2011-06-28T00:01:20 *** okay is now known as okayzed 2011-06-28T00:13:24 *** janzert has quit IRC (Ping timeout: 255 seconds) 2011-06-28T00:15:07 *** Zaphus has joined #aichallenge 2011-06-28T00:17:18 *** Zaphus has quit IRC (Client Quit) 2011-06-28T00:25:53 *** ltriant has quit IRC (Ping timeout: 250 seconds) 2011-06-28T00:27:25 *** filter has joined #aichallenge 2011-06-28T00:27:51 *** filter_ has quit IRC (Ping timeout: 244 seconds) 2011-06-28T00:28:21 *** janzert has joined #aichallenge 2011-06-28T00:34:22 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-06-28T00:38:03 *** ltriant has joined #aichallenge 2011-06-28T00:38:23 *** janzert has joined #aichallenge 2011-06-28T00:39:06 janzert: http://ants.zeroviz.us/EToM2/ :) 2011-06-28T00:49:51 looks good 2011-06-28T00:49:55 what about a mirror one? 2011-06-28T00:59:11 *** Kingpin13 has quit IRC (Quit: this used to be the life, but I don't need another one) 2011-06-28T00:59:20 *** McLeopold has joined #aichallenge 2011-06-28T01:11:57 amstan: how many players? 2011-06-28T01:12:02 McLeopold: why don't you need the directions anymore? 2011-06-28T01:12:06 4 2011-06-28T01:12:49 antimatroid: I can detect all symmetries automatically now 2011-06-28T01:13:42 you can make maps with more than 1 valid symmetry as well 2011-06-28T01:14:16 amstan: http://ants.zeroviz.us/LiFHn/ http://ants.zeroviz.us/K7xie/ http://ants.zeroviz.us/Iwb88/ 2011-06-28T01:14:22 one of each kind of rotational symmetry 2011-06-28T01:14:50 antimatroid: only the last one does rotational 2011-06-28T01:15:12 i include a diagonal flip/transpose as a rotation of a grid 2011-06-28T01:15:26 including a horizontal/vertical flip 2011-06-28T01:15:36 then my other maps i call tile/tessellation 2011-06-28T01:15:50 sorry, i'm terrible at picking names non-mathematicians will understand :P 2011-06-28T01:16:04 McLeopold: soo... when can the maps go up on the server? :P 2011-06-28T01:16:18 i'm just saying, you're only doing rotational on the last one, the first 2 are mirrors 2011-06-28T01:17:06 that's still a rotation :P 2011-06-28T01:17:22 if you consider the z axis, sure.. 2011-06-28T01:17:46 that's how people do symmetries of polygons etc :P 2011-06-28T01:18:30 *** okayzed is now known as okay 2011-06-28T01:19:26 antimatroid: make sure you include command line switches for all the variables of map generation 2011-06-28T01:19:55 McLeopold: how do i do that? :P 2011-06-28T01:20:03 import argparse 2011-06-28T01:20:11 i've set it up so there are constructer functions 2011-06-28T01:20:16 so it shouldn't be too hard 2011-06-28T01:20:27 antimatroid: see the worker script 2011-06-28T01:20:28 no, that works nicely with switches 2011-06-28T01:24:07 here's a bunch of maps i generated without throwing any out, do either of you have any complaints/suggestions as to the design of them...? 2011-06-28T01:24:13 http://pastebin.com/snhq0aAq 2011-06-28T02:37:01 *** okay is now known as okayzed 2011-06-28T02:42:30 McLeopold: how far do i go to checking the validity of parameters? 2011-06-28T02:42:57 like do i make sure rows/cols are positive on top of making sure say for tile maps the parameters will work? 2011-06-28T02:47:24 *** V3rG1L has joined #aichallenge 2011-06-28T02:52:02 *** V3rG1L has left #aichallenge 2011-06-28T02:52:39 antimatroid: The server will want a list of parameters and the ranges it is allowed to choose. Integers are best. 2011-06-28T02:53:03 *** gobang has joined #aichallenge 2011-06-28T02:53:18 the feasible parameters changes depending on your type of symmetry 2011-06-28T02:53:29 hmm... 2011-06-28T02:53:32 like you can only do 2,4,8 player maps for rotationally symmetric maps 2011-06-28T02:54:22 and if i'm doing a block map, the dimensions need to be divisible by like 2*block_size 2011-06-28T02:54:24 I would consider them different map generators then. 1 parameter will be 'type' 2011-06-28T02:54:33 the rest of the parameter list comes from that 2011-06-28T02:55:25 can't the engine just call the generator for a random map? 2011-06-28T02:55:33 then deal with the proportions of what types in the map generator 2011-06-28T02:55:36 i would have thought that'd be easier 2011-06-28T02:56:04 I want the server to control types generated 2011-06-28T02:56:21 And to calculate how spread out the parameters will be 2011-06-28T02:56:49 it's going to need to know all the requirements for what is needed for symmetries to work 2011-06-28T02:56:50 then different map generators don't need their own logic for spreading out the parameters. 2011-06-28T02:57:43 Is a list of parameter names and a numeric range for each not good enough? 2011-06-28T02:57:44 like if you ask for a rotationally symmetric map with a block size of 4 and 10 columns, it's not going to look nice 2011-06-28T02:58:25 what if i make the min and max levels of these things input parameters? 2011-06-28T02:58:41 Yes, that's what a range is 2011-06-28T02:58:42 i have those as fixed parameters atm 2011-06-28T02:58:52 i'll do that 2011-06-28T02:58:55 ok 2011-06-28T02:59:28 so, there should be a list of types, and each type has it's own list of parameters, and each parameter has an acceptable range of values 2011-06-28T02:59:45 yeah, i think i'm good now 2011-06-28T02:59:46 The "range" can be a list, maybe, like for rot sym, [2, 4, 8] 2011-06-28T03:00:24 so, tuples for actual ranges, lists for specific values 2011-06-28T03:00:39 I'll want a get_types function I can call 2011-06-28T03:01:42 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-06-28T03:04:24 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-28T03:04:39 antimatroid: return something like this {'rotational': {'players': [2,4,8], 'block_size': (2,5)}, 'transposition': {'players': (2,10), 'rows': (20,100)} } 2011-06-28T03:05:59 yep sure 2011-06-28T03:06:24 antimatroid: do you know what sigh did when 2 bots can get to a space at the same time, for food spawning? 2011-06-28T03:06:51 McLeopold: i can guess 2011-06-28T03:07:18 ok 2011-06-28T03:07:33 i find a basis, which can be larger than total_land/no_players, because of first access squares, but each square in the basis can't "get to another basis element" by going between the symmetric locations 2011-06-28T03:08:24 some of the basis elements will have less than n symmetric locations associated, just take the set that doesn't double elements 2011-06-28T03:08:27 then spawn food on that set 2011-06-28T03:08:30 does that make sense? 2011-06-28T03:08:45 uh, so don't spawn in the middle? 2011-06-28T03:08:51 no you can 2011-06-28T03:08:59 just spawn 1 food item there, rather than n items 2011-06-28T03:09:17 same as if you have 4 players, you can spawn on the diagonals, just don't spawn twice on the same location that turn 2011-06-28T03:09:20 so, a set can be smaller than the number of players 2011-06-28T03:09:24 yes 2011-06-28T03:09:34 the set of squares symmetrically equivalent to a basis element 2011-06-28T03:09:37 what if I just queued up 2 foods? 2011-06-28T03:09:48 and no two basis elements are in the same set of squares that are equivalent to each toehr 2011-06-28T03:10:10 McLeopold: you could do that, but if one bot gets there then they get an advantage? 2011-06-28T03:10:18 yeah, they woould 2011-06-28T03:10:18 and the food spawn info is going from starting locations... 2011-06-28T03:10:24 i would spawn just 1 2011-06-28T03:10:32 but, spawning less messes with the spawn rate 2011-06-28T03:10:37 yeah 2011-06-28T03:10:44 can you just queue up locations 2011-06-28T03:10:54 I have to make sure I have enough middle spawns to equal the number of players 2011-06-28T03:11:23 i would have a queue of symmetric sets to spawn on 2011-06-28T03:11:23 It's like a partial set, I have to have enough to make a full set 2011-06-28T03:11:46 and while the fronts size + spawned food <= amount_to_spawn, spawn from the front? 2011-06-28T03:12:14 not perfect, but still symmetric? 2011-06-28T03:12:24 the spawn rate wouldn't be perfectly constant, but close enough 2011-06-28T03:12:27 I'm saving left over food to spawn, I think I could just play with the rates 2011-06-28T03:12:53 *** gobang has joined #aichallenge 2011-06-28T03:13:18 if you have spawn_rate = 5 food per turn or whatever, then you have amount_to_spawn = turn*spawn_rate, and then keep amount_spawned and a queue of to spawn sets 2011-06-28T03:13:52 so, I have "total food to spawn", right now I'm doing a mod, but instead I could just pull the next set, which is a variable size, and only spawn if it is not more than leftover food. 2011-06-28T03:14:27 I think we are saying the same thing... 2011-06-28T03:15:15 okay, variable sized spawn sets it is, which will be very close to calculating mod, just slightly off if a middle set gets chosen. 2011-06-28T03:15:29 i don't think blockable regions exist with my maps either 2011-06-28T03:15:40 they'd be quite rare if they did 2011-06-28T03:15:53 they are looking nice 2011-06-28T03:16:10 when I finish the better food spawn code, we can put them on the server 2011-06-28T03:16:20 yep 2011-06-28T03:16:23 i'm doing a rewrite now 2011-06-28T03:16:25 cleaning it all up 2011-06-28T03:16:32 i'm going to scrap the random walk stuff 2011-06-28T03:16:40 I'm glad I did symmetry detection, much better than in the map 2011-06-28T03:16:51 this is way faster than the symmetric walk stuff too 2011-06-28T03:17:01 did you see the 1000x1000 map? :P 2011-06-28T03:17:04 Don't scrap random walk, it's still good 2011-06-28T03:17:06 no 2011-06-28T03:17:15 http://ants.zeroviz.us/vhf1b/ 2011-06-28T03:17:34 that took a few seconds 2011-06-28T03:17:48 whoa, it crashed chrome :p 2011-06-28T03:17:57 i have it loaded in chrome :P 2011-06-28T03:18:08 i can only see one player in normal view 2011-06-28T03:18:19 pink is near the bottom right if you zoom in, i don't know where the third player is 2011-06-28T03:19:01 I just found him :) 2011-06-28T03:19:10 Let's make that the final map :) 2011-06-28T03:19:12 it's off the left 2011-06-28T03:19:14 aha 2011-06-28T03:19:43 i'm looking forward to crazy maps tcp 2011-06-28T03:19:55 Hopefully we won't need it. 2011-06-28T03:19:59 single width mazes, ridiculously large maps etc. 2011-06-28T03:21:57 aichallenge: McLeopold epsilon * rd119816 / ants/ants.py : added map symmetry detection - http://bit.ly/mCxPUi 2011-06-28T03:21:57 aichallenge: McLeopold epsilon * r62c8f3f / (3 files): added symmetry test maps - http://bit.ly/ln7eHe 2011-06-28T03:21:58 aichallenge: McLeopold epsilon * r568a216 / (5 files in 5 dirs): Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://bit.ly/ixs2qA 2011-06-28T03:22:06 antimatroid: pull that update to test your map symmetries 2011-06-28T03:22:28 later :P 2011-06-28T03:22:29 type "ants.py mapfile.map" 2011-06-28T03:22:32 i'm rewriting now 2011-06-28T03:24:15 another problem is it stalling 2011-06-28T03:24:27 :( 2011-06-28T03:24:31 if min_starting_distance is too high, it can't pick starting locations for players 2011-06-28T03:24:35 like an infinite loop? 2011-06-28T03:24:36 it's fine if the parameters are alright 2011-06-28T03:24:48 but if you mess up the parameters, it's just going to loop trying to find starting locations 2011-06-28T03:24:51 yeah 2011-06-28T03:25:08 give it a max tries and raise an error if reached 2011-06-28T03:25:11 i can make it quit out probably, but you wont get a map back 2011-06-28T03:25:23 I'll write the map generator stuff to retry on failure 2011-06-28T03:25:45 As long as you give me parameters that will usually produce a map, it's okay 2011-06-28T03:26:03 :P 2011-06-28T03:26:03 I had to tweak min_distance a lot on my cell mazes 2011-06-28T03:26:03 how do you want errors done? 2011-06-28T03:26:08 is there a file i can look at? :P 2011-06-28T03:26:41 no output would be fine 2011-06-28T03:26:58 I'll just run the map through the symmetric detection to look for failure. 2011-06-28T03:27:08 easy 2011-06-28T03:27:41 or, you could restart the map generation yourself, either way 2011-06-28T03:28:13 can i assume [min,max] players will always include 2, 4 or 8 in its range? 2011-06-28T03:28:22 no 2011-06-28T03:28:28 you give me a range, or a list 2011-06-28T03:28:54 actually, nevermind 2011-06-28T03:28:55 so, you will *know* what values I may select 2011-06-28T03:30:08 @later tell amstan ai-contest.com is down again 2011-06-28T03:30:08 McLeopold: Ready to serve. 2011-06-28T03:30:45 weird, now it's back 2011-06-28T03:31:19 *** aerique has joined #aichallenge 2011-06-28T03:43:40 McLeopold: i can't take the type or rotational symmetry from you without the exact number of players.. 2011-06-28T03:43:44 of* 2011-06-28T03:50:28 make a matrix of acceptable type and player_count combos 2011-06-28T03:52:57 there is 1-5 for 2 players, 1-3 for 4 players and 1 for 8 players 2011-06-28T03:55:19 McLeopold: how about if no_players is fixed, but you give me a range for dimensions? 2011-06-28T04:00:51 antimatroid: what ever you think is best, I just want a list of parameters and acceptable values 2011-06-28T04:03:21 *** amstan has quit IRC (Ping timeout: 255 seconds) 2011-06-28T04:08:58 *** stocha has joined #aichallenge 2011-06-28T04:11:49 *** Palmik has joined #aichallenge 2011-06-28T04:11:52 @later tell antimatroid: yes i have 6. Three are anchors (StoSnake it's a snake bleh, Sto_greed_002 - the one i gave on the forum, sto_random - weak reference bot). The other three, i use to see how different strategy ranks. 2011-06-28T04:11:52 stocha: I come to serve. 2011-06-28T04:13:53 @later tell antimatroid: so i really have three account (with only one, there would be stosnake by the way). Why three ? Because ranking takes ages to reach. Sto003 may become a sort of anchor also. (it's very different from my other bots, but i don't want to work any more on this style of bot) 2011-06-28T04:13:53 stocha: I think that worked... 2011-06-28T04:15:21 @later tell antimatroid: what would be could would to be able to use the beta to actually really test my bots. But because of the slow delay before having a trustworthy rank, i would need probably a few dozen account to do that (not that i can't test at home, because i have no free processor time available) 2011-06-28T04:15:21 stocha: Yes master! 2011-06-28T04:29:45 nice top level game this one : http://aichallengebeta.hypertriangle.com/visualizer.php?game=42561&user=7 2011-06-28T04:30:00 at least i have the impression i understand what's going on a bit :) 2011-06-28T04:31:13 *** smippy is now known as smellyhippy 2011-06-28T04:36:13 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-28T04:36:34 *** gobang has joined #aichallenge 2011-06-28T04:36:46 *** onensora has joined #aichallenge 2011-06-28T04:48:59 *** Naktibalda has joined #aichallenge 2011-06-28T05:10:52 2011-06-28T05:14:15 2011-06-28T05:15:07 2011-06-28T05:15:10 , 2011-06-28T05:15:12 ? 2011-06-28T05:19:28 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-28T05:19:45 *** sYnfo has joined #aichallenge 2011-06-28T06:03:03 *** stocha has joined #aichallenge 2011-06-28T06:03:07 http://aichallengebeta.hypertriangle.com/visualizer.php?game=42641&user=21 2011-06-28T06:03:20 bots are weak, but the game is worth looking at once, i feel :) 2011-06-28T06:03:35 (greentea playing, and other similar bots) 2011-06-28T06:07:52 *** stocha has quit IRC (Client Quit) 2011-06-28T06:16:02 *** sigh has joined #aichallenge 2011-06-28T07:02:50 *** olexs has joined #aichallenge 2011-06-28T07:19:09 *** mleise has joined #aichallenge 2011-06-28T07:21:06 *** boegel has joined #aichallenge 2011-06-28T07:50:21 *** Accoun has quit IRC (Read error: Connection reset by peer) 2011-06-28T07:50:45 *** noid16388 has joined #aichallenge 2011-06-28T07:53:11 *** smiley1993 has joined #aichallenge 2011-06-28T07:55:25 contestbot: tcp 2011-06-28T07:55:26 smiley1993: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2011-06-28T07:57:06 *** stocha has joined #aichallenge 2011-06-28T07:57:44 davidd: davidjliu bot was expelled from the top 10 by some new maps. It shows how much the ranking depends on maps. 2011-06-28T08:00:54 *** stocha has quit IRC (Client Quit) 2011-06-28T08:20:20 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-06-28T08:26:27 *** smiley1993 has quit IRC (Quit: leaving) 2011-06-28T08:26:27 *** noid16388 has quit IRC (Read error: Connection reset by peer) 2011-06-28T08:26:44 *** noid16388 has joined #aichallenge 2011-06-28T08:27:58 *** smiley1983 has joined #aichallenge 2011-06-28T08:38:21 *** onensora has quit IRC () 2011-06-28T08:38:38 *** onensora has joined #aichallenge 2011-06-28T08:46:35 *** Zaphus has joined #aichallenge 2011-06-28T08:51:20 how would i express 'attackradius2' in a bitboard world ? 2011-06-28T08:58:05 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-28T09:04:14 McLeopold: you up? 2011-06-28T09:04:55 @topic 2011-06-28T09:04:55 antimatroid: Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || Code Repo: http://github.com/aichallenge/aichallenge || Beta testers needed: http://aichallengebeta.hypertriangle.com/ (amstan) || Launch Preparation Meeting http://bit.ly/kYYbD4 (amstan) 2011-06-28T09:24:06 *** eashoka has joined #aichallenge 2011-06-28T09:27:17 hi All! 2011-06-28T09:29:29 Got to bother you with a question. sorry if this one is already discussed many times. Why do we have to explicitly call "destinations.add(location)" if we are not moving an Ant (location)? all the sample code has it. can anybody please explain why it's needed ? 2011-06-28T09:34:13 if we are already checking for destination.isOccuied() before an issueOrder, then why do we need to explicitly keep track of where idle ants are ? 2011-06-28T09:34:49 even if an ant doesn't move, it'll be at some square next turn, so it's a destination in a way 2011-06-28T09:35:13 the python function unnocupied just compare the type value of the square 2011-06-28T09:35:41 it would be enougth to check 2011-06-28T09:36:55 how would i express 'attackradius2' in a bitboard world ? 2011-06-28T09:37:02 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-28T09:37:03 checking both occupied square and destination prevent us from colliding ants (moving or not) 2011-06-28T09:42:36 tx . what I meant was : 2011-06-28T09:44:04 the current tiles of the ants are already marked isn't it. (with the Ilk type) so, if we are not moving the ant, why do we need to specifically remember it's location ? 2011-06-28T09:44:47 the reason is, I'm getting a nasty error 'duplicate order' wonder thether this has something to do with it.... 2011-06-28T10:10:54 *** olexs has quit IRC (Quit: Leaving.) 2011-06-28T10:11:45 *** stocha has joined #aichallenge 2011-06-28T10:13:32 *** olexs has joined #aichallenge 2011-06-28T10:15:16 @later tell McLeopold: a larger view radius may actually hurt bots with better path finding 2011-06-28T10:15:16 antimatroid: Ready to serve. 2011-06-28T10:15:32 @later tell McLeopold: at least those who make use of information outside of view 2011-06-28T10:15:32 antimatroid: As you wish. 2011-06-28T10:16:38 eashoka: you're welcome to write your bot however you like :) 2011-06-28T10:16:45 the starter bots are just a place to get started 2011-06-28T10:17:26 @later tell gobang what do you mean by bitboard world ? How is attackradius2 a problem in there ? 2011-06-28T10:17:26 stocha: Ready to serve. 2011-06-28T10:17:35 stocha; second that :) 2011-06-28T10:17:51 i think my map code is finished 2011-06-28T10:17:58 i'ma put it on github soon 2011-06-28T10:18:54 antimatroid: good :) I'm worrying about how much it will impact current game feeling. But you don't progress without some big revolution. I think the maze map, will make the game quite different. Even if it is only one map over ten. 2011-06-28T10:19:22 stocha: i'm kind of planning to make this the map gen 2011-06-28T10:19:33 antimatroid: that would be very bad choice indeed. 2011-06-28T10:19:36 http://codepad.org/MoaTzzro 2011-06-28T10:19:40 in my opinion :) 2011-06-28T10:19:44 that's the map gen atm, with a sample map at the bottom 2011-06-28T10:19:52 Map generator choice, is really the main rule choice for the contest. 2011-06-28T10:20:04 maz maps are a very particular thing. They lack diversity. 2011-06-28T10:20:32 stocha: i can tweak the parameters to open the maps up more 2011-06-28T10:20:59 you'll notice they aren't "perfect mazes" 2011-06-28T10:21:05 in that there are many paths between some places 2011-06-28T10:21:08 antimatroid: let's try to put a few in there. But i doubt people will support the supression of the random walk. Which quite a good game . 2011-06-28T10:21:21 random walk is a crap strategy 2011-06-28T10:21:29 antimatroid: it's your generator, so feel free to make it great :) 2011-06-28T10:21:35 *** olexs has quit IRC (Quit: Leaving.) 2011-06-28T10:21:58 antimatroid: the resulting maps are good. So if you find a better way to have the same set of map, feel free to do it. 2011-06-28T10:22:53 What i'm sure of, is that the 100% random walk map, and the 100% maze map are different kind of ant games. 2011-06-28T10:23:11 Winning bot will be different. Efficient features, and best strategy will differ much. 2011-06-28T10:23:36 So chosing only one, is really imposing the major rule choice. 2011-06-28T10:23:52 But you can mix them (which arguably will make the game even more complicated) 2011-06-28T10:24:21 then you'll have to choose the ratio. But i think it's less a game breaker than having 0% of some map types. 2011-06-28T10:24:52 i think the idea of random walk maps can be better achieved with opening up the maze maps 2011-06-28T10:24:56 and they look better 2011-06-28T10:25:00 antimatroid: put a few labyrinth on the test server (while the old maps in there remains of course) 2011-06-28T10:25:02 and see how people react. 2011-06-28T10:25:12 i can't just do it myself 2011-06-28T10:25:18 and we need to update the food spawn code 2011-06-28T10:25:23 once it's ready we'll add a bunch 2011-06-28T10:25:25 ah. have the proper person do that then. 2011-06-28T10:25:44 when do you think it will be ready ? 2011-06-28T10:26:10 it will be very intersting to observer how people react during the first two week of maze adding :) 2011-06-28T10:26:21 initially they will be happy. 2011-06-28T10:26:36 then some will realize that their bot isn't doing what they though it was doing :p 2011-06-28T10:26:44 it will be the interesting point. 2011-06-28T10:27:22 That is a major rule change in fact. 2011-06-28T10:27:26 stocha: http://ants.zeroviz.us/UnQuI/ 2011-06-28T10:27:53 it looks very nice. 2011-06-28T10:28:05 it is however very corridor oriented maps. 2011-06-28T10:28:11 can you do very open ones ? 2011-06-28T10:29:02 if the contest was using this map generator only, i guess i will have a very big advantage all of a sudden :) 2011-06-28T10:29:12 because the corridor map are those i like the best. 2011-06-28T10:29:26 But ... a lot of people want to have open maps also. I'm quite sure of that. 2011-06-28T10:29:39 with just a few lakes 2011-06-28T10:30:14 stocha: http://ants.zeroviz.us/kLWoS/ 2011-06-28T10:30:17 Random walk is very good at diversity. That's why i think it will be nice to still use it. Whatever the percentage. So people don't take too much for granted. 2011-06-28T10:30:20 you end up with something like that 2011-06-28T10:30:41 antimatroid: it's interesting. But certainly not a very open map :) 2011-06-28T10:30:51 you should maybe kill some water plans. 2011-06-28T10:31:13 i could "bomb" patches of the map 2011-06-28T10:31:18 but still, it will be different from the random walk generator. Which are good and flexible (but won't generate good maze of course :) ) 2011-06-28T10:31:30 antimatroid: yeah. But i think the maze map is really good as it is 2011-06-28T10:31:33 like, after making the maze, bomb circles in places 2011-06-28T10:31:36 it does it's job very well 2011-06-28T10:31:42 i think the maze mapgen is the way to go though 2011-06-28T10:31:46 then open maps up from this 2011-06-28T10:31:48 (although you need to add open rooms also etc) 2011-06-28T10:31:52 it's way faster and more reliable too 2011-06-28T10:32:29 antimatroid: the random walk generator is very flexible, and good for diversity. I think it's a very important feature. The mazes will add greatly to those properties 2011-06-28T10:32:52 because i think good reliably corridor maps are a very usefull addition 2011-06-28T10:32:55 *** UncleVasya has joined #aichallenge 2011-06-28T10:33:09 stocha; i will have another go at making a random walk one with my new generator at some point 2011-06-28T10:33:11 antimatroid: assume it will be 50 of each. 2011-06-28T10:33:59 antimatroid: if you come up with a yet more diverse and flexible way than the current one, feel free to go there :) Let's keep the old one in the meantime, and mix it with those corridors 2011-06-28T10:34:10 there will probably will be many ways of using your current maze generator. 2011-06-28T10:34:20 and then there might be other ways to generate mazes. 2011-06-28T10:34:34 (which maybe you will want to dig in through, who knows) 2011-06-28T10:34:57 i'm not sure i really understand your algorithm :) 2011-06-28T10:35:10 *** Kingpin13 has joined #aichallenge 2011-06-28T10:35:33 (why is there a grid of lake in the most open maps, that puzzles me :p ) 2011-06-28T10:36:20 antimatroid: i think i would have done it this way. Buid 3x3 cells separated by full (1width) lines of water. 2011-06-28T10:37:11 antimatroid: then apply the wall painter algorithm. Whenever there is a wall you let the 1width water line be. Where there is not, you remplace it by 1x3 segment of emptyness. 2011-06-28T10:37:32 (it would be clearer with a picture maybe :p ) 2011-06-28T10:38:09 the idea is to apply the algorithm wich revolve about empty cells having four wall. While a cell share all of it's wall with the 4 neighbors. 2011-06-28T10:38:26 so walls are 1x3 water. and empty cells are 3x3 emptyness. 2011-06-28T10:39:33 obviously, you can make the wall width a parameter. 1x3 walls are hard to carve :) 2011-06-28T10:40:16 stocha: look at the wikipedia page for maze generation 2011-06-28T10:40:24 i'm doing the recursive back track algo from there 2011-06-28T10:40:52 you do use cells and walls right. 2011-06-28T10:42:16 antimatroid: the walls and cells concept are independant of wich algorithm you use :) 2011-06-28T10:42:38 antimatroid: i simply says : you can have wall width a parameter. It seems you always your 3 for wall width. 2011-06-28T10:43:05 stocha: i will experiment with other maze generation algos 2011-06-28T10:43:25 it should be a matter of replacing the backtrack function in the python thing i linked 2011-06-28T10:43:36 antimatroid: make sure there is one that definetly run well with the contest first though :) 2011-06-28T10:44:07 *** sigh_ has joined #aichallenge 2011-06-28T10:44:46 antimatroid: i think that maybe 1 or 3 very different generator will be enougth for the contest. So we may already have two. The random walk one (or a futur declinaison) and now your maze one (or a futur evolution) 2011-06-28T10:44:57 so we may add one very different thing. And we are done :) 2011-06-28T10:45:19 Though i have no idea what the third way could be. 2011-06-28T10:46:02 *** sigh__ has joined #aichallenge 2011-06-28T10:47:00 *** sigh has quit IRC (Ping timeout: 276 seconds) 2011-06-28T10:47:24 Adding 3rd dimension is the next thing 2011-06-28T10:47:27 :) 2011-06-28T10:47:52 UncleVasya: maybe you could generator some 3d form, and slice it for the map ? 2011-06-28T10:48:34 Slice map generator. Still you would have to generated a good 3d form. Which i don't want to being to think about 2011-06-28T10:48:39 *** sigh_ has quit IRC (Ping timeout: 246 seconds) 2011-06-28T10:48:45 UncleVasya: i would love to do n-dimensional tron :) 2011-06-28T10:49:01 sigh__: I haz new maps :) 2011-06-28T10:49:32 stocha: sorry, can't understand your sentence. 2011-06-28T10:50:37 http://aichallengebeta.hypertriangle.com/visualizer.php?game=43198&user=107 Two bots managed to have the same score (for the last rank) 2011-06-28T10:51:07 antimatroid: What do you think about PW with map bounds like in ants? I mean when you fly out of top boundary you come to the down. 2011-06-28T10:52:31 UncleVasya: I say. Generate a 3d form, with two substance (let's call them gold and iron). Then you slice the 3d form. When you encounter gold, you do a water cell. And when you encounter iron, you do a ground cell. But you would have to have a very special 3d form for it to generate maps. It's most probably not an interesting idea. 2011-06-28T10:52:35 *** sigh__ has quit IRC (Remote host closed the connection) 2011-06-28T10:54:03 *** slkjcna has quit IRC (Ping timeout: 240 seconds) 2011-06-28T10:54:21 antimatroid: As ships can move only through planets I think it it isn't a very big change in the gameplay. Am I wrong? 2011-06-28T10:54:53 *** slkjcna has joined #aichallenge 2011-06-28T10:54:56 UncleVasya: hyper torus? 2011-06-28T10:55:00 stocha: I am totally uncompetent person to discuss this stuff :) 2011-06-28T10:55:17 UncleVasya: so do i in fact 2011-06-28T10:56:38 when we remake tron 2011-06-28T10:56:41 *** carlop has joined #aichallenge 2011-06-28T10:56:43 that will need to be on a torus imo 2011-06-28T10:56:56 carlop: http://codepad.org/MoaTzzro this is my current map code 2011-06-28T10:56:57 torus are nice 2011-06-28T10:57:01 with a map generated at the bottom 2011-06-28T10:57:06 stocha: the ants maps are a torus :) 2011-06-28T10:57:31 antimatroid: so do most civilization maps so i heard 2011-06-28T10:57:56 carlop: you now have command line options for requesting map types, like "python mapgen.py --symmetry rotational --no_players 4" 2011-06-28T10:58:43 is this map 3x3 compliant ? http://aichallengebeta.hypertriangle.com/visualizer.php?game=43155&user=107 2011-06-28T10:58:48 hi... i've done some experimenting with generation symmetry, this is my grid.py (that work with your pattern generator): http://codepad.org/93JJ1KuF 2011-06-28T10:58:55 antimatroid: yeah, torus (thanks, now I know a new word). 2011-06-28T10:59:12 UncleVasya: do you get how the maps are a torus? 2011-06-28T10:59:20 *** olexs has joined #aichallenge 2011-06-28T10:59:30 currently generate 10 type of symmetry for 8 player map 2011-06-28T10:59:31 Hi, 2011-06-28T10:59:59 it's REALLY slow if you tell it to generate a tile symmetric map with a specified number of players 2011-06-28T11:00:04 i dunno how to fix that 2011-06-28T11:00:09 it's just hard to do that 2011-06-28T11:00:20 how slow ? 2011-06-28T11:00:52 antimatroid: the most important part of my code is the list of symmetry at row 32 2011-06-28T11:00:57 not that slow, i will just tell you i failed if it can't find valid parameters within a few seconds 2011-06-28T11:01:09 antimatroid: you said ants maps are a toruses so I think torus means "go one direction and you'll come to your starter position from the other side". 2011-06-28T11:01:12 UncleVasya: are you an experimented functional developer ? 2011-06-28T11:01:33 Anyway, I go to wiki to read about this wotd more. 2011-06-28T11:02:37 UncleVasya: think of a garden hose, if fold the top and bottom over the back onto each other, you get a cylinder/hose, then if you wrap the hose around into a donut shape (joining the sides) then you get a torus 2011-06-28T11:03:06 stocha: My bot is the "Hello world" for me in the functional world. I've never read functional code before Smiley's satarter pack. 2011-06-28T11:03:26 UncleVasya: you have pretty good ranking. 2011-06-28T11:03:39 He is expirienced fun-guy and he loves it. 2011-06-28T11:03:51 antimatroid: this is an example of 8player symmetry not previously generated: http://ants.zeroviz.us/vhax8/, and this is a list of antpast link for other symmety I can generate: http://pastebin.com/Eq1u00fS 2011-06-28T11:03:56 stocha: I have no life :) 2011-06-28T11:04:35 So if we compare my time invested to the bot and with others... 2011-06-28T11:04:40 carlop: that looks nice :) 2011-06-28T11:04:51 UncleVasya: is that synonym for divorced ? 2011-06-28T11:05:38 UncleVasya: you don't know for sure how much time others have put into this. 2011-06-28T11:06:12 there is the time you use for typing. And then all the unsubstantial time your use while you are doing (or supposed to) other stuff. And sleeping etc. 2011-06-28T11:06:28