-10-30T08:05:31 so we could watch a stalemate? 2010-10-30T08:05:42 not very interesting 2010-10-30T08:06:46 the outcome is wrong though 2010-10-30T08:06:59 i think the band of radius idea is best 2010-10-30T08:12:45 Top 10 players: bocsimacko(4073), rsergio(4040), Accoun(4040), GreenTea(4007), felixcoto(3982), rs5147(3957), dmj111(3889), Slin-.-(3868), joakim_mjardner(3856), shangas(3825) 2010-10-30T08:16:21 band of radius? 2010-10-30T08:18:13 so is stupid the new retard 2010-10-30T08:18:16 just not as retarded 2010-10-30T08:21:33 *** perestrelka has quit IRC (Ping timeout: 252 seconds) 2010-10-30T08:26:02 i think it might be more retarded :P 2010-10-30T08:27:04 heh 2010-10-30T08:28:27 *** nullkuhl_ has quit IRC (Ping timeout: 245 seconds) 2010-10-30T08:28:36 i think i managed to fix a bunch of bugs in my bot and get it into a decent shape again 2010-10-30T08:28:51 i need to work on my first turn strategy though 2010-10-30T08:28:57 on some maps i just pick bad starting planets 2010-10-30T08:29:00 and get overrun 2010-10-30T08:34:31 *** mceier has joined #aichallenge 2010-10-30T08:34:48 *** Sarin has joined #aichallenge 2010-10-30T08:35:00 yeah, whenever i try to make the bot smart for itself it just falls to pieces 2010-10-30T09:00:27 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-30T09:02:02 *** rangzen has joined #aichallenge 2010-10-30T09:05:16 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-10-30T09:12:46 Top 10 players: rsergio(4072), bocsimacko(4070), Accoun(4068), GreenTea(4060), felixcoto(3982), rs5147(3981), joakim_mjardner(3939), dmj111(3935), Slin-.-(3928), shangas(3870) 2010-10-30T09:16:56 *** Cold-Phoenix has joined #aichallenge 2010-10-30T09:17:22 *** delt0r has joined #aichallenge 2010-10-30T09:18:02 *** delt0r___ has quit IRC (Ping timeout: 245 seconds) 2010-10-30T09:28:41 *** phreeza has joined #aichallenge 2010-10-30T09:49:18 *** yasith has joined #aichallenge 2010-10-30T09:52:53 *** Olathe has joined #aichallenge 2010-10-30T10:12:46 Top 10 players: GreenTea(4114), rsergio(4106), bocsimacko(4102), Accoun(4102), felixcoto(4020), rs5147(3989), joakim_mjardner(3972), Slin-.-(3963), dmj111(3954), shangas(3909) 2010-10-30T10:15:04 *** sigh has quit IRC (Remote host closed the connection) 2010-10-30T10:15:32 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-10-30T10:21:55 antimatroid: I don't know about the band of radius idea, but to me it looks like all games people complain about show a higher production over the last 50 or so turns for the player who lost. While the winner could have turned the match he was about to loose around, he didn't attempt to in such a long time and lost his advantage. It is fair to say that he lost 'by points' as in boxing. 2010-10-30T10:26:43 While the game rules must not become complicated it would be good to see a solution. How about this: IF a player has more production than the other, continue until he also has more ships. Any comments? 2010-10-30T10:28:50 the limit of turns pus another dimension to the game 2010-10-30T10:29:10 imo, it requires better AI to handle it 2010-10-30T10:29:35 s/pus/puts 2010-10-30T10:29:45 it just means you need to try and execute wins faster 2010-10-30T10:30:02 oh boy, here goes, a fairly large new thing i'm about to run 2010-10-30T10:30:04 it'll probably crash 2010-10-30T10:30:07 so ^^ 2010-10-30T10:30:16 i don't dare run mine yet 2010-10-30T10:30:19 yep 2010-10-30T10:31:04 i had to draw some circles and triangles on a piece of paper to work out my strategy 2010-10-30T10:31:12 im still unsure on how to pick to attack 2010-10-30T10:31:23 but i think somehow inversing my heuristic for defense should work 2010-10-30T10:31:51 yeah pen and paper help 2010-10-30T10:32:15 my "moves" involve multiple sources and a single destination, and i might send many fleeets from a single planet at different points in time 2010-10-30T10:32:27 yea 2010-10-30T10:32:32 pooling resources 2010-10-30T10:33:01 and then a pure strategy is just a vector of such moves 2010-10-30T10:33:10 yep 2010-10-30T10:33:15 that was my initial idea 2010-10-30T10:33:15 then mixed strategy assigns probability distribution over them, now i need solutions :() 2010-10-30T10:33:20 :)* 2010-10-30T10:33:34 i've generalised the problem somewhat 2010-10-30T10:33:57 it runs fine in my head 2010-10-30T10:34:12 unforuntatley its still a while from running on my system :p 2010-10-30T10:34:23 i can't imagine how many segfaults will be there :P 2010-10-30T10:34:25 *** roginn has joined #aichallenge 2010-10-30T10:42:29 anyone of the people doing payoff matrices awake? 2010-10-30T10:42:32 antimatroid? 2010-10-30T10:45:47 *** Mathnerd314 has joined #aichallenge 2010-10-30T10:48:03 I am 2010-10-30T10:48:07 no luck so far 2010-10-30T10:49:05 i'm thinking of doing that too now... but i want to come up with a way to have incomplete matrices... sortof guessing which will be the best entry iteratively 2010-10-30T10:49:39 a bit like minimax on a 2ply tree, where each players moves are a ply 2010-10-30T10:52:24 phreeza: sure what's up? 2010-10-30T10:52:30 i am just testing it out now though 2010-10-30T10:52:49 phreeza: do you have any idea how to do that? 2010-10-30T10:52:49 have you tried something like i described above? incomplete matrices? 2010-10-30T10:53:01 there's a number of way's i have tried things, nothing has done well, i'll list them off 2010-10-30T10:53:20 yea i have some ideas... but i'm sure there must be some research done about it 2010-10-30T10:53:27 so i'm looking for some keywords ;) 2010-10-30T10:54:48 the simplest way would be do the diagonal first and then do the next entry with inices (i,j) = argmax_i , argmin_j P_ij 2010-10-30T10:54:54 decision trees (these work best for me), i make one move per child at each node, then make moves off new information 2010-10-30T10:54:54 game trees, each player makes one move per turn and state doesn't increment, keep playing each other in that turn, pick moves from an entire path 2010-10-30T10:54:54 each player makes all their moves for the turn at each node (use decision tree? it's slow) and state increments every second layer 2010-10-30T10:54:54 payoff matrix, each player's pure strategy is a vector of moves and payoffs are done based off state from both players pure strategies 2010-10-30T10:55:14 oh you want to solve it? you can use iterated strict dominance to reduce the size 2010-10-30T10:55:22 *** wvdschel has joined #aichallenge 2010-10-30T10:56:19 minimax is fairly efficient on zero sum payoff matrices if you've kept your payoff function zero sum? 2010-10-30T10:56:50 but there might not be any saddle points, at which point you need another choice, even then just playing a ne isn't going to always do you well 2010-10-30T10:57:25 i can expand on anything there if you want? 2010-10-30T10:57:36 cool, let me read 2010-10-30T10:57:39 one thing i wouldn't mind playing around with yet and haven't worked out is some kind of minimax regret 2010-10-30T10:58:07 what do you mean exactly by pure strategy? 2010-10-30T10:58:08 which sort of takes into account how much you regret not making a different move for each of theirs 2010-10-30T10:58:28 pure strategy are the actions you have available to you, a mixed strategy is just assigning probabilities over them 2010-10-30T10:58:53 so in rock, paper scissors, those are your pure strategies, a mixed strategy is saying that for any game you will play each with a 1/3 probability 2010-10-30T10:58:56 ok like game theory 2010-10-30T10:58:59 yep 2010-10-30T10:59:04 yep 2010-10-30T10:59:12 i thought maybe you mean something higher level 2010-10-30T10:59:19 like attack planet, snipe 2010-10-30T10:59:24 i'm just throwing whatever I can at it from game theory at this point 2010-10-30T11:00:02 mega1: have you done things any differently :P 2010-10-30T11:00:19 other than your magical killTheOponnent() function 2010-10-30T11:00:20 oops, missed a page of conversation 2010-10-30T11:00:30 my bot does something similar to what you called game trees 2010-10-30T11:00:52 but its really dumb... i think my move generation is really weak 2010-10-30T11:00:54 which one? one move per node? or multiple moves per node and multiple "time" stages? 2010-10-30T11:00:56 its very random 2010-10-30T11:01:10 so a node for me is one players moves 2010-10-30T11:01:10 yeah, i can't get that to work at all 2010-10-30T11:01:23 my decision tree bot always does best, but it makes some really stupid moves 2010-10-30T11:01:23 so it only gets evaluated every other layer 2010-10-30T11:02:05 and for payoff, i advance state at the leaf nodes until there are no more fleets 2010-10-30T11:02:08 to make my zero sum game, i use a decision tree for both players and pull moves from all terminal nodes (or best x) as pure strategies, then generate the game matrix from that 2010-10-30T11:02:24 and then take the standard, projected fleets at turn 200 2010-10-30T11:02:26 yeah i do that, and possibly even "play nodes" where players just defend and attack 2010-10-30T11:02:51 ah but how do you choose what to attack? 2010-10-30T11:02:52 i think i need to refine my underlying choices more though 2010-10-30T11:03:01 pick move functions 2010-10-30T11:03:10 each move has multiple sources and one destination 2010-10-30T11:03:25 okay 2010-10-30T11:03:33 interesting 2010-10-30T11:03:38 i have that the other way roun 2010-10-30T11:03:39 d 2010-10-30T11:04:18 *** irchs has joined #aichallenge 2010-10-30T11:04:35 hm and the payoff matrix, you have sets of moves like that as the indices? 2010-10-30T11:04:45 i start with a planet i want to take, then move time outwards, and add usable ships from me inside that time limit, if it goes above what i need to make a decent move, i return that 2010-10-30T11:04:59 yeah, each pure strategy is just a vector 2010-10-30T11:05:36 aha. thats what i have for now, too... but i think i want to move it to the next layer of abstraction 2010-10-30T11:05:45 define a hierarchy of goals 2010-10-30T11:05:55 so hard 2010-10-30T11:05:59 and then generate pure strategies based on that 2010-10-30T11:06:07 i've considered trying to partition the graph, but can't work out a good way to do it 2010-10-30T11:06:08 *** irchs has quit IRC (Client Quit) 2010-10-30T11:06:11 some goals are simole 2010-10-30T11:06:19 but others are hard 2010-10-30T11:06:41 which graph 2010-10-30T11:06:48 the map? 2010-10-30T11:06:57 yeah 2010-10-30T11:07:07 sorry, i've just thought of it as a graph the whole time 2010-10-30T11:07:09 yea clustering maybe? 2010-10-30T11:07:26 yea the game tree is a graph though too ;) strictly speaking 2010-10-30T11:07:43 yeah, noncyclic :P 2010-10-30T11:08:09 some people use minimum spanning trees i think 2010-10-30T11:08:18 i don't like that 2010-10-30T11:08:19 not really grpah partitioning 2010-10-30T11:08:27 *** virdo has joined #aichallenge 2010-10-30T11:08:30 but graph theory ;) 2010-10-30T11:08:54 the only real reason (i think) for enforcing trees on games is for perfect recall 2010-10-30T11:08:59 they do a good job at preventing overextending themselves from what i've seen 2010-10-30T11:09:24 perfect recall? 2010-10-30T11:09:37 take any node, you know the exact path of play to get there 2010-10-30T11:10:29 ah 2010-10-30T11:10:36 the problem with doing gametree's with this is it's a simultaneous game and one doesn't really take account of the imperfect information when using minimax 2010-10-30T11:10:40 well, one of the problems 2010-10-30T11:11:04 THE problem is probably how many god damn possible moves there are 2010-10-30T11:11:13 yup... 2010-10-30T11:11:31 combinatorial explosion 2010-10-30T11:11:40 games are amazing like that 2010-10-30T11:11:41 is that the english word? 2010-10-30T11:12:07 even the game tree for 4x4 tic tac toe is HUGE 2010-10-30T11:12:46 Top 10 players: GreenTea(4125), Accoun(4122), bocsimacko(4115), rsergio(4100), felixcoto(4022), Slin-.-(4002), joakim_mjardner(3988), rs5147(3988), dmj111(3966), shangas(3923) 2010-10-30T11:13:00 16! 2010-10-30T11:13:01 ? 2010-10-30T11:13:26 2*10^13 2010-10-30T11:13:47 gah, i always forget how to calculate it 2010-10-30T11:13:54 *** Olathe has quit IRC (Ping timeout: 240 seconds) 2010-10-30T11:14:06 3^(4 * 4) 2010-10-30T11:14:28 max number of states 2010-10-30T11:14:44 i have tried to work out how to eliminate moves already tried 2010-10-30T11:14:48 but it's hard to remove ALL of them 2010-10-30T11:14:56 but that's including symmetries 2010-10-30T11:16:56 16! is an upper bound on the size of the game tree 2010-10-30T11:17:55 but if you merge branches it will probably get closer to 3^16 2010-10-30T11:19:22 *** wvdschel_ has joined #aichallenge 2010-10-30T11:19:50 any of you know how to properly remove symmetric moves? like say player one moves top left and top right in moves 1 and 3, but switched each time and player 2 moves in the centre 2010-10-30T11:19:56 i need the game tree to node have that twice 2010-10-30T11:20:11 to not* ha 2010-10-30T11:22:11 hm 2010-10-30T11:22:30 hashing game state is one way to do it 2010-10-30T11:22:47 rotation invariant hashing 2010-10-30T11:22:48 *** wvdschel has quit IRC (Ping timeout: 240 seconds) 2010-10-30T11:22:48 ? 2010-10-30T11:22:58 yeah, how to i do the invariant part? :P 2010-10-30T11:23:12 ah right, hehe 2010-10-30T11:23:15 minimise the distance to first filled square? 2010-10-30T11:23:19 across rows? 2010-10-30T11:23:37 and rotate the object till found? 2010-10-30T11:23:45 maybe even generate a function to transform that? 2010-10-30T11:23:54 hash field + field' + field '' + field ''' 2010-10-30T11:23:55 ? 2010-10-30T11:23:55 what you can do is rotate evertyhing before you hash 2010-10-30T11:24:01 but that could be costly of course.. 2010-10-30T11:24:12 rotation is slow, i'd use a function to find the rotation that minises it 2010-10-30T11:24:14 find a normal form 2010-10-30T11:24:17 then apply the right transformation 2010-10-30T11:24:20 that'd do it 2010-10-30T11:24:36 just search out from each corner to find min distance to player 1 i think? 2010-10-30T11:24:44 may have issues when that conflicts 2010-10-30T11:24:50 are you talking about planetwars 2010-10-30T11:24:54 or tic tac toe 2010-10-30T11:24:55 ? 2010-10-30T11:24:56 :D 2010-10-30T11:24:58 tic tac toe 2010-10-30T11:25:00 ok 2010-10-30T11:25:04 phew 2010-10-30T11:25:04 :D 2010-10-30T11:25:07 :p 2010-10-30T11:25:21 i'm a huge game theory gan :P 2010-10-30T11:25:23 fan** 2010-10-30T11:25:29 rotation in tic tac toe is not so costly hehe 2010-10-30T11:25:37 i like it too but i'm a total amateur 2010-10-30T11:25:45 but what you can do to save time, is to keep say 8 instances of the board 2010-10-30T11:25:51 Commandir: the point is to blow the dimensions out and solve larger and larger games 2010-10-30T11:25:55 and just update them each time you make a move 2010-10-30T11:26:00 i'm semi making a game theory library in my spare time 2010-10-30T11:26:19 i pulled most of my stuff for zero sum games out of that 2010-10-30T11:26:26 and fixed it for passing game state information in 2010-10-30T11:26:57 ok 2010-10-30T11:28:01 well, transpositions might be tricky to handle for arbitrary games, in a general way, if that's what you mean to do 2010-10-30T11:28:42 *** seisatsu has joined #aichallenge 2010-10-30T11:28:51 hmm, maybe I got off topic there 2010-10-30T11:28:59 what makes it trickier? 2010-10-30T11:29:05 non zero sum? 2010-10-30T11:29:17 simple way for the hashing: calulate all 8 hashes, including mirrored, then sort them lexically 2010-10-30T11:29:31 writing a function to get all pure subgame perfect nash equilibria for n-player extensive form games was a total pain in the ass 2010-10-30T11:29:33 not super efficient but does the trick 2010-10-30T11:29:53 you can do it incrementally 2010-10-30T11:30:07 do what? :P 2010-10-30T11:30:11 chess and go engines use zobrist hashing 2010-10-30T11:30:27 that way you only need to xor one value each time you make a move 2010-10-30T11:30:48 so you can do it incrementally very quickly 2010-10-30T11:31:17 *** p4p4 has joined #aichallenge 2010-10-30T11:31:46 in zobrist hashing you have a random number for each game piece for each position on the board, and each move you simply xor the hash value 2010-10-30T11:34:20 which also means you get a very nice distribution in your hash table 2010-10-30T11:34:54 i want to play around with equivalence classes for different game structures at some point 2010-10-30T11:35:09 it would be cool to have a metric for how similar 2 game states are 2010-10-30T11:35:12 start with 2x2 normal form games and work my way out 2010-10-30T11:35:15 across maps etc 2010-10-30T11:35:39 i've considered trying to change my evaluation of game states, but i have no idea how else to do it 2010-10-30T11:35:49 but its hard because a small difference in location/fleet size can have big effects 2010-10-30T11:35:55 i'm basically doing what you are now, trying to maximise ships at the end 2010-10-30T11:36:24 seems like the most obvious way to do it... 2010-10-30T11:36:43 i started out optimizing growth_1-growth_2 2010-10-30T11:38:08 i had some horrible function like growth/ships 2010-10-30T11:38:16 which was time till it got me positive growth 2010-10-30T11:38:32 but then beyond that didn't take into account growth 2010-10-30T11:38:48 *** roginn has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100423140709]) 2010-10-30T11:39:27 ah 2010-10-30T11:39:40 mine has a very bizarre effect 2010-10-30T11:39:47 *** seisatsu has quit IRC (Quit: Leaving) 2010-10-30T11:39:48 if there is a single planet left of the enemy 2010-10-30T11:39:54 it will sometimes keep that alive 2010-10-30T11:40:06 so that it can continue growing till move 200 2010-10-30T11:40:18 since that optimizes the goal 2010-10-30T11:40:42 *** irchs has joined #aichallenge 2010-10-30T11:40:43 my payoff function assigns huge scores to a win 2010-10-30T11:40:54 and adds maxturns-timewewon to the payoff 2010-10-30T11:40:59 that way it tries to take the earliest win 2010-10-30T11:41:47 yea i should do that too, but right now thats a secondary goal because it rarely even gets that far 2010-10-30T11:41:47 that worked wonderfully in tron 2010-10-30T11:41:56 the bot would pwn the other bot the moment it could 2010-10-30T11:42:32 here, if you are in a position to do that, you will probably win anyway... 2010-10-30T11:42:36 *** wvdschel_ has quit IRC (Quit: Leaving) 2010-10-30T11:42:38 things were a lot more exact for tron, in some ways i miss it, in other ways this is more awesome 2010-10-30T11:43:49 i didn't do tron 2010-10-30T11:45:24 @seen j3camero 2010-10-30T11:45:24 phreeza: j3camero was last seen in #aichallenge 3 weeks, 4 days, 13 hours, 25 minutes, and 39 seconds ago: amstan: yes, but no names, haha. 2010-10-30T11:45:31 heh 2010-10-30T11:46:16 ok i have to go home. good talk, thanks antimatroid Commandir 2010-10-30T11:46:31 np 2010-10-30T11:46:36 i'm goign to put two bots on tcp 2010-10-30T11:46:57 bot are using a zero sum game, they're picking their move from the game matrix differently though 2010-10-30T11:47:35 mine are running, but as i said they suck 2010-10-30T11:47:38 dimwit 2010-10-30T11:47:42 is what they are called 2010-10-30T11:47:59 *** perror has joined #aichallenge 2010-10-30T11:50:04 it's not doing very well already aha 2010-10-30T11:50:19 *** davidjliu has joined #aichallenge 2010-10-30T11:50:59 what is it calles? 2010-10-30T11:51:04 called 2010-10-30T11:51:04 zerosum1 and 2 2010-10-30T11:51:20 timeout 2010-10-30T11:51:29 it keeps timing out 2010-10-30T11:51:50 :P 2010-10-30T11:52:04 ok i'm off 2010-10-30T11:52:06 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-30T12:04:33 *** Naktibalda_ has joined #aichallenge 2010-10-30T12:07:08 *** wvdschel has joined #aichallenge 2010-10-30T12:12:47 Top 10 players: Accoun(4170), bocsimacko(4168), GreenTea(4136), rsergio(4135), felixcoto(4060), Slin-.-(4058), rs5147(4013), dmj111(3982), joakim_mjardner(3965), Hazard(3958) 2010-10-30T12:14:05 Accoun: congrats :P 2010-10-30T12:14:56 picking nash equilibria is a bad idea i think 2010-10-30T12:15:23 *** phreeza has joined #aichallenge 2010-10-30T12:15:55 *** Bobng has left #aichallenge ("Leaving") 2010-10-30T12:19:14 ahh, i have another idea 2010-10-30T12:19:38 have players repeatedly do best responses and hope it converges on a decent move 2010-10-30T12:25:06 *** edcba has quit IRC (Ping timeout: 240 seconds) 2010-10-30T12:25:45 *** Bobng has joined #aichallenge 2010-10-30T12:29:45 antimatroid: so you're actually trying and implementing all these strategies? 2010-10-30T12:32:09 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-10-30T12:37:36 yeah, it's fairly easy for me to switch between them now 2010-10-30T12:37:45 i don't think my underlying decisions are good enough yet though 2010-10-30T12:38:59 Zannick: ping? 2010-10-30T12:39:10 janzert: pong 2010-10-30T12:39:37 hi, working on getting golang setup and trying to get the compile_anything stuff correct for it 2010-10-30T12:40:32 it basically compiles sorta like java, with one command taking all the .go files and producing an intermediate 2010-10-30T12:40:46 afaik it should work out of the box 2010-10-30T12:40:47 then the linker takes the intermediate file and produces the executable 2010-10-30T12:40:54 it's not :/ 2010-10-30T12:41:10 :/ 2010-10-30T12:41:18 it seems right now compile_anything expects the first command to produce MyBot directly 2010-10-30T12:42:05 oh, i'm an idiot 2010-10-30T12:42:17 I think it just needs a second ExternalCompiler line maybe? 2010-10-30T12:42:21 yeah, the language is missing its second step 2010-10-30T12:42:32 ok 2010-10-30T12:43:19 but there is no glob for the second so just something like ([""], ExternalCompiler(comp_args["Go"][1])) right? 2010-10-30T12:44:06 yeah 2010-10-30T12:44:44 thanks, just wanted to work through it with you to make sure I was understanding it correctly 2010-10-30T12:45:16 *** wvdschel_ has joined #aichallenge 2010-10-30T12:45:16 *** wvdschel has quit IRC (Read error: Connection reset by peer) 2010-10-30T13:04:15 *** antimatroid has quit IRC (Ping timeout: 252 seconds) 2010-10-30T13:12:47 Top 10 players: Accoun(4162), bocsimacko(4161), GreenTea(4138), rsergio(4127), Slin-.-(4100), felixcoto(4068), dmj111(4031), rs5147(4003), joakim_mjardner(3983), Hazard(3945) 2010-10-30T13:15:13 *** mceier has quit IRC (Quit: leaving) 2010-10-30T13:18:05 *** mceier has joined #aichallenge 2010-10-30T13:19:55 *** antimatroid has joined #aichallenge 2010-10-30T13:25:02 does no stupid questions mean no question is stupid or stupid questions aren't allowed? yes I realise this is a stupid question 2010-10-30T13:26:14 :) 2010-10-30T13:28:44 lavalamp_: ping? 2010-10-30T13:31:57 hi is qqqt1 @ tcp around here? 2010-10-30T13:32:11 plz msg me 2010-10-30T13:33:07 *** demerzel_ has joined #aichallenge 2010-10-30T13:33:43 man i thought i finished rudimentary target selection but they are all zero lol 2010-10-30T13:36:09 advance apologies to whoever is going to match me today on tcp 2010-10-30T13:40:30 qqqt1 anyone knows him? :S 2010-10-30T13:48:03 man i cannot figure out why my 'available ships' works but then zeroes out before i can use it lol 2010-10-30T13:49:53 debugging future states is going to be fun too - already see one bug for some reason it sees turn 200 as the optimal turn to attack lol 2010-10-30T13:50:50 sounds like your problem is using a language that allows side effects :p 2010-10-30T13:53:12 except this language is so amazingly fast it's blowing my mind 2010-10-30T13:53:38 i calc state for every planet every turn and scan each one for a ratio then run it back through and calc mine/enemy fleet max 2010-10-30T13:53:46 turns take <.001s lol 2010-10-30T13:54:34 i might actually be able to achieve my dream of a true minmax bot in c++ after all 2010-10-30T13:54:37 c# was hitting the limit 2010-10-30T13:55:25 kinda funny that you say youre an optimization freak and just came across c++ now :P 2010-10-30T13:55:33 well i haven't touched it in forever 2010-10-30T13:55:53 i mean i played with it before i ever became a pro dev and none of my jobs ever let me use it 2010-10-30T13:56:17 when you just play with it is a lot different than having a structured idea of what you want 2010-10-30T13:56:20 http://shootout.alioth.debian.org/u32q/which-programming-languages-are-fastest.php 2010-10-30T13:56:29 yea, true 2010-10-30T13:56:43 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-10-30T13:56:52 wow java is pretty impressive 2010-10-30T13:57:00 i always thought it ws slower than .net 2010-10-30T13:57:02 *** Apophis has joined #aichallenge 2010-10-30T13:57:08 what are they benching on 2010-10-30T13:57:13 yea i'm surprised too :) 2010-10-30T13:57:21 the machine? 2010-10-30T13:57:41 oh mono 2010-10-30T13:57:43 nm lol 2010-10-30T13:58:26 *** p4p4 has joined #aichallenge 2010-10-30T13:58:47 2 things strike me on that - ada is still around and how the hell is scala faster than java 2010-10-30T13:59:16 isn't it using the java vm 2010-10-30T13:59:25 better bytecode optimizations? might also come down to coding style 2010-10-30T14:00:07 remember the second line on the page "No. Which programming language implementations have the fastest benchmark programs?" :P 2010-10-30T14:00:21 too bad they didn't bench js 2010-10-30T14:00:21 actually scala and java have almos6t the same medan 2010-10-30T14:00:24 madian 2010-10-30T14:00:27 i almost did a js bot just to be esoteric 2010-10-30T14:00:33 go is pretty slow! 2010-10-30T14:00:37 what stopped me is i have no idea what it uses for stdin/out 2010-10-30T14:00:45 i thought it was a fast language 2010-10-30T14:00:56 hehe document.write? 2010-10-30T14:00:58 no idea 2010-10-30T14:01:34 holy shit python really is slow as hell i thought it was exaggerated 2010-10-30T14:01:40 now i really don't understand its popularity 2010-10-30T14:02:15 running speed is very rarely a factor in real life 2010-10-30T14:02:22 python is the most pleasant language to code in 2010-10-30T14:02:46 most of the time in most programs is spent waiting on I/O anyway 2010-10-30T14:02:48 i know almost all the c(whatevers), initial BASIC/PASCAL and php/js 2010-10-30T14:02:55 and python is totally alien to me 2010-10-30T14:03:00 its used for stuff where developer time is the bottleneck, not computer time 2010-10-30T14:03:15 you should give it a try 2010-10-30T14:03:20 its an eye opener 2010-10-30T14:03:25 if i just wanted implicit typing and foreach i'd use VB.NET lol 2010-10-30T14:03:52 i really don't understand its draw - how is it easier than say C# 2010-10-30T14:04:12 ignoring the fact the syntax is totally alien to me 2010-10-30T14:04:45 not to mention the python console while nice doesn't even begin to compare to vs debugging 2010-10-30T14:04:51 ive never used c++ but the whole language just feels right. the way OO is done for example 2010-10-30T14:05:09 rot12("ada") 2010-10-30T14:05:21 altho i guess they made some variant for vs now 2010-10-30T14:05:30 i thought ada was long dead 2010-10-30T14:05:34 had no idea it was still around 2010-10-30T14:05:55 grouped it with COBOL and PASCAL as just obsolete 2010-10-30T14:06:23 do you guys know map ids with the starting bases close? 2010-10-30T14:06:27 lots of legacy systems around... i think you can make a good living as a cobol guru 2010-10-30T14:06:53 mega1: 1, iirc? 2010-10-30T14:07:05 mega 1 i pulled all the maps down to a db haha 2010-10-30T14:07:05 but i might be wrong 2010-10-30T14:07:09 i can check if you want 2010-10-30T14:07:44 thanks 2010-10-30T14:07:46 anyone knows who qttt1 on tcp is? 2010-10-30T14:07:55 i actually pulled all top 100 games and moves down too 2010-10-30T14:08:02 but have never used it -- too many variables haha 2010-10-30T14:08:49 mega1: 118 on tcp server 2010-10-30T14:09:04 http://72.44.46.68/canvas?game_id=780153 2010-10-30T14:09:50 that's one 2010-10-30T14:10:04 30 and 700 2010-10-30T14:10:06 there was one I used to lose on with the bases at minimum distance 2010-10-30T14:10:49 700 is not close at all 2010-10-30T14:11:56 mega1: are you qqqt1? 2010-10-30T14:12:01 Do any of you own a VPS? 2010-10-30T14:12:12 Error323: no, my bot is bocsimacko 2010-10-30T14:12:15 I just want to ask a few questions if you do 2010-10-30T14:12:17 hold on a sec lol 2010-10-30T14:12:18 ah k 2010-10-30T14:12:48 here 2010-10-30T14:12:48 http://pastebin.com/9CRPE46Y 2010-10-30T14:12:48 Top 10 players: bocsimacko(4193), GreenTea(4170), Slin-.-(4152), Accoun(4143), rsergio(4129), felixcoto(4071), joakim_mjardner(4044), dmj111(4043), rs5147(4013), Hazard(3962) 2010-10-30T14:13:04 ordered by distance 2010-10-30T14:13:23 i have a linux vps why 2010-10-30T14:13:31 runs my minecraft server i never play on haha 2010-10-30T14:14:08 wow, thank you 2010-10-30T14:14:35 sure 2010-10-30T14:14:41 i've never tested that data but looks right 2010-10-30T14:14:46 demerzel_: nice 2010-10-30T14:14:53 demerzel_: can I ask you a few questions about it? 2010-10-30T14:15:00 sure 2010-10-30T14:15:09 How much do you pay for it and what specs is it? 2010-10-30T14:15:09 altho my knowledge of linux is functional at best 2010-10-30T14:15:25 Oh, its only about the price. Me and a friend want to start selling VPS servers 2010-10-30T14:15:32 i pay $25/mo for a 1mbps unmetered, 2gig ram and 400gig space 2010-10-30T14:15:36 it's a cheap deal tho 2010-10-30T14:15:39 i hunted for a while 2010-10-30T14:15:47 normally that setup is around 50 2010-10-30T14:15:58 Bobng i think thats a very competitive market... linode rackspace amazon... 2010-10-30T14:16:10 Yeah demerzel_, that is a pretty good deal 2010-10-30T14:16:12 yea i know the guys i get from are reselling 2010-10-30T14:16:20 but idc really 2010-10-30T14:16:22 Well could you take a look at the prices we drew up? 2010-10-30T14:16:23 http://69.42.214.138:85/VPSForge.com/services.php 2010-10-30T14:16:48 phreeza: true, or though amazon has a different pricing structure and market 2010-10-30T14:17:07 the plans in the link seem to beat 99% of other plans I have seen 2010-10-30T14:17:09 that looks like a good deal 2010-10-30T14:17:12 feel free to correct me though 2010-10-30T14:17:13 altho i don't see what the line speed is 2010-10-30T14:17:31 yea they seem quite cheap 2010-10-30T14:17:36 there are 2 types of vps customers imo 2010-10-30T14:17:46 the ones that will never use it and the others that want unmetered for as cheap as possible 2010-10-30T14:17:52 *** nullkuhl_ has joined #aichallenge 2010-10-30T14:18:04 demerzel_: the line speed is in the table 2010-10-30T14:18:12 it says port speed 2010-10-30T14:18:12 "Port speed", the line speed overall is 2gb/s 2010-10-30T14:18:21 on a vps nobody stops you from hammering the cpu right? 2010-10-30T14:18:24 ah shared then 2010-10-30T14:18:34 nope phreeza 2010-10-30T14:18:39 most vps set a hard limit on how much cpu you can use 2010-10-30T14:18:41 but there is a token credit system 2010-10-30T14:18:43 ie 50% or whatever 2010-10-30T14:18:45 we dont set a limit 2010-10-30T14:18:48 ah ok 2010-10-30T14:18:49 we give you priority 2010-10-30T14:19:12 so if the CPU is unused you can use all of it, but if a larger plan is using it then you get reduced priority over that one 2010-10-30T14:19:21 yea the main reason mine is cheap is actually cuz of the ram 2010-10-30T14:19:28 i don't get why 1gig->2gig is a 3x increase 2010-10-30T14:19:32 not picking on you that's a vps standard 2010-10-30T14:19:44 yeah, well they dont sell 2010-10-30T14:19:59 but idk 2010-10-30T14:20:06 ram is the premium I guess 2010-10-30T14:20:24 whats a seedbox? 2010-10-30T14:20:26 mega1: what do you use to test with outside of tcp 2010-10-30T14:20:36 well, if this does take off the ground and we do start selling I was going to donate a super instance or two to #aichallenge 2010-10-30T14:20:40 in return for some publicity. 2010-10-30T14:20:44 i used iouri_'s visualizer for a while till i realized it had bugs causing me to hunt for nonexistent bugs in mine 2010-10-30T14:20:49 phreeza: it does what it says on the tin - used for torrents. 2010-10-30T14:20:56 ahh 2010-10-30T14:21:03 you get a restricted shell to run rtorrent 2010-10-30T14:21:04 demerzel_: shell scripts and http://github.com/albertz/planet_wars-cpp.git 2010-10-30T14:21:06 and lots of HD space. 2010-10-30T14:21:17 ok 2010-10-30T14:21:24 mega1: dammit i'm on a windows machine lol 2010-10-30T14:21:32 albertz scares me a little or i would get his 2010-10-30T14:21:50 so you'll get a ton of dmca notices then i guess? 2010-10-30T14:22:07 idk why i haven't built my own by now, i've written turn logic so many times now i could do it blindfolded in 5 languages 2010-10-30T14:22:41 *** eburnette_ has quit IRC (Ping timeout: 265 seconds) 2010-10-30T14:22:47 lol why does albert scare you? 2010-10-30T14:22:58 so what does everyone here think about adding a comment sent to bots that gives the turns remaining in the game something like "# : turns_remaining 123" 2010-10-30T14:23:00 idk just way too eager and bad english lol 2010-10-30T14:23:04 i'm paranoid tho 2010-10-30T14:23:04 having beers with him in an hour 2010-10-30T14:23:06 hahaha 2010-10-30T14:23:22 hes a good guy :) 2010-10-30T14:23:25 are you really lol 2010-10-30T14:23:28 yea 2010-10-30T14:23:28 can i come work at your job 2010-10-30T14:23:29 *** p4p4 has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2010-10-30T14:23:36 were students 2010-10-30T14:23:42 oh damn lol 2010-10-30T14:23:47 who was the guy who worked in c++ shop 2010-10-30T14:23:51 lava oops 2010-10-30T14:25:01 Bobng: i like the link at the top to save people the scrolling for seedboxes haha 2010-10-30T14:25:13 is that for people with 640x480 2010-10-30T14:25:20 yep :) 2010-10-30T14:25:25 i should write a barebones game engine i think 2010-10-30T14:25:29 or some SEO stuff? 2010-10-30T14:25:31 yea your ui is pretty nice 2010-10-30T14:25:34 I showed it to someone and they didn't see the seedboxes 2010-10-30T14:25:38 they were like "where is it??" 2010-10-30T14:25:40 haha 2010-10-30T14:25:45 so I thought I would put a link just in case. 2010-10-30T14:25:50 ab testing ftw 2010-10-30T14:26:21 phreeza: DMCA shouldn't worry us 2010-10-30T14:26:26 if all our users use private trackers 2010-10-30T14:26:31 else yeah we will have to shut them down 2010-10-30T14:26:55 aha, i'm not really familiar with all the torrent stuff. 2010-10-30T14:27:18 i have torrented for almost a decade lol 2010-10-30T14:27:34 i wish i understood freenet tho it's too complicated for me 2010-10-30T14:27:45 tor is nice but way too damn slow 2010-10-30T14:27:52 i used to, until our student residence got raided that scared me hehe 2010-10-30T14:28:08 install truecrypt problem solved 2010-10-30T14:28:15 now i use megaupload 2010-10-30T14:28:18 etc 2010-10-30T14:28:28 that's way more trackable than torrents 2010-10-30T14:28:35 just set your torrent port to 80 ;) 2010-10-30T14:28:39 yea but youre not liable 2010-10-30T14:28:57 in germany they get you for seeding 2010-10-30T14:29:01 not downloading 2010-10-30T14:29:03 afaik 2010-10-30T14:29:04 someone debug my code for me 2010-10-30T14:29:15 i am unmotivated but i want a semi-working bot :( 2010-10-30T14:31:04 i feel bad using the tcp server as a debugger lol 2010-10-30T14:32:09 janzert: seems better to permanently fix the turn limit at say 200 in the specs imho 2010-10-30T14:36:28 yea if you change it it will jack a lot of stuff up 2010-10-30T14:36:35 for me anyway lol 2010-10-30T14:36:41 or rather if it's dynamic 2010-10-30T14:36:46 i can change the constant obv 2010-10-30T14:37:02 demerzel_: I don't think there is any chance it will become dynamic 2010-10-30T14:37:15 well that would be the point of a turns remaining wouldn't it 2010-10-30T14:37:19 but there is some debate about changing the length 2010-10-30T14:37:39 it let's you know what the length of the game actually is 2010-10-30T14:38:33 it should always be decreasing by 1 every move 2010-10-30T14:38:55 right but why would you need to state it 2010-10-30T14:39:08 i assume most bots that knwo what turn it is also know how many are left 2010-10-30T14:39:09 because it might start at 200 currently and 300 later 2010-10-30T14:39:22 well that would be dynamic then 2010-10-30T14:39:23 later as in a different game 2010-10-30T14:40:13 it is just more in line with the protocol 'philosophy' to send it every move instead of just at the beginning of the game 2010-10-30T14:40:37 *** perror has quit IRC (Quit: Bye all !) 2010-10-30T14:41:02 it preserves the full game state being given at every move that way 2010-10-30T14:41:21 ah i see ok 2010-10-30T14:42:07 wouldn't hurt anything to add it iirc the starter packs would just ignore it 2010-10-30T14:42:11 as would mine 2010-10-30T14:42:24 right, all the starter packs will ignore 2010-10-30T14:42:31 as will any bot that follows the spec 2010-10-30T14:42:52 just make it a T xxx after planets 2010-10-30T14:43:05 but is it really worth mucking around with halfway through the contest? 2010-10-30T14:43:09 that could break bots that follow the spec 2010-10-30T14:43:14 why is that 2010-10-30T14:43:21 it just looks for P or F or go\n 2010-10-30T14:43:28 it's none of those so it would just pass by it 2010-10-30T14:43:52 the spec doesn't say anything about ignoring unexpected lines 2010-10-30T14:44:12 the spec does give a comment line we can reuse for this information 2010-10-30T14:44:25 oh god don't do that just add a sentence to the spec ffs lol 2010-10-30T14:44:33 i.e. anything starting with a # should already be ignored 2010-10-30T14:44:56 and we can do "# :" for any additional 'meta' information 2010-10-30T14:44:59 that would be way more annoying 2010-10-30T14:45:10 to have to check only the first planet or whatever for a # then whatever 2010-10-30T14:45:13 than just a T xxx 2010-10-30T14:45:14 that's the only way this will happen I'm sure 2010-10-30T14:45:33 I know there are bots that will die with a T xxx 2010-10-30T14:45:53 well none of the starter packs and we're talking like 30 seconds of code 2010-10-30T14:46:16 iirc the starter packs dont' keep track of turns at all 2010-10-30T14:46:18 gotta run, time to eat 2010-10-30T14:46:22 nope they don't 2010-10-30T14:46:30 nothing broken then 2010-10-30T14:51:42 *** mceier has quit IRC (Quit: leaving) 2010-10-30T15:03:57 oh boy my first stack corruption 2010-10-30T15:04:01 sweet 2010-10-30T15:04:29 As long as I'm not forced to change my command line options for a "-t 200" that is ok. 2010-10-30T15:08:25 just remember you don't get any command line options on the servers 2010-10-30T15:08:28 *** wvdschel_ has quit IRC (Read error: Connection reset by peer) 2010-10-30T15:08:31 *** wvdschel has joined #aichallenge 2010-10-30T15:12:48 Top 10 players: GreenTea(4151), bocsimacko(4149), Accoun(4143), Slin-.-(4125), rsergio(4109), felixcoto(4079), joakim_mjardner(4071), dmj111(4056), rs5147(4027), luksian(3971) 2010-10-30T15:13:09 @seen mega1 2010-10-30T15:13:09 Mathnerd314: mega1 was last seen in #aichallenge 52 minutes and 5 seconds ago: demerzel_: shell scripts and http://github.com/albertz/planet_wars-cpp.git 2010-10-30T15:13:19 @seen Accoun 2010-10-30T15:13:19 Mathnerd314: Accoun was last seen in #aichallenge 3 days, 19 hours, 1 minute, and 24 seconds ago: perfect 2010-10-30T15:13:58 interesting. 2010-10-30T15:14:01 *** yasith has quit IRC (Ping timeout: 265 seconds) 2010-10-30T15:17:44 lol just tried to track down a stack corruption bug for like half an hour but was my stupid fault in trying to concat string instead of << 2010-10-30T15:18:13 bad when your debug log is bugged lol 2010-10-30T15:18:31 Mathnerd314: looking for me? 2010-10-30T15:19:17 *** wvdschel has quit IRC (Ping timeout: 245 seconds) 2010-10-30T15:19:19 not unless you're giving away your strategy :p 2010-10-30T15:19:32 *** wvdschel has joined #aichallenge 2010-10-30T15:20:02 oh he is he just doesn't know it yet 2010-10-30T15:20:10 my database will tell me all his secrets! mwahaha 2010-10-30T15:20:29 i just need to factor in every possible variable 2010-10-30T15:20:32 should be done in an hour or so 2010-10-30T15:24:30 *** wvdschel has quit IRC (Ping timeout: 252 seconds) 2010-10-30T15:24:57 mega1: you seem to be beating all of the competition (despite the flux in rankings) 2010-10-30T15:25:18 Mathnerd314: I think my bot's timing out 2010-10-30T15:25:27 in the sense that it doesn't finish thinking 2010-10-30T15:25:44 it's always running on server #0 according to janzert. 2010-10-30T15:26:12 *** yasith has joined #aichallenge 2010-10-30T15:26:19 Which is strange because it's quite quick. 2010-10-30T15:26:44 *** McLeopold has joined #aichallenge 2010-10-30T15:27:01 and also: http://code.google.com/p/ai-contest/issues/detail?id=207 2010-10-30T15:28:16 frankly, it's much better on tcp 2010-10-30T15:28:22 any plans to rewrite in another language? 2010-10-30T15:28:27 no way 2010-10-30T15:28:45 why should I? 2010-10-30T15:28:46 your loss :p 2010-10-30T15:29:17 I guess they get around to installing sbcl on the cloud servers. 2010-10-30T15:29:41 and server #0 won't be magically faster if I rewrite it 2010-10-30T15:30:05 *** wvdschel has joined #aichallenge 2010-10-30T15:30:05 *** wvdschel has joined #aichallenge 2010-10-30T15:31:20 yeah. 2010-10-30T15:32:17 but I'm guessing that getting languages working at all (e..g Go) is going to take more priority than fixing bugs 2010-10-30T15:35:04 alternately, we could have switched to CPU time back in the beginning of the contest... 2010-10-30T15:35:11 I'm more worried about not knowing anything about how they plan to run the final tournament. 2010-10-30T15:35:23 switch cpu back?? 2010-10-30T15:35:31 ah 2010-10-30T15:35:33 sorry, can't read 2010-10-30T15:36:01 that would make more sense 2010-10-30T15:36:10 he meant "read cpu time used instead of wall time" 2010-10-30T15:36:21 *** yasith has quit IRC (Ping timeout: 272 seconds) 2010-10-30T15:36:47 *** nullkuhl_ has quit IRC (Ping timeout: 245 seconds) 2010-10-30T15:37:43 well at least they know you and might actually make an attempt to fix lisp lol 2010-10-30T15:38:27 installing sbcl on the cloud server is not a big deal 2010-10-30T15:38:36 but maybe it's the compilation 2010-10-30T15:38:55 are the cloud servers different from server #0? 2010-10-30T15:39:05 (not binary compatible)? 2010-10-30T15:39:16 i have no idea how they work the cloud servers and i get teh feeling i'm not alone lol 2010-10-30T15:39:27 i'm pretty sure 0 is their actual server 2010-10-30T15:40:14 explains why it's slow and different from the others 2010-10-30T15:40:18 right. 2010-10-30T15:40:49 I get the feeling the contest is bigger than expected 2010-10-30T15:41:33 just like tron 2010-10-30T15:42:37 yeah. so how does your bot time out? it just doesn't finish the computation? 2010-10-30T15:43:10 *** nullkuhl_ has joined #aichallenge 2010-10-30T15:43:18 ^ mega1 2010-10-30T15:44:14 Mathnerd314: http://ai-contest.com/visualizer.php?game_id=6359004 2010-10-30T15:44:35 there is an uncharacteristics stretch of no orders around turn 10 2010-10-30T15:44:41 *** wvdschel_ has joined #aichallenge 2010-10-30T15:44:58 that's what happens with the current code if it doesn't finish computation within 0.8s 2010-10-30T15:45:27 I can make it smarter and return the best move found so far, but it takes about 0.1s on my laptop. 2010-10-30T15:45:57 *** wvdschel has quit IRC (Ping timeout: 252 seconds) 2010-10-30T15:46:00 Still I have to check that I'm not off the mark. 2010-10-30T15:47:51 why does it take so long to find the move after your search? are you you not looking at orders? 2010-10-30T15:47:53 people have said timers are off 2010-10-30T15:48:06 i saw one guy said if he set his 'move timer' to more than 0.1s he'd time out 2010-10-30T15:48:43 s/you you/you/ 2010-10-30T15:49:03 * Mathnerd314 is multitasking 2010-10-30T15:49:27 Mathnerd314: I don't understand the question. 2010-10-30T15:49:44 demerzel_: that is 100ms and NOT 1000ms-100ms, right?! 2010-10-30T15:49:49 *** Azrathud has quit IRC (Ping timeout: 264 seconds) 2010-10-30T15:49:58 yes 2010-10-30T15:50:34 cpu cycles != wall clock time 2010-10-30T15:50:44 for whatever reason it's only getting 10% cycles is what i gather from that 2010-10-30T15:51:03 You say his move timer was cpu time based? 2010-10-30T15:51:21 *** McLeopold has quit IRC (Ping timeout: 252 seconds) 2010-10-30T15:51:35 i'd guess it is - most people don't even know how to check wall clock 2010-10-30T15:51:50 if it was wall clock time 0.1s then that would make no sense at all 2010-10-30T15:51:51 mega1: if you were doing some sort of minimax search, you would be looking at the orders you could give, and could quickly return the best set of orders found and print them out. 2010-10-30T15:52:14 *** wvdschel has joined #aichallenge 2010-10-30T15:52:18 is that a joke? when you enter timing in google you almost always find wall time articles first 2010-10-30T15:52:49 *** wvdschel_ has quit IRC (Read error: Connection reset by peer) 2010-10-30T15:52:51 some languages don't even come with cpu time functions i believe. 2010-10-30T15:52:51 Mathnerd314: yes, I could be smarter about it. But I didn't see the need, because it's plenty fast on my laptop. 2010-10-30T15:52:57 regardless of that it doesn't make sense as wall clock 2010-10-30T15:53:16 i ... hope so ^ 2010-10-30T15:54:54 what? that mosquito got concious again? i hit you with a 250g package of mixed nuts, you should be DEAD! 2010-10-30T15:55:13 i would review the sandbox code but python is so hard to read for me 2010-10-30T15:57:30 *** wvdschel has quit IRC (Ping timeout: 252 seconds) 2010-10-30T15:59:51 *** hellman has quit IRC (Remote host closed the connection) 2010-10-30T16:00:14 *** wvdschel has joined #aichallenge 2010-10-30T16:00:14 *** wvdschel has joined #aichallenge 2010-10-30T16:02:46 *** Mekanik has joined #aichallenge 2010-10-30T16:12:48 Top 10 players: Slin-.-(4191), GreenTea(4164), Accoun(4163), bocsimacko(4157), rsergio(4138), felixcoto(4090), dmj111(4067), joakim_mjardner(4063), rs5147(4035), luksian(3983) 2010-10-30T16:16:53 *** nullkuhl_ has quit IRC (Ping timeout: 272 seconds) 2010-10-30T16:20:31 *** nullkuhl_ has joined #aichallenge 2010-10-30T16:21:14 *** Accoun has quit IRC () 2010-10-30T16:25:34 *** Mekanik has quit IRC (Quit: WeeChat 0.3.3) 2010-10-30T16:27:57 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-30T16:35:21 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-30T16:36:07 *** Todd200 has joined #aichallenge 2010-10-30T16:36:18 hi people 2010-10-30T16:36:53 *** Accoun has joined #aichallenge 2010-10-30T16:36:58 yo todd 2010-10-30T16:37:18 * Todd200 trying to build his first decient bot 2010-10-30T16:38:35 its a bot mostly based on guessing the future and acting accordingly, it wont have any other strategy for now 2010-10-30T16:40:50 gotta start somewhere 2010-10-30T16:41:35 not much going on on tcp right now 2010-10-30T16:42:09 anyone up for sparring? 2010-10-30T16:42:59 Todd200: usually bot has a strategy and doesn't try to guess future 2010-10-30T16:43:54 Todd200: and a lot doesn't have a stategy and doesn't try to guess future 2010-10-30T16:44:20 true 2010-10-30T16:44:37 oh, I thought that even RageBot has a strategy 2010-10-30T16:45:01 Todd200: but the ~top500, they all have future guessing and have strategy 2010-10-30T16:45:19 and top50 have A LOT of strategy 2010-10-30T16:45:22 its more of a proof of concept to tie into its battle strategy later 2010-10-30T16:45:35 im 14, ill go slowly 2010-10-30T16:46:09 most of top 10 use supply trains 2010-10-30T16:46:39 With the time of translation in my head I thought that you rank 14 :) With no strategy, whoua! 2010-10-30T16:47:02 lol, im rank 2300 or so 2010-10-30T16:47:25 Understand why supply train is good then implement it :) 2010-10-30T16:47:30 that's rather high compared to number of bots (about 4000) 2010-10-30T16:47:40 *** phreeza has joined #aichallenge 2010-10-30T16:48:14 i made two tweaks to the starter package, 2010-10-30T16:48:21 accually 4 2010-10-30T16:48:27 2300 you have a start of strategy but probably no defense, no growthrate warning before attack, etc. 2010-10-30T16:48:35 yep 2010-10-30T16:49:02 I know cause i'm rank 2000 :) 2010-10-30T16:49:20 it has a max of 35 fleets, it gets most cost effive panet and sends 1/4 of its strongest planets ships to it 2010-10-30T16:49:49 and it likes using ships from it smaller planets insted of its large ones 2010-10-30T16:50:10 oh you can stop sending "part" of planets, you need to calculate exactly how much ships you send 2010-10-30T16:50:19 true 2010-10-30T16:50:38 *** Azrathud has joined #aichallenge 2010-10-30T16:50:48 check the forum, there is 3 very good strategy guide with a lot of ideas 2010-10-30T16:50:49 thats something i need to add 2010-10-30T16:50:54 i did 2010-10-30T16:51:12 ill add them to my future bot later 2010-10-30T16:51:33 *** phreeza has quit IRC (Client Quit) 2010-10-30T16:51:51 Your bot run so the fun part is now, you'll implement quick and grow up in rank quickly, it's fun before the top 500 2010-10-30T16:52:32 ? 2010-10-30T16:52:40 check my bot out : 2010-10-30T16:52:44 :) 2010-10-30T16:52:49 http://ai-contest.com/profile.php?user_id=6765 2010-10-30T16:52:52 I was in top50 before I decided to rewrite all in another language and to stay in the ranks, you need improvment everyday. Very stressing :) 2010-10-30T16:53:02 fun thought 2010-10-30T16:53:10 though 2010-10-30T16:53:39 Python -> Java without any experience in Java :) Pain in the ass!!! 2010-10-30T16:53:55 im using python, 2010-10-30T16:54:05 Keep it! 2010-10-30T16:54:09 lol 2010-10-30T16:54:14 Why did you decide to rewrite it? 2010-10-30T16:54:34 Performance reasons? 2010-10-30T16:54:41 Learn a new langage, that's all. 2010-10-30T16:54:46 Neat :) 2010-10-30T16:55:25 *** demerzel_ has quit IRC (Ping timeout: 265 seconds) 2010-10-30T16:55:49 I was stuck to improve my bot (implementing queue and supply train) without complete rewriting so I thought why not in another langage :) 2010-10-30T16:56:01 the problem is with attacking a planet with exactly the amount of ships i need is that a good defense can snipe my target 2010-10-30T16:56:34 Todd200: you're talking about taking neutral planets over? 2010-10-30T16:57:35 yep, it could send 6 ships to claim what ever i took 2010-10-30T16:58:42 im going to make my bot a bit more aggressive and resend it 2010-10-30T17:00:24 Test on the tcp server, if you can defend yourself there, you'll rock on the official server :) 2010-10-30T17:01:29 im yet to make the tcp work 2010-10-30T17:02:19 at first you can test against example bots 2010-10-30T17:02:22 my bot always crashes 2010-10-30T17:02:38 im yet to make that work too 2010-10-30T17:02:55 the only place my bot ever works is where i upload it 2010-10-30T17:03:58 * Todd200 tries for the 20th time 2010-10-30T17:04:01 How do you test your bot locally on your computer? 2010-10-30T17:04:12 i dont 2010-10-30T17:04:23 i wish i could 2010-10-30T17:04:32 What error do you get when you try? 2010-10-30T17:05:03 i cant make the java command run my python program 2010-10-30T17:05:17 *** wh1teside has joined #aichallenge 2010-10-30T17:05:20 paste your command here. 2010-10-30T17:06:31 for tcp: tcp.exe 72.44.46.68 995 Toddk -p password C:\Python22\python.exe MyBot.py 2010-10-30T17:07:07 well, meant more about the local testing against sample bots. 2010-10-30T17:07:09 it connects then it says an error with main() in my bot 2010-10-30T17:07:21 which came with the starter pack to submit my move 2010-10-30T17:08:11 you are using python 2.2? 2010-10-30T17:09:27 accually that was a copy paste 2010-10-30T17:09:36 it doesnt if i use 25 eather 2010-10-30T17:10:07 i have 22, 25, and 27 installed 2010-10-30T17:10:43 *bot timed out, nice it ran on tcp 2010-10-30T17:10:50 but it stall failed 2010-10-30T17:10:53 still 2010-10-30T17:11:07 What about the sample bots? 2010-10-30T17:12:06 *** phreeza has joined #aichallenge 2010-10-30T17:12:48 Top 10 players: Slin-.-(4192), bocsimacko(4162), Accoun(4159), rsergio(4157), GreenTea(4151), felixcoto(4109), rs5147(4064), dmj111(4060), joakim_mjardner(4048), luksian(4029) 2010-10-30T17:12:49 ill try now 2010-10-30T17:16:34 *** delt0r___ has joined #aichallenge 2010-10-30T17:17:05 *** fluffluff1 has joined #aichallenge 2010-10-30T17:17:20 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-30T17:18:19 *** delt0r has quit IRC (Ping timeout: 272 seconds) 2010-10-30T17:18:43 *** fluffluff has quit IRC (Ping timeout: 276 seconds) 2010-10-30T17:19:26 *** phreeza has joined #aichallenge 2010-10-30T17:20:47 found my first problem 2010-10-30T17:21:31 'python is not reconized as an internal/external command or batch file 2010-10-30T17:22:51 \bin\python.exe ? 2010-10-30T17:23:07 copy/paste your command here... if python in not in your path, do the same way you did it for tcp (c:\python\python.exe) 2010-10-30T17:23:21 *** phreeza has quit IRC (Client Quit) 2010-10-30T17:23:24 i tried just python in cmd 2010-10-30T17:23:41 create a bat file 2010-10-30T17:23:43 no luck 2010-10-30T17:23:57 you're doing the "java -jar tools/PlayGame.jar ..." command? 2010-10-30T17:24:01 bot call must be a single filename for TCP, not two 2010-10-30T17:24:07 at first the unix file system layout confused me, but now I kinda like that everything user-executable resides in /usr/bin 2010-10-30T17:24:30 im on windows xp 2010-10-30T17:24:50 although my hacked ipod has a unix filesystem 2010-10-30T17:25:26 before playing on the tcp server, I think you might as well get your bot working on your home environment using the playgame.jar ;) 2010-10-30T17:25:56 but it works on their server, which is all i really care about :p 2010-10-30T17:26:54 you enjoy waiting long periods of time before seeing your bot in action? or submitting it non stop? 2010-10-30T17:30:13 i dont really mind... 2010-10-30T17:31:25 *** wvdschel has quit IRC (Read error: Connection reset by peer) 2010-10-30T17:32:58 *** wvdschel has joined #aichallenge 2010-10-30T17:32:58 *** wvdschel has joined #aichallenge 2010-10-30T17:34:25 *** phreeza has joined #aichallenge 2010-10-30T17:35:03 *** phreeza has quit IRC (Client Quit) 2010-10-30T17:39:06 *** wvdschel has quit IRC (Read error: Connection reset by peer) 2010-10-30T17:40:55 *** wvdschel has joined #aichallenge 2010-10-30T17:40:55 *** wvdschel has joined #aichallenge 2010-10-30T17:44:43 *** Todd200 has quit IRC (Ping timeout: 265 seconds) 2010-10-30T17:44:44 *** Olathe has joined #aichallenge 2010-10-30T17:45:25 *** irchs has quit IRC (Quit: irchs) 2010-10-30T17:49:21 *** McLeopold has joined #aichallenge 2010-10-30T17:50:17 *** alch has joined #aichallenge 2010-10-30T17:50:19 *** McLeopold has quit IRC (Client Quit) 2010-10-30T17:51:34 *** fluffluff1 has quit IRC (Quit: Leaving.) 2010-10-30T17:54:56 *** fluffluff has joined #aichallenge 2010-10-30T17:57:36 *** Gudradain has joined #aichallenge 2010-10-30T18:05:51 *** nullkuhl_ has quit IRC (Remote host closed the connection) 2010-10-30T18:12:50 Top 10 players: Slin-.-(4189), bocsimacko(4158), Accoun(4156), rsergio(4153), GreenTea(4147), felixcoto(4106), rs5147(4060), dmj111(4057), luksian(4035), joakim_mjardner(4023) 2010-10-30T18:20:45 *** phreeza has joined #aichallenge 2010-10-30T18:26:06 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-30T18:29:32 *** Commandir has left #aichallenge 2010-10-30T18:32:59 *** coofamani has joined #aichallenge 2010-10-30T18:33:59 *** Bobng has quit IRC (Quit: Leaving) 2010-10-30T18:37:24 *** Bobng has joined #aichallenge 2010-10-30T18:51:16 *** demerzel_ has joined #aichallenge 2010-10-30T18:51:25 *** Sarin has quit IRC (Remote host closed the connection) 2010-10-30T18:51:52 rangzen: what language did you go from/to? i remember you on tcp 2010-10-30T18:52:52 *** wvdschel has quit IRC (Read error: Connection reset by peer) 2010-10-30T18:53:10 *** wvdschel has joined #aichallenge 2010-10-30T18:53:10 *** wvdschel has joined #aichallenge 2010-10-30T18:54:28 *** sigh has joined #aichallenge 2010-10-30T18:54:29 *** wvdschel has quit IRC (Read error: Connection reset by peer) 2010-10-30T18:55:47 hahaha http://en.wikipedia.org/wiki/Extreme_programming 2010-10-30T18:55:58 eXtreme programming - we do whatever 2010-10-30T18:57:28 demerzel_: from Python to Java 2010-10-30T18:57:40 why java 2010-10-30T18:58:19 Read a lot about it but never tried so … 2010-10-30T18:59:08 ah just curious - i went to c++ cuz of the c# issues but now that i have i am glad i did because i can do an insane amount of calcs in c++ 2010-10-30T18:59:25 although it's day 3 or 4 and i still don't have a working bot 2010-10-30T18:59:28 The library is fine, Eclipse really impressive but it's so verbose compare to Python :( 2010-10-30T18:59:30 at least its state is there 2010-10-30T18:59:53 ah python is the one i don't know 2010-10-30T19:00:00 it just looks crazy weird to me 2010-10-30T19:00:55 started from october but I have not enough time per day :) 2010-10-30T19:00:59 *** Gudradain has quit IRC (Quit: Page closed) 2010-10-30T19:01:38 i bet its hard to port from python 2010-10-30T19:01:41 implicit typing and foreachs everywhere 2010-10-30T19:01:43 *** Bobng has quit IRC (Ping timeout: 255 seconds) 2010-10-30T19:01:59 Very hard to discover a langage from start after almost 10 years of Python. But with already a good bot, it's easy to translate tools, logging, etc. 2010-10-30T19:02:18 yea i started on this contest when i was unemployed got a job a few weeks back and my dev time has dropped a lot 2010-10-30T19:02:29 but then i got to do it at work cuz we're waiting on docs haha 2010-10-30T19:03:04 java is a good lang to start on imo it's portable and the syntax is close enough to a lot of other langs 2010-10-30T19:03:51 I miss Python a lot for short code but now I'm glad to understand a little more why Java is used on big project, you can't code fast, you need to think a lot 2010-10-30T19:04:32 lol what a slogan 2010-10-30T19:04:36 python: you don't need to think much 2010-10-30T19:05:25 The thing I miss the most in Python is : nb_my_ships = sum([f.NumShips() for f in pw.Fleets() if f.Owner() == 1 and f.DestinationPlanet() == p.PlanetID()]) 2010-10-30T19:05:35 so long in Java … 2010-10-30T19:05:43 to have the same 2010-10-30T19:06:18 you can think fast, it's a bit different :) 2010-10-30T19:06:40 foreach(f in pw.Fleets(nb_myships += f.Owner()==1 ? f.Numships() : 0) 2010-10-30T19:06:46 *** superflit has joined #aichallenge 2010-10-30T19:07:00 oops brackets not parens but w/e 2010-10-30T19:07:51 what does that even do 2010-10-30T19:08:00 the f.dest = planetid makes me think that's some weird index retrieval 2010-10-30T19:08:08 you don't have the filter on fleet's target and you have to declare all that :) 2010-10-30T19:08:29 it sums the ships from 1 planet if it's yours? 2010-10-30T19:08:33 that's a weird approach 2010-10-30T19:09:01 no you sum all ships from all your fleet going to a particular planet 2010-10-30T19:09:29 oh oh fleets 2010-10-30T19:09:48 about the only thing that's different then is the sum() vs a int nb_my_ships = 0; 2010-10-30T19:09:53 it's a bit strange the first time but when you use list everywhere it's so comfortable :) 2010-10-30T19:10:20 nb is a good abbr for inbound i'm gonna steal it 2010-10-30T19:11:24 i can't believe there are so few people that don't have a fleet struct 2010-10-30T19:11:32 i don't 2010-10-30T19:12:41 also why nobody uses arrays when the dims are known is weird to me too 2010-10-30T19:12:49 Top 10 players: bocsimacko(4208), Slin-.-(4194), felixcoto(4191), GreenTea(4183), Accoun(4175), rsergio(4148), dmj111(4101), rs5147(4062), Hazard(4044), protocolocon(4039) 2010-10-30T19:12:58 how do you know that? 2010-10-30T19:13:01 in c++ at least i can kinda understand it because you have to track the length i 2010-10-30T19:13:23 Naktibalda: know what 2010-10-30T19:13:45 who is Slin- 2010-10-30T19:13:48 all the things that nobody does :) 2010-10-30T19:13:58 because i see code snippets all the time 2010-10-30T19:14:39 that is of course not mentioning that most people don't change anything beyond doturn 2010-10-30T19:15:16 i was going to offer up my c# adv starter pack but i dont' want to give false hope lol 2010-10-30T19:15:23 gave it to a couple coworkers at least 2010-10-30T19:38:30 *** Cold-Phoenix has quit IRC () 2010-10-30T19:41:39 *** RainCT has quit IRC (Remote host closed the connection) 2010-10-30T19:41:42 no fleet struct? ships incoming to a planet in a certain turn is a core structure of the game i think 2010-10-30T19:44:13 *** alpha-aquilae has joined #aichallenge 2010-10-30T19:49:49 *** phreeza has joined #aichallenge 2010-10-30T19:49:52 *** alpha-aquilae has left #aichallenge 2010-10-30T19:54:23 int[201] myinc int[201] enemy inc 2010-10-30T19:54:41 incatturn(int turn) { return enemyinc[turn]; } 2010-10-30T19:55:24 that's all you need 2010-10-30T19:55:52 i did try to just add it in but it's too hard with neutral stuff you have to track whose fleet it is while it's a neutral 2010-10-30T20:03:48 *** seisatsu has joined #aichallenge 2010-10-30T20:08:01 *** phreeza has quit IRC (Quit: kthxbai) 2010-10-30T20:12:50 Top 10 players: Slin-.-(4224), bocsimacko(4216), GreenTea(4196), felixcoto(4191), Accoun(4166), dmj111(4148), rsergio(4144), Hazard(4072), rs5147(4067), joakim_mjardner(4053) 2010-10-30T20:17:00 who is Slin-._ 2010-10-30T20:17:04 never seen him before 2010-10-30T20:17:08 beating mega1 even 2010-10-30T20:28:04 *** TravisD has joined #aichallenge 2010-10-30T20:38:42 *** Frontier has quit IRC (Ping timeout: 240 seconds) 2010-10-30T20:40:02 *** rangzen has quit IRC (Quit: Quitte) 2010-10-30T20:56:57 *** Frontier has joined #aichallenge 2010-10-30T21:06:47 *** Tobu has quit IRC (Ping timeout: 245 seconds) 2010-10-30T21:10:34 *** Tobu has joined #aichallenge 2010-10-30T21:12:55 Top 10 players: bocsimacko(4211), Slin-.-(4207), GreenTea(4176), felixcoto(4171), Accoun(4144), dmj111(4129), rsergio(4101), Hazard(4059), rs5147(4041), joakim_mjardner(4036) 2010-10-30T21:19:09 oh no my bot is a failure 2010-10-30T21:19:17 at least it makes a single move now lol 2010-10-30T21:24:39 demerzel_: I'm pity for you. I just read about so called God's alrorithms and I try to implement on for a partial problem of PlanetWars. 2010-10-30T21:25:01 *algorithms, *one 2010-10-30T21:25:35 my bot surprises me sometimes 2010-10-30T21:25:36 http://72.44.46.68/canvas?game_id=785417 2010-10-30T21:26:16 i'm up to 90% against ragebot 2010-10-30T21:26:24 now that it's finally running 2010-10-30T21:29:04 *** Naktibalda_ has quit IRC (Remote host closed the connection) 2010-10-30T21:29:12 There may be scenarios where you can only reach a stalemate against RageBot. 2010-10-30T21:30:27 i'm not sure if these are them 2010-10-30T21:30:27 yes the scenario where you do nothing 2010-10-30T21:30:33 i will look into them, though 2010-10-30T21:30:37 you always draw if you just do nothing 2010-10-30T21:30:48 bhasker: nice 2010-10-30T21:31:03 once in bluemoon! 2010-10-30T21:31:05 my dev is at a standstill http://72.44.46.68/canvas?game_id=785445 2010-10-30T21:31:06 in a 2010-10-30T21:31:09 time to make myself food while the gauntlet runs 2010-10-30T21:31:10 it is pissed at me for missing a move on turn #11 2010-10-30T21:31:20 ouch 2010-10-30T21:31:22 and so keeps recommending that as the best move forever lol 2010-10-30T21:31:27 bhasker: Your opponent lacked a second layer defense. He had enough ships to defend that planet, but didn't move them as if they were needed for some other defense. 2010-10-30T21:31:43 Frontier: that is #1 player right now lol 2010-10-30T21:31:46 yea well the good thing is my opponnent is #1 2010-10-30T21:31:49 lol:d 2010-10-30T21:31:55 @rankings 2010-10-30T21:31:57 bhasker: Top 10 players: bocsimacko(4220), Slin-.-(4212), GreenTea(4183), felixcoto(4177), Accoun(4141), dmj111(4128), rsergio(4118), Hazard(4057), rs5147(4048), joakim_mjardner(4044) 2010-10-30T21:32:02 ^^ 2010-10-30T21:32:04 haha, great 2010-10-30T21:32:10 bosimacko? scrub 2010-10-30T21:32:34 well that scrub beats everyone on TCP :-\ 2010-10-30T21:32:37 my test guy is bogussomething or other 2010-10-30T21:32:39 like 96% of the time 2010-10-30T21:32:44 if i can beat him every time i know i have a decent bot 2010-10-30T21:32:51 heh 2010-10-30T21:32:55 bogusbrainpower? 2010-10-30T21:33:00 yea 2010-10-30T21:33:19 janstaunton or whatever was way too easy 2010-10-30T21:33:23 bogus is a good midlevel benchmark 2010-10-30T21:33:30 heh 2010-10-30T21:33:38 an annoying bot to beat 2010-10-30T21:33:41 is obligatory shampoo 2010-10-30T21:33:46 or cosmic kangaroo 2010-10-30T21:33:53 both are like 250ish elo rated bots 2010-10-30T21:33:55 i gotta get my bot to make more than one move haha 2010-10-30T21:33:58 and beat some very good bot 2010-10-30T21:34:05 bhasker: I see now, he actually *thought* he didn't lose that planet, but the distance calculation went wrong. 2010-10-30T21:34:05 at least the one move it makes isn't bad lol 2010-10-30T21:34:21 which planet are you talking about 2010-10-30T21:34:49 The ships he sent to defend the lower planet you attacked took a route that was too long. 2010-10-30T21:34:55 ah yea 2010-10-30T21:35:07 tbh you won because his 'safe expansion' backfired when you took that upper planet 2010-10-30T21:35:23 most of his losses are due to his expansion being checked early 2010-10-30T21:36:21 he has a nice linear expansion but it means that if you jump a couple planets ahead without leaving yourself open you can check him 2010-10-30T21:37:13 obv easier said than done ;) 2010-10-30T21:37:20 i still have a lot of issues figuring out optimal early expansion 2010-10-30T21:37:30 same here 2010-10-30T21:37:35 i am trying out a few different heuristics 2010-10-30T21:37:36 very fine line between too conservative and too aggressive 2010-10-30T21:37:39 for optimal selection 2010-10-30T21:37:52 but all of them have some maps on which i end up picking wrong start planet 2010-10-30T21:37:56 and its downhill from there 2010-10-30T21:38:03 yep it's hard to say 2010-10-30T21:38:11 i thought oldman's method of go for broke was a good one 2010-10-30T21:38:21 then i realized if i did better locally i could easily check him when he got to me 2010-10-30T21:38:35 now i have no idea lol 2010-10-30T21:38:47 i was going to emulate mega1's expansion but even that has vulnerabilities 2010-10-30T21:39:34 Is mega1's expansion worth a look at? 2010-10-30T21:39:50 oh yea he has a nice flowing expansion 2010-10-30T21:39:53 it's very nice 2010-10-30T21:40:08 he is looking 2 - 4 ply ahead for sure 2010-10-30T21:41:04 not to say that all of us aren't looking ahead but he's actually planning ahead 2-4 ply 2010-10-30T21:41:07 I'm going to waste some time trying to look 200 ply ahead and realizing that it would take forever. Wish me luck. 2010-10-30T21:41:19 well it depends on what you're doing 2010-10-30T21:41:24 i calc all 200 turns every turn it takes almost zero time 2010-10-30T21:41:29 but trying to chain moves is another story 2010-10-30T21:43:15 It does include lots of chained moves and branches ^^. I know it is silly, but somehow I have to try it. 2010-10-30T21:45:07 if i ever get 1-ply down i'll be taking it as far as i can 2010-10-30T21:45:11 i'm still under 50ms 2010-10-30T21:46:06 and there's a few while loops left i can get rid of lol 2010-10-30T21:48:30 Anyone know if Groovy is installed yet? 2010-10-30T21:48:48 At some point I decided to write my own server code and plug my bot in as a dll. It was too annoying to debug the bot through PlayGame.jar and cerr output. My old bot is also under 50ms I guess. 2010-10-30T21:52:11 yea i wrote my own server code a while back but i really need the visualizer to see where i screwed up lol 2010-10-30T21:52:33 i should just grab the js and do it locally but i'm already 5 projects deep lol 2010-10-30T21:53:35 so you guys are writing in c++? 2010-10-30T21:55:00 oh no, my bot is losing to copybot sometimes 2010-10-30T21:55:36 i am writing in c++ now yea 2010-10-30T21:55:42 was C# 2010-10-30T21:55:52 this bot is a total rewrite this week lol 2010-10-30T21:56:24 heh i should implement a minimax version just to see how it does 2010-10-30T21:56:28 *** sigh has quit IRC (Remote host closed the connection) 2010-10-30T21:56:29 just watch mine 2010-10-30T21:56:30 but lazy to rewrite from scratch 2010-10-30T21:56:38 i'm stuck on minimax because it's all i know 2010-10-30T21:56:53 game theory concepts are gibberish to me i don't understand how a relative number can be more useful than an absolute one 2010-10-30T21:57:11 doesn't mean they're wrong that's just my brain's response 2010-10-30T21:58:12 i cannot figure out why my bot won't do jack past the first turn it has targets >:( 2010-10-30T21:59:23 lol 2010-10-30T21:59:31 only thing it will do is launch after my planets are taken to the taken planets 2010-10-30T21:59:33 bizarre 2010-10-30T22:00:48 * TravisD wonders if he should write his own server to help debug 2010-10-30T22:01:44 nah i code the tcp shit directly into my bot now lol 2010-10-30T22:02:02 i do wish the tcp vis had planet numbers tho sucks pasting it into that one every battle 2010-10-30T22:02:08 demerzel_: I whish it would be no work to write a visualizer on top of my server that takes 'drawables' from my bot so I can visually log stuff like a bar showing how many % of a planets ships are currently locked down for defense or assigned to a future mission. 2010-10-30T22:02:53 *** ghi has joined #aichallenge 2010-10-30T22:05:05 ah well i just log that shit to a file 2010-10-30T22:05:47 And how do you get the log for a certain round you look at in ShowGame.jar? ... 2010-10-30T22:07:24 i don't use ShowGame.jar 2010-10-30T22:07:37 but i do log turn# with every msg 2010-10-30T22:07:54 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2010-10-30T22:07:56 And the visualizer you use shows you the turn number? 2010-10-30T22:08:05 all the js visualizers do 2010-10-30T22:08:20 if not just have it stderr the turn # 2010-10-30T22:08:26 on whatever you're using 2010-10-30T22:09:03 argh now i completely broke it 2010-10-30T22:09:34 can I run a js visualizer stand-alone, like the ShowGame.jar? 2010-10-30T22:11:17 or... do you concatenate a html file with the visualizer and your server output to a new html file that you display in your browser ? Oo 2010-10-30T22:11:18 *** Azrathud has quit IRC (Ping timeout: 250 seconds) 2010-10-30T22:12:50 Top 10 players: bocsimacko(4206), Slin-.-(4180), GreenTea(4164), felixcoto(4156), Accoun(4129), dmj111(4129), rsergio(4097), Hazard(4067), rs5147(4049), joakim_mjardner(4031) 2010-10-30T22:14:24 *** antimatroid1 has joined #aichallenge 2010-10-30T22:14:41 Why don't you guys just use the JBotManager that is posted on the forums somewhere? 2010-10-30T22:14:52 i finally downloaded that yesterday 2010-10-30T22:15:29 Because we want to debug our bot from inside our favorite IDE. 2010-10-30T22:16:29 *** antimatroid has quit IRC (Ping timeout: 240 seconds) 2010-10-30T22:16:34 *** superflit has quit IRC (Quit: superflit) 2010-10-30T22:16:48 That doesn't work by pressing 'debug' if the bot is running in a second process. 2010-10-30T22:17:12 Seems like more effort than it's worth 2010-10-30T22:17:14 i actually built a proxy bot that just read from a file and passed to stdio 2010-10-30T22:17:46 i use tcp as debug now even tho i feel a little guilty about it 2010-10-30T22:18:17 i spent hours and hours tryng to debug errors that didn't exist using iouri's visualizer 2010-10-30T22:18:27 sorta soured me on the whole local game thing 2010-10-30T22:18:32 as long as you don't need to stop for more than one second or have reconstructable game states. 2010-10-30T22:18:43 hah http://72.44.46.68/canvas?game_id=786064 2010-10-30T22:19:49 TravisD: I was also able to help a newbie with his bot, by debugging it in my server. He sent two 'go' lines after each turn. 2010-10-30T22:21:20 that was nice of you 2010-10-30T22:21:33 TravisD: Also you can run tests (matches on all 100 maps for example) within one process. Things that took minutes to complete, run in seconds now 2010-10-30T22:22:54 *** n9thbit has quit IRC (Quit: n9thbit) 2010-10-30T22:23:06 JBotManager doesn't offer any batch mode, iirc. 2010-10-30T22:23:48 ...so it is a bit more than just the 'debug' button ;) 2010-10-30T22:24:39 Frontier: its overkill to write your own server to have the batch mode. Its easy enough to accomplish that with the provided PlayGame.jar file and a small script 2010-10-30T22:24:44 But I don't want to argue, lol 2010-10-30T22:26:42 *** Tobu has quit IRC (Ping timeout: 240 seconds) 2010-10-30T22:27:34 yea but if you have a serious bot you probably have written the turn logic 2 or 3 times now 2010-10-30T22:27:42 it's literally something like 20 lines 2010-10-30T22:27:42 *** dmj111 has joined #aichallenge 2010-10-30T22:27:47 not a huge effort 2010-10-30T22:29:01 while loops are the devil 2010-10-30T22:29:15 *** sigh has joined #aichallenge 2010-10-30T22:29:16 haha my shitbot is playing mega1 2010-10-30T22:29:29 i got this 2010-10-30T22:29:30 *** Tobu has joined #aichallenge 2010-10-30T22:29:39 haha, you'll be pantsed :P 2010-10-30T22:29:48 it doesn't evne have defense yet 2010-10-30T22:29:58 altho i do calc it i haven't figured out if i want to make it optional or not 2010-10-30T22:31:23 *** Azrathud has joined #aichallenge 2010-10-30T22:31:24 demerzel_: paste replay? 2010-10-30T22:31:29 defense is the first thing i do 2010-10-30T22:31:51 then i use trees for deciding expansionary and offensive moves, then pick redistribution at the end 2010-10-30T22:32:01 my bot sucks though 2010-10-30T22:32:19 oh it crapped out invalid move 2010-10-30T22:32:26 yea i've always made defense automatic 2010-10-30T22:33:03 voidptr said something that made me consider making it optional tho - busy defending a 1 growth planet instead of capping the enemy who left theirs 2010-10-30T22:33:14 it's really damn hard to figure out how to make it optional though 2010-10-30T22:45:16 *** TravisD has quit IRC (Quit: TravisD) 2010-10-30T22:49:16 *** Tobu has quit IRC (Ping timeout: 255 seconds) 2010-10-30T22:52:44 *** TravisD has joined #aichallenge 2010-10-30T22:54:18 *** seisatsu is now known as notseisatsu 2010-10-30T23:04:28 *** rebelxt_ has joined #aichallenge 2010-10-30T23:05:33 *** pgpaskar_ has quit IRC (Quit: leaving) 2010-10-30T23:06:30 *** eburnette_ has joined #aichallenge 2010-10-30T23:06:41 *** TravisD has quit IRC (Quit: TravisD) 2010-10-30T23:12:51 Top 10 players: bocsimacko(4207), Slin-.-(4182), GreenTea(4180), felixcoto(4163), dmj111(4147), Accoun(4138), rsergio(4086), Hazard(4072), rs5147(4051), joakim_mjardner(4031) 2010-10-30T23:15:56 I need a better function to score the current game state 2010-10-30T23:16:22 don't we all :P 2010-10-30T23:17:42 I tried one that gives a bonus for getting planets closer to the enemy but that causes it to become overextended 2010-10-30T23:18:23 antimatroid1 I was wondering are you using antimatroids in your bot or was that just a random name? 2010-10-30T23:21:52 The new donna and boris bots are doing pretty good; guess it's not too late for new people to enter 2010-10-30T23:23:32 i agree that it is not too late. On tcp though, its always hard to tell who is who. 2010-10-30T23:26:54 i finally completed my bot 2010-10-30T23:27:04 and it is looooooooooosing 2010-10-30T23:27:10 yay 2010-10-30T23:27:54 mega1 can you go on vacation or something for a couple weeks so the rest of us can catch up? :) 2010-10-30T23:28:04 *** janzert has quit IRC (Read error: Connection reset by peer) 2010-10-30T23:32:09 *** hellman has joined #aichallenge 2010-10-30T23:33:00 omfg 2010-10-30T23:33:21 i had a sign backwards 2010-10-30T23:33:26 in a rather important case 2010-10-30T23:34:05 eburnette: random that, although they do actually look like they could be used 2010-10-30T23:37:34 @rankings 2010-10-30T23:37:36 antimatroid1: Top 10 players: GreenTea(4169), bocsimacko(4161), felixcoto(4155), Slin-.-(4139), Accoun(4118), dmj111(4114), rsergio(4062), Hazard(4039), joakim_mjardner(4024), rs5147(4019) 2010-10-30T23:38:09 the server should implode whenever a java bot gets on top 2010-10-30T23:39:10 *** McLeopold has joined #aichallenge 2010-10-30T23:40:15 i need to implode 2010-10-30T23:40:16 *** narnach has quit IRC (Ping timeout: 276 seconds) 2010-10-30T23:41:16 i can't work on my bot today anyway, i have an exam tomorrow 2010-10-30T23:41:37 meh, i need to work on my bot... my brain is just waking up though 2010-10-30T23:43:24 java is great for debugging 2010-10-30T23:45:30 *** notseisatsu has quit IRC (Quit: Leaving) 2010-10-30T23:45:39 *** Rubicon-|-Cross has joined #aichallenge 2010-10-30T23:48:59 Ballsy first move by bocs on this game: http://72.44.46.68/canvas?game_id=786712 2010-10-30T23:50:09 very ballsy 2010-10-30T23:56:19 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-10-30T23:56:53 *** Olathe has quit IRC (Ping timeout: 265 seconds) 2010-10-30T23:57:49 ugh, each time I try to edit my bot, it gets worse...