2010-11-11T00:01:45 I can't tell if this is good or bad? http://72.44.46.68/canvas?game_id=958489 2010-11-11T00:02:46 antimatroid1: the map generator tests each planet as it places it 2010-11-11T00:03:09 hmmm, there are a number of different ways i could do summetries, should i make a bunch of generators? 2010-11-11T00:03:18 should i draw some examples? 2010-11-11T00:03:34 do you have a white board? 2010-11-11T00:04:14 i have paint 2010-11-11T00:04:29 McLeopold: that is an interesting game. which part are you asking about? 2010-11-11T00:04:46 well, i survived which is good 2010-11-11T00:04:53 *** Utkarsh has joined #aichallenge 2010-11-11T00:04:58 but, it think it was a poor game on my part 2010-11-11T00:05:25 yes. the end looked a little funny, but if you are looking for the 200th turn, maybe it knew it had it won? 2010-11-11T00:05:33 nah, it was a great game! no mistakes at all. I think its good enough. you can go watch TV now 2010-11-11T00:05:40 I don't look for turn 200 2010-11-11T00:05:46 (the yes was for " you won which is good.") 2010-11-11T00:06:40 I have an urge to watch tv... 2010-11-11T00:06:46 i like the funneling you are doing. mine is too simplistic. 2010-11-11T00:06:55 it was crazy 2010-11-11T00:07:01 I got to tone it down some 2010-11-11T00:09:17 I'm sending stuff backwards and crossing paths 2010-11-11T00:09:31 so, which part did you like? 2010-11-11T00:09:40 http://imgur.com/B6QND.png just shows a couple of examples 2010-11-11T00:09:47 but there are lots of different possible symmetries 2010-11-11T00:10:02 oh wait, the bottom one was the same aha 2010-11-11T00:10:13 flip the blue ones for the bottom 2010-11-11T00:10:33 McLeopold: i have been thinking about working at pushing stuff from the back to support the front planets under attack, allowing the front to launch. as of yet, I don't do anythin like that. 2010-11-11T00:10:44 and then the other portions don't need to be symmetric with the sections starting with players planets etc. 2010-11-11T00:10:46 allowing the front to launch *more*. 2010-11-11T00:11:01 antimatroid1: that's really still just two symmetries 2010-11-11T00:11:06 dmj111: i am halfway through writing such a function :P 2010-11-11T00:11:16 Zannick: a couple of examples :P 2010-11-11T00:11:23 the left ones are horizontal symmetry, and the other two are rotational 2010-11-11T00:11:23 ahhh good. i didn't spill the beans then. 2010-11-11T00:12:15 yeah, I solved that awhile ago 2010-11-11T00:12:23 McLeopold: this is more aesthetic, but I liked all of the neutrals being captured on turn 15. 2010-11-11T00:12:34 It's picking good planets to attack that I have trouble with. 2010-11-11T00:12:40 *** delt0r___ has joined #aichallenge 2010-11-11T00:12:50 dmj111: huh? what do you mean you don't send ships from the back? 2010-11-11T00:13:07 it looks like you do? 2010-11-11T00:13:16 sigh: I do, but I attack from the back first, then move ships to the front after I don't have anything to attack. 2010-11-11T00:13:40 But, I always keep enough on a planet to keep from losing it to incoming ships, if I have enough to start with. 2010-11-11T00:13:48 oh, i see what you mean 2010-11-11T00:14:09 dmj111: the only advantage to not doing that is informational 2010-11-11T00:14:24 *** delt0r has quit IRC (Ping timeout: 264 seconds) 2010-11-11T00:14:33 wait, no it's not 2010-11-11T00:14:36 nevermind 2010-11-11T00:14:53 i was doing it just for the informational advantage, but i think it helps otherwise too 2010-11-11T00:16:03 dmj111: hoping for a match! 2010-11-11T00:16:04 only way is to try it and see 2010-11-11T00:16:27 Top 10 players: bocsimacko(4221), _iouri_(4088), george(4081), GreenTea(4076), dmj111(4049), rsergio(4043), felixcoto(4030), bix0r4ever(4028), Accoun(4017), rs5147(4016) 2010-11-11T00:16:36 maybe if we syncronize the start time, well have better luck matching up? 2010-11-11T00:17:52 I'll give it a try. 2010-11-11T00:18:11 I have a game at the moment, I kill my script after its over. 2010-11-11T00:18:12 *** decipher has quit IRC (Quit: Page closed) 2010-11-11T00:18:13 I'm using JBotManager, so I can hit a button 2010-11-11T00:18:19 make two really crap bots so their rank is like -999... then put your actual bots in and they'll play for sure :P 2010-11-11T00:18:49 why not just play them under two new names? :P 2010-11-11T00:18:52 sigh: it think it has more to do with timing 2010-11-11T00:18:53 yep. 2010-11-11T00:19:22 i'm tired.. not sure who I am agreeing with there :) 2010-11-11T00:19:40 east coast? 2010-11-11T00:20:12 just say when... 2010-11-11T00:20:12 yes. stayed up too late last night too, so I am going to have to crash soon. 2010-11-11T00:20:25 button on standby 2010-11-11T00:21:14 ok..... on your mark. 2010-11-11T00:21:18 now 2010-11-11T00:21:29 yes 2010-11-11T00:21:30 ha! 2010-11-11T00:22:15 it's not looking good 2010-11-11T00:22:38 hmmm, i'm just going to do the original different symmetry i intended, too much effort to randomise between types of symmetry 2010-11-11T00:22:45 i just seen the numbers, so I wasn't sure there for a while. 2010-11-11T00:22:53 i just saw... 2010-11-11T00:23:03 want to try again? 2010-11-11T00:23:06 yep 2010-11-11T00:23:36 ok... in a few seconds. 2010-11-11T00:23:42 now 2010-11-11T00:23:53 2 in a row 2010-11-11T00:24:02 we are like pros at this. 2010-11-11T00:24:20 holy crap, why is my bot doing that! 2010-11-11T00:24:40 it take a planet down to 4 ships and wraps it up in a bow for you! 2010-11-11T00:24:55 merry christmas! >:( 2010-11-11T00:24:57 i appreciate it :/ 2010-11-11T00:25:12 for some reason my bot does that sometimes 2010-11-11T00:25:14 oh wait... 2010-11-11T00:25:21 I forgot a command line switch... 2010-11-11T00:25:27 maybe not all the way to 4 2010-11-11T00:25:33 McLeopold: I designate you as official irc commentator during the final contest 2010-11-11T00:25:37 but it will hit the planet a little and then snipe from the enemy >.> 2010-11-11T00:25:59 I'll get some cuss words on hot keys... 2010-11-11T00:26:30 dmj111: 1 more, but let me turn off the suck switch. 2010-11-11T00:26:30 i'm heading to sleep. one more time (with the right switches?) 2010-11-11T00:26:36 hold on 2010-11-11T00:26:39 ha. let me know when you are ready. 2010-11-11T00:27:23 okay, on standby 2010-11-11T00:27:36 ok. 3 ... 2. 2010-11-11T00:27:38 go 2010-11-11T00:27:43 crap 2010-11-11T00:27:56 I got shampooed 2010-11-11T00:28:07 at least you will be clean. 2010-11-11T00:29:23 okay, ready to try again, but my elo is dropping... 2010-11-11T00:29:42 i'm going to lose to a 106 elo here in a second. 2010-11-11T00:29:58 I lost to philic, at 1 elo earlier 2010-11-11T00:30:13 I still have a major bug I'm not sure how to solve yet. 2010-11-11T00:30:21 mine is not consistent. i'll give the go in 5 seconds. 2010-11-11T00:30:24 It depends on the map type when it bites me. 2010-11-11T00:30:32 go 2010-11-11T00:30:38 oldman 2010-11-11T00:30:58 that is obligatory. 2010-11-11T00:31:00 at least I'm going to win this 2010-11-11T00:31:10 no, that's shampoo 2010-11-11T00:31:42 he's pretty high up there too 2010-11-11T00:31:50 sound thrashing 2010-11-11T00:32:04 mines almost done. 2010-11-11T00:32:16 i can try one more if you want. 2010-11-11T00:32:23 just one more match 2010-11-11T00:32:29 ok. 2010-11-11T00:32:33 ready in 5 seconds? 2010-11-11T00:32:36 between us 2010-11-11T00:32:40 yes, just say go 2010-11-11T00:32:42 go 2010-11-11T00:32:48 arrg 2010-11-11T00:33:25 this is amusing 2010-11-11T00:33:30 I might take down accoun now 2010-11-11T00:33:41 nope 2010-11-11T00:33:42 i have him too. i'm going to lose mine though. 2010-11-11T00:33:52 badly. 2010-11-11T00:34:21 oh good, just gave him a 67, snipe for 5 opportunity 2010-11-11T00:34:34 that's my bug, I gotta fix that 2010-11-11T00:34:52 ok. one more try. then I must pass out. We had thanksgiving lunch at work today, so on top of working on this contest all the time, I had a bunch of turkey today. 2010-11-11T00:34:55 now he's toying with me 2010-11-11T00:35:13 he's just poking my planets 2010-11-11T00:35:25 okay, ready 2010-11-11T00:35:33 ok ... 5 seconds. 2010-11-11T00:35:39 go 2010-11-11T00:35:43 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-11T00:35:50 do. 2010-11-11T00:35:52 doh. 2010-11-11T00:35:59 we should have wrote that peer to peer client 2010-11-11T00:36:05 *** sigh has joined #aichallenge 2010-11-11T00:36:11 I got a -248 elo guy 2010-11-11T00:36:13 done 2010-11-11T00:36:41 good night all 2010-11-11T00:36:45 later! 2010-11-11T00:36:49 *** McLeopold has quit IRC (Quit: Leaving) 2010-11-11T00:38:51 *** Utkarsh has quit IRC (Read error: Connection reset by peer) 2010-11-11T00:49:11 *** hellman has quit IRC (Remote host closed the connection) 2010-11-11T00:49:33 *** davidjliu has quit IRC (Quit: Page closed) 2010-11-11T00:51:24 *** virdo has quit IRC (Ping timeout: 276 seconds) 2010-11-11T00:55:41 *** tapwater has quit IRC (Quit: tapwater) 2010-11-11T01:00:39 *** Accoun has quit IRC (Ping timeout: 255 seconds) 2010-11-11T01:15:57 *** Gudradain has quit IRC (Quit: Page closed) 2010-11-11T01:16:27 Top 10 players: bocsimacko(4236), george(4086), dmj111(4054), GreenTea(4048), rsergio(4035), rs5147(4033), felixcoto(4022), bix0r4ever(4022), _iouri_(4019), Accoun(4013) 2010-11-11T01:25:57 *** sigh_ has joined #aichallenge 2010-11-11T01:25:58 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-11T01:28:23 *** yasith has joined #aichallenge 2010-11-11T01:28:42 *** yasith_ has joined #aichallenge 2010-11-11T01:29:01 *** Accoun has joined #aichallenge 2010-11-11T01:32:58 *** sigh_ has quit IRC (Remote host closed the connection) 2010-11-11T01:43:09 *** Zoost has joined #aichallenge 2010-11-11T01:51:54 hmmm, there is only two types of symmetry aye 2010-11-11T01:52:07 i can switch between that easily enough 2010-11-11T01:53:17 *** kincody has joined #aichallenge 2010-11-11T02:09:33 what do you mean 2010-11-11T02:10:48 *** Accoun has quit IRC (Ping timeout: 260 seconds) 2010-11-11T02:14:19 there are two ways to generate all possible 2 player symmetric maps randomly, 2010-11-11T02:14:19 pick both players theta (rotation from one of the axis) and radius the same 2010-11-11T02:14:19 then for each neutral, pick theta and radius 2010-11-11T02:14:19 1 neutral is (r, (ourTheta+theta)%360) the other is either 2010-11-11T02:14:19 (r, (theirTheta+theta)%360) or (r, abs(theirTheta-theta)%360)) 2010-11-11T02:14:27 *** Accoun has joined #aichallenge 2010-11-11T02:16:27 Top 10 players: bocsimacko(4227), george(4076), dmj111(4054), GreenTea(4040), rs5147(4028), rsergio(4018), felixcoto(4015), bix0r4ever(4013), _iouri_(4010), Accoun(4002) 2010-11-11T02:21:58 *** narnach has joined #aichallenge 2010-11-11T02:21:58 *** narnach is now known as narnach_mb 2010-11-11T02:51:11 *** janzert1 has joined #aichallenge 2010-11-11T02:51:56 *** janzert has quit IRC (Ping timeout: 252 seconds) 2010-11-11T03:15:47 *** p4p4 has joined #aichallenge 2010-11-11T03:16:27 Top 10 players: bocsimacko(4227), george(4076), dmj111(4054), GreenTea(4040), rs5147(4028), rsergio(4018), felixcoto(4015), bix0r4ever(4013), _iouri_(4010), Accoun(4002) 2010-11-11T03:19:48 not too long off having this map generator done, i think it should theoretically be generating random maps from the set of all possible 2 player symmetric maps with specified parameters about maps 2010-11-11T03:45:46 *** aerique has joined #aichallenge 2010-11-11T03:49:36 what's the easiest way to convert an integer to a string? 2010-11-11T03:49:40 in c++ 2010-11-11T03:54:21 *** amstan has quit IRC (Ping timeout: 255 seconds) 2010-11-11T03:55:55 *** amstan has joined #aichallenge 2010-11-11T03:55:55 *** ChanServ sets mode: +o amstan 2010-11-11T04:02:29 *** krokkrok has quit IRC (Quit: Page closed) 2010-11-11T04:04:48 *** Cold-Phoenix has quit IRC (Ping timeout: 240 seconds) 2010-11-11T04:05:48 they don't look like they're coming out symmetrically :| 2010-11-11T04:08:31 *** Cold-Phoenix has joined #aichallenge 2010-11-11T04:09:47 *** kincody_ has joined #aichallenge 2010-11-11T04:13:15 *** kincody has quit IRC (Ping timeout: 265 seconds) 2010-11-11T04:15:34 *** Cold-Phoenix has quit IRC (Ping timeout: 240 seconds) 2010-11-11T04:15:36 hmm, what would be a good test for a map to be symmetric? 2010-11-11T04:16:27 Top 10 players: bocsimacko(4227), george(4076), dmj111(4054), GreenTea(4040), rs5147(4028), rsergio(4018), felixcoto(4015), bix0r4ever(4013), _iouri_(4010), Accoun(4002) 2010-11-11T04:16:46 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-11-11T04:18:52 *** Cold-Phoenix has joined #aichallenge 2010-11-11T04:19:03 *** kincody_ has quit IRC (Ping timeout: 265 seconds) 2010-11-11T04:19:05 *sniff* 2010-11-11T04:19:46 rabidus:? 2010-11-11T04:19:53 i almost have this generator done 2010-11-11T04:19:56 the maps are much cooler 2010-11-11T04:21:06 *** ColdPhoenix has joined #aichallenge 2010-11-11T04:21:10 i was thinking today a planet wars game where planets moves 2010-11-11T04:21:22 just a bit more realistic 2010-11-11T04:22:11 haha, fuck everything about that 2010-11-11T04:22:17 want to do me a favour? 2010-11-11T04:22:18 http://codepad.org/g7KPgLCr 2010-11-11T04:22:26 run that map, does it look properly symmetric to you? 2010-11-11T04:23:09 sry, i'm on work 2010-11-11T04:23:18 ahh fair enough 2010-11-11T04:23:20 sigh:! 2010-11-11T04:23:32 isn't here 2010-11-11T04:23:41 bhasker: ! 2010-11-11T04:23:53 Error323_:! 2010-11-11T04:24:03 :( 2010-11-11T04:25:01 *** Cold-Phoenix has quit IRC (Ping timeout: 240 seconds) 2010-11-11T04:26:54 *** ColdPhoenix has quit IRC (Ping timeout: 240 seconds) 2010-11-11T04:28:26 gah, i keep getting some nonsymmetric maps anyway 2010-11-11T04:32:37 *** Naktibalda has joined #aichallenge 2010-11-11T05:08:30 *** decipher has joined #aichallenge 2010-11-11T05:08:35 anyone else having trouble uploading? 2010-11-11T05:13:27 wtf, it wont output an integer to file correctly now :| 2010-11-11T05:14:17 what wont.. my program doesnt do that and it still wont let it up load 2010-11-11T05:14:39 Warning: mkdir() [function.mkdir]: Permission denied in /var/www/vhosts/ai-contest.com/www/check_submit.php on line 26 Warning: move_uploaded_file(/home/contest/ai-contest/planet_wars/submissions/186873/entry.zip) [function.move-uploaded-file]: failed to open stream: No such file or directory in /var/www/vhosts/ai-contest.com/www/check_submit.php on line 50 Warning: move_uploaded_file() [function.move-uploaded-file]: Unable 2010-11-11T05:16:28 Top 10 players: bocsimacko(4225), george(4048), dmj111(4046), GreenTea(4036), rsergio(4012), bix0r4ever(4007), felixcoto(3993), rs5147(3987), _iouri_(3974), Accoun(3969) 2010-11-11T05:17:41 decipher: i'm making a new map generator 2010-11-11T05:21:37 oh are you behind the contest 2010-11-11T05:21:49 no 2010-11-11T05:21:56 i just wanted more interesting maps 2010-11-11T05:21:59 oh ok 2010-11-11T05:22:07 they don't always seem to be coming out symmetric atm though 2010-11-11T05:22:10 i thought that was the answer to why the program wouldnt upload 2010-11-11T05:22:18 the idea is that they aren't all just symmetric across the centre 2010-11-11T05:31:23 decipher: try again now 2010-11-11T05:31:43 Just saw the forum post that it was broken and fixed it 2010-11-11T05:31:59 *** janzert1 is now known as janzert 2010-11-11T05:32:13 awesome thanks ill try now 2010-11-11T05:38:42 yeah it worked awesome now i can go to sleep ^^ thanks for the fix have a good night 2010-11-11T05:39:18 *** decipher has quit IRC (Quit: Page closed) 2010-11-11T05:44:13 gah, i cannot for the life of me work out why this wont work :( 2010-11-11T05:44:34 it works most of the time, but ocasionally 2 joint lots of planets will be out by a distance of 1 2010-11-11T05:45:53 *** yasith_ has quit IRC (Ping timeout: 240 seconds) 2010-11-11T05:46:30 *** yasith has quit IRC (Ping timeout: 276 seconds) 2010-11-11T05:47:29 *** Ari2 has quit IRC (Ping timeout: 250 seconds) 2010-11-11T06:04:47 *** p4p4 has joined #aichallenge 2010-11-11T06:10:13 i'm going to test if the first 100 maps are symmetric with this test 2010-11-11T06:11:57 *** Ari2 has joined #aichallenge 2010-11-11T06:13:38 *** boegel has joined #aichallenge 2010-11-11T06:13:55 I can't believe how close this was: http://ai-contest.com/visualizer.php?game_id=6945149 2010-11-11T06:16:29 Top 10 players: bocsimacko(4229), george(4073), dmj111(4068), GreenTea(4066), Raschi(4051), rsergio(4023), bix0r4ever(4017), felixcoto(4009), rs5147(3995), hutabhu(3988) 2010-11-11T06:20:17 well the starting maps seem fine, so why is mine retarded :( 2010-11-11T06:25:58 fixed 2010-11-11T06:25:59 yay 2010-11-11T06:26:07 i'll generate a set of maps now 2010-11-11T06:35:59 morning 2010-11-11T06:36:38 morning 2010-11-11T06:36:50 bhasker: are you in a position to play a map? 2010-11-11T06:37:01 as in ? 2010-11-11T06:37:14 i made a map generator 2010-11-11T06:37:19 hold on, i'll find a coolish one 2010-11-11T06:39:32 http://codepad.org/Ni6C6Nhp 2010-11-11T06:39:37 that sort of gets my point across 2010-11-11T06:39:45 in that it's not symmetric the same way the rest are 2010-11-11T06:41:47 looks perfectly symmetric to me? 2010-11-11T06:42:15 yes, but the other maps being used have players on the same line splitting the middle planet yeah? 2010-11-11T06:43:03 notice the difference? 2010-11-11T06:43:20 hmm let me look at it 2010-11-11T06:43:51 can't make out in my viewer 2010-11-11T06:43:55 looks identical to other maps 2010-11-11T06:44:11 *** mega1 has joined #aichallenge 2010-11-11T06:45:56 http://imgur.com/H2dAz.png 2010-11-11T06:46:01 that's what i am referring to 2010-11-11T06:46:53 ah k 2010-11-11T06:47:52 i want maps like this included in the final tourny 2010-11-11T06:48:33 mega1: i have a new map generator if you're interested in some new maps? 2010-11-11T06:48:42 they have less restrictions on symmetry 2010-11-11T06:49:17 antimatroid1: any word on the final tournament map generator? 2010-11-11T06:49:41 no, i'm hoping it might resemble something like i just made 2010-11-11T06:49:58 I hope it might resemble something that we actually test on :-) 2010-11-11T06:50:00 http://codepad.org/Ni6C6Nhp that's one map i generated, i'm just cleaning the code up 2010-11-11T06:50:07 how will the bots for the final tournament be picked? 2010-11-11T06:50:10 this is generating random maps 2010-11-11T06:50:16 boegel: all of them 2010-11-11T06:50:18 antimatroid1: but to answer the question yes I'm interested 2010-11-11T06:50:19 will it be like just the top 100 or something? 2010-11-11T06:50:25 antimatroid1: ah, cool, good news 2010-11-11T06:50:43 antimatroid1: and will there be updates of the ranking after the submission deadline? 2010-11-11T06:50:46 http://imgur.com/H2dAz.png that's what i've made it to do 2010-11-11T06:51:13 boegel: last time there wasn't a live leaderboard while it was running, but i have petitioned as much as possible for that to be different this time 2010-11-11T06:51:21 otherwise results are viewable at the end 2010-11-11T06:52:22 uhu, cool, thx 2010-11-11T06:52:41 * boegel hopes he'll have time to weed out the final bugs from his bot and tune it fully 2010-11-11T06:53:18 i just finished exams, so i have nothing but the ai comp to work on 2010-11-11T06:53:28 i took the night out to make the new map generator for better testing maps 2010-11-11T06:54:02 i started a complete rewrite a few days ago :) 2010-11-11T06:55:11 ugh 2010-11-11T06:56:24 do people mostly ignore middle planets on the first move? 2010-11-11T06:56:39 one thing generators don't really do yet is multiple "middle" planets 2010-11-11T06:58:21 *** Apophis has joined #aichallenge 2010-11-11T07:01:05 *** nann has quit IRC (Ping timeout: 240 seconds) 2010-11-11T07:12:16 mega1: https://docs.google.com/leaf?id=0B5joEFt1EpwZNzY2MGNiMTMtZGFhYi00MGIwLTkzMmEtMTljZjg5MGViNzgy&sort=name&layout=list&num=50 2010-11-11T07:12:25 i hope that link works for you? there's 100 maps 2010-11-11T07:12:53 thanks, downloaded it 2010-11-11T07:13:17 i'd be interested to see if you think it'd be good or bad to use maps like these in the final contest 2010-11-11T07:13:22 as i plan to push for them to be 2010-11-11T07:14:14 they don't have any 0 growth planets, and min distance is 3 2010-11-11T07:14:21 but that kind of stuff is all set from parameters 2010-11-11T07:16:29 Top 10 players: bocsimacko(4248), GreenTea(4071), dmj111(4060), george(4059), bix0r4ever(4029), rsergio(4015), Accoun(4009), felixcoto(4007), shangas(4002), oldman(3999) 2010-11-11T07:22:03 mega1: 76 is pretty cool 2010-11-11T07:22:11 you sort of have to go off and capture before doing anything 2010-11-11T07:22:30 anyone else about? 2010-11-11T07:22:36 bhasker: play with my maps :P 2010-11-11T07:22:52 get dhartmei to put them up on tcp:) 2010-11-11T07:25:22 i'll send him an email, i'm not 100% sure they are without fault, i did a test about symmetry though 2010-11-11T07:29:16 proud of my bot in this game http://72.44.46.68/canvas?game_id=964450 2010-11-11T07:30:01 i'm going to shoot of this email then get back to work on my bot :| 2010-11-11T07:30:15 i added some abandon ship code 2010-11-11T07:30:19 lets see how that works out 2010-11-11T07:30:21 only like a 4 hour side track, that took way longer than expected aha 2010-11-11T07:30:40 antimatroid1: nice maps 2010-11-11T07:34:25 boegel: i'm glad someones trying them :P 2010-11-11T07:35:10 i might generate 1000 and chuck a link in the forums 2010-11-11T07:35:43 antimatroid1: drop me an email, will you :) 2010-11-11T07:35:57 I'm planning to automatically tune my bot, so lot's of maps is good :) 2010-11-11T07:36:15 antimatroid1: kenneth dot hoste funny-sign ugent dot be 2010-11-11T07:41:49 boegel: https://docs.google.com/leaf?id=0B5joEFt1EpwZYTNhOTY1ZWYtYmRlYS00N2RmLWJjOTAtOWY4MDAzNzEzOTg0&sort=name&layout=list&num=50 2010-11-11T07:43:55 http://tinyurl.com/antimaps or alternatively 2010-11-11T07:48:08 antimatroid http://apple.slashdot.org/story/10/11/11/022255/Australian-State-Govt-To-Fund-iPads-For-Doctors 2010-11-11T07:48:09 lol? 2010-11-11T07:50:48 i don't pay attention to australian politics, it's possibly a bigger joke than us politics atm 2010-11-11T07:51:19 hehe 2010-11-11T07:51:25 <--not an american though:) 2010-11-11T07:51:29 yet* 2010-11-11T07:51:48 i didn' say you are :P 2010-11-11T07:52:18 where in australia do you live? 2010-11-11T07:52:18 antimatroid1: i'll run them too in my tests, do you see any significant behaviour on those maps? 2010-11-11T07:52:33 bhasker: tasmania, the island at the bottom :) 2010-11-11T07:52:45 ah 2010-11-11T07:52:48 Lithosphere: what do you mean by significant behaviour? 2010-11-11T07:52:50 i guess most bots will behave more or less the same on these? 2010-11-11T07:53:00 i haven't really got a good running bot atm, i'm in the middle of a complete rewrite 2010-11-11T07:53:02 the place where i live the street is named "tasman" after the tasman sea 2010-11-11T07:53:33 they i think the strategies might switch up a bit more with these 2010-11-11T07:53:42 but* i think 2010-11-11T07:54:02 tasmania is awesome, there isn't lots of people and plenty of places to go camping/hiking 2010-11-11T07:54:39 the uni isn't great, but i'm fairly happy with the maths department 2010-11-11T07:57:32 nice half a million people in total 2010-11-11T07:57:55 *** espes has joined #aichallenge 2010-11-11T07:58:02 looks very pretty 2010-11-11T08:01:26 hey guys 2010-11-11T08:01:40 i'm out of ammo :( 2010-11-11T08:01:49 antimatroid1: does that include the same maps as before? 2010-11-11T08:04:09 boegel: no 2010-11-11T08:04:11 sorry 2010-11-11T08:04:36 Error323_: pew pew pew 2010-11-11T08:04:42 thrust harder :) 2010-11-11T08:04:54 hehe 2010-11-11T08:10:19 hmm if i build orders for each planet with same availabilty queue and then try to score and pick the best orders 2010-11-11T08:10:35 its gonna be a really hard to identify which orders can be executed simultaneously 2010-11-11T08:12:38 is a simple pick highest score order, followed by the next one which can be executed simultaneously a good solution? 2010-11-11T08:12:47 do you guys think it will give a near optimal solution 2010-11-11T08:13:29 *** delt0r has joined #aichallenge 2010-11-11T08:13:30 not sure sorry 2010-11-11T08:14:28 *** delt0r___ has quit IRC (Ping timeout: 265 seconds) 2010-11-11T08:16:30 Top 10 players: bocsimacko(4263), GreenTea(4086), dmj111(4077), Raschi(4075), george(4062), bix0r4ever(4049), rsergio(4035), protocolocon(4015), felixcoto(4012), Accoun(4011) 2010-11-11T08:19:45 bhasker: heh this is the basis of my minimax 2010-11-11T08:19:54 however, its teh fail 2010-11-11T08:20:51 i've given up on minimax :P 2010-11-11T08:21:28 heh 2010-11-11T08:21:35 i am just bored of tuning my heuristic bot 2010-11-11T08:21:37 its so boring 2010-11-11T08:21:42 to keep adding rules 2010-11-11T08:21:52 if this don't do this 2010-11-11T08:21:53 bah 2010-11-11T08:22:31 i agree 2010-11-11T08:22:39 hence i stopped doing that 2010-11-11T08:22:41 Error323_: did you get a copy of my maps? :P 2010-11-11T08:22:53 antimatroid1: no? 2010-11-11T08:22:58 http://tinyurl.com/antimaps 2010-11-11T08:23:38 thanks 2010-11-11T08:23:49 let me know what you think :P 2010-11-11T08:24:00 they should be reasonably different to other ones 2010-11-11T08:25:48 okay 2010-11-11T08:26:04 they're still symmetric, but with less restrictions on it 2010-11-11T08:26:18 more then 23 planets? 2010-11-11T08:26:21 no 2010-11-11T08:26:35 ok 2010-11-11T08:26:47 all 23 planets, and still only 1 planet equal distance between them at the start, but they aren't on a line joining the centre planet anymore 2010-11-11T08:27:11 min distance is 3, min starting distance between players is 7 and growth is between 1-5 2010-11-11T08:27:30 *** Migi32 has joined #aichallenge 2010-11-11T08:27:52 hi everyone 2010-11-11T08:28:00 hello 2010-11-11T08:28:07 i made som maps if you're interested :) http://tinyurl.com/antimaps 2010-11-11T08:28:28 *** jaspervdj has joined #aichallenge 2010-11-11T08:28:29 hah the master returns 2010-11-11T08:28:35 You LOSE against bocsimacko-b0.57 2010-11-11T08:28:37 :( 2010-11-11T08:28:40 lol 2010-11-11T08:28:50 I love how the capital letters scream it at me 2010-11-11T08:28:54 LOSE 2010-11-11T08:28:56 you FAIL 2010-11-11T08:28:59 you SUCK 2010-11-11T08:29:03 xD 2010-11-11T08:29:04 lol 2010-11-11T08:29:30 I dropped down to #29 after a massive losing streak 2010-11-11T08:29:33 :( 2010-11-11T08:29:45 Do hope it was temporarily 2010-11-11T08:29:51 heh, I know the overall game isn't really worth watching but the 5-fleet sent at turn 41 here was a total superhero: http://72.44.46.68/canvas?game_id=965091 2010-11-11T08:29:53 the rankings are like that 2010-11-11T08:30:59 yeah 2010-11-11T08:31:01 i hate that 2010-11-11T08:31:07 :P 2010-11-11T08:31:21 tcp has become more difficult 2010-11-11T08:31:26 i used to be 300 elo 2010-11-11T08:31:35 but now down to 250 or sth 2010-11-11T08:31:44 after a night of play 2010-11-11T08:31:50 yup 2010-11-11T08:31:57 bot are constantly improving 2010-11-11T08:32:00 its a losing battle lol 2010-11-11T08:32:03 its like the arms race 2010-11-11T08:32:04 :-\ 2010-11-11T08:32:17 unless you are mega1, the only player with the nuke 2010-11-11T08:32:32 yeah 2010-11-11T08:32:36 I can't stand it 2010-11-11T08:32:46 I can to 2 ply with massive branching factor 2010-11-11T08:32:51 it should be epic 2010-11-11T08:33:17 Its just that I really don't know a good eval function 2010-11-11T08:33:17 you should wokr on a better eval function maybe? 2010-11-11T08:33:39 yeah 2010-11-11T08:33:46 but I haven't got a solid idea 2010-11-11T08:33:49 find some math major friends of yours 2010-11-11T08:33:54 to find you a good eval function 2010-11-11T08:33:54 this is my current: 2010-11-11T08:34:11 find some game theory expert to do it for tyou:) 2010-11-11T08:34:13 and just stick it in 2010-11-11T08:34:15 * The heuristic score is defined as follows: 2010-11-11T08:34:15 * / f_e(p) + 1 \ / f_m(p) + 1 \ 2010-11-11T08:34:18 * h(n) = |Sum_{p in P_w(n)} ---------- * g(p)| - |Sum_{p in P_l(n)} ---------- * g(p)|, 2010-11-11T08:34:21 * \ f_m(p) + 1 / \ f_e(p) + 1 / 2010-11-11T08:34:24 * 2010-11-11T08:34:27 * Where P_w(n) is the set of planets won at node n, P_l is the set of 2010-11-11T08:34:29 * planets lost at node n, f_m(p) are the amount of fleets on my side send 2010-11-11T08:34:32 * to this planet, f_e(p) are the amount of enemy fleets send to this 2010-11-11T08:34:34 * planet and g(p) is the planet's growth rate. 2010-11-11T08:34:54 bhasker: hehe well I could maybe but that wouldn't be my work :P 2010-11-11T08:35:07 well most of it will still be yours:) 2010-11-11T08:35:13 only the eval function:) 2010-11-11T08:35:56 so you basically do this for every planet 2010-11-11T08:36:00 and sum up the scores? 2010-11-11T08:36:09 for a given node? 2010-11-11T08:36:17 only the planets that the actionStack had influence on 2010-11-11T08:36:54 so the delta's given by the path down to the terminal node 2010-11-11T08:37:54 where do f_m(p)/f_e(p) fit into the h(n) function 2010-11-11T08:38:08 i am sorry but i am not the most mathematically inclined person 2010-11-11T08:38:19 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-11-11T08:38:32 well the idea is this 2010-11-11T08:39:13 if I win a planet by sending e.g. 1000 ships to it while the enemy just sends 1 ship its netto gain is (1 / 1000) * growth(p) 2010-11-11T08:39:17 which isn't good 2010-11-11T08:39:36 but if I send less ships then the enemy it score goes up 2010-11-11T08:39:42 *** p4p4 has joined #aichallenge 2010-11-11T08:39:42 and vise versa for planets i've lost 2010-11-11T08:40:10 but I should prolly take into account the number of ships on the planet somehow 2010-11-11T08:40:29 so P_w(n) uses the f_e(p)/f_m(p) to calculate the cost of acquisition of the planet 2010-11-11T08:41:11 P_w is the set of planets that I've won because of my actions 2010-11-11T08:41:31 so i sum over all the planets in that set 2010-11-11T08:42:03 what about if you don't win /lose anything 2010-11-11T08:42:05 adding ((f_e(p) + 1) / (f_m(p) + 1)) * growth 2010-11-11T08:42:07 does the score for that goto 0 2010-11-11T08:42:30 yes thats the problem, i don't really know howto deal with that 2010-11-11T08:42:49 maybe look at ship difference 2010-11-11T08:42:55 if you are ahead on ships 2010-11-11T08:42:59 i give it something similar if its my planet or the enemies 2010-11-11T08:42:59 give it a higher score than zero? 2010-11-11T08:43:25 but i should specifically add scorings for sniping in it etc 2010-11-11T08:43:26 also i think you should factor in how much ahead in ships you are 2010-11-11T08:43:33 if you are a lot ahead 2010-11-11T08:43:37 yes i do have that 2010-11-11T08:43:49 then it won't matter if you sent a ton of ships 2010-11-11T08:43:51 to a planet 2010-11-11T08:44:03 h(n) = A - B + (myNumShips - enemyNumShips) 2010-11-11T08:44:18 also if you send a ton of ships to a planet which is close to enemy planets it should score higher than 2010-11-11T08:44:24 say a planet that is far awa 2010-11-11T08:44:31 even if you/enemy send same number of ships 2010-11-11T08:44:51 yeah, this is good i think 2010-11-11T08:45:11 but i do want something elegant 2010-11-11T08:45:25 not flesh out every single heuristic i know and score that :P 2010-11-11T08:45:33 like sniping etc 2010-11-11T08:45:51 i don't think sniping should be a heuristic 2010-11-11T08:45:59 indeed 2010-11-11T08:46:08 i've never had it as one 2010-11-11T08:46:26 no me neither and my simple action version did apply it on its own 2010-11-11T08:46:36 i'm in the process of writing a completely new bot without making stupid short cuts/approximations all over the place 2010-11-11T08:47:42 Error323 i think distribution of your ships matters 2010-11-11T08:47:45 when evaluating 2010-11-11T08:47:50 you should evaluate your spread vs enemies 2010-11-11T08:48:23 i just redistribute any ships i haven't used at the end of my turn 2010-11-11T08:48:26 bhasker: Indeed, cool :) 2010-11-11T08:49:08 you should probably see if for the node you are much mroe spread out than the enemy 2010-11-11T08:49:18 since strategically, its a losing situation 2010-11-11T08:49:34 if you have same number of ships and enemy has them closer on a bunch of planets 2010-11-11T08:49:36 not really, i will have already allocated ships needed to defend anything 2010-11-11T08:49:41 he can use them more effectively 2010-11-11T08:49:43 or that is planned to be used for expansion 2010-11-11T08:49:47 not really 2010-11-11T08:49:48 example 2010-11-11T08:49:56 lets say its a map with planets on the corners 2010-11-11T08:50:01 and something in the central 2010-11-11T08:50:05 the person who controls the central 2010-11-11T08:50:12 surely has a much higher advantage 2010-11-11T08:50:20 his time to travel to attack anything is much shorter 2010-11-11T08:50:30 if i attack a corner, you have to reinforce from really far away 2010-11-11T08:50:50 thats why on most games if you see, the person who usually takes teh central tends to win 2010-11-11T08:50:52 yes, but i will have already allocated ships to defend any attack necessary there 2010-11-11T08:51:03 hence it will be part of used ships, even if it's not sent 2010-11-11T08:51:15 you can't defend everything 2010-11-11T08:51:19 when enemy is staging to center 2010-11-11T08:51:22 with same number of ships 2010-11-11T08:51:26 *** boegel has quit IRC (Quit: Leaving) 2010-11-11T08:51:33 no, but i can defend what i expect i might have to 2010-11-11T08:51:51 you have to defend everying if you are even 2010-11-11T08:51:55 lets taken an example 2010-11-11T08:52:00 not always 2010-11-11T08:52:16 player A) has bottom right corner 2010-11-11T08:52:17 and middle 2010-11-11T08:52:19 you can potentially expand and save yourself in the time that they get to you 2010-11-11T08:52:25 player B) has the other 3 corners 2010-11-11T08:52:26 otherwise no one would expand on the first move ever 2010-11-11T08:52:37 player B has the other 3 corners i.e 2010-11-11T08:52:47 lets assume they are even on growth 2010-11-11T08:52:54 whose position do you think is stronger 2010-11-11T08:53:00 i can reinforce my planets in 1/2 the time 2010-11-11T08:53:11 if you do attack 2010-11-11T08:53:17 but if i attack any of the corners 2010-11-11T08:53:22 you won't be able to prevent the attack 2010-11-11T08:53:48 and i can hold my center even if you launch from all 3 sides 2010-11-11T08:54:14 an example like that doesn't work very well with words :P it very much depends on distances 2010-11-11T08:54:23 assuming its all equidistant 2010-11-11T08:54:31 the corner planets are equidistant from center 2010-11-11T08:54:36 lets take a simple map 2010-11-11T08:54:41 a 5 growth in every corner 2010-11-11T08:54:49 well that won't work 2010-11-11T08:54:55 yea nm 2010-11-11T08:54:58 5 growth in corners 2010-11-11T08:55:01 and a 10 in the center 2010-11-11T08:55:12 player A controls 3 corners 2010-11-11T08:55:19 player B controls center and 1 corner 2010-11-11T08:55:25 would be interesting to see 2010-11-11T08:55:39 or maybe 2 growths in corner and 2 growth in middle 2010-11-11T08:55:56 can your map generator generate a map like that 2010-11-11T08:56:13 4 growth in the middle i.e 2010-11-11T08:56:43 just make it by hand, wouldn't take very long 2010-11-11T08:56:48 mine is generating random maps 2010-11-11T08:56:54 ah k 2010-11-11T08:57:01 <--lazy:p 2010-11-11T08:57:17 oh well anyway i think any evaluation function should factor in position/spread of ships 2010-11-11T08:57:25 I agree 2010-11-11T08:57:32 its very important 2010-11-11T09:00:09 has anyone ever succeeded in changing their "about me"? 2010-11-11T09:00:26 i was originally going to be in a pair/team, but i'm doing it all by myself now 2010-11-11T09:00:42 also i believe if your bot requires staging code then your evaluation function isn't correct 2010-11-11T09:00:49 staging should emerge as a result of evaluation 2010-11-11T09:01:03 its clearly more desirable to stage than not to 2010-11-11T09:01:08 so if your function is correct 2010-11-11T09:01:14 it should not require any staging code 2010-11-11T09:03:01 i would say error try to remove your staging code if you have any 2010-11-11T09:03:10 and see if you can tune your eval function to stage :) 2010-11-11T09:03:26 what is "staging"? 2010-11-11T09:03:40 funnelling/staging 2010-11-11T09:03:41 bhasker: that would just explode my actionspace :( 2010-11-11T09:03:45 sending ships to the front 2010-11-11T09:03:52 i call it feeding 2010-11-11T09:03:55 in my code :P 2010-11-11T09:03:58 yea same thing 2010-11-11T09:03:59 redistribution? 2010-11-11T09:04:03 yup 2010-11-11T09:04:05 ye 2010-11-11T09:04:23 if your eval function is correct it should favour redistribution in a steady state 2010-11-11T09:04:32 if the redistribution helps improve your position 2010-11-11T09:04:33 i have safety/defensive/offensive/expansionary and redistributive moves 2010-11-11T09:04:40 example 2010-11-11T09:04:47 having 50 ships on two planets one behind the other 2010-11-11T09:04:51 is clearly not desirable 2010-11-11T09:04:56 compared to 100 in front 2010-11-11T09:05:04 yep 2010-11-11T09:05:13 but it explodes my branching factor 2010-11-11T09:05:21 and i'm already pushing the boundaries currently 2010-11-11T09:05:36 well actually i should reduce it even more in some states 2010-11-11T09:06:54 hmm but most of the redistribution nodes can probably be pruned 2010-11-11T09:07:02 like sending ships back to a planet thats protected 2010-11-11T09:07:04 can be pruned 2010-11-11T09:07:19 so if you build like a map of which planet can protect which ones etc 2010-11-11T09:07:24 true, but the problem is more sending the right amount 2010-11-11T09:07:25 you could probably prune a lot of the moves 2010-11-11T09:07:29 ah 2010-11-11T09:08:13 anyway all the best with your minmax version:) 2010-11-11T09:08:47 =] 2010-11-11T09:08:48 ty 2010-11-11T09:09:07 may the force be with you 2010-11-11T09:09:14 and may the dark side(mega1) not win 2010-11-11T09:09:49 Error323_, what's a ply actually in your minimax? A turn? A move (1 move = 1 target planets, n source planets)? Something else? 2010-11-11T09:10:17 Error323_: redistribution is makes my bot ~10x slower 2010-11-11T09:10:32 but it's crucial 2010-11-11T09:10:49 Migi32: a ply is an action for a player 2010-11-11T09:11:28 ok but then what is an action? Sending ships from 1 planet to 1 other planet? 2010-11-11T09:11:44 Migi32: yes for multiple sources 2010-11-11T09:11:51 to multiple targets 2010-11-11T09:12:49 mega1: yeah i really don't see how i can do that without creating a giant problem in my branching factor :/ 2010-11-11T09:14:33 mega1: did you every get minimax working? 2010-11-11T09:14:42 i've decided to work on a purely heuristic bot 2010-11-11T09:14:51 technically yes 2010-11-11T09:15:02 well? 2010-11-11T09:15:13 but at 1 ply it can hardly be called minimax 2010-11-11T09:15:20 and at 2 plies it sucks 2010-11-11T09:15:27 yeah 2010-11-11T09:15:29 too slowN 2010-11-11T09:15:31 what do you mean by 1 ply? just you making moves? 2010-11-11T09:15:35 too slow? 2010-11-11T09:15:35 well i'm going two ply 2010-11-11T09:15:36 yes 2010-11-11T09:15:45 with alpha beta, it doesn't suck that bad 2010-11-11T09:15:49 being slow is one thing, but it's worse too 2010-11-11T09:15:55 but it ain't great either 2010-11-11T09:15:57 that's strange 2010-11-11T09:16:10 it's not conservative enough 2010-11-11T09:16:24 Migi32: nah, you are creating heavy assumptions about the enemy only making the exact same moves as you can 2010-11-11T09:16:31 Top 10 players: bocsimacko(4313), GreenTea(4097), dmj111(4091), bix0r4ever(4085), george(4082), shangas(4053), rsergio(4049), asavis(4041), protocolocon(4032), oldman(4031) 2010-11-11T09:16:37 which if that were so would be a good model 2010-11-11T09:16:55 but its not, and then its just locally optimal 2010-11-11T09:16:59 within your actionspace 2010-11-11T09:17:02 but 1 ply minimax, isn't that just generating a list of possible actions and then picking the highest scoring one? 2010-11-11T09:17:12 yea 2010-11-11T09:17:32 well I do that too. 2010-11-11T09:17:57 mega1: I'm truely amazed by your branching factor being ~100 2010-11-11T09:18:00 _o_ 2010-11-11T09:18:07 why? 2010-11-11T09:18:13 *** Azrathud has quit IRC (Quit: Meh) 2010-11-11T09:18:17 its so small 2010-11-11T09:18:32 you are basically examining 100 moves 2010-11-11T09:18:32 it's small because it's preordered 2010-11-11T09:18:47 and picking the best given the eval 2010-11-11T09:18:54 yes 2010-11-11T09:19:12 preordered? 2010-11-11T09:19:17 oh you generate more moves you mean? 2010-11-11T09:21:15 mega1: how much worse is your bot when going two ply? 2010-11-11T09:21:23 actually, I've just checked and preordering does not factor here 2010-11-11T09:21:28 it's still ~100 2010-11-11T09:21:32 +-50 2010-11-11T09:21:46 damn 2010-11-11T09:21:53 those are highly compound actions 2010-11-11T09:22:11 yes 2010-11-11T09:22:17 that must be a beast of a function :P 2010-11-11T09:22:18 what are the numbers for you? 2010-11-11T09:22:28 is most work in the move generation or the move evaluation so far? 2010-11-11T09:22:28 3000 2010-11-11T09:22:28 and for what kind of moves? 2010-11-11T09:22:38 moves of my frontline 2010-11-11T09:22:45 Migi32: more in the eval 2010-11-11T09:22:49 the rest is staging 2010-11-11T09:25:18 *** mceier has joined #aichallenge 2010-11-11T09:26:05 let me see what i produce in terms of actionsets on an average game 2010-11-11T09:28:35 mega1: how much worse is your bot when going 2 ply? 2010-11-11T09:28:51 if you don't wanna share i understand though 2010-11-11T09:28:53 :) 2010-11-11T09:29:15 my trees have never made it past processing a couple of hundred nodes 2010-11-11T09:29:23 I _think_ after tuning it was about 60% loss 2010-11-11T09:29:30 and much slower 2010-11-11T09:29:32 my move selection is too inefficient 2010-11-11T09:29:56 the nash player is about equal 2010-11-11T09:29:57 mega1: even against your own version? 2010-11-11T09:29:59 but much slower too 2010-11-11T09:30:12 Error323_: yes, against that 2010-11-11T09:30:30 hmmmm 2010-11-11T09:30:30 I haven't tried it on tcp 2010-11-11T09:30:40 that's kind of weird 2010-11-11T09:30:43 or well 2010-11-11T09:30:49 mega1: you can test with other usernames to avoid public humiliation :P 2010-11-11T09:30:52 it's entirely possible that I screwed something up 2010-11-11T09:31:20 that, or the actions are locally optimal 2010-11-11T09:31:29 i should probably try 1 ply 2010-11-11T09:31:32 anyway, i'm going to bed so i can get up early and get some actual bot work done 2010-11-11T09:31:43 good night 2010-11-11T09:32:10 *** hellman has joined #aichallenge 2010-11-11T09:35:43 mega1: I sometimes have an average branching factor of 3500 during one turn 2010-11-11T09:36:17 ew 2010-11-11T09:36:21 without alphabeta it would be impossible :P 2010-11-11T09:36:28 yea.. 2010-11-11T09:39:36 but this is way too much for the official server 2010-11-11T09:41:09 especially when my eval function becomes more sophisticated 2010-11-11T09:41:36 *** Azrathud has joined #aichallenge 2010-11-11T09:41:46 *** amstan has quit IRC (Ping timeout: 245 seconds) 2010-11-11T09:44:16 bhasker: your bot is above me 2010-11-11T09:44:24 it should be doing much better at official imo 2010-11-11T09:44:29 did you upload it yet? 2010-11-11T09:44:31 iouri.37 with 967 elo :-) 2010-11-11T09:44:39 yea but i guess i am just not getting enough games on official 2010-11-11T09:44:45 whoops, not anymore 2010-11-11T09:44:46 1 game 2010-11-11T09:44:51 100% win rate 2010-11-11T09:44:51 haha 2010-11-11T09:45:28 and just because my bot can beat yours doesn't guarantee that it does better against others:-\ 2010-11-11T09:45:40 true 2010-11-11T09:45:47 but given that its elo > mine 2010-11-11T09:46:43 yea but # of bots on tcp is very low 2010-11-11T09:46:51 whereas # of bots on official is a lot more 2010-11-11T09:46:55 ye 2010-11-11T09:47:00 thats prolly it 2010-11-11T09:47:14 you've been overfitting at tcp :P 2010-11-11T09:47:22 nah 2010-11-11T09:47:36 i just look at bugs in my heuristics and try to fix them 2010-11-11T09:47:51 though its gotten to a point where i don't know what else to fix 2010-11-11T09:47:52 :-\ 2010-11-11T09:47:57 everything i fix 2010-11-11T09:48:00 breaks something else 2010-11-11T09:48:11 yeah... i know that feeling 2010-11-11T09:49:40 and i still lose to shampoo 2010-11-11T09:50:09 and sometimes on random maps to -ve rated bots 2010-11-11T09:50:11 :-\ 2010-11-11T09:50:35 but then my bot surprises me like this 2010-11-11T09:50:35 http://72.44.46.68/canvas?game_id=965717 2010-11-11T09:55:19 *** RainCT has joined #aichallenge 2010-11-11T09:56:05 bhasker: you don't use routing anymore? 2010-11-11T09:56:23 as in routing via my other planets? 2010-11-11T09:56:28 i do use it 2010-11-11T09:56:34 ye 2010-11-11T09:56:36 ok 2010-11-11T09:56:57 in that game you can see that i am funnelling through top and bottom separately 2010-11-11T09:57:01 after like turn 61 2010-11-11T09:57:45 at turn 61 the 4 growthrate planet goes over the 5 growthrate 2010-11-11T09:57:53 while the 5 growthrate is in its path 2010-11-11T09:57:58 oh thats because i don't always route through 2010-11-11T09:58:05 i stage to the closest to enemy 2010-11-11T09:58:22 oh 2010-11-11T09:58:24 and i only route to next hop if its covers this planet completely 2010-11-11T09:58:32 doesn't that defeat its purpose? 2010-11-11T09:58:43 otherwise i route to the next next hop 2010-11-11T09:58:46 i route if the detour <= 1 2010-11-11T09:59:34 i guess i could do that 2010-11-11T09:59:47 also because it looks pretty :P 2010-11-11T09:59:56 yea that it does lol 2010-11-11T10:00:09 let me add that in and see 2010-11-11T10:00:55 hmm it will be hard to change my code 2010-11-11T10:09:04 *** JamesMG has joined #aichallenge 2010-11-11T10:11:39 *** hellman has quit IRC (Remote host closed the connection) 2010-11-11T10:13:21 *** demerzel has joined #aichallenge 2010-11-11T10:16:32 Top 10 players: bocsimacko(4440), bix0r4ever(4162), GreenTea(4146), george(4120), dmj111(4096), rsergio(4070), shangas(4061), asavis(4038), hutabhu(4020), Accoun(4015) 2010-11-11T10:22:27 *** aerique has quit IRC (Quit: ...) 2010-11-11T10:22:53 *** superflit has quit IRC (Quit: superflit) 2010-11-11T10:27:24 *** Frontier has joined #aichallenge 2010-11-11T10:30:21 *** Utkarsh has joined #aichallenge 2010-11-11T10:39:44 does TCP have a map extractor? 2010-11-11T10:41:21 *** Azrathud has quit IRC (Ping timeout: 245 seconds) 2010-11-11T10:41:57 lavalamp linked a zipfile of all the TCP maps here a day or two ago 2010-11-11T10:42:03 *** vpetro has joined #aichallenge 2010-11-11T10:48:08 *** tapwater has joined #aichallenge 2010-11-11T10:54:04 *** Azrathud has joined #aichallenge 2010-11-11T10:55:46 *** superflit has joined #aichallenge 2010-11-11T11:11:22 *** hellman has joined #aichallenge 2010-11-11T11:14:24 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-11-11T11:16:32 Top 10 players: bocsimacko(4463), george(4178), GreenTea(4158), bix0r4ever(4140), dmj111(4120), rsergio(4096), shangas(4087), asavis(4051), Accoun(4044), oldman(4043) 2010-11-11T11:19:57 *** Azrathud has quit IRC (Ping timeout: 276 seconds) 2010-11-11T11:21:46 *** Sylph has quit IRC (Ping timeout: 272 seconds) 2010-11-11T11:23:40 *** Sylph has joined #aichallenge 2010-11-11T11:25:37 *** rebelxt has joined #aichallenge 2010-11-11T11:29:01 *** lavalamp_ has joined #aichallenge 2010-11-11T11:34:47 *** yasith has joined #aichallenge 2010-11-11T11:34:52 *** yasith_ has joined #aichallenge 2010-11-11T11:39:16 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-11-11T11:44:07 *** braddunbar has quit IRC (Quit: leaving) 2010-11-11T11:44:09 *** Sylph has quit IRC (Ping timeout: 255 seconds) 2010-11-11T11:45:19 *** Sylph has joined #aichallenge 2010-11-11T11:55:05 *** georgeSebastian has joined #aichallenge 2010-11-11T12:00:19 *** hellman has quit IRC (Remote host closed the connection) 2010-11-11T12:03:24 *** Azrathud has joined #aichallenge 2010-11-11T12:05:16 *** rebelxt has left #aichallenge 2010-11-11T12:13:16 *** Bobng has joined #aichallenge 2010-11-11T12:16:33 Top 10 players: bocsimacko(4423), george(4218), GreenTea(4167), Raschi(4136), dmj111(4131), shangas(4108), rsergio(4090), oldman(4084), felixcoto(4067), bix0r4ever(4059) 2010-11-11T12:19:45 *** bhasker has quit IRC (Quit: bhasker) 2010-11-11T12:21:04 *** p4p4 has joined #aichallenge 2010-11-11T12:30:42 *** Coaxecva has joined #aichallenge 2010-11-11T12:31:39 *** Coaxecva has quit IRC (Client Quit) 2010-11-11T12:44:14 *** demerzel has quit IRC (Quit: Page closed) 2010-11-11T12:45:18 weird... my bot is back down to the 200's... I guess the ranks vary by up to 120 places or so 2010-11-11T12:45:33 I broke mine :) 2010-11-11T12:46:52 *** Azrathud has quit IRC (Ping timeout: 265 seconds) 2010-11-11T12:48:39 *** bhasker has joined #aichallenge 2010-11-11T12:54:38 my "optimization" last night broke mine, too 2010-11-11T12:55:36 *** Utkarsh has quit IRC (Read error: Connection reset by peer) 2010-11-11T12:55:49 it also didn't speed it up much 2010-11-11T12:55:58 *** Utkarsh has joined #aichallenge 2010-11-11T12:56:35 I sped mine up a decent amount, maybe it won't crash on the official server now 2010-11-11T12:57:38 barely holding on to my positive ELO on the TCP server 2010-11-11T12:57:51 I lost mine... :/ 2010-11-11T12:58:16 but to be fair my best bot is vulnerable to my other bots so they may be keeping it artificially low 2010-11-11T13:04:43 *** RainCT has quit IRC (Ping timeout: 260 seconds) 2010-11-11T13:05:35 *** RainCT has joined #aichallenge 2010-11-11T13:16:38 Top 10 players: bocsimacko(4433), george(4222), Raschi(4207), GreenTea(4177), bix0r4ever(4142), dmj111(4138), shangas(4127), oldman(4102), rsergio(4098), Accoun(4067) 2010-11-11T13:20:06 *** hellman has joined #aichallenge 2010-11-11T13:26:00 *** hellman has quit IRC (Remote host closed the connection) 2010-11-11T13:35:31 does TCP only play 199 moves instead of 200? 2010-11-11T13:35:51 I mean turns, not moves 2010-11-11T13:43:18 its funny CCCF has the same problem as my more correct bot 2010-11-11T13:43:22 it just stops doing anything at all 2010-11-11T13:43:26 once it loses the advantage 2010-11-11T13:43:30 http://72.44.46.68/canvas?game_id=969118 2010-11-11T13:48:36 *** georgeSebastian has quit IRC (Ping timeout: 264 seconds) 2010-11-11T13:49:06 E323 whats new in 17.2? 2010-11-11T13:58:27 *** fawek has joined #aichallenge 2010-11-11T14:00:30 *** georgeSebastian has joined #aichallenge 2010-11-11T14:08:20 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-11-11T14:11:15 *** Bobng has quit IRC (Quit: Leaving) 2010-11-11T14:11:54 *** Apophis has joined #aichallenge 2010-11-11T14:14:43 *** Accoun has quit IRC () 2010-11-11T14:16:35 Top 10 players: bocsimacko(4407), george(4223), GreenTea(4156), Raschi(4123), dmj111(4120), shangas(4114), oldman(4094), rsergio(4067), Accoun(4056), asavis(4052) 2010-11-11T14:29:17 *** Accoun has joined #aichallenge 2010-11-11T14:35:56 *** choas has joined #aichallenge 2010-11-11T14:36:54 *** Bobng has joined #aichallenge 2010-11-11T14:46:49 *** xzaramurd has joined #aichallenge 2010-11-11T14:48:00 *** xzaramurd has quit IRC (Client Quit) 2010-11-11T14:48:43 *** davidd has quit IRC (Read error: Operation timed out) 2010-11-11T14:52:21 *** davidd has joined #aichallenge 2010-11-11T14:56:34 *** olafje has joined #aichallenge 2010-11-11T14:57:53 *** Frontier has quit IRC (Ping timeout: 260 seconds) 2010-11-11T15:01:36 hi, is there anyone else having problems creating an account? (Failed to send confirmation email. Try again in a few minutes.) 2010-11-11T15:03:48 *** Zoost has quit IRC (Ping timeout: 245 seconds) 2010-11-11T15:04:15 looks like the site is busted at the moment 2010-11-11T15:04:27 *** Zoost has joined #aichallenge 2010-11-11T15:10:39 *** Zoost has quit IRC (Ping timeout: 240 seconds) 2010-11-11T15:12:19 Thanks, I'll try again another day 2010-11-11T15:13:59 I just uploaded my bot again. It said "success!" but my bot has been "ready to be compiled" for almost half an hour now 2010-11-11T15:14:29 *** Apophis has quit IRC (Ping timeout: 265 seconds) 2010-11-11T15:15:55 *** Zoost has joined #aichallenge 2010-11-11T15:16:17 http://ai-contest.com/game_stats.php 2010-11-11T15:16:35 went from 32 games / hour to 6 games / hour somewhere in the last 24 hours 2010-11-11T15:16:35 Top 10 players: bocsimacko(4412), george(4229), Raschi(4194), GreenTea(4161), dmj111(4126), shangas(4120), oldman(4100), bix0r4ever(4098), rsergio(4073), Accoun(4070) 2010-11-11T15:17:06 sorry games/minute instead of games/hour 2010-11-11T15:22:54 *** fawek is now known as fawek__ 2010-11-11T15:22:57 *** fawek__ is now known as fawek 2010-11-11T15:23:26 lol, http://ai-contest.com/advanced_strategy_guide.php 2010-11-11T15:27:06 ok my bot's submission failed to compile but I received no email. What do I do now? Keep uploading? 2010-11-11T15:33:38 I hear dancing naked on your roof helps 2010-11-11T15:37:22 or buying us a better server 2010-11-11T15:44:13 *** yasith has quit IRC (Ping timeout: 245 seconds) 2010-11-11T15:44:41 *** yasith_ has quit IRC (Ping timeout: 245 seconds) 2010-11-11T15:50:37 *** JensTi has joined #aichallenge 2010-11-11T15:55:39 ok, changing MyBot.cpp to MyBot.cc made it recognize the submission as c++ (.cpp is the most used c++ extension btw), but it still fails. I'll now try renaming all .cpp files to .cc 2010-11-11T15:57:15 good idea 2010-11-11T15:57:24 @repo 2010-11-11T15:57:25 lavalamp_: repo = http://code.google.com/p/ai-contest/ 2010-11-11T15:57:41 you can look there for the script that actually does the compiling if you have more issues 2010-11-11T16:04:29 *** hellman has joined #aichallenge 2010-11-11T16:07:58 *** Apophis has joined #aichallenge 2010-11-11T16:13:11 *** delt0r___ has joined #aichallenge 2010-11-11T16:14:32 *** delt0r has quit IRC (Ping timeout: 240 seconds) 2010-11-11T16:16:36 Top 10 players: bocsimacko(4392), george(4201), bix0r4ever(4186), GreenTea(4128), dmj111(4106), Raschi(4096), shangas(4095), oldman(4090), Accoun(4053), felixcoto(4031) 2010-11-11T16:17:03 *** irchs has joined #aichallenge 2010-11-11T16:17:09 *** irchs is now known as janstaunton 2010-11-11T16:25:12 *** georgeSebastian has quit IRC (Ping timeout: 264 seconds) 2010-11-11T16:25:29 *** olafje has quit IRC (Quit: Page closed) 2010-11-11T16:31:48 bhasker: new version? =) 2010-11-11T16:32:46 lol, 1 elo below me @ tcp 2010-11-11T16:32:58 its gonna be interesting 2010-11-11T16:33:11 if the freakin servers were up 2010-11-11T16:33:29 *** boegel has joined #aichallenge 2010-11-11T16:33:55 no more servers active right now? :( 2010-11-11T16:34:01 http://ai-contest.com/game_stats.php 2010-11-11T16:42:28 *** hellman has quit IRC (Remote host closed the connection) 2010-11-11T16:45:59 *** boegel has quit IRC (Quit: This computer has gone to sleep) 2010-11-11T16:52:20 *** Mathnerd314 has joined #aichallenge 2010-11-11T16:59:58 *** boegel has joined #aichallenge 2010-11-11T17:00:16 *** amstan has joined #aichallenge 2010-11-11T17:00:16 *** ChanServ sets mode: +o amstan 2010-11-11T17:02:18 *** sigh has joined #aichallenge 2010-11-11T17:06:57 *** choas has quit IRC (Quit: leaving) 2010-11-11T17:08:02 *** lavalamp_ has quit IRC (Ping timeout: 265 seconds) 2010-11-11T17:10:57 *** Apophis has quit IRC (Ping timeout: 276 seconds) 2010-11-11T17:12:58 hey amstan, could you check what's wrong with the mailing system? My code won't compile and after 4 submissions I've still received no mail. 2010-11-11T17:13:25 Migi32: did you check your spam? 2010-11-11T17:13:52 ahh ok it went to spam. Thanks :) 2010-11-11T17:14:18 hehe 2010-11-11T17:14:31 gmail? 2010-11-11T17:15:21 *** npoektop has quit IRC (Quit: bye) 2010-11-11T17:15:30 Migi32: the only reason it's there because people don't unmark the spam 2010-11-11T17:15:49 yes, gmail. I will unmark it. 2010-11-11T17:16:35 Top 10 players: bocsimacko(4377), george(4198), bix0r4ever(4161), GreenTea(4104), dmj111(4093), shangas(4074), Raschi(4072), oldman(4068), Accoun(4031), asavis(4006) 2010-11-11T17:16:50 ok, apparently your version of gcc doesn't support writing numbers as 0b000101 2010-11-11T17:18:09 o_O 2010-11-11T17:18:37 i didn't know that was possible 2010-11-11T17:18:42 no problem, I can write it as 1U<<3 2010-11-11T17:19:00 its binary? 2010-11-11T17:19:03 i presume 2010-11-11T17:19:16 yes 2010-11-11T17:20:08 there's also 0x0123456789ABCDEF for hexadecimal but I guess you knew that 2010-11-11T17:20:22 why did I lose this gme according to the list of games in my profile? http://ai-contest.com/visualizer.php?game_id=6969582 2010-11-11T17:20:26 yes i did 2010-11-11T17:20:42 I don't see a crashed warning message, and I do seem to have won 2010-11-11T17:21:36 good question 2010-11-11T17:21:38 =) 2010-11-11T17:21:51 *** Rubicon-|-Cross has joined #aichallenge 2010-11-11T17:23:08 seems like a bug in the system, if you ask me :) 2010-11-11T17:29:09 *** boegel has quit IRC (Quit: This computer has gone to sleep) 2010-11-11T17:32:58 *** codefun64 has joined #aichallenge 2010-11-11T17:33:16 There seems to be a problem with perl scripts on the server... anyone else notice that? 2010-11-11T17:33:28 I mean, I run them on my computer and they work fine, they also win the battles 2010-11-11T17:33:43 but then I upload them and all the battles result in me not starting/timing out/crashing. How is this possible? 2010-11-11T17:33:51 *** Azrathud has joined #aichallenge 2010-11-11T17:33:52 codefun64: runtime exceptions? 2010-11-11T17:34:25 Well that woudl be alot of exceptions since I've had 4 battles with this latest upload, all of which resulted in a "crash" but worked fine on my laptop 2010-11-11T17:34:35 and previous uploads had the same problem. 2010-11-11T17:34:55 1k ranks down the toilet ftw! 2010-11-11T17:35:14 codefun64: yes.. do you have the same perl version? 2010-11-11T17:35:30 I'm pretty sure I do, but I'll have to check. 2010-11-11T17:35:38 5.8.8 2010-11-11T17:37:01 Mine's 5.12.2 2010-11-11T17:37:17 Not aware of changes between the two that would cause a complete runtime failure, though. 2010-11-11T17:37:38 perl is scary 2010-11-11T17:38:04 Usually it's not >_> 2010-11-11T17:38:27 The perl starter that I saw uses regexes to parse the state, when all you need is a split 2010-11-11T17:38:45 that's scary :P 2010-11-11T17:41:00 :o 2010-11-11T17:41:05 Expensive regexes 2010-11-11T17:42:30 i'm having the hardest time figuring out a formula that represents the goodness of a gamestate >_< 2010-11-11T17:43:21 * codefun64 chews on a piece of gamestate 2010-11-11T17:43:28 that piece is pretty good ;) 2010-11-11T17:43:41 gamestate goodness! :D 2010-11-11T17:43:49 lol 2010-11-11T17:44:52 * codefun64 has to download an old version of perl to make his bot 2010-11-11T17:44:57 Evil perl 5.8.8 :O 2010-11-11T17:47:31 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-11-11T17:47:48 :o 2010-11-11T17:47:53 Why use perl? 2010-11-11T17:48:22 Because that's my favorite language? 2010-11-11T17:48:55 and because C# is reported to not work on the server? lol 2010-11-11T17:49:45 yeah, don't use C# 2010-11-11T17:50:20 Yeah well since I'm 14 I don't have a huge repertiore of languages... perl and C# are the only ones I know really well. 2010-11-11T17:50:41 repertoire* 2010-11-11T17:50:46 I'm only talking about the contest, the mono version on the official server has some issues 2010-11-11T17:51:43 morning 2010-11-11T17:52:04 hi antimatroid1 2010-11-11T17:52:06 Morning? Where do you live, man? ;) 2010-11-11T17:52:15 australia :) 2010-11-11T17:52:25 Awesome sauce 2010-11-11T17:52:43 have you guys tried the maps i made? 2010-11-11T17:52:50 i'm curious what other people thought? 2010-11-11T17:52:52 Didn't know you made maps! 2010-11-11T17:53:01 http://tinyurl.com/antimaps 2010-11-11T17:53:07 there's 1000 2010-11-11T17:53:31 are these the ones with different symmetries? 2010-11-11T17:53:39 Why so many maps..? 2010-11-11T17:53:49 codefun64: i just told it to make a bunch 2010-11-11T17:54:26 sigh: sort of, there's really only one "type of symmetry" although the angles at which the players start from the origin are picked randomly and they aren't on a line joining the centre planet 2010-11-11T17:55:21 awww... as a math student you should be able to come up with more exotic symmetries :P 2010-11-11T17:55:51 *** davidjliu has joined #aichallenge 2010-11-11T17:55:56 i think these maps are pretty cool 2010-11-11T17:56:01 :) 2010-11-11T17:56:05 running them all now 2010-11-11T17:56:16 I'll have to try some of them 2010-11-11T17:56:35 different maps might break my bot. Glitch fixing ftw :O 2010-11-11T17:56:48 they all have 23 planets with a centre planet 2010-11-11T17:57:05 min distance is 3, 7 for players starting planets, and growth is between 1-5 2010-11-11T17:57:16 You should've made some of the maps random 2010-11-11T17:57:26 they are all random under those restrictions 2010-11-11T17:57:33 this many maps might break my processor :) 2010-11-11T17:57:33 i will clean the map generator up and post that soon 2010-11-11T17:57:35 Ahhhhh, cool 2010-11-11T17:57:47 James, what kind of processor are you runnin'? o_O 2010-11-11T17:57:55 although i probably did a lot of things in embarrasing ways :) 2010-11-11T17:58:01 it's my laptop 2010-11-11T17:58:20 * codefun64 has a ridiculously overpowered laptop 2010-11-11T17:58:29 running 1102 maps with each of the sample bots 2010-11-11T17:59:44 sigh: i'll try to make some maps satisfying some other symmetries at some point, other things i would like is more than one planet starting at an equal distance from both players 2010-11-11T17:59:56 possible, all neutrals starting like that 2010-11-11T18:00:14 1102? there should have been just 1000 2010-11-11T18:00:14 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-11-11T18:00:34 I let it run the starter 100 and a couple of McLeopold's maps too 2010-11-11T18:00:34 map59 looks like some thing from space invaders 2010-11-11T18:00:47 Hehe 2010-11-11T18:01:20 Here's a question, why build a map generator? Why not just use the ones the kits come with? 2010-11-11T18:01:41 they don't display symmetrically with jBotManager, but i'm 99% sure they are, i had troubles for a while where some pairs would be out by a distance of 1, but i'm pretty sure they're ALL symmetric now 2010-11-11T18:01:49 because they are different 2010-11-11T18:02:02 codefun64: as a general ruel, to test your bot in different conditions and make it more robust 2010-11-11T18:02:10 all the maps in the starter pack have the players starting planets on a line joining the centre planet 2010-11-11T18:02:16 final tournament won't use the current maps 2010-11-11T18:02:26 Oh cool 2010-11-11T18:02:31 i'm hoping i can push for maps like these to be included 2010-11-11T18:02:37 although i wont if people didn't like them 2010-11-11T18:02:38 What's the final tournament like? Do they restart the scores? 2010-11-11T18:02:39 hassel j3camero some more 2010-11-11T18:02:43 hassle 2010-11-11T18:02:48 dare I ask why the java sample bots use 10MB of RAM? 2010-11-11T18:02:55 he was receptive to the idea 2010-11-11T18:03:02 Because Java sucks? 2010-11-11T18:03:05 *** Azrathud has quit IRC (Read error: Connection reset by peer) 2010-11-11T18:03:06 they weren't reset last time i don't think, but i have stated that they should be 2010-11-11T18:04:00 Could prove interesting 2010-11-11T18:04:22 and blah. My bot still runs fine on my laptop... but crashes on the server. 2010-11-11T18:04:39 *** p4p4 has quit IRC (Remote host closed the connection) 2010-11-11T18:05:42 turn 1 crashes, or later in the game? 2010-11-11T18:05:56 turn 1, crashes at the very start before the first move. 2010-11-11T18:06:08 oh, good luck with that 2010-11-11T18:06:17 And I just tested again with 100 mseconds per turn, and didn't crash at all even near the end. 2010-11-11T18:06:25 So it's not a problem with the code -_- 2010-11-11T18:07:24 codefun64: c++? 2010-11-11T18:07:29 i've had such problems 2010-11-11T18:07:31 perl. 2010-11-11T18:07:40 i have problems like that with c++\ 2010-11-11T18:07:57 It didn't used to do this but the last few days have crashed my bot with no cause 2010-11-11T18:08:53 How do I solve a problem I don't have? 2010-11-11T18:14:09 I like them antimatroid1 2010-11-11T18:14:44 JamesMG: :) 2010-11-11T18:14:56 i think they make for some more interesting games 2010-11-11T18:15:10 codefun64: you are using strict and warnings, right? 2010-11-11T18:15:48 *** JamesMG has quit IRC (Quit: Leaving.) 2010-11-11T18:16:36 Top 10 players: bocsimacko(4366), george(4182), bix0r4ever(4157), Raschi(4133), shangas(4121), oldman(4110), GreenTea(4104), dmj111(4087), asavis(4024), felixcoto(4007) 2010-11-11T18:19:55 *** Azrathud has joined #aichallenge 2010-11-11T18:20:56 *** davidjliu has joined #aichallenge 2010-11-11T18:21:05 *** vpetro has quit IRC (Ping timeout: 240 seconds) 2010-11-11T18:22:11 http://72.44.46.68/canvas?game_id=972760 2010-11-11T18:22:15 * janstaunton likes those kinds of maps 2010-11-11T18:23:00 codefun64: http://pastebin.ca/1988605 2010-11-11T18:23:35 janzert: are you interested in new maps for official server games? 2010-11-11T18:24:01 where the players starting planets aren't necessarily on a line joining the centre planet, but are still symmetrical at the start 2010-11-11T18:24:12 hmm, if you get some sort of consensus that people would like them on the forums I don't have any problem with using a different generator 2010-11-11T18:24:45 cool, i shall clean the generator up, post it and get peoples opinions 2010-11-11T18:24:50 I'm a bit worried about changing to a different generator at this stage though, but if the players don't mind I'd go for it 2010-11-11T18:24:58 i expect i'm doing some stuff fairly retardedly, but that can be fixed 2010-11-11T18:25:27 the maps are still random, and i've kept 23 planets 2010-11-11T18:25:30 also I'd prefer that it was in python to match the rest of the codebase 2010-11-11T18:25:44 * janzert has no idea what you're using 2010-11-11T18:25:50 ahhh, someone will need to rewrite it then :P i did it in c++ 2010-11-11T18:26:03 i don't even have python installed :) 2010-11-11T18:26:21 it's real easy ;) 2010-11-11T18:26:37 * codefun64 prays for his latest upload to work 2010-11-11T18:29:22 boegel: btw regarding http://ai-contest.com/visualizer.php?game_id=6969582 2010-11-11T18:29:41 you lose because yo take the neutral 58 ship planet on the last turn 2010-11-11T18:29:57 remember the visualizer doesn't show the result of the last turn 2010-11-11T18:31:51 heeheehee! My bot didn't crash but the other guy's bot did! XD 2010-11-11T18:31:55 *** amstan_ has joined #aichallenge 2010-11-11T18:31:55 *** ChanServ sets mode: +o amstan_ 2010-11-11T18:32:30 nice going :) 2010-11-11T18:32:59 *** pgpaskar1 has joined #aichallenge 2010-11-11T18:33:08 Tahnks janzert, didn't occur to me to get rid of that module :) 2010-11-11T18:33:13 Thanks, too 2010-11-11T18:34:41 *** Mathnerd314 has quit IRC (Ping timeout: 245 seconds) 2010-11-11T18:35:19 welcome 2010-11-11T18:37:34 Well later guys - got a paper to rewrite -_- 2010-11-11T18:37:44 Best of luck with your bot wars! 2010-11-11T18:38:06 *** codefun64 has quit IRC (Quit: Page closed) 2010-11-11T18:41:02 *** amstan has quit IRC (*.net *.split) 2010-11-11T18:41:03 *** pgpaskar_ has quit IRC (*.net *.split) 2010-11-11T18:57:38 *** Utkarsh has quit IRC (Ping timeout: 260 seconds) 2010-11-11T19:07:56 E323 our bots are dancing around each other 2010-11-11T19:08:02 and it depends on who plays shampoo more 2010-11-11T19:08:04 lol 2010-11-11T19:08:39 *** Mjothvitnir has joined #aichallenge 2010-11-11T19:09:13 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-11-11T19:09:26 *** bhasker_ has joined #aichallenge 2010-11-11T19:10:26 *** bhasker__ has joined #aichallenge 2010-11-11T19:13:39 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2010-11-11T19:14:00 *** bhasker_ has quit IRC (Ping timeout: 255 seconds) 2010-11-11T19:16:37 Top 10 players: bocsimacko(4393), Raschi(4208), george(4205), shangas(4143), bix0r4ever(4130), oldman(4116), GreenTea(4112), dmj111(4098), _Astek_(4054), asavis(4049) 2010-11-11T19:16:52 *** superflit has quit IRC (Quit: superflit) 2010-11-11T19:17:18 *** mceier has quit IRC (Quit: leaving) 2010-11-11T19:17:32 *** Mjothvitnir has quit IRC (Quit: Page closed) 2010-11-11T19:34:38 *** fawek has quit IRC (Ping timeout: 245 seconds) 2010-11-11T19:43:29 *** janstaunton has quit IRC (Quit: janstaunton) 2010-11-11T19:46:09 dude, this contest has become so much harder in the past couple months. The submission of kcm1700 that was top 1 for a fairly long time is now at rank 144 2010-11-11T19:47:30 *** eburnette_ has joined #aichallenge 2010-11-11T19:48:09 Hi everybody, what's up 2010-11-11T19:49:06 *** amstan_ is now known as amstan 2010-11-11T19:50:28 I wrote an article about the challenge at http://www.zdnet.com/blog/burnette/its-youth-vs-experience-in-the-google-ai-challenge-and-the-kids-are-winning/2123 2010-11-11T19:52:16 And I finished my umpteenth rewrite and released a new bot called Houdour 2010-11-11T19:53:24 you should really use evolutionary algorithms so you can call the next one houndoom 2010-11-11T19:54:13 That's what I'm planning, I've done Charmeleon, Monferno, and their evolved forms 2010-11-11T19:54:14 looks good. I'll read it tomorrow 2010-11-11T19:54:53 who the heck is monferno? is it one of those crazy newfangled Pokemon? 2010-11-11T19:55:05 ouch eburnette you picked bockismacko vs dmj 2010-11-11T19:55:25 the graph looks rather depressing 2010-11-11T19:55:33 should have picked some games that draw 2010-11-11T19:56:29 Make me a list of your faves and I'll update it 2010-11-11T19:56:43 will do 2010-11-11T19:57:12 but mega1 has made all the other bots look like a joke :-( 2010-11-11T19:57:44 I got a whole bunch of ideas today so you guys are in trouble :) 2010-11-11T19:58:08 two of the games from some of my old games I think are interesting: 2010-11-11T19:58:10 http://ai-contest.com/visualizer.php?game_id=5303057 2010-11-11T19:58:17 http://ai-contest.com/visualizer.php?game_id=5713053 2010-11-11T19:58:20 ps what happened to Ruilov? 2010-11-11T19:58:30 like, he kinda... disappeared 2010-11-11T19:59:53 what's with the youth vs experience claim? 2010-11-11T20:01:16 Well, I noticed a ton of students are playing. I wonder who the youngest is? 2010-11-11T20:01:33 yea actually mega1 are you a student or a professional? 2010-11-11T20:01:46 codefun64 was in here, said he was 14 :) 2010-11-11T20:02:59 there are only 151 pokemon 2010-11-11T20:03:15 you haven't watched much pokemon lately have you? 2010-11-11T20:03:37 what are you talking about, there are only 151 2010-11-11T20:03:41 ;) 2010-11-11T20:03:45 I agree with antimatroid1 2010-11-11T20:03:57 just like there only exists one matrix movie 2010-11-11T20:04:01 exactly 2010-11-11T20:04:06 I'd say there are 251 2010-11-11T20:04:14 you would be wrong 2010-11-11T20:04:19 clearly generation 2 counts 2010-11-11T20:04:23 *** bhasker has joined #aichallenge 2010-11-11T20:04:56 eburnette: are you interested in different testing maps? 2010-11-11T20:05:02 Migi32: I think this is mega1's blog: http://quotenil.com/ 2010-11-11T20:05:06 i generated 1000 with less restrictions on symmetry 2010-11-11T20:05:20 seems he has quire a lot of experience in AI :) 2010-11-11T20:05:35 sigh: think? 2010-11-11T20:05:45 it has his resume and everything 2010-11-11T20:05:52 he did ai for his masters 2010-11-11T20:06:01 yup 2010-11-11T20:06:21 *** bhasker__ has quit IRC (Ping timeout: 245 seconds) 2010-11-11T20:06:50 There are over 600 pokemon now 2010-11-11T20:08:22 his blog started in 2008. That means he's at least 2 ;) 2010-11-11T20:08:39 antimatroid1 I saw your maps but I'm having enough trouble with the existing ones :) 2010-11-11T20:08:51 you guys aren't very good at cognitive dissonance 2010-11-11T20:09:01 if i have that word right 2010-11-11T20:09:12 antimatroid1: huh? 2010-11-11T20:09:16 nope 2010-11-11T20:09:53 i was trying to use some smart word for choosing to ignore existing information about something, but i think i chose poorly :P 2010-11-11T20:10:15 Migi32: ahaha 2010-11-11T20:10:31 i'm stil laughing aha 2010-11-11T20:10:35 I still don't know what your point is 2010-11-11T20:10:44 there are only 151 pokemon 2010-11-11T20:10:51 that is my point 2010-11-11T20:11:08 oh... I thought you were in the mega1 conversation 2010-11-11T20:11:12 bilateral symmetry is exciting 2010-11-11T20:11:30 i was in both 2010-11-11T20:11:44 for that particular statement 2010-11-11T20:11:52 bilateral symmetry? 2010-11-11T20:12:11 it's confusing when people keep making up their own terms for things, yet i'm probably one of the worst offenders 2010-11-11T20:12:19 mirror-image symmetry maps 2010-11-11T20:12:25 maybe it's just a biology thing? 2010-11-11T20:12:47 all maps have been haven't they? 2010-11-11T20:13:14 the only real difference with mine is that the starting angles of the players aren't 180degrees apart 2010-11-11T20:13:40 59 has to be my current favourite, only because it looks like a space invader 2010-11-11T20:13:46 all maps have been radially symmetric, I think 2010-11-11T20:13:52 but yeah, the maps are pretty exciting 2010-11-11T20:14:26 *** Migi32 has quit IRC (Quit: www.ai-contest.com) 2010-11-11T20:14:43 oh yeah, i don't think you can do radially symmetric with starting angles not 180 degrees apart 2010-11-11T20:14:46 is that right? 2010-11-11T20:14:51 yeah 2010-11-11T20:14:59 i was going to have it switch between them randomly when generating, but it didn't work 2010-11-11T20:15:58 i wanted to have it randomly picking noPlanets between 20 and 25, but it seems to screw things up, at least when i try to play as a human on jbotmanager it always crashes the human on the first turn for some maps 2010-11-11T20:16:37 Top 10 players: bocsimacko(4365), george(4193), shangas(4153), Raschi(4124), oldman(4109), _Astek_(4089), GreenTea(4087), dmj111(4076), bix0r4ever(4073), asavis(4035) 2010-11-11T20:22:31 *** rebelxt has joined #aichallenge 2010-11-11T20:23:52 some new names on the TCP: nesselrode, metternich 2010-11-11T20:24:03 any idea whose bots these are? ;-) 2010-11-11T20:25:19 I'm relying on home planets to be on a line that intersects the center, is that a safe assumption? 2010-11-11T20:25:52 ha, those are me and me 2010-11-11T20:26:04 http://en.wikipedia.org/wiki/Congress_of_Vienna#Participants 2010-11-11T20:27:01 castlereagh's around too, but he's a little lower 2010-11-11T20:28:21 davidjliu, hahaha 2010-11-11T20:28:30 your bots are doing great 2010-11-11T20:28:33 I get bored with names sometimes 2010-11-11T20:28:55 heh, thanks... on the real server they have timeout issues sometimes though 2010-11-11T20:29:27 although I think they're fast enough now 2010-11-11T20:29:52 cool. i've heard you want to bail out at 0.8s wall time to be safe 2010-11-11T20:30:42 Heh... I'm currently using the unofficial python kit which apparently uses some functionality in python 2.7 to do that 2010-11-11T20:30:53 eburnette_: why? 2010-11-11T20:31:01 but yeah, at this point I'd rather see that I time out and try to make it faster than watch it do sill ystuff 2010-11-11T20:31:01 timeisup exception? yep it requires 2.6 2010-11-11T20:31:10 i was planning to push to have other maps included, but if there is decent opposition i will leave it alone 2010-11-11T20:31:42 I think it would probably screw up a few peoples' bots... some people may have some code with targeting planet 0, or something 2010-11-11T20:31:52 I draw a line between them and center activity around that line, seems to help some 2010-11-11T20:31:55 planet 0 is still the centre planet 2010-11-11T20:32:28 but now it's not always as important because it's not directly between the two home planets 2010-11-11T20:32:37 wait, let me get an example 2010-11-11T20:34:08 eburnette_: http://imgur.com/KSIDE.png is that red line what you mean? 2010-11-11T20:34:16 and can't you still do it, possibly having to change the code 2010-11-11T20:34:38 the orange line in the black graph 2010-11-11T20:35:03 hmmmm, i really like the less restrictive maps :( 2010-11-11T20:35:13 it requires a more general bot too 2010-11-11T20:35:20 *** rebelxt has left #aichallenge 2010-11-11T20:36:14 maybe a separate category in the contest for more general maps? 2010-11-11T20:36:56 I'm quite a fan of them too 2010-11-11T20:37:06 but for some reason it seems like my new bot times out on all of them on the first turn 2010-11-11T20:37:17 eburnette_: how much does the performance of your bot rely on that code? 2010-11-11T20:37:56 oh wait, it's just because my bot's really slow 2010-11-11T20:37:57 jk 2010-11-11T20:38:01 hard to say exactly but not much 2010-11-11T20:38:03 it's all good 2010-11-11T20:38:18 this is why specifics of the competition need to be concrete at the start 2010-11-11T20:38:24 +1 2010-11-11T20:38:29 but i'm pretty sure jeff always wanted maps like these 2010-11-11T20:38:32 he gave me the idea 2010-11-11T20:38:52 ooh, playing bocsi... 2010-11-11T20:39:00 not for long :P 2010-11-11T20:39:17 s i g h 2010-11-11T20:39:51 75 turns 2010-11-11T20:43:35 yo yo 2010-11-11T20:43:49 yo yo 2010-11-11T20:44:06 j3camero: hello, i made less restrictive symmetric maps 2010-11-11T20:44:08 hayo 2010-11-11T20:44:32 *** jeff_ has joined #aichallenge 2010-11-11T20:44:39 yo yo 2010-11-11T20:44:46 o_O 2010-11-11T20:44:51 wuht 2010-11-11T20:44:53 *** jeff_ is now known as Guest39644 2010-11-11T20:45:10 k I think I fixed my nick now. 2010-11-11T20:45:12 nice nick 2010-11-11T20:45:13 damn 2010-11-11T20:45:15 lol 2010-11-11T20:45:27 God damn it how do I fix my nick? 2010-11-11T20:45:37 /nickerv release nick password 2010-11-11T20:45:39 I try /nick j3camero. I'm such a newb guys. 2010-11-11T20:45:43 /nick 2010-11-11T20:46:01 /msg nickserv ghost 2010-11-11T20:46:02 *** bhasker_ has joined #aichallenge 2010-11-11T20:46:07 or that 2010-11-11T20:46:11 then /nick as before 2010-11-11T20:46:25 if it's on another computer, it might try to come back on, though 2010-11-11T20:46:29 some clients do that 2010-11-11T20:47:04 That's messed. I don't think I have irssi open in any other screen... 2010-11-11T20:47:27 Oh no wait here I am 2010-11-11T20:47:31 *** j3camero has left #aichallenge 2010-11-11T20:47:43 O_o 2010-11-11T20:47:48 lol 2010-11-11T20:47:54 *** Guest39644 is now known as release 2010-11-11T20:48:04 lol 2010-11-11T20:48:09 holy hell 2010-11-11T20:48:13 /nickserv not /nixk 2010-11-11T20:48:15 This is so embarassing, lol 2010-11-11T20:48:16 /nickserv not /nick 2010-11-11T20:48:34 you should be able to do /nick j3camero 2010-11-11T20:48:35 now 2010-11-11T20:48:36 though 2010-11-11T20:48:44 unless it has locked you 2010-11-11T20:48:52 MAybe 2010-11-11T20:49:06 I got a message back from nickserv saying j3camero is not a registered nickname. 2010-11-11T20:49:08 release: http://imgur.com/KSIDE.png are those the sort of maps you wanted a few weeks ago? 2010-11-11T20:49:21 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2010-11-11T20:49:25 oh... then in that case... just /nick j3camerio 2010-11-11T20:49:31 */nick j3camero 2010-11-11T20:49:34 lol antimatroid1... this nick better not stick :-P 2010-11-11T20:49:53 if you haven't registered, ignore everything I've said 2010-11-11T20:50:03 I do /nick j3camero but alas nothing happens. My understanding of irssi just broke. 2010-11-11T20:50:14 what does it say? 2010-11-11T20:50:43 I'll try :) 2010-11-11T20:50:48 *** sigh is now known as j3camero_ 2010-11-11T20:50:54 *** antimatroid1 is now known as sigh 2010-11-11T20:50:58 "Error(433): j3camero Nickname is already in use." 2010-11-11T20:51:05 *** sigh has quit IRC (Disconnected by services) 2010-11-11T20:51:10 haha 2010-11-11T20:51:15 *** j3camero_ is now known as sigh 2010-11-11T20:51:19 hehe 2010-11-11T20:51:25 Identity theft! 2010-11-11T20:51:29 all you did was leave the channel with /part 2010-11-11T20:51:32 you didn't /quit 2010-11-11T20:51:37 *** release has quit IRC (Quit: leaving) 2010-11-11T20:51:46 *** antimatroid has joined #aichallenge 2010-11-11T20:51:52 but you apparently found the right screen 2010-11-11T20:52:00 * Zannick facepalm 2010-11-11T20:52:01 wb antimatriod 2010-11-11T20:52:04 :P 2010-11-11T20:52:22 @click 2010-11-11T20:52:23 antimatroid: No! 2010-11-11T20:52:34 *** j3camero has joined #aichallenge 2010-11-11T20:52:39 Ah I'm me again. 2010-11-11T20:52:45 hi release 2010-11-11T20:52:48 oh... release was the wrong command all that time... I as thinking of recover 2010-11-11T20:52:49 That was a good tip on the /quit 2010-11-11T20:53:02 so yeah, most of my advice was stupid anyway 2010-11-11T20:53:16 Anyways yeah. You guys would probably be interested in seeing this: http://ai-contest.com/game_stats.php 2010-11-11T20:53:31 more servers? 2010-11-11T20:53:33 woaaaaaaaaaaaaaaaaaah 2010-11-11T20:53:35 Yeah 2010-11-11T20:53:36 nice 2010-11-11T20:53:37 :) 2010-11-11T20:53:41 And they should stay up from now on. 2010-11-11T20:53:53 Instead of just burning out 30 minutes after each time I kick them. 2010-11-11T20:53:54 cool 2010-11-11T20:53:58 woooooooooooooooah 2010-11-11T20:53:59 j3camero: what about maps? :P 2010-11-11T20:54:03 We're at 90 games per minute and climbing. 2010-11-11T20:54:19 @roulette 2010-11-11T20:54:19 *** davidjliu was kicked by contestbot (BANG!) 2010-11-11T20:54:20 * contestbot reloads and spins the chambers. 2010-11-11T20:54:24 *** bhasker_ has quit IRC (Ping timeout: 276 seconds) 2010-11-11T20:54:26 haha 2010-11-11T20:54:30 antimatroid: I looked at the img you sent me. What property of the maps are you highlighting there? 2010-11-11T20:54:37 *** davidjliu has joined #aichallenge 2010-11-11T20:54:50 the fact that the line joining home planets doesn't intersect the centre planet 2010-11-11T20:55:09 Ahhh 2010-11-11T20:55:13 the symmetry is done the other direction too, as radial symmetry doesn't work with randomly chosen starting angles 2010-11-11T20:55:47 The left map seems to not have this property. 2010-11-11T20:56:00 the left map is from the official site 2010-11-11T20:56:06 But the right one does. 2010-11-11T20:56:09 i'm pretty sure none of the older maps have that property 2010-11-11T20:56:15 it's to show the difference 2010-11-11T20:56:31 current generator is all 180 degree rotation 2010-11-11T20:56:53 i'm interested in having some of these different maps included, although some people seem to rely on it being like the first map 2010-11-11T20:57:08 Fuck em 2010-11-11T20:57:12 I don't think people should be able to rely on that 2010-11-11T20:57:12 :S 2010-11-11T20:57:12 Your maps are better 2010-11-11T20:57:13 :D 2010-11-11T20:57:21 *** wh1teside_ has joined #aichallenge 2010-11-11T20:57:42 The map on the right is not quite symmetric right? 2010-11-11T20:57:52 it is, jbotmanager just sucks at drawing them 2010-11-11T20:57:57 no, the line isn't quite on the axis 2010-11-11T20:58:04 it is symettric 2010-11-11T20:58:20 sigh: look at the 89+4 planets near home 2010-11-11T20:58:34 ah, yeah 2010-11-11T20:58:40 It looks like red is closer to the centre than green. That could be an issue. Or is it just an optical illusion? 2010-11-11T20:58:45 jbotmanager just sucks at drawing, i've tested maps both when generating them and counting turns in the program to see fleets actually take the same time 2010-11-11T20:58:55 aspect ratio, I think 2010-11-11T20:58:59 ? 2010-11-11T20:59:00 it's jBotManager not rendering them properly i think 2010-11-11T20:59:23 Ah okay. So the map actually is symmatric? 2010-11-11T20:59:32 j3camero: there are 1000 of them here http://tinyurl.com/antimaps (59 looks like a space invader) 2010-11-11T20:59:56 There's nothing special or well thought out about the current map generator. I wrote it on a greyhound bus on day when I was first starting to think about Galcon. It really wasn't supposed to last this long. 2010-11-11T20:59:57 but i wouldn't add those to official ones, one would prefer to test to make sure they are perfect first 2010-11-11T21:00:14 that's why i wrote this, no one else was 2010-11-11T21:00:43 I am sure they're better than what we have. What we have was never supposed to get past the stage of "I need a map with more than 2 planets just to make sure the game logic is right" 2010-11-11T21:01:12 *** wh1teside has quit IRC (Ping timeout: 264 seconds) 2010-11-11T21:01:27 what about final tournament, could we have an entirely new set of maps generated just for that? 2010-11-11T21:01:41 I think that would be wise, yes. 2010-11-11T21:01:47 antimatroid++ 2010-11-11T21:02:05 Can you create a branch in the repo and replace the map generator with yours? Put it up for review and I'll review it. 2010-11-11T21:02:32 i wouldn't have a clue how to do that :P plus it's in c++ atm, and i think janzert is right that it should be in python 2010-11-11T21:03:13 If that's what he wants that's cool. 2010-11-11T21:03:22 there will need be decisions made on parameters too 2010-11-11T21:03:29 Python gives me a boner, but C++ will always be my native language. 2010-11-11T21:04:29 It's not as rigorous or scientific as you think. I'm pretty sure that anything would be better than what we have now. 2010-11-11T21:04:53 Oh, we did promise people that the minimum travel time would be 2.0. We gotta make sure we abide by that. 2010-11-11T21:05:03 i think they need to be made for: min growth, max growth, min starting distance between home planets, min distance between planets, max distance between planets (preferably this would be defined as an even number) and i think 23 planets should be kept 2010-11-11T21:05:21 if i allow 0 growth, i seem to get quite a few of them 2010-11-11T21:05:41 I hate 0 growth. 2010-11-11T21:05:51 I know they have a use and all, but they don't visualize well. 2010-11-11T21:05:55 I second the motion to remove 0 growth 2010-11-11T21:06:02 keep max 5 growth? 2010-11-11T21:06:08 Sure. 2010-11-11T21:06:15 I see no reason to change the max 5 growth 2010-11-11T21:06:27 Min travel time == 2.0 is improtant. 2010-11-11T21:06:30 what about starting ships at a neutral? 0-100? 1-100? 2010-11-11T21:06:40 Cartesian distance has to be > 1.0 in other words. 2010-11-11T21:06:46 j3camero: yep 2010-11-11T21:06:51 antimatroid: I would support changing it 2010-11-11T21:07:01 sigh: changing what? 2010-11-11T21:07:19 I'm not sure. Check and see what I put in the existing map gen for the population of neutral planets. 2010-11-11T21:07:28 oh sorry, thought you were talking about starting planet ships 2010-11-11T21:07:39 no, that is 100 ships and 5 growth, i'm not altering that 2010-11-11T21:07:45 ok, sure 2010-11-11T21:07:57 I picked 100 for the home planets starting population because I think that's what Galcon uses. 2010-11-11T21:08:05 ahh 2010-11-11T21:09:09 the other thing is, currently there is always one planet that starts an equal distance from both players, i can change that to being a randomly chosen uneven number of planets at an even starting distance 2010-11-11T21:09:10 don't forget to do adjustments to avoid rounding errors 2010-11-11T21:09:35 eburnette_: i only add a pair of neutrals if their distance and their distance to all other planets is >= min distance 2010-11-11T21:09:46 Existing map generator: http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/map_generator.py 2010-11-11T21:10:01 antimatroid: eburnette_ is talking about distances that are too close to an integer 2010-11-11T21:10:03 I think 2010-11-11T21:10:49 I thought about that before. Could be a good idea. 2010-11-11T21:10:50 wait, can someone explain that further 2010-11-11T21:11:24 If the distance between two planets is 4.99999999 it could cause problems because some languages might think the travel time is 5 and others might think its 4. 2010-11-11T21:11:41 PlanetWars.Distance takes ceil of the real distance, so yeah what j3camero said 2010-11-11T21:11:55 yup, so check if distance is greater than epsilon of an integer 2010-11-11T21:11:59 yep 2010-11-11T21:12:00 easy 2010-11-11T21:12:12 Like if you have two planets: one at (0,0) and another at (sqrt(2), sqrt(2)) 2010-11-11T21:12:13 from every planet to every other planet 2010-11-11T21:12:14 alright, i'ma download python and write it up 2010-11-11T21:12:27 Sick. 2010-11-11T21:12:30 *** Bobng has quit IRC (Ping timeout: 255 seconds) 2010-11-11T21:12:30 does python generate random doubles for me easily? 2010-11-11T21:12:37 Do you see the existing map gen? 2010-11-11T21:12:41 yeah 2010-11-11T21:12:49 you should be able to just copy most of that :) 2010-11-11T21:12:56 import random; r = random.random(); print r; 2010-11-11T21:13:20 and it should keep the part where it makes planets not be on top of each other (not just starting planets) 2010-11-11T21:13:33 eburnette_: yeah... antimatroid already has that 2010-11-11T21:13:34 that should happen from min sitance 2010-11-11T21:13:43 distance* 2010-11-11T21:13:51 *** superflit has joined #aichallenge 2010-11-11T21:13:55 Python's random is so bitchin. You can even do "list = [2, 3, 5, 7, 11, 13, 17]; r = random.choose(list); print r;" 2010-11-11T21:14:07 oh, one last thing, i am still allowed to compete and write the used map generator yes? 2010-11-11T21:14:19 Yeah 2010-11-11T21:14:21 lol 2010-11-11T21:14:24 :P 2010-11-11T21:14:32 just making sure :) 2010-11-11T21:14:45 You're only banhammered if you have root on a contest server or something like that. Contributing to the open source project is fine. 2010-11-11T21:14:45 that would be the ultimate troll 2010-11-11T21:15:01 aha sigh 2010-11-11T21:15:14 Thanks for the map gen dude. Now gtfo. 2010-11-11T21:15:57 what's the best ide for python? 2010-11-11T21:16:14 edlin 2010-11-11T21:16:38 Top 10 players: bocsimacko(4390), asavis(4225), george(4207), Raschi(4193), shangas(4155), _Astek_(4114), oldman(4112), GreenTea(4105), dmj111(4077), Slin-.-(4019) 2010-11-11T21:16:53 emacs of course 2010-11-11T21:17:05 also notepad.exe 2010-11-11T21:17:18 antimatroid: unless you have a preferred text editor, just use the on that comes with python (idle) 2010-11-11T21:17:39 But real men use emacs. 2010-11-11T21:18:13 I'm sure emacs has a very nice python interpreter, but hte guy just needs a text editor 2010-11-11T21:18:43 :P 2010-11-11T21:18:53 Oh snap. 2010-11-11T21:18:59 you guys are too mature for trolling. 2010-11-11T21:19:01 Damn. 2010-11-11T21:19:04 http://jpcameron.com/blog/?p=123 2010-11-11T21:22:33 93 games per minute and rising. Yes! 2010-11-11T21:23:00 The servers should crap out soon. But we're gonna set up the server-kicking script as a cron job soon. 2010-11-11T21:23:15 why do they crap out? 2010-11-11T21:28:03 hmm, how do i compile a python file? :P 2010-11-11T21:29:18 ie. command? 2010-11-11T21:29:54 *** JensTi has quit IRC (Quit: Leaving.) 2010-11-11T21:30:23 just run: python filename.py 2010-11-11T21:30:29 oh, that's neat, i just run the file 2010-11-11T21:30:35 yes :P 2010-11-11T21:30:41 welcome to python aha 2010-11-11T21:32:14 what did you end up using to write it? 2010-11-11T21:32:20 c++ 2010-11-11T21:32:39 no, what did you use to write the python script? 2010-11-11T21:32:47 codeblocks aha 2010-11-11T21:33:58 ok, don't know that one... if it is a decent editor you should be able to run the script without leaving it :) 2010-11-11T21:35:27 oh, what's a good minimum starting distance? 2010-11-11T21:35:50 I'd wait for McLeopold's opinion :P 2010-11-11T21:36:31 yep, it's easily changed, i'll leave it at 7 for the moment 2010-11-11T21:36:50 what is it in the official? 2010-11-11T21:38:12 repo 2010-11-11T21:38:16 @repo 2010-11-11T21:38:17 sigh: repo = http://code.google.com/p/ai-contest/ 2010-11-11T21:38:40 sigh: i think it's different, in that you aren't now on opposing sides of the map 2010-11-11T21:53:01 does math.sin/cos take arguments in degrees or radians? 2010-11-11T21:54:14 radians o_O 2010-11-11T21:54:23 what kind of crazy lib would take degrees 2010-11-11T21:54:40 hmm :P well i hate converting 2010-11-11T21:55:11 math.radians 2010-11-11T22:15:10 Hope you don't take out 0 growth planets, esp. at center 2010-11-11T22:15:38 why? 2010-11-11T22:15:43 ? 2010-11-11T22:15:48 (btw, currently 0-growth can only occur in the center) 2010-11-11T22:15:58 they make a nice launching point, like on map 1 2010-11-11T22:16:11 and some bots don't handle them well 2010-11-11T22:16:15 i will give you 0 centre planets for the moment 2010-11-11T22:16:39 Top 10 players: bocsimacko(4356), asavis(4284), george(4216), shangas(4198), Raschi(4155), oldman(4147), GreenTea(4096), _Astek_(4089), dmj111(4077), _iouri_(4040) 2010-11-11T22:18:18 *** superflit has quit IRC (Ping timeout: 260 seconds) 2010-11-11T22:19:31 *** virdo has joined #aichallenge 2010-11-11T22:27:22 *** Prillicy has joined #aichallenge 2010-11-11T22:28:03 I understand we're permitted to use our own files, does this mean I may store information in a file and write to it during the game? 2010-11-11T22:28:37 no, writing to files is specifically prohibited in the rules 2010-11-11T22:28:50 good thing I asked then :) 2010-11-11T22:28:53 you can upload any precomputed data files you want though 2010-11-11T22:29:12 as long as you stay under the upload size limit anyway :) 2010-11-11T22:29:14 no file can predetermine the average distance of the planets before the game begins though 2010-11-11T22:29:24 which means I have to do so via a nested loop 2010-11-11T22:29:34 every round 2010-11-11T22:29:51 unless I'm mistaken about how the game works 2010-11-11T22:29:53 just do it on the first move 2010-11-11T22:30:03 the bot keeps running for the whole game 2010-11-11T22:30:11 that doesn't make sense 2010-11-11T22:30:34 ah, I c 2010-11-11T22:30:36 thanks 2010-11-11T22:34:08 I thought it was run once each round 2010-11-11T22:34:17 a more challenging, but less logical test for an ai 2010-11-11T22:36:44 janzert: i'm about halfway through rewriting the map generator in python 2010-11-11T22:36:52 sweet 2010-11-11T22:36:55 i'm learning python as i do it though, so it's not happening terrible fast 2010-11-11T22:37:35 half way already is pretty good 2010-11-11T22:38:23 i'm using the old one as a base 2010-11-11T22:39:13 what would people like as a map radius? 2010-11-11T22:39:17 it's currently set at 12 2010-11-11T22:39:21 so max distance is 24 2010-11-11T22:39:32 current seems good 2010-11-11T22:39:51 the maps i genreated last night had like max distance of 40 2010-11-11T22:39:51 keep whatever the current generator does 2010-11-11T22:39:55 okay 2010-11-11T22:40:51 some algs slow down if map gets bigger because of distances till ships lands, also some bots have hardcoded future time lookaheads (like mine) 2010-11-11T22:40:52 i think the map generating code used for final maps should be open to anyone to view as well 2010-11-11T22:41:05 i think that's already the case yeah? 2010-11-11T22:41:11 yes 2010-11-11T22:41:14 yes 2010-11-11T22:41:28 * wishes he could get rid of the _ after his name 2010-11-11T22:41:31 i wasn't sure cause not everyone can see the discussions 2010-11-11T22:44:05 leave the max radius at 12, yeah 2010-11-11T22:49:27 done 2010-11-11T22:55:03 am I the only one who's noticed a random factor involved in the battles? 2010-11-11T22:55:19 what do you mean by random factor? 2010-11-11T22:55:28 playing the same battle multiple times occasionally turns out a different result 2010-11-11T22:55:34 Prillicy: games are deterministic unless you are explicitly using random numbers 2010-11-11T22:55:39 are you using a tree? 2010-11-11T22:55:49 no, basic formulas 2010-11-11T22:55:56 there should be no difference 2010-11-11T22:56:05 although the original visualiser was buggy 2010-11-11T22:56:19 Prillicy: never seen that, and I use that fact that it is deterministic in my testing 2010-11-11T22:56:21 there's an article on deterministic bots somewhere 2010-11-11T22:57:01 bots can be nondeterministic based on cpu time they're given 2010-11-11T22:57:29 this is especially true of traditional minimax chess bots 2010-11-11T22:57:42 Zannick: sure, but mine will just timeout at the moment and that is easy to detect :) 2010-11-11T22:58:08 sure 2010-11-11T22:59:10 I just know that I saw it with mybot, even the testbotts when facing each other would occasionally result in (while similar) different results, and often a different winner 2010-11-11T22:59:22 *** sigh has quit IRC (Remote host closed the connection) 2010-11-11T22:59:36 *** sigh has joined #aichallenge 2010-11-11T22:59:40 hey that's clever, my bot is trying to use the oponents ships to attack him 2010-11-11T22:59:51 eh? 2010-11-11T23:00:03 funny bug 2010-11-11T23:00:24 ahh :) 2010-11-11T23:00:31 I'm still in the process of creating a value formula 2010-11-11T23:00:34 I thought you had found some new strategy :P 2010-11-11T23:00:56 yeah, corrupting the server state to change his fleets into yours 2010-11-11T23:01:06 that would be a pretty useful strategy 2010-11-11T23:02:05 *** codefun64 has joined #aichallenge 2010-11-11T23:02:10 Anyone know how the points system works? 2010-11-11T23:02:40 I've won 6 battles and lost 1 but haven't advanced in the ranks almost at all, whereas other people have done similarly well and gotten to like rank 11 etc. 2010-11-11T23:03:10 it depends on who you won and lost against 2010-11-11T23:03:35 just win against bocsi and you'll be fine 2010-11-11T23:03:41 yep :) 2010-11-11T23:04:22 Wishful thinking - my bot's not even done yet :) 2010-11-11T23:04:25 *** Prillicy has quit IRC (Quit: I be bangin') 2010-11-11T23:04:35 Plus I dropped like 3k ranks with updates that failed miserably -_- 2010-11-11T23:05:27 yeah, disadvantage of rating of an old submission being used to start off a new one 2010-11-11T23:06:09 janzert: I thought you weren't doing that atm? 2010-11-11T23:07:00 Dear god I hate filter keys. 2010-11-11T23:07:01 the new matchup code that went in today uses an old submissions rating for a new unranked submission to determine an appropriate pairing 2010-11-11T23:07:12 was listening to music then BEEEEEEP wanna turn filter keys on? 2010-11-11T23:07:15 ah, cool :) 2010-11-11T23:07:39 That sucks, Janzert o_O 2010-11-11T23:08:14 so it doesn't really keep the old ranking at all but it does tend to pair it at a more appropriate spot in the rankings initially 2010-11-11T23:08:45 codefun64: heh, you weren't around to hear all the, err, whining about the old method 2010-11-11T23:09:14 janzert: one way around that disadvantage is to stick to the old method for bots below the average 2010-11-11T23:09:44 or just push everyone back towards the center instead of at their old ranking 2010-11-11T23:10:11 but isn't that what people were complaining about? 2010-11-11T23:10:17 but it still shouldn't take a bot that long to climb back up anyway 2010-11-11T23:10:33 does it still start at bottom most ranking? 2010-11-11T23:10:36 If the bot is *good* >_> <_< 2010-11-11T23:10:38 it took my bot over 24 hours to settle 2010-11-11T23:10:48 before it always dropped you in at the same rank 2010-11-11T23:11:22 I assume that was 0 elo (before offset)? 2010-11-11T23:11:23 didn't it used to start you at bottom? 2010-11-11T23:11:37 nah, it started you in the middle of the pack 2010-11-11T23:12:31 and now it starts you where you were? or it starts you in the middle and pairs you with a battle at where you were? 2010-11-11T23:12:49 I think it pairs you up with someone close to your previous rank 2010-11-11T23:13:00 yeah, that's the idea... I think 2010-11-11T23:13:05 codefun64: yes, it does now 2010-11-11T23:13:17 but not before today 2010-11-11T23:13:40 what happens if you lose the first battle? 2010-11-11T23:13:57 then you are not worthy! 2010-11-11T23:14:00 ? it's going to drop your rank 2010-11-11T23:14:03 What I'd like to know is why janzert 6 wins and a runtime error gets me only from rank 3900 to 3647... seems to be a little slow to me 2010-11-11T23:14:10 wtf? 2010-11-11T23:14:15 I did NOT type janzert's name in there... 2010-11-11T23:14:23 hehe 2010-11-11T23:14:30 this sentence has janzert six words 2010-11-11T23:14:35 stupid key must've cycled through names -_- 2010-11-11T23:15:08 because the bots you are playing against at the moment are really bad and you lost against one 2010-11-11T23:15:23 so the rating system will take a while to let you overcome that lost 2010-11-11T23:15:44 so let's say ranks go from 1 (bocsi) to 3000, and I was at 100 and resubmit. I start at 1500th place and get paired with somebody at 100th place. If I win, I jump up close to 100th place, and if I lose I go to 2000th place? 2010-11-11T23:15:47 Well that's another thing, I think it's annoying to have a runtime error set you back in ranks when you've had other battles that don't have that glitch 2010-11-11T23:15:57 my general impression is that most of the bots down there have lots and lots of games so a very certain ranking 2010-11-11T23:16:09 I also find it interesting it even had that error, it didn't give me any on my end and the other battles didn't glitch >_> 2010-11-11T23:16:15 since they were basically just submitted and then abandoned 2010-11-11T23:16:21 the thing is you don't go up or down ranks based on your opponent's rank on one match 2010-11-11T23:16:40 Top 10 players: bocsimacko(4352), asavis(4297), george(4202), shangas(4194), oldman(4154), Raschi(4150), GreenTea(4095), dmj111(4086), _Astek_(4065), _iouri_(4055) 2010-11-11T23:16:47 this is a common misconception 2010-11-11T23:16:56 elo is not per-game, it's overall 2010-11-11T23:17:03 *** bhasker has joined #aichallenge 2010-11-11T23:17:12 then my rank should be awesome not pitiful! :D 2010-11-11T23:17:15 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-11T23:17:29 so if you resubmit, and are paired against someone ranked 100, and win, then elo sees only that 2010-11-11T23:17:35 you weren't really at rank 1500 before 2010-11-11T23:17:47 plus it evaluates everyone at the same time 2010-11-11T23:18:01 so 6 wins against people with lots of losses isn't going to boost you very much 2010-11-11T23:18:04 oh wait 7 wins - missed one... 2010-11-11T23:18:08 I see 2010-11-11T23:18:10 Ok, so suppose boscsi resubmits and happens to get unlucky and lose his first battle, how long does it take him to get back to 1st place? 2010-11-11T23:18:27 who knows 2010-11-11T23:18:29 depends on who he lost to 2010-11-11T23:18:32 not very long 2010-11-11T23:18:46 depends on who he loses to, and who he beats 2010-11-11T23:19:01 if he somehow lost to to the 4000th rank bot then a very long time 2010-11-11T23:19:20 *makes him lose to the 4000th bot* 2010-11-11T23:19:22 janzert: presumably the first game is against someone high ranked, say top 10 2010-11-11T23:19:22 :D 2010-11-11T23:19:33 Well if it's with George he'd probably lose 2010-11-11T23:19:37 sigh: then probably not that long 2010-11-11T23:19:37 only if he gets paired against people all the way down there 2010-11-11T23:20:00 but it really depends on what bayeselo determines for him 2010-11-11T23:20:07 Zannick: right it would be quite unlikely for him to even get that pairing of course 2010-11-11T23:20:25 the only history bayeselo has at that point is that one game 2010-11-11T23:20:36 * sigh offers a sacrifice to the bayeselo black box... please look upon me favorably 2010-11-11T23:21:01 who/what is bayeselo? 2010-11-11T23:21:12 the scoring algorithm 2010-11-11T23:21:14 there was probably a whole discussion about this but why not keep the old history? 2010-11-11T23:21:16 Ah 2010-11-11T23:21:17 the arbiter 2010-11-11T23:21:35 "What would you have your arbiter do?" 2010-11-11T23:21:44 an impartial judge who cannot be bribed 2010-11-11T23:21:50 except by admins 2010-11-11T23:21:57 i was about to say 2010-11-11T23:21:58 (and virgins) 2010-11-11T23:22:04 o_O; 2010-11-11T23:22:12 eburnette_: because it's a different bot, if it has a fatal bug and crashes 50% of the time it doesn't deserve to get credit for all the wins of the last bot 2010-11-11T23:22:35 and it will screw up the ranks for other people 2010-11-11T23:22:56 if bosci resubmits copybot, it doesn't deserve to keep first...much less top 200 2010-11-11T23:23:13 codefun64: it seems like you timeout on maps where there is a zero-growth planet. Do you ever divide by a planet's growth rate? 2010-11-11T23:23:48 Well hypothetically sure but that seems unlikely, and even it it happened the new bot would start dropping 2010-11-11T23:24:14 eburnette_: but then it will also affect anyone who has played bocsimacko 2010-11-11T23:24:36 oic 2010-11-11T23:24:37 inconsistent bots in general will wreak havoc on the rankings 2010-11-11T23:25:40 like ObligatoryShampoo, who seemingly can lose to anybody and beat anybody on TCP 2010-11-11T23:26:17 did we ever figure out who Shampoo was, btw? 2010-11-11T23:26:34 also bayeselo in general does not handle a single entity that changes strength over the course of the game history it is given exceptionally well 2010-11-11T23:26:53 one of it's basic assumptions is that this does not occur actually 2010-11-11T23:26:56 if (random() > .25) win() else lose(); 2010-11-11T23:27:07 janzert: you don't have decay over time? 2010-11-11T23:27:09 that's not a very good line 2010-11-11T23:27:15 should be .3, maybe .4 2010-11-11T23:27:19 >.> 2010-11-11T23:27:38 bayeselo doesn't have a concept of time at all in it's calculations 2010-11-11T23:27:44 oh, ok 2010-11-11T23:27:44 not going to complain about my missing semi? :) 2010-11-11T23:28:22 is there still a limit on the games it considers? iirc, only from the last 3 days 2010-11-11T23:28:31 yes, there is 2010-11-11T23:28:36 eburnette_: care to share your win() method? 2010-11-11T23:28:53 I'd like to try and get rid of that limit in the next week 2010-11-11T23:29:05 I had one but my power went out and I didn't have a backup sorry 2010-11-11T23:29:27 but database performance is a huge issue, it already takes 4 minutes to query the 3 days of games 2010-11-11T23:29:33 void win() { solve_game(); } 2010-11-11T23:30:06 wow, 164 games per minute 2010-11-11T23:31:08 void solve_game() { win(); } 2010-11-11T23:31:32 I love the gpm when it's going through the bottom of the rankings :P 2010-11-11T23:33:26 *** superflit has joined #aichallenge 2010-11-11T23:33:56 My rank is still the same after reuploading code with a fix >_> 2010-11-11T23:35:23 codefun64: that's just to start you off... wait for some games 2010-11-11T23:36:24 the rankings only update every 10 minutes btw, so don't expect instantaneous change after a game either 2010-11-11T23:36:34 10 minutes currently that is 2010-11-11T23:36:36 i know i'm not working on improving my bot...but i like watching it when it wins 2010-11-11T23:36:39 makes me so produ 2010-11-11T23:36:41 proud * 2010-11-11T23:36:43 Why not? I want my instant gratification now! ;) 2010-11-11T23:36:43 http://ai-contest.com/visualizer.php?game_id=6974215 2010-11-11T23:37:09 tcp = instant 2010-11-11T23:44:32 i think i'm done 2010-11-11T23:49:12 http://codepad.org/tTLNbALt anyone want to take a look at that for me? 2010-11-11T23:51:06 antimatroid: you random.randint instead of rolling your own :P 2010-11-11T23:51:24 ? i just used what jeff cameron did for random ints 2010-11-11T23:52:31 I think I saw something weird before getting a random int somewhere in the codebase 2010-11-11T23:52:40 oh sorry, you wanted doubles 2010-11-11T23:52:50 then you both should have used random.uniform 2010-11-11T23:52:53 :P 2010-11-11T23:53:06 does it matter now? :P 2010-11-11T23:53:26 why have a min starting distance instead of comparing p1 and p2 and p for mindistance? 2010-11-11T23:54:08 I think he wanted a larger distance for starting planets 2010-11-11T23:54:18 over and above the min_distance 2010-11-11T23:54:33 it can be made smaller, but given you aren't on opposite sides of the map, i think if you start too close the game might be a trivial draw 2010-11-11T23:54:47 not necessarily on opposite sides of the map* 2010-11-11T23:54:58 i guess...? 2010-11-11T23:55:16 I think it's a wise decision 2010-11-11T23:55:19 the minimum on the other maps would be 4, i think, since it's at least 2 to center 2010-11-11T23:55:24 i think 7 is a bit much 2010-11-11T23:55:52 any time the home planets are closer than cost of the nearest neutral / growth rate it's a 'trivial' draw right? 2010-11-11T23:56:01 ooh, also 2010-11-11T23:56:05 but it seems an awful lot of bots still don't recognize that 2010-11-11T23:56:16 random() generates numbers in [0,1) 2010-11-11T23:56:27 so you can use 360 instead of 359.999 2010-11-11T23:56:33 and not hit 360 2010-11-11T23:56:40 last time maps that resulted in lots of draws were removed from the pool of maps 2010-11-11T23:56:45 we could reduce it and do something like that again 2010-11-11T23:57:01 if I have the formula right above then instead of a hardcoded distance for the home planets you could base it off that 2010-11-11T23:57:01 it will never get to 360? 2010-11-11T23:57:23 okay cool 2010-11-11T23:57:30 right random.random will never return 1 2010-11-11T23:57:39 your random function only returns max if random() returns 1, which won't happen 2010-11-11T23:57:58 yep 2010-11-11T23:58:42 oh 2010-11-11T23:58:51 Zannick: interesting... reading the python docs on ranodm.uniform 2010-11-11T23:58:58 http://docs.python.org/library/random.html 2010-11-11T23:59:05 why the check that p1[x] isn't within epsilon of round(p1[x])? 2010-11-11T23:59:34 "The end-point value b may or may not be included in the range depending on floating-point rounding in the equation a + (b-a) * random()." << python library is not willing to make the claim you are :)