2011-02-04T00:05:08 Mathnerd314: it was surely a joke 2011-02-04T00:52:35 McLeopold: are maps always to start with one ant per team? 2011-02-04T00:57:42 *** Mathnerd314 has quit IRC (Remote host closed the connection) 2011-02-04T01:05:04 amstan, j3camero: either of you about? 2011-02-04T01:05:24 i'm going to write a map generator, atm it looks like you guys are using .png files for maps? 2011-02-04T01:19:41 *** boegel has joined #aichallenge 2011-02-04T01:21:30 *** boegel has quit IRC (Client Quit) 2011-02-04T01:47:56 oh, interesting 2011-02-04T02:06:19 *** mega1 has joined #aichallenge 2011-02-04T02:27:09 Frontier: how would your visualiser go with smaller squares/ants but without hp/harvest? 2011-02-04T02:28:45 McLeopold: give me a yell when you're about, i want to talk input/output :) 2011-02-04T03:28:34 *** smellyhippy has joined #aichallenge 2011-02-04T03:28:34 *** smellyhippy has joined #aichallenge 2011-02-04T03:30:00 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-02-04T03:40:32 *** aerique has joined #aichallenge 2011-02-04T03:41:24 *** Palmik has joined #aichallenge 2011-02-04T03:58:35 *** boegel has joined #aichallenge 2011-02-04T04:24:53 *** pyro- has quit IRC (Ping timeout: 240 seconds) 2011-02-04T04:26:33 *** Naktibalda has joined #aichallenge 2011-02-04T04:26:35 *** Meatkat has joined #aichallenge 2011-02-04T04:29:54 *** pyro- has joined #aichallenge 2011-02-04T04:30:51 *** pyro- has quit IRC (Changing host) 2011-02-04T04:30:51 *** pyro- has joined #aichallenge 2011-02-04T05:23:35 *** mega1 has quit IRC (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) 2011-02-04T05:23:42 *** mega1 has joined #aichallenge 2011-02-04T06:56:07 *** Blkt has joined #aichallenge 2011-02-04T06:57:44 good day everyone 2011-02-04T07:09:07 Blkt: hi Blkt 2011-02-04T07:57:04 *** delt0r_ has joined #aichallenge 2011-02-04T07:58:45 hi aerique 2011-02-04T08:00:37 *** Palmik_ has joined #aichallenge 2011-02-04T08:00:39 *** amstan has joined #aichallenge 2011-02-04T08:00:39 *** ChanServ sets mode: +o amstan 2011-02-04T08:01:26 *** Palmik has quit IRC (Ping timeout: 250 seconds) 2011-02-04T08:29:23 *** amstan has quit IRC (Read error: Operation timed out) 2011-02-04T09:14:18 *** mceier has joined #aichallenge 2011-02-04T09:52:35 *** iFire` has joined #aichallenge 2011-02-04T09:54:33 antimatroid: It depends on how small the squares would be. There is no point in animating the ants when they are on 3x3 px squares. 2011-02-04T09:56:13 The same goes for the background graphics. They would be too distracting I think. 2011-02-04T09:58:13 *** Zannick has quit IRC (Ping timeout: 240 seconds) 2011-02-04T09:58:59 *** iFire has quit IRC (Read error: Connection reset by peer) 2011-02-04T09:59:19 *** Zannick has joined #aichallenge 2011-02-04T10:01:47 Frontier: it depends on what dimensions get used i guess 2011-02-04T10:02:27 i'm waiting to talk to McLeopold about io and maps to do a map generator 2011-02-04T10:04:00 it would be fairly easy to make a symmetric torus map for general n player games if each player got a ring section of the torus, or if 2|n, then give each player half a ring section, but the symmetry for that is a bit more iffy i think 2011-02-04T10:14:43 Wow, lots of action on the forum recently. 2011-02-04T10:15:55 I'd say that would mean that the next competition will be announced soon, but I think it means people *want* the next competition to be announced soon. 2011-02-04T10:17:29 How do you propose to make a visualizer for a torus shaped map? 2011-02-04T10:18:10 It's nothing different. It only means that the units can move out on one side and come back in on the other 2011-02-04T10:18:37 Oh, right 2011-02-04T10:19:52 Ants ought to be on a Moebius strip. 2011-02-04T10:21:33 Then just wall off the left and the right side :) 2011-02-04T10:21:51 *** bhasker has joined #aichallenge 2011-02-04T10:22:04 8-D 2011-02-04T10:25:38 Frontier: not quite :P 2011-02-04T10:25:47 you wrap the left bottom to right top 2011-02-04T10:26:22 otherwise you just get a cylinder 2011-02-04T10:29:07 I don't know if I like the idea of wrapping the map. I think it adds too many negative affects in the form of a barriers for noobies understanding the game (and feeling confident enough to make a player) than it adds in gameplay enhancement. 2011-02-04T10:31:08 it's not that complicated 2011-02-04T10:31:22 and makes it easier to do symmetry for n>4 players 2011-02-04T10:31:27 says the Masters student? or PhD student? 2011-02-04T10:31:38 just finished undergrad 2011-02-04T10:33:21 It does make symmetry easier for large number of players. But like I said, I think it throws up an unnecessary barrier to people just starting out. 2011-02-04T10:34:29 But if it were specified that there would be a large number of players per map it would probably be necessary. 2011-02-04T10:35:15 it's fairly easy to write a "get direction" function using an array of directions 2011-02-04T10:35:36 you just translate and modulo 2011-02-04T10:35:53 it could even be included in the starter bots 2011-02-04T10:36:32 If it's abstracted out in the started bots that'd be okay. 2011-02-04T10:36:59 But that just adds more work to the people making the starter bots. 2011-02-04T10:37:20 Frontier: i still much prefer to ants game we had going :( that would would with more players on a torus too 2011-02-04T10:37:41 Meatkat: that shouldn't be an issue 2011-02-04T10:37:50 hell i'll write the c++ one 2011-02-04T10:39:10 Okay... and the other 10? 2011-02-04T10:39:35 it usually starts with 4-5, then competitors write the rest 2011-02-04T10:39:47 it really is like a <10 line function 2011-02-04T10:40:12 I just don't find it logical to say that man-hour concerns aren't an issue... 2011-02-04T10:40:47 it's not expensive relative to a lot of other things 2011-02-04T10:40:52 making it fairly insignificant 2011-02-04T10:41:35 You sound like a congressmen... lol 2011-02-04T10:41:50 i did do an economics degree :P 2011-02-04T10:42:08 $1 million there $1 million there, it's fairly insignificant in the grand scheme of things 2011-02-04T10:45:46 Have we settled with moving pixels that have an influence on every square around them already? 2011-02-04T10:46:05 Frontier: ?? 2011-02-04T10:46:26 Or is there still a chance the ants have a 'direction' they look? 2011-02-04T10:46:35 i don't like their current visualiser 2011-02-04T10:46:43 i'm hoping you'll save the day :P 2011-02-04T10:47:02 haha, out of question mr businessman :p 2011-02-04T10:47:15 ergh, i'm not a businessman 2011-02-04T10:47:24 i would make a terrible businessman 2011-02-04T10:47:37 social economics then? 2011-02-04T10:47:46 bleh :P 2011-02-04T10:47:58 i just have fun learning micro theory and game theory 2011-02-04T10:48:20 micro is pretty much just game theory these days 2011-02-04T10:49:10 @Frontier I'd say yes (no more direction), all the new versions seem to have that in common 2011-02-04T10:49:11 Meatkat: User error -- Replace user. 2011-02-04T10:49:39 @rainbow shutup contesbot 2011-02-04T10:49:40 antimatroid: shutup contesbot 2011-02-04T10:51:47 *** aerique has quit IRC (Quit: ...) 2011-02-04T10:52:12 :-( we'll have to live with that then. My visualizer just looks like that because we worked closely together when designing the game engine and visualizer 2011-02-04T10:52:35 i still prefer that game 2011-02-04T10:52:40 it had a lot to it 2011-02-04T10:53:42 harvest could even go straight to the bases and there could be more players, each player gets a point for every hp they remove of another player, stepping on an enemy base kills every one of their ants and you get all the hp from their remaining ants 2011-02-04T10:53:42 Yeah, it gave the ants some life and decisions to make. 2011-02-04T10:53:45 player with the most points wins 2011-02-04T10:54:07 Well you could always write a starter bot and get someone to throw up a TCP server. If it took off the admins would take notice 2011-02-04T10:54:29 Meatkat: there was a working engine, visualiser and bot, people didn't seem that keen :\ 2011-02-04T10:55:12 *working* engine, it was fairly inefficient 2011-02-04T10:56:29 Well you could always write a starter bot and get someone to throw up a TCP server. If it took off the admins would take notice. 2011-02-04T10:56:39 basically, your ants spawned on your home base, you had to go out and harvest grains to spawn new ants (depositing it back home, but that was optinal), each ant started with 5 hp, you could attack adjacent ants and ants landing on the same square were fused/fought 2011-02-04T10:56:58 remember how the ants used to dance in circles *sigh* the good old times 2011-02-04T10:57:17 Frontier: how hard would it be for you to run the old games on that page again? 2011-02-04T10:57:26 i'd be keen for jeff to at least see the game 2011-02-04T10:57:29 i don't think he ever did 2011-02-04T10:58:06 The site is on my laptop. Right now it is online 2011-02-04T10:58:20 oh easy, can you remind me of the link? :P 2011-02-04T10:58:31 http://marco-leise.homedns.org/antgui/v3/ 2011-02-04T10:58:33 j3camero: are you about? 2011-02-04T11:00:14 j3camero: at some point, go to http://marco-leise.homedns.org/antgui/v3/ and watch some of those games, the objective is to kill off their ants and you win if you get an ant onto their base, that could also work for 4+ players? 2011-02-04T11:00:43 Meatkat: what do you think? :P 2011-02-04T11:01:58 It hasn't loaded yet 2011-02-04T11:02:13 it shouldn't take very long 2011-02-04T11:02:22 I'm in Africa 2011-02-04T11:04:03 i'm in tasmania :P 2011-02-04T11:04:28 It is only 140 kb, come on :p 2011-02-04T11:05:03 It should have loaded by now, Chromium keeps telling me that the page is unresponsive and wants to kill it 2011-02-04T11:05:17 Any known problems with Linux/Chromium? 2011-02-04T11:05:18 loads for me in chrome :\ 2011-02-04T11:05:22 I hope you didn't press any buttons while the page was loading 2011-02-04T11:05:23 i'm on xp though 2011-02-04T11:05:32 refresh? 2011-02-04T11:05:36 Fronteir: nope 2011-02-04T11:05:49 Yeah, a couple times 2011-02-04T11:06:35 There is a bug I never fixed. The page locks up when you load a new game while another is still loading. 2011-02-04T11:07:18 I should deactivate the buttons then. 2011-02-04T11:07:48 Loading it on Firefox, I'll see if that works 2011-02-04T11:09:00 Weeee!!! it loaded 2011-02-04T11:09:31 *** amstan has joined #aichallenge 2011-02-04T11:09:31 *** ChanServ sets mode: +o amstan 2011-02-04T11:09:33 Very nicely done sir 2011-02-04T11:10:00 Frontier made it look nice 2011-02-04T11:10:36 Indeed he did 2011-02-04T11:12:25 I think that deserves a new thread on the forums. If you had the motivation I'd say you should write a starter bot (or just modify the current to be more readable) and post that along with it 2011-02-04T11:12:41 And if you're REALLY motivated, get a TCP server running 2011-02-04T11:12:52 I check the forums often and never saw that, did you post the link? 2011-02-04T11:12:55 it's already been posted in the old ants thread 2011-02-04T11:13:04 i'll wait to see what jeff thinks when he does actually see it 2011-02-04T11:13:34 i think they want 4 player, which could be done, but i'd rather not bother unless it was a viable option 2011-02-04T11:14:07 there's also too many ants being spawned 2011-02-04T11:14:15 but that's easyish to fix 2011-02-04T11:17:06 combat is far less trivial than with simplified ants 2011-02-04T11:17:20 antimatroid: which one? 2011-02-04T11:17:34 http://marco-leise.homedns.org/antgui/v3/ 2011-02-04T11:17:54 way too slow and too many things on screen 2011-02-04T11:19:42 amstan: slowness can be fixed by changing game parameters/rules 2011-02-04T11:19:54 what parameters? 2011-02-04T11:20:04 ie. harvest goes straight to base without being "returned" and not as many resources so not a ridiculous number of ants 2011-02-04T11:20:50 that also removes "harvest carrying" for each ant 2011-02-04T11:21:08 then you're left with ant hp, which could also be removed, just have an attack kill the attacked ant 2011-02-04T11:21:26 that's pretty much simplified ants then 2011-02-04T11:21:35 except new ants come from the base 2011-02-04T11:21:43 and battles are solved by who attacks first 2011-02-04T11:21:54 yeah, and you still have "invade the base" objective 2011-02-04T11:22:07 i prefer having hp because then attacks are more complicated 2011-02-04T11:22:16 as you want to attack them from more directions at once 2011-02-04T11:23:13 you could also score games by how many hp of enemies they take from attacks, if you land on their base, you get the hp from all their live ants and that player is wiped from the game 2011-02-04T11:23:21 there are many variations 2011-02-04T11:24:15 it's fairly easy to write a semi intelligible stupid starter bot that people can make a few simple improvements too and see a decent improvement in their botst 2011-02-04T11:24:37 yet allows plenty of room for strategy variation near the top etc., I think it's a fairly ideal game, but i have always been biased on that 2011-02-04T11:25:31 the time is pi 2011-02-04T11:36:41 *** boegel has quit IRC (Read error: Operation timed out) 2011-02-04T12:06:59 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-02-04T12:09:52 murr. 2011-02-04T12:10:01 my desktop is busted for the weekend so suddenly i have to be productive 2011-02-04T12:35:05 *** bhasker has quit IRC (Quit: bhasker) 2011-02-04T12:40:12 antimatroid: I'm here 2011-02-04T12:40:41 McLeopold: i'll be off soonish... 2011-02-04T12:40:57 i wanted to know what your io is for ants 2011-02-04T12:41:07 and your thoughts on maps 2011-02-04T12:43:49 i'm not really comfortable with players > 4 playing on a map that doesn't have wrapped edges 2011-02-04T12:44:14 if each player gets a single ring on a torus, then maps can be generated for any n 2011-02-04T12:44:42 an alternative is to split each ring up into x sections for x*d players 2011-02-04T12:45:25 then i'm also not really sure of how symmetry to make everything, one option is to fill in 4 squares for every players starting region 2011-02-04T12:45:51 symmetric** 2011-02-04T12:47:14 sry, catching up on last night 2011-02-04T12:47:22 all good 2011-02-04T12:47:32 i'm going to go to bed in a bit, will be up again later 2011-02-04T12:47:59 okay, I thought about io last night while trying to sleep 2011-02-04T12:48:10 I think I want to etherpad this 2011-02-04T12:48:42 http://ietherpad.com/TR0zbEqvx1 2011-02-04T12:49:57 antimatroid: ^^ 2011-02-04T12:50:08 yeah, i'm watching 2011-02-04T12:50:15 oh, typein your name 2011-02-04T12:52:47 what's the deal with .png maps? 2011-02-04T12:52:56 haha 2011-02-04T13:00:08 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2011-02-04T13:05:24 *** delt0r___ has joined #aichallenge 2011-02-04T13:06:49 *** delt0r_ has quit IRC (Ping timeout: 255 seconds) 2011-02-04T13:58:55 *** Blkt has quit IRC (Ping timeout: 246 seconds) 2011-02-04T13:59:39 *** bhasker has joined #aichallenge 2011-02-04T14:00:43 *** dlila has joined #aichallenge 2011-02-04T14:03:56 *** Blkt has joined #aichallenge 2011-02-04T14:04:53 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2011-02-04T14:10:45 *** bhasker has joined #aichallenge 2011-02-04T14:24:53 *** Accoun has quit IRC () 2011-02-04T14:34:08 *** Meatkat has quit IRC (Ping timeout: 255 seconds) 2011-02-04T14:37:51 *** Accoun has joined #aichallenge 2011-02-04T14:53:46 *** boegel has joined #aichallenge 2011-02-04T15:07:32 *** Naktibalda has joined #aichallenge 2011-02-04T15:13:30 *** boegel has quit IRC (Quit: Leaving) 2011-02-04T15:31:18 *** Blkt` has joined #aichallenge 2011-02-04T15:32:30 *** Blkt has quit IRC (Ping timeout: 240 seconds) 2011-02-04T15:56:12 *** ryan_m has joined #aichallenge 2011-02-04T16:04:20 *** ryan_m has quit IRC (Quit: Leaving) 2011-02-04T16:22:34 jmcarthur: I'm thinking about move resolution 2011-02-04T16:22:44 I'd like your perspective 2011-02-04T16:23:18 or antimatroid 2011-02-04T16:23:26 heck, even mega1 2011-02-04T16:23:32 like you have an idea or you want suggestions? 2011-02-04T16:23:39 both 2011-02-04T16:23:54 let's hear yours. i need to get my mind warmed up before i can contribute anything 2011-02-04T16:23:56 so, to resolve all moves, i'm going to make a copy of the current map without any ants 2011-02-04T16:24:06 then all moves create ants on the new map 2011-02-04T16:24:21 this will allow you to bump your own forces 2011-02-04T16:24:29 the bot still controls the order of the moves 2011-02-04T16:25:12 wait, no 2011-02-04T16:25:13 arg 2011-02-04T16:25:22 okay, so you mean it's order dependent? 2011-02-04T16:25:29 so, if 2 colors go for the same spot, they kill each other 2011-02-04T16:25:31 which isn't necessarily a bad thing 2011-02-04T16:25:38 mkay 2011-02-04T16:25:49 if 2 red and 1 yellow go for the same spot, what happens? 2011-02-04T16:25:52 1 red? 2011-02-04T16:26:11 if I resolve red first, one red bumps, then the other gets killed by yellow 2011-02-04T16:26:26 we could really throw things off by making the resolution random :P 2011-02-04T16:26:29 reversed, the first red kills yellow, then the second one succeeds? 2011-02-04T16:26:53 or, I could mark the spot as having dead ants, then bump the second red. 2011-02-04T16:27:00 then it is always deterministic 2011-02-04T16:27:20 modulo ordering of commands 2011-02-04T16:27:43 a single colors ordering matter for it 2011-02-04T16:28:00 but I want to make sure the engines ordering of players doesn't matter 2011-02-04T16:28:18 (why is "modulo" even appropriate for things like that? i never understood its colloquial usage) 2011-02-04T16:28:20 I think blocking on attacked spaces will work 2011-02-04T16:28:25 i think so too 2011-02-04T16:28:30 okay 2011-02-04T16:28:31 and makes sense 2011-02-04T16:28:33 *** amstan has joined #aichallenge 2011-02-04T16:28:33 *** ChanServ sets mode: +o amstan 2011-02-04T16:28:42 this also allows ants to swap places 2011-02-04T16:28:43 however... 2011-02-04T16:29:01 ... 2011-02-04T16:29:49 I hope I'm waiting for a really good thought... 2011-02-04T16:30:25 what about recursive situations? like one of the red ants is actually trying to move somewhere but gets blocked, then is killed by an incoming yellow, making a body, then that blocks another red? but what if the thing blocking the original red is only due to yet another move? i think this could cause cycles, is my point 2011-02-04T16:30:43 which makes it complicated to implement 2011-02-04T16:31:33 might be nicer to just kill all ants that try to move to the same spot, regardless of number or color, and they just create a single body 2011-02-04T16:31:36 each bot would have their ant moves done separatly 2011-02-04T16:32:02 so, resolve all red moves 2011-02-04T16:32:07 then resolve all yellow moves 2011-02-04T16:32:34 hmm... with 3 players, that does get messy 2011-02-04T16:32:43 resolve all red moves, resolve all yellow moves (creates some bodies!), resolve all blue moves (wtf!) 2011-02-04T16:32:47 player 1 and 2 go for the spot, die, then player 3 might block 2011-02-04T16:32:51 blue gets screwed then 2011-02-04T16:33:02 or maybe red and yellow do 2011-02-04T16:33:04 either way, unfair 2011-02-04T16:33:10 yeah, okay, all ants die 2011-02-04T16:33:26 only fair way that doesn't involve too much code 2011-02-04T16:33:53 the way i would do it is: for each square, gather all incoming ants. if it's just one ant, that's the new occupant of the square. if it's more than one, the new occupant is a body. 2011-02-04T16:33:57 well, I could have a separate map for each color, and include the blocking, then overlay them all 2011-02-04T16:34:49 I'd like to try and get a single color to block itself, just to make it easier for the programmers not to accidentally lose ants 2011-02-04T16:35:50 two points about that: 1) yes, it's perhaps a bit nicer behavior for when you make an AI mistake, but 2) it's very tough to code 2011-02-04T16:35:50 so, if I mark a spot as a conflict, only 1 ant of each color can die 2011-02-04T16:36:32 either way, I need to mark spots as having conflicts 2011-02-04T16:36:51 the blocking way, I also need to mark which colors have attempted to enter the spot, not too much more 2011-02-04T16:36:57 actually, i was thinking that if you had blocking for same-colored ants then *both* of them should be blocked 2011-02-04T16:37:06 that would make resolution easy again 2011-02-04T16:37:22 well 2011-02-04T16:37:25 this is turning into diplomacy 2011-02-04T16:37:31 forget i said that :) 2011-02-04T16:37:34 it's not actually that easy 2011-02-04T16:37:42 i'm just not so sure about blocking in the first place 2011-02-04T16:37:52 only own color 2011-02-04T16:38:08 i personally think it makes sense that if you make a dire error then suffering a penalty for it is fine 2011-02-04T16:38:10 and it is deterministic based on the bot's own move ordering 2011-02-04T16:39:13 Okay, but think about a bot where each ant is using it's own neural network. It's simpler for the coder if you don't have to worry about killing your own ants. 2011-02-04T16:39:27 the coder could just stick in a failsafe 2011-02-04T16:40:08 plus, I'm devising a scoring method where you get a point for each ant of yours that dies, so I don't want you to be able to kill your own ants without a fight with an opponent 2011-02-04T16:40:34 that seems to be a counterintuitive scoring method... ^_^ 2011-02-04T16:40:53 it rewards aggresiveness, but not greed 2011-02-04T16:41:19 to prevent cherry pickers from getting a good score 2011-02-04T16:41:25 so you just want your opponent to kill as many of your ants as possible? 2011-02-04T16:41:38 without killing theirs? 2011-02-04T16:41:40 yes, but you also have to harvest them 2011-02-04T16:41:57 and if you are wiped out, you loose the chance for more score 2011-02-04T16:42:28 maybe it should just be kills, but then that gets tricky to assign to players 2011-02-04T16:42:35 well, sounds tough, but i don't know whether i like the implication it has one the game... that the only way you should be able to lose ants is in combat 2011-02-04T16:42:41 *on the game 2011-02-04T16:42:59 *** Naktibalda has quit IRC (Remote host closed the connection) 2011-02-04T16:43:17 how else? 2011-02-04T16:43:26 order mistakes? 2011-02-04T16:43:28 illegal moves, walk into water, etc. 2011-02-04T16:43:40 Oh, I've changed from water to wall. 2011-02-04T16:43:42 *** rebelxt1 has left #aichallenge 2011-02-04T16:43:52 oh, so it's just land and wall now? 2011-02-04T16:43:58 I'm kinda against the map changing during the game. 2011-02-04T16:44:02 yeah 2011-02-04T16:44:18 the walls are movable or no? 2011-02-04T16:44:21 no 2011-02-04T16:44:26 but bodies are? 2011-02-04T16:44:30 no bodies 2011-02-04T16:44:33 oh 2011-02-04T16:44:38 it's simpler 2011-02-04T16:44:50 let's move to the etherpad 2011-02-04T16:44:55 that was one of the things i was liking though :\ 2011-02-04T16:46:38 *** Blkt` has quit IRC (Ping timeout: 276 seconds) 2011-02-04T16:58:37 *** amstan has quit IRC (Remote host closed the connection) 2011-02-04T17:18:33 *** amstan has joined #aichallenge 2011-02-04T17:18:33 *** ChanServ sets mode: +o amstan 2011-02-04T17:32:49 *** Mathnerd314 has joined #aichallenge 2011-02-04T17:34:55 *** mega1 has quit IRC (Ping timeout: 240 seconds) 2011-02-04T18:09:12 *** chris__0076 has joined #aichallenge 2011-02-04T18:12:07 *** Chris_0076 has quit IRC (Ping timeout: 240 seconds) 2011-02-04T18:22:51 *** dlila has quit IRC (Quit: Leaving) 2011-02-04T18:30:33 *** j3camero has quit IRC (Quit: leaving) 2011-02-04T18:35:16 *** Palmik_ has quit IRC (Remote host closed the connection) 2011-02-04T18:36:20 McLeopold: I'm up again 2011-02-04T18:42:54 *** holmar has joined #aichallenge 2011-02-04T18:47:09 *** holmar has quit IRC (Client Quit) 2011-02-04T18:48:20 hello 2011-02-04T18:48:22 antimatroid 2011-02-04T18:48:27 i'm here 2011-02-04T18:48:34 what's up 2011-02-04T18:48:39 i added a tiny bit to your mapline info sent to the bot 2011-02-04T18:48:57 ok 2011-02-04T18:49:15 with your resolving ants moving into the same square, could you not have an array numbering the number of kids going in for each player then argmax that? taking no argmax if it's a tie? 2011-02-04T18:49:51 so, the most win? 2011-02-04T18:50:20 yes 2011-02-04T18:50:44 the idea being that one ant from each time participates in each round of fighting 2011-02-04T18:50:53 leaving the player with the most ants with any left 2011-02-04T18:51:01 and all friend ants left after fighting fuse into 1 ant 2011-02-04T18:51:47 sure, I can make that an option 2011-02-04T18:51:56 i have an idea for maps to run past you... 2011-02-04T18:52:00 jmcarthur and I were thinking the all die 2011-02-04T18:52:10 ok 2011-02-04T18:53:51 so, what's the idea for the maps? 2011-02-04T18:54:17 each player (say n players) needs a region in a rectangular grid with edges wrapped, so if I take the largest factor f of n > 1 and preferably < n but not strictly, then that gives the number of rings on the torus, then n/f players partition each ring? 2011-02-04T18:54:54 what if n is prime? 2011-02-04T18:55:03 each player gets a single ring on the torus 2011-02-04T18:55:16 hence the but not strictly 2011-02-04T18:55:24 technically, a ring and a slice are the same thing on a torus 2011-02-04T18:55:26 in that case f is n 2011-02-04T18:55:41 yeah, i think you know what I mean though yeah? 2011-02-04T18:55:52 that way a map can be generated for anything up to 26 players 2011-02-04T18:56:04 if you go back to the rectangle, assuming edges wrap, each player gets its own rectangle of the same size, arranged in any grid fasion 2011-02-04T18:56:19 also, is the engine currently passing information to players and having all their turns run at once? I think it's important that becomes sequential, particularly with >2 players 2011-02-04T18:56:35 so 24 players can be 1x24 or 2x12 or 3x8 or 4x6 2011-02-04T18:57:08 yeah 2011-02-04T18:57:16 yes, everything is simultaneous 2011-02-04T18:57:55 i think players should make their moves with computation sequentially (but all passed information from the same state) then process all moves simultaneously 2011-02-04T18:58:31 we also need to force the other processes to not get cpu time during other players turns 2011-02-04T18:59:26 a better option is to time bots on cpu time, but it's seemed rather hard to coordinate with all the different languages 2011-02-04T18:59:52 I don't understand the computation sequentially comment 2011-02-04T19:00:17 And, I'm not planning to implement any bot timing or anything. That's up to Zannick to do. 2011-02-04T19:00:37 instead of : pass players the information and give them all 1 second to return their moves 2011-02-04T19:00:38 pass player 1 information, get their moves, pass player 2 information, get their moves, ..., process player 1,... , n's moves 2011-02-04T19:00:54 oh, ok 2011-02-04T19:01:04 that's for zannick to fix as well 2011-02-04T19:01:12 otherwise atm each player gets at worst 1/n seconds per move 2011-02-04T19:01:21 - overhead 2011-02-04T19:01:26 He'll probably agree with you. 2011-02-04T19:01:35 that could double game time, but it sounds fairly reasonable besides 2011-02-04T19:01:43 We could probably even lessen it to 0.5 seconds if we did that. 2011-02-04T19:01:54 it may more than double the game time, but i think it's necessary :) 2011-02-04T19:02:03 if we go multiplayer, even more 2011-02-04T19:02:12 Right now, ants will probably have longer run times than planet wars. 2011-02-04T19:02:19 but it'd be bad to run them all at the same time anyway 2011-02-04T19:02:24 exactly, i'd rather get half the time under that condition rather than the actual time be fairly dynamic 2011-02-04T19:02:43 smaller maps will make for faster games 2011-02-04T19:02:54 i was thinking each players starting region might be static? 2011-02-04T19:03:05 like 15x15 or something? 2011-02-04T19:03:06 yeah, but I don't want to limit what we see at first like we did with planetwars 2011-02-04T19:03:32 I want all the crazy maps ideas to come out first, then the community can choose what is working best. 2011-02-04T19:03:44 Then we avoid the map issue from last time, somewhat. 2011-02-04T19:03:58 what if i take all non-paired unique factors of n then pick randomly from that to make up the ring density and no. rings on the torus? 2011-02-04T19:04:21 McLeopold: that's why you haven't already got a map generator from me, I'm trying to think it through first 2011-02-04T19:04:34 this method works for n in [2, 26] 2011-02-04T19:04:39 this is like when you where doing polar coordinates and I was doing cartesian for planetwars :) 2011-02-04T19:04:44 so players in a game could also be dynamic 2011-02-04T19:05:33 i sat down and worked the problem out in polar, it was just easier for me to program it like that :P 2011-02-04T19:06:29 what's a non-paired unique factor? 2011-02-04T19:06:42 say you have 6, then 1, 2 and 3 2011-02-04T19:06:50 wait, 1 and 2 2011-02-04T19:06:59 okay, I know what factors are 2011-02-04T19:07:01 6 is paired with 1 and 3 paired with 2 2011-02-04T19:07:16 there is probably a better name, but i don't know it off the top of my head :) 2011-02-04T19:07:19 because their product is n? 2011-02-04T19:07:26 yep 2011-02-04T19:07:37 and i can get it from 6/first factor 2011-02-04T19:08:02 so i take the set 1,2 and pick one of those, then the regions are set up like first factor x second factor 2011-02-04T19:08:19 then with edges wrapped, i can make it symmetric except for relative positions to other players 2011-02-04T19:08:28 so, with an area of a (or n players) how many unique rectangles can you make? 2011-02-04T19:08:45 area of a? 2011-02-04T19:08:49 6 2011-02-04T19:08:59 unique rectangles? 2011-02-04T19:09:10 1x6 and 2x3 2011-02-04T19:09:28 is 2x3 each players starting region? 2011-02-04T19:09:33 yes 2011-02-04T19:09:38 oh no 2011-02-04T19:09:45 2x3 grid of starting regions 2011-02-04T19:10:17 say you split the map for 6 players up into regions of 2x3 2011-02-04T19:10:32 I think we are saying the same thing. 2011-02-04T19:10:36 each region might be say 15x15, so you have a final grid of 30x45 2011-02-04T19:10:43 I was just pointing out our different ways of thinking 2011-02-04T19:10:49 yes 2011-02-04T19:11:04 then i pick a starting location for the first region, and place that symmetrically in each other region for other players starting region 2011-02-04T19:11:18 and, favor the width being longer, since monitors are usually longer width 2011-02-04T19:11:49 i had an idea, you could have "ant hill creation" which occurs if a single player has 8 ants surrounding a square? 2011-02-04T19:12:14 but may complicate things too much 2011-02-04T19:12:30 *** mceier has quit IRC (Quit: leaving) 2011-02-04T19:12:36 favouring it being longer just has regions split by nx1 2011-02-04T19:12:48 1xn** 2011-02-04T19:13:09 n x m where n <= m 2011-02-04T19:13:19 okay sure 2011-02-04T19:13:31 but that's easy, just start at 1 and calc the factors 2011-02-04T19:13:34 i'll just take the set of smaller factors and pick n from that 2011-02-04T19:13:39 *** amstan has quit IRC (Read error: Operation timed out) 2011-02-04T19:14:16 I've got an idea for a map generator. Let me look for the link... 2011-02-04T19:14:16 for the love of god, nobody should write any code indexing stuff [x][y] rather than [row][col] 2011-02-04T19:14:23 it's aweful, and not just because it's what i'm used to 2011-02-04T19:14:34 hmm... 2011-02-04T19:14:40 eg. map files in .txt form 2011-02-04T19:14:40 maybe in the next version... 2011-02-04T19:14:51 indexed by [x][y] you need to rotate it to print 2011-02-04T19:15:48 the next issue is how symmetric to make each region? 2011-02-04T19:16:10 http://weblog.jamisbuck.org/2011/1/12/maze-generation-recursive-division-algorithm 2011-02-04T19:16:12 should each player have squares placed symmetrically for accessing adjacent regions? 2011-02-04T19:16:30 let's not limit ourselves at first 2011-02-04T19:16:46 Let's try and stretch the limit of maps for the start of the contest. 2011-02-04T19:17:00 i plan to 2011-02-04T19:17:06 i have wall placement worked out 2011-02-04T19:17:48 there will be min/max density perameters to play around with 2011-02-04T19:18:18 atm i think i need to place 4 squares per region for every wall/resource square picked out 2011-02-04T19:18:39 it can still be symmetric placing less, but "less symmetric" 2011-02-04T19:19:05 i'm doing a physics unit this semester entirely devoted to symmetry stuff with lie groups :D 2011-02-04T19:19:22 it's a post grad unit and i've done no physics, i'm banking on it mostly being maths 2011-02-04T19:20:35 over my head 2011-02-04T19:21:34 oh, what are your thoughts on vision radius? 2011-02-04T19:21:49 my old ants game had that set at 6 and i ensures players home bases were outside others field of vision 2011-02-04T19:22:10 also, it might be easy to just assume where players start is also an ant hill/ home base? 2011-02-04T19:22:37 *** iFire` has quit IRC (Read error: Connection reset by peer) 2011-02-04T19:23:03 I hadn't put fog into the current version. Players get the full map. 2011-02-04T19:23:22 i'm going to leave the feature in there for the moment 2011-02-04T19:23:25 fog is cooler imo 2011-02-04T19:23:42 it's fairly easy to paint out which squares each player can see for passing information 2011-02-04T19:23:49 I think it makes for a game of less skill. 2011-02-04T19:24:03 hmmm, i'd disagree 2011-02-04T19:24:09 *** iFire has joined #aichallenge 2011-02-04T19:24:35 you both need to scout out the map and maximise your visibility while ensuring you have friends close enough to create good offensive structures when attacking 2011-02-04T19:25:13 it also means you need to maintain visibility with all paths to your base 2011-02-04T19:25:23 so you can see incoming intruders and block them before they invade 2011-02-04T19:26:00 I'm not saying it involves less tactics, just that you have less control over your survival. 2011-02-04T19:26:10 More randomness 2011-02-04T19:26:26 you could also potentially not even tell people how many players there were then 2011-02-04T19:27:02 okay, back to code... 2011-02-04T19:27:23 i don't think i can hide the opponents home bases very well 2011-02-04T19:27:31 i think people will have a pretty decent idea of where they all are 2011-02-04T19:27:38 ant hills* 2011-02-04T19:27:41 i am trying to convert i swear 2011-02-04T19:28:30 we need to work out how to pass game strings to the visualiser 2011-02-04T19:28:39 particularly, for Frontier's visualiser 2011-02-04T20:34:17 oh i'm drunk again. what is it? 24 players in one match? no way 2011-02-04T20:34:36 24 is pretty excessive 2011-02-04T20:34:49 i have it picking randomly from 4-16 atm 2011-02-04T20:34:55 but that can easily be tweaked 2011-02-04T20:35:00 there aren't 24 distinct colors even to visualize them 2011-02-04T20:35:20 *** amstan has joined #aichallenge 2011-02-04T20:35:20 *** ChanServ sets mode: +o amstan 2011-02-04T20:37:04 don't try to make the game fit the visualizer. the idea must be fun, the graphics come secondary 2011-02-04T20:37:23 what do you think would be a realistic max dimensions? 2011-02-04T20:37:36 with smaller ants and no hp/harvest 2011-02-04T20:37:38 i'm worried more about implementation and efficiency 2011-02-04T20:37:53 atm i think the maps are too large for short games anyway 2011-02-04T20:38:56 maybe 16x16 would be a gfx size that still looks ok. now take your average laptop screen of 1280x800 and cut off some space for the browser 2011-02-04T20:39:35 16x16? 2011-02-04T20:39:46 i was thinking maybe 50xsomething? 2011-02-04T20:39:48 the next step would be just points moving around 2011-02-04T20:40:05 16x16 pixels per square that is 2011-02-04T20:41:03 ah okay 2011-02-04T20:43:45 amstan is right, the gameplay on the visualizer is slow. i noticed that today. 2011-02-04T20:44:21 Frontier: yeah.. like.. i wanna see a game end within one minute 2011-02-04T20:44:52 I *could* of course change the timers -.- 2011-02-04T20:46:17 there is less to see so i was kind of expecting that to happen :) 2011-02-04T20:46:20 I think two speeds can do the trick. One speed, just to get the idea and have the animation end in a few seconds and another speed to see all the details. 2011-02-04T20:46:51 Frontier: being able to progress through with arrow keys is sufficient for watching slowly imo, just speed it up :) 2011-02-04T20:47:34 i don't think this engine has ant ids atm either 2011-02-04T20:48:14 done deal - or whatever you native speakers say ^^ 2011-02-04T20:49:43 :) 2011-02-04T21:02:47 no, my version doesn't have ant id's 2011-02-04T21:03:25 and, you would have to infer which ants died unless you went through the resolution yourself from turn to turn 2011-02-04T21:04:02 i added it to mine just for the visualiser 2011-02-04T21:04:03 Zannick: I'm working on the efficiency of a 100x100 map. I think I can get the engine pretty fast. 2011-02-04T21:04:09 i think it's worth it 2011-02-04T21:04:25 McLeopold: what are the bottlenecks? 2011-02-04T21:04:49 possibleNoPlayers = [4, 6, 8, 9, 10, 12, 14, 15, 16] 2011-02-04T21:04:52 i'm now using that 2011-02-04T21:04:54 the main bottlenecks are actually the players themselves and the number of players 2011-02-04T21:05:03 i'm not worried about the engine 2011-02-04T21:05:10 but it's good for that to be efficient, too :) 2011-02-04T21:05:27 give each player 0.5 seconds to start with i say 2011-02-04T21:05:30 *** delt0r_ has joined #aichallenge 2011-02-04T21:05:51 while each game may take longer, you also get n people getting a game at a time 2011-02-04T21:05:56 so it's really not all that different? 2011-02-04T21:06:33 *** delt0r___ has quit IRC (Ping timeout: 240 seconds) 2011-02-04T21:16:40 Zannick: with smaller map sizes, we can probably run more games per minute on the server, but I'd like to keep a large variable of map sizes to start with, and then we can see through trial an error where the sweet spot is. 2011-02-04T21:16:54 The sweet spot being the most fun. 2011-02-04T21:16:58 sounds good to me 2011-02-04T21:17:23 After that, then we can tune for games per minute or whatever gives players enough games to do improvements. 2011-02-04T21:17:55 so, I figure start with map sizes varying from 20x20 to 200x200 to start with 2011-02-04T21:18:18 and we can slowly narrow the constraints from there 2011-02-04T21:18:31 actually, 200x200 might be kinda big. 2011-02-04T21:18:58 if you figure a 1000 turn starting limit, that gets you 5 times around the map max on the longest game 2011-02-04T21:20:09 antimatroid: the bottleneck is choosing to index the ants by location, and having to loop through the list to find neighbors, or having a matrix and having to scan it to find ants. 2011-02-04T21:20:21 I'm trying a system that does both right now. 2011-02-04T21:24:03 200x200 doesn't work with frontiers visualiser 2011-02-04T21:24:15 and i think is larger than necessary anyway 2011-02-04T21:24:22 i'm still playing around with the mapgen stuff atm 2011-02-04T21:24:29 i'll see what maps look like once that is done 2011-02-04T21:25:00 can I change X to '#' for walls? 2011-02-04T21:28:47 # would be for comments 2011-02-04T21:30:02 that's what // is for :P 2011-02-04T21:30:21 i'll leave it as X 2011-02-04T21:32:21 i'm a fan of player numbers starting from 0 too 2011-02-04T21:32:50 they should start from 1 2011-02-04T21:32:58 0 is land 2011-02-04T21:34:19 i was using . for land 2011-02-04T21:34:54 I mean for in the engine, and starter bots. 2011-02-04T21:35:07 '.' for land in the map files. 2011-02-04T21:38:50 should maps be symmetric in all four directions? or just symmetric if everyone played exactly the same? 2011-02-04T21:39:24 just enough that each players starting grid is similar 2011-02-04T21:39:39 yet maybe not with the same orientation 2011-02-04T21:40:08 but say you had different opponents left and right of you 2011-02-04T21:40:17 should it be symmetric for each way you went 2011-02-04T21:40:23 or just if everyone went "right"? 2011-02-04T21:40:56 just if everyone when 1 directions, right for some, maybe left or up for others 2011-02-04T21:42:20 so, for a 1x4 grid, you could have <><> or <<<< but not <<>> 2011-02-04T21:43:01 for a 2 x 2 grid 2011-02-04T21:43:09 >v 2011-02-04T21:43:13 ^< 2011-02-04T21:43:22 or 2011-02-04T21:43:23 >< 2011-02-04T21:43:25 >< 2011-02-04T21:43:34 if that makes sense 2011-02-04T21:51:43 not at all 2011-02-04T21:51:51 i'm not doing 1xanything anymore 2011-02-04T21:54:18 *** Cyndre has joined #aichallenge 2011-02-04T22:16:20 *** chris__0076 is now known as chris_0076 2011-02-04T22:50:57 the sequential thing is insufficient. if i'm doing some sort of search i could keep running that between sending commands and receiving the next game state 2011-02-04T22:52:30 in fact, i did that in my tron bot 2011-02-04T22:56:21 *** elimisteve has joined #aichallenge 2011-02-04T22:56:55 I've read the roadmap. Any estimates on when the contest may begin? 2011-02-04T22:57:38 very soon, j3camero has committed to getting stuff going this weekend 2011-02-04T22:57:48 okay thanks 2011-02-04T22:58:01 that is, getting the game defined 2011-02-04T22:58:08 sounds great 2011-02-04T23:00:25 *** jmpespxoreax has quit IRC (Ping timeout: 260 seconds) 2011-02-04T23:31:25 jmcarthur: at the beginning of a turn, there should be no way that players have ant next to each other 2011-02-04T23:31:34 ? 2011-02-04T23:51:26 *** McLeopold has left #aichallenge 2011-02-04T23:51:54 *** McLeopold_ has joined #aichallenge 2011-02-04T23:52:31 *** McLeopold_ has left #aichallenge