2011-11-20T00:02:02 *** delt0r_ has joined #aichallenge 2011-11-20T00:07:28 *** helkafen has joined #aichallenge 2011-11-20T00:08:40 how are people doing battle resolution in their own code? 2011-11-20T00:09:25 i'm thinking i can just do it off locations 2011-11-20T00:09:32 but i need to pass in hill locations then 2011-11-20T00:11:20 thestinger: that'd do it! 2011-11-20T00:12:51 . 2011-11-20T00:30:46 *** TheLinker has quit IRC (Quit: Bye) 2011-11-20T00:41:42 later tell janzert any objection to boost for c++? 2011-11-20T00:41:50 @later tell janzert any objection to boost for c++? 2011-11-20T00:41:50 McLeopold: Job's done. 2011-11-20T00:44:35 @contestbot 2011-11-20T00:44:36 roflmao: User error -- Replace user. 2011-11-20T00:44:39 @later 2011-11-20T00:44:40 roflmao: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2011-11-20T00:44:55 @later tell me that cute girl loves me 2011-11-20T00:44:55 roflmao: OK 2011-11-20T00:44:59 thank you 2011-11-20T00:47:52 *** Antimony has joined #aichallenge 2011-11-20T00:48:46 *** roflmao has quit IRC (Quit: roflmao) 2011-11-20T00:48:56 *** roflmao has joined #aichallenge 2011-11-20T00:50:15 *** bmh has joined #aichallenge 2011-11-20T00:56:16 *** McLeopold is now known as me 2011-11-20T00:56:28 . 2011-11-20T00:56:38 *** me is now known as McLeopold 2011-11-20T00:56:51 evening 2011-11-20T00:57:00 @later tell roflmao I totally stole you girl 2011-11-20T00:57:00 McLeopold: I think that worked... 2011-11-20T00:57:10 !? 2011-11-20T00:57:33 (9:56:31 PM) contestbot: (notice) Sent 11 minutes ago: that cute girl loves me 2011-11-20T00:57:48 Clearly some missing context 2011-11-20T00:57:56 yeah, but just by ten minutes 2011-11-20T01:01:31 *** dlila has quit IRC (Quit: Leaving) 2011-11-20T01:04:14 *** twymer has quit IRC (Ping timeout: 260 seconds) 2011-11-20T01:05:56 *** bmh has quit IRC (Changing host) 2011-11-20T01:05:56 *** bmh has joined #aichallenge 2011-11-20T01:17:30 *** Antimony has quit IRC (Ping timeout: 276 seconds) 2011-11-20T01:27:00 *** Antimony has joined #aichallenge 2011-11-20T01:31:43 *** antimatroid has quit IRC (Read error: No route to host) 2011-11-20T01:31:54 *** antimatroid has joined #aichallenge 2011-11-20T01:35:38 *** antimatroid1 has joined #aichallenge 2011-11-20T01:35:58 *** antimatroid has quit IRC (Client Quit) 2011-11-20T01:37:00 *** antimatroid1 has quit IRC (Client Quit) 2011-11-20T01:37:01 *** antimatroid has joined #aichallenge 2011-11-20T01:39:15 *** JorgeB has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-11-20T01:39:39 thestinger: what language are you using? 2011-11-20T01:41:52 *** bmh has quit IRC (Quit: bmh) 2011-11-20T01:41:53 C++ 2011-11-20T01:42:04 I learned it for this (I did know C before though) 2011-11-20T01:42:43 *** ficode has joined #aichallenge 2011-11-20T01:43:55 python was just too slow 2011-11-20T01:45:25 thestinger: well I tried BFS from the boundary of my know world and it gets far too slow far too quickly - I'm using Java 2011-11-20T01:45:48 can anyone suggest a good way to test / attach to a mono process from monodevelop when loading from playgame.py ? 2011-11-20T01:45:55 well, my BFS is very specialized for this 2011-11-20T01:46:12 the only dynamic allocation I do is for the fifo queue (a std::queue) 2011-11-20T01:46:38 I store the distances in a 2d array (pre-allocated), and use that to avoid revisiting nodes 2011-11-20T01:46:51 *** u_ has quit IRC (Quit: u_) 2011-11-20T01:47:27 it's less than a dozen lines of code, so it was pretty easy to get it very optimized 2011-11-20T01:47:56 thestinger: yeah - it's tough in Java to not allocate stuff when you need it! 2011-11-20T01:48:07 I might have to implement some re-use pools 2011-11-20T01:48:34 you can pass a 2d array by reference though, right? 2011-11-20T01:48:41 and just clear it and reuse it whenever you do a BFS 2011-11-20T01:48:52 and then use it for pathfinding/distances (that's what I'm doing atm) 2011-11-20T01:49:24 well, I use a 1d array of boolean for explored tracking 2011-11-20T01:50:33 I'm not really calculating distances - only subtracting 1 from the 'value' of each point whenever I go one step deeper in the BFS 2011-11-20T01:50:53 which gives you the same thing I suppose 2011-11-20T01:51:49 ah 2011-11-20T01:51:59 I start off most things at '0' (as in, 0 distance from themselves :P) 2011-11-20T01:52:11 then +1 to the old value with each step of the BFS 2011-11-20T01:52:23 and I do +4 instead when going through an ant 2011-11-20T01:52:39 that last part doesn't work very well atm 2011-11-20T01:53:18 I think I just have a bug or two in my movement code because it used to work well 2011-11-20T01:53:56 I guess for the collaborative diffusion thing you start off with big numbers and diffuse them over the map 2011-11-20T01:54:04 I'm sort of doing the opposite 2011-11-20T01:54:20 ahh, so your ants flow downwards then - it's the same thing though really, just mine flow 'up' trying to maximize the value, right? 2011-11-20T01:54:29 yeah 2011-11-20T01:54:38 same thing, as long as your value is big enough that it covers the whole map 2011-11-20T01:55:13 I'm not doing any 'diffusion' as far as I know - I think that means you run some kind of convolution on the grid 2011-11-20T01:55:48 I like doing it this way b/c I get distance measurements in addition to pathfinding (I guess you get something like that too though) 2011-11-20T01:56:14 I wonder about that anyway - surely that diffusion method means it takes time for the values to spread and find your ants, rather than instant like with BFS? 2011-11-20T01:56:23 *** Fandekasp has joined #aichallenge 2011-11-20T01:56:27 dunno, I haven't tried it 2011-11-20T01:57:02 *** treeform has quit IRC (Remote host closed the connection) 2011-11-20T01:57:11 I think your +4 for ants will only create a small shadow behind the ant - because won't the other values from round the sides swallow over it? 2011-11-20T01:57:40 yeah, it just makes them willing to go sideways rather than "following the leader" 2011-11-20T01:57:57 they will still occasionally follow the ant in front of them, just not always like they used to 2011-11-20T01:58:22 and if I get a big clog in a chokepoint, it will increase the distance of that path a lot 2011-11-20T01:58:26 so they'll take alternate routes 2011-11-20T01:58:29 *** roflmao has quit IRC (Remote host closed the connection) 2011-11-20T01:58:40 ahh ok - I thought it was so a slightly farther object would seem more appealing since the most appealing already has an ant between you and it 2011-11-20T01:59:28 yeah, that can happen too 2011-11-20T01:59:30 but as soon as the ant steps out of the shadow it would start to go for the closest one again 2011-11-20T01:59:34 if I put multiple targets on the map 2011-11-20T01:59:46 I was just talking about a map with 1 target though (like 1 enemy hill) 2011-11-20T02:00:14 I do a separate BFS for every single food because doing them together didn't work well 2011-11-20T02:00:46 I might port my Java into C, or Objective-C if it gets supported 2011-11-20T02:00:49 doing it from ~7000 targets on one 2d array takes ~1ms (which is what happens for exploration) 2011-11-20T02:01:37 it used to take ~40ms before I switched to multi-source BFS 2011-11-20T02:04:08 *** antimatroid1 has joined #aichallenge 2011-11-20T02:04:32 http://ants.fluxid.pl/replay.6821 I need to fix this xD 2011-11-20T02:04:34 *** antimatroid has quit IRC (Read error: Connection reset by peer) 2011-11-20T02:05:35 I fixed a few bugs that caused stalemates but I need to make the kill:death ratio goal for combat dynamic 2011-11-20T02:05:49 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-11-20T02:06:13 *** Antimony has joined #aichallenge 2011-11-20T02:10:53 yeah - they need to know when they have greater numbers than the opponent so they know it is safe to sacrifice some ants in order to break through a barrier 2011-11-20T02:16:10 *** Flort has quit IRC (Ping timeout: 244 seconds) 2011-11-20T02:30:43 *** avdg1 has joined #aichallenge 2011-11-20T02:33:33 *** Antimony has quit IRC (Ping timeout: 276 seconds) 2011-11-20T02:33:35 *** avdg has quit IRC (Ping timeout: 248 seconds) 2011-11-20T02:41:39 thestinger: This is my state at 27 turns - bot times out around this many nodes :( 2011-11-20T02:41:50 thestinger: http://dl.dropbox.com/u/9957688/imgbin/ai-state-vis006.png 2011-11-20T02:42:24 *** ficode has quit IRC (Remote host closed the connection) 2011-11-20T02:44:24 *** UncleVasya has joined #aichallenge 2011-11-20T02:45:38 *** retybok_ has joined #aichallenge 2011-11-20T02:52:58 *** yoden has quit IRC (Quit: Leaving.) 2011-11-20T02:53:50 *** helkafen has quit IRC (Ping timeout: 265 seconds) 2011-11-20T02:54:10 jab_bott: just with a single BFS? sounds like you have a bug 2011-11-20T02:59:13 I'm off to bed, cya 2011-11-20T02:59:25 nite 2011-11-20T03:03:35 *** epicmonkey has joined #aichallenge 2011-11-20T03:04:19 *** Yexo_ has joined #aichallenge 2011-11-20T03:05:20 *** Yexo__ has joined #aichallenge 2011-11-20T03:08:27 *** Yexo has quit IRC (Ping timeout: 244 seconds) 2011-11-20T03:08:49 *** Yexo_ has quit IRC (Ping timeout: 260 seconds) 2011-11-20T03:09:08 *** Yexo has joined #aichallenge 2011-11-20T03:10:13 ahh, it's because the BFS is multi-point start - I need to mark nodes as explored as soon as they're added to the frontier, not when they're expanded (I think) 2011-11-20T03:10:25 *** retybok_ has quit IRC (Ping timeout: 244 seconds) 2011-11-20T03:10:36 *** Yexo__ has quit IRC (Ping timeout: 276 seconds) 2011-11-20T03:12:16 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-11-20T03:12:20 *** Fandekasp has quit IRC (Quit: leaving) 2011-11-20T03:13:37 *** Yexo has quit IRC (Ping timeout: 244 seconds) 2011-11-20T03:14:59 *** Fandekasp has joined #aichallenge 2011-11-20T03:15:55 *** HaraKiri has joined #aichallenge 2011-11-20T03:16:27 HaraKiri: heya - have you tried the map info stuff yet? 2011-11-20T03:16:46 good morning! yes, works like a charm <3 2011-11-20T03:17:09 a bit cumbersome that you can't use spaces inside the strings, but it works 2011-11-20T03:17:57 *** ALplus has quit IRC (Ping timeout: 245 seconds) 2011-11-20T03:26:40 HaraKiri: I spent ages trying to figure that out- something mauls the JSON to pass it into the HTML and a JS String. 2011-11-20T03:26:53 *as a 2011-11-20T03:27:11 *** CowTipperVirus has quit IRC (Ping timeout: 265 seconds) 2011-11-20T03:27:54 hmm ok, i'm afraid i can't help you with that :/. but it works fine like it is i guess 2011-11-20T03:29:01 HaraKiri: Check it: http://dl.dropbox.com/u/9957688/imgbin/ai-state-vis006.png 2011-11-20T03:29:43 fancy :D. looks like some bfs + blue things are the fringe of the visible territory? 2011-11-20T03:34:25 yeah - they are the start-points for the BFS 2011-11-20T03:35:38 cool, i did something similar, but with more colors and probably less cleer :D http://i.imgur.com/oQSLn.png 2011-11-20T03:36:50 *** speakeasy has joined #aichallenge 2011-11-20T03:38:17 wow - awesome! 2011-11-20T03:38:50 the arrows should be fixed now 2011-11-20T03:39:44 *** retybok_ has joined #aichallenge 2011-11-20T03:39:48 yea they looked a bit awkward :D. the hover thing makes looking into some ants brain so much easier haha 2011-11-20T03:45:44 i have very early battle set up :D 2011-11-20T03:45:49 bedtime :( - too tired to code any more 2011-11-20T03:45:53 *** jab_bott has left #aichallenge 2011-11-20T03:46:06 well, i have it making moves 2011-11-20T03:46:16 i haven't done anything about making sure it's working within the time limit yet 2011-11-20T03:49:40 antimatroid1: Do you have already some nfo about seed of your battle resolution? What is the biggest amount of ants in the battle you can handle in 500ms? 2011-11-20T03:49:57 UncleVasya: like i said, i haven't done anything about time yet 2011-11-20T03:50:12 atm it's partitioning ants up into sets that can affect each other from battle 2011-11-20T03:50:27 then it's doing a kind of minimax thing to pick moves for friendly ants in each partition 2011-11-20T03:50:45 the plan is to next do each partition incrementally so as to get it running under the time limit 2011-11-20T03:50:50 ah, I started to write my question before you wrote about time. And during typing I look at the keyboard :) 2011-11-20T03:52:15 Yep, a set of ants that can affect each other - it's my definition of Battle type. It consist 2 lists: my_battle_ants and enemy_battle_ants. 2011-11-20T03:52:41 i already partition it up a bit 2011-11-20T03:52:46 but they could still be too large 2011-11-20T03:53:27 But when the amount of ants goes to 10+ I have no time for this even with prunning (or I have to tune up this). 2011-11-20T03:54:02 So I' thinking about the idea to split big battles into few small ones. 2011-11-20T03:55:37 Solving them will not be as accurate as solving a big one completely (because some affects do not count) but maybe it will be better with the time limit. 2011-11-20T03:56:57 *** flag has quit IRC (Read error: Operation timed out) 2011-11-20T04:00:03 antimatroid1: Also since my moves do not affect enemy moves isn't it faster to build once a list of all possible enemy positions in current battle and after that apply them to my positions? 2011-11-20T04:00:36 Instead of generating enemy moves for every your move. 2011-11-20T04:00:51 UncleVasya: i do what i think you're saying 2011-11-20T04:01:44 generate moves for me and my oponents, then for each of my lot of moves implement those to a copy of the state, then for each of the enemies moves implement those to the partial state until i know i wouldn't want to play that move and continue searching through the moves 2011-11-20T04:02:13 ok, I misunderstood when you said minimax. I thought you're talking about recursion but you mean only the way you're solving positions 2011-11-20T04:02:36 it's the same as the recursion way i'm just not using recursion 2011-11-20T04:02:48 yep, the loop. 2011-11-20T04:02:50 someone once told me recursion is bad and it stuck 2011-11-20T04:04:10 heh, somewhen I even saw some humans were writing a lot of loops to avoid recursion :) 2011-11-20T04:04:48 Something like 3 or 4 inner loops. 2011-11-20T04:06:56 wouldn't surprise me if that was me aha 2011-11-20T04:08:29 'until i know i wouldn't want to play that move' So my 'if cur_pos_score <= best_pos_score' isn't something I must proud for, it's just a standart :( 2011-11-20T04:12:57 antimatroid1: Do you still have a folder with your PW-bot sources? 2011-11-20T04:13:10 i do 2011-11-20T04:13:30 Not the only last version. 2011-11-20T04:13:39 all of them? 2011-11-20T04:13:41 it's a real mess aha 2011-11-20T04:14:34 i don't even know what's my final bot in there aha 2011-11-20T04:15:06 Maybe you remember I told you I wanna see how you was making it step by step. 2011-11-20T04:16:13 *** Euphony has joined #aichallenge 2011-11-20T04:17:16 From the very beginning (adding simple defence, stoling planets) to the cool features like opponent modelling, etc. 2011-11-20T04:17:32 antimatroid1: Can you upload it somewhere? 2011-11-20T04:18:58 *** emiel_ has joined #aichallenge 2011-11-20T04:19:07 you'd find that hard to work out 2011-11-20T04:19:25 and i can't really be bothered going through and deleting compiled versions from folders 2011-11-20T04:19:36 i'll send you one of my finalish copies if you want? 2011-11-20T04:19:39 hmm, ok. I trust you :) 2011-11-20T04:21:35 I have a final version (from the forums). 2011-11-20T04:31:51 hi 2011-11-20T04:32:19 does the ai apply to the ants themselves or as a sort of god of ants? 2011-11-20T04:32:44 thats a bit confusing sorry 2011-11-20T04:41:21 Real ants are already pretty smart, and they do have a sort of God (their queen), so yeah 2011-11-20T04:42:08 but does the bot apply to the ants individualy or to the colony as a whole? 2011-11-20T04:42:12 speakeasy: there is a "god" (your bot) that controls all ants and to whom all ants report what they see 2011-11-20T04:42:34 aha, that makes things very clear thanks 2011-11-20T04:42:46 hi,One instance of your bot contrlos all the ants you have. So bot some kind of playing an RTS game. 2011-11-20T04:43:10 also whats a treeset? 2011-11-20T04:43:16 thanks uncle 2011-11-20T04:44:27 set means no duplicates. tree means sorted. yar 2011-11-20T04:45:05 so whats the difference between a treeset and an arraylist? 2011-11-20T04:45:29 sorry, i probs should have said im working in java 2011-11-20T04:46:24 http://lmgtfy.com/?q=treeset+arraylist+in+java 2011-11-20T04:46:31 *** antimatroid1 has quit IRC (Quit: Leaving.) 2011-11-20T04:46:32 *** antimatroid has joined #aichallenge 2011-11-20T04:47:42 ok haha lol 2011-11-20T04:48:40 i think one of the sample bots might suck time wise 2011-11-20T04:49:36 i think i should probably learn some more java before trying this any more, im a little confused :s 2011-11-20T04:50:02 yeah lefty bot has an issue 2011-11-20T04:50:40 it'll eat up just under the entire turn limit on like turn 9 in this game i have :\ 2011-11-20T04:50:49 speakeasy: nothing wrong with learning on the job as they say... 2011-11-20T04:52:06 is there any docs on things i can call, e.g. move ant x north or attack with ant y 2011-11-20T04:52:07 ? 2011-11-20T04:52:08 also java is a little more forgiving that some other "core" langs... where core here is defined as "used a lot in industry" 2011-11-20T04:52:21 lol 2011-11-20T04:52:52 speakeasy: the javadocs describe things a bit. Then wiki/google would help with the rest 2011-11-20T04:53:14 where can i get a copy of the javadocs? 2011-11-20T04:53:14 but the starter bot source is its documentation 2011-11-20T04:53:29 you can download them --java api 2011-11-20T04:53:30 ah ok 2011-11-20T04:54:07 javadoc is a standard of commenting code... then you run a javadoc program that output a set of html pages that is now your api docs 2011-11-20T04:54:25 the core java api docs are very good 2011-11-20T04:54:32 python has something similar 2011-11-20T04:55:10 but C++/C even tho has doxygen don't seem to use it for there own api 2011-11-20T04:55:12 I believe there are comments in the code for Ants.java and other important starter bot classes. The javadocs just use those comments before the function and class declarations. 2011-11-20T04:55:16 ah ok 2011-11-20T04:55:32 thanks for being so helpful guys 2011-11-20T04:56:14 otherwise a tree set internally orders things in a balanced tree so most operations are ln(n) 2011-11-20T04:57:21 speakeasy: one of the fetures of this comp is the "learning" or newb coders can join and lean... so feel free to ask for help 2011-11-20T04:57:46 but whatever you do DONT go into the ##java channel 2011-11-20T04:58:10 nothing but trolls 2011-11-20T04:58:37 haha lol 2011-11-20T04:58:42 will take note of that 2011-11-20T04:59:01 you would get more help for java in a C# channel 2011-11-20T04:59:48 haha 2011-11-20T05:00:07 im reading through the tutorial section and trying to understand the code 2011-11-20T05:00:15 erm there is a bit that says getAnts 2011-11-20T05:00:28 however i cant find it in the ants.java file 2011-11-20T05:00:39 whatsgoingon? 2011-11-20T05:00:47 that is calling a method in another class 2011-11-20T05:01:03 aka another file 2011-11-20T05:01:22 which file? 2011-11-20T05:01:33 don't remember 2011-11-20T05:01:36 or should i just accept that it does what it says on the tin :L 2011-11-20T05:01:39 if you use a IDE 2011-11-20T05:01:45 it would find it for you 2011-11-20T05:01:50 otherwise there is grep 2011-11-20T05:01:59 I would recomend useing a ide 2011-11-20T05:02:03 like eclipse 2011-11-20T05:02:06 how do import all these files into netbeans? 2011-11-20T05:02:17 don't use netbeans 2011-11-20T05:02:32 but my guess there is a import under file 2011-11-20T05:02:45 so you need to start a project with no files 2011-11-20T05:02:55 ah, it just imports "project file" format 2011-11-20T05:02:57 then import them... or just copy them, 2011-11-20T05:03:21 eclipse uses ~/workspace/ under linux 2011-11-20T05:03:30 don't know about netbeans 2011-11-20T05:04:05 ah ok 2011-11-20T05:04:07 thanks 2011-11-20T05:06:46 *** UncleVasya has quit IRC (Ping timeout: 244 seconds) 2011-11-20T05:07:54 will the bot run faster on a better pc? 2011-11-20T05:11:05 *** besh has joined #aichallenge 2011-11-20T05:11:10 *** pairofdice has joined #aichallenge 2011-11-20T05:14:54 *** besh has quit IRC (Client Quit) 2011-11-20T05:15:01 when i come to upload a bot what format should it be in? 2011-11-20T05:18:35 *** Euphony has quit IRC (Quit: Page closed) 2011-11-20T05:21:28 *** speakeasy has quit IRC (Ping timeout: 240 seconds) 2011-11-20T05:21:52 *** jasox has quit IRC (Ping timeout: 240 seconds) 2011-11-20T05:25:13 *** Yexo has joined #aichallenge 2011-11-20T05:37:22 *** retybok_ has quit IRC (Ping timeout: 252 seconds) 2011-11-20T05:41:41 *** foRei has joined #aichallenge 2011-11-20T05:46:37 *** retybok_ has joined #aichallenge 2011-11-20T05:47:02 *** ikaros has joined #aichallenge 2011-11-20T05:51:41 *** jasox has joined #aichallenge 2011-11-20T05:55:43 *** antimatroid has quit IRC (Ping timeout: 248 seconds) 2011-11-20T05:58:57 *** retybok_ has quit IRC (Ping timeout: 276 seconds) 2011-11-20T06:06:55 *** antimatroid has joined #aichallenge 2011-11-20T06:07:07 bah i hate bugs 2011-11-20T06:08:04 they are everwhere :) 2011-11-20T06:08:43 *** retybok_ has joined #aichallenge 2011-11-20T06:17:36 okay it now seems to be running without error 2011-11-20T06:17:52 i just need to make it run more battle stuff in the given time 2011-11-20T06:18:04 reduce ALL the partitions 2011-11-20T06:20:14 *** retybok_ has quit IRC (Ping timeout: 252 seconds) 2011-11-20T06:20:40 *** mcstar has joined #aichallenge 2011-11-20T06:33:52 *** jasox has quit IRC (Quit: Leaving) 2011-11-20T06:35:59 *** g0llum has joined #aichallenge 2011-11-20T06:42:00 mcstar: i have semi working battle code 2011-11-20T06:42:14 it's slow 2011-11-20T06:42:17 hence semi working 2011-11-20T06:42:28 and i still need to play around with the evaluation function 2011-11-20T06:44:05 yay 2011-11-20T06:45:03 im back on square 1 2011-11-20T06:45:20 ? 2011-11-20T06:45:43 i got angry with her yesterday 2011-11-20T06:45:59 i didnt like its exploration coed 2011-11-20T06:46:06 or her 2011-11-20T06:46:35 it did well on open terrain but was mediocre in mazes 2011-11-20T06:48:01 its soo crazy that some simple things turn out to be extremely good 2011-11-20T06:50:10 *** xathis has joined #aichallenge 2011-11-20T06:51:38 *** Conorach has joined #aichallenge 2011-11-20T06:52:59 wtf 2011-11-20T06:53:08 i cant properly parameterize playgame.py 2011-11-20T06:53:20 stupid thing keeps saying i must specify a map file 2011-11-20T06:53:49 what parameters are you using? 2011-11-20T06:54:43 all of them 2011-11-20T06:54:46 f 2011-11-20T06:54:48 im angryű 2011-11-20T06:54:59 i tried to change food spawning 2011-11-20T06:55:20 just added --food_turn --food_rate 2011-11-20T06:55:26 and stopped working 2011-11-20T06:56:50 --food_rate: invalid integer value: '--food_turn=10' 2011-11-20T06:56:56 wtf? 2011-11-20T06:57:01 i added 2011-11-20T06:57:10 --food_rate=0 --food_turn=10 2011-11-20T06:57:14 for example 2011-11-20T06:57:23 why does it not see the 0? 2011-11-20T06:58:12 --food_rate is two numbers 2011-11-20T06:58:34 --food_rate=FOOD_RATE 2011-11-20T06:58:37 so is food_turn 2011-11-20T06:58:37 Numerator of food per turn per player rate 2011-11-20T06:58:42 --food_turn=FOOD_TURN 2011-11-20T06:58:44 Denominator of food per turn per player rate 2011-11-20T06:59:05 i dont get it 2011-11-20T06:59:06 http://aichallenge.org/game_settings.php 2011-11-20T06:59:14 one is the numerator the other is the denominator 2011-11-20T06:59:25 of a single quantity: food per turn 2011-11-20T06:59:54 pairofdice: wtf? 2011-11-20T06:59:57 i see thx 2011-11-20T06:59:58 mcstar: what's your username on the main site? 2011-11-20T07:00:01 but i dont get this 2011-11-20T07:00:20 antimatroid: let me calm down 2011-11-20T07:00:47 pairofdice: have you read ./playgame.py's usage message? 2011-11-20T07:01:11 how on earth should i deduce from that that those need 2 numbers? 2011-11-20T07:01:29 I don't know :) 2011-11-20T07:01:43 It is weird 2011-11-20T07:02:05 antimatroid: agentsmith 2011-11-20T07:02:17 but pls dont watch my bot its very shy 2011-11-20T07:04:08 *** retybok_ has joined #aichallenge 2011-11-20T07:04:17 i cant believe this 2011-11-20T07:04:28 :p 2011-11-20T07:04:36 that stupid spreading behavior comes from the simplest thing 2011-11-20T07:04:40 antimatroid: your spreading 2011-11-20T07:04:49 how so? 2011-11-20T07:05:04 you just pathfind to the first "not yet seen" tile right? 2011-11-20T07:05:07 and move your bot there 2011-11-20T07:05:27 I add unseen tiles to the same goal list my food goes 2011-11-20T07:05:33 omg 2011-11-20T07:05:36 :> 2011-11-20T07:05:49 i would never thought about this before 2011-11-20T07:05:52 i just tried ti 2011-11-20T07:06:07 i mean you antimatroid always said that 2011-11-20T07:06:12 but i didnt actually believe you 2011-11-20T07:06:37 i thought there must be something more going on 2011-11-20T07:07:17 im gonna combine this with my previous one 2011-11-20T07:07:17 mcstar: i have a collectUnseen and collectNotRecentlySeen function 2011-11-20T07:07:45 and it'll do bfs's from movable ants to find an appropriate square and move the ant on that path and "bomb" the vision information at its target 2011-11-20T07:07:52 so it doesn't send lots of ants to the same patch 2011-11-20T07:09:04 *** mleyen has joined #aichallenge 2011-11-20T07:09:07 i tried a supercell approach before but it didnt work out 2011-11-20T07:10:26 but i might use it for something other than navigation 2011-11-20T07:10:58 *** retybok_ has quit IRC (Ping timeout: 258 seconds) 2011-11-20T07:10:58 i had a collection of nodes, that connect to many pixels of the original one, like 16 2011-11-20T07:11:10 and i can collect information from those cells 2011-11-20T07:11:16 and store that in the supercell 2011-11-20T07:11:27 so i can get a genral idea whats where 2011-11-20T07:11:33 *** nle85 has joined #aichallenge 2011-11-20T07:11:58 i worked out a scheme that removes degeneracies in this supercell structure 2011-11-20T07:12:13 and rearranges the pixels 2011-11-20T07:12:22 anyway, this is great 2011-11-20T07:13:14 i still haven't found a good, fast and elegant solution to the exploring problem :/ 2011-11-20T07:14:10 just try this one 2011-11-20T07:14:16 you will be pleased 2011-11-20T07:14:28 you mean what antimatroid said? 2011-11-20T07:14:47 *** yk has joined #aichallenge 2011-11-20T07:15:10 not necessarily 2011-11-20T07:15:18 you can start with a simpler one 2011-11-20T07:15:35 like pairofdice said 2011-11-20T07:16:00 closest unseen tile leads to horrible behaiour 2011-11-20T07:16:20 ? 2011-11-20T07:16:24 noway 2011-11-20T07:16:28 its beautiful 2011-11-20T07:16:41 they behave like children with good manners 2011-11-20T07:16:50 when i tried that i had alot of jittering ants in many situations 2011-11-20T07:16:56 exactly 2011-11-20T07:17:00 no 2011-11-20T07:17:01 they oscillate between two places 2011-11-20T07:17:02 I probably set the rate at which they recheck unseen tiles too fast 2011-11-20T07:17:03 thats soemthing else 2011-11-20T07:17:06 *** raemde has joined #aichallenge 2011-11-20T07:17:16 we are not talking about the same thing 2011-11-20T07:17:28 *** mikewintermute has joined #aichallenge 2011-11-20T07:17:30 the closest never explored tile then? 2011-11-20T07:17:34 i pathfind till i find a tile that noone has seen before 2011-11-20T07:17:47 that is better, but then a lot of ants go to the same spot 2011-11-20T07:17:52 not really 2011-11-20T07:18:05 There's ways around that 2011-11-20T07:18:10 well if 3 ants spawn in 3 rounds, then they will probably all go to the same target 2011-11-20T07:18:18 no 2011-11-20T07:18:30 HaraKiri, what you could try out is this: http://forums.aichallenge.org/viewtopic.php?f=24&t=1776 2011-11-20T07:18:30 i dont preserve paths 2011-11-20T07:19:38 its a different approach to exploration.. but i found it to work very good. at least until you have explored the map or loops 2011-11-20T07:20:43 i altered it this way that if an ant leaves the positive "highway" it begins to stumble this works very good in combination 2011-11-20T07:21:22 i've seen this thread, and i'm generating highways too, but frankly the general idea wasn't clear to me, so i gave up reading halfway 2011-11-20T07:21:33 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-11-20T07:22:18 the idea is pretty simple 2011-11-20T07:22:31 an efficient implementation is a bit fiddling i think 2011-11-20T07:22:52 http://ants.fluxid.pl/replay.6351 wanderer is one of my experimentals .. it uses the linked strategy 2011-11-20T07:23:12 the bot isnt very good but im satisfied with the exploration so far 2011-11-20T07:23:37 hmm yes at least looks much better than any results i had 2011-11-20T07:24:51 what didnt you understand with the strategy? 2011-11-20T07:26:10 you use your hills as origins which are set to 0 2011-11-20T07:26:41 then update all visible values to the real distance to the origin (bfs or something) 2011-11-20T07:26:55 it's probably understandable of you read up what hill climbing is, i just didnt do that, and then the text makes little sense because it delves into details 2011-11-20T07:27:07 i know and do bfs's 2011-11-20T07:27:23 hill climbing in this context just means your ant always takes the greatest step 2011-11-20T07:27:27 *** cyphase has quit IRC (Ping timeout: 258 seconds) 2011-11-20T07:27:44 it never goes to a tile with a smaller value than which it is sitting on 2011-11-20T07:28:03 well that's the general idea behind using bfs for pathfinding, right? 2011-11-20T07:28:16 the point is in detecting loops and dead ends.. and the very elegant idea of flipping signs then 2011-11-20T07:28:24 yea kinda 2011-11-20T07:28:41 detecting dead ends speeds up the whole thing, but has nothing to do with the exploration problem itself as i see it 2011-11-20T07:28:43 the difference is perhaps that you dont have real targets here 2011-11-20T07:29:12 oh it has. if your ants dont enter dead ends they explore the unknown faster 2011-11-20T07:29:19 same with loops 2011-11-20T07:30:13 on occasion you will want to enter loops or dead ends. you have to ignore that map then 2011-11-20T07:30:30 but for initial exploration the algorithm is very good. especially in mazes 2011-11-20T07:31:01 hmm maybe i'll look into it. but it "feels" forced and unnatural 2011-11-20T07:31:53 Setting your hills as origins your ants try to get just sounds like a bad idea to me 2011-11-20T07:32:00 get away from* 2011-11-20T07:32:31 why that 2011-11-20T07:32:52 i can see how not exploring dead ants can help, but that's really just a tiny optimazion. i'm still optimistic that there is a simple and elegant algorithm to the general problem 2011-11-20T07:33:03 dead ants = dead ends lol :D 2011-11-20T07:33:17 well in fact you "explore" them.. you must see it 2011-11-20T07:33:25 you just dont go in a second time 2011-11-20T07:33:31 and leave them the fastest way 2011-11-20T07:34:04 *** yk has quit IRC (Quit: Page closed) 2011-11-20T07:37:33 *** AntDroid has quit IRC (Ping timeout: 265 seconds) 2011-11-20T07:40:45 *** cyphase has joined #aichallenge 2011-11-20T07:44:14 *** retybok_ has joined #aichallenge 2011-11-20T07:46:19 *** zyberkiddy has joined #aichallenge 2011-11-20T07:48:40 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-20T07:56:41 *** zyberkiddy has quit IRC (Quit: Page closed) 2011-11-20T08:01:24 *** delt0r_ has joined #aichallenge 2011-11-20T08:11:09 *** TheLinker has joined #aichallenge 2011-11-20T08:19:58 *** willem has joined #aichallenge 2011-11-20T08:20:46 I had a question about the replayformat. For the ants, it seems that the length of the moves sometimes equals to the end turn minus the beginning turn but sometimes it seems to differ by one. 2011-11-20T08:21:07 How is this possible? 2011-11-20T08:25:00 nvm I solved the problem. 2011-11-20T08:30:31 *** Redgis has joined #aichallenge 2011-11-20T08:36:19 *** Bluedgis has quit IRC (*.net *.split) 2011-11-20T08:36:19 *** Clex has quit IRC (*.net *.split) 2011-11-20T08:36:19 *** sorki has quit IRC (*.net *.split) 2011-11-20T08:36:19 *** mawif has quit IRC (*.net *.split) 2011-11-20T08:36:19 *** ruser has quit IRC (*.net *.split) 2011-11-20T08:36:19 *** BoAnd has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** lericson has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** elderwolf has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** jbroman has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** bysin has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** Ttech has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** welterde has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** jhannah has quit IRC (*.net *.split) 2011-11-20T08:36:23 *** blueprints has quit IRC (*.net *.split) 2011-11-20T08:36:26 *** mawif has joined #aichallenge 2011-11-20T08:36:26 *** BoAnd has joined #aichallenge 2011-11-20T08:36:27 *** jhannah_ has joined #aichallenge 2011-11-20T08:36:27 *** elderwol1 has joined #aichallenge 2011-11-20T08:36:28 *** lericson has joined #aichallenge 2011-11-20T08:36:30 *** ruser has joined #aichallenge 2011-11-20T08:36:33 *** Bluedgis has joined #aichallenge 2011-11-20T08:36:35 *** sorki has joined #aichallenge 2011-11-20T08:36:36 *** sorki has joined #aichallenge 2011-11-20T08:36:38 *** jbroman has joined #aichallenge 2011-11-20T08:36:38 *** blueprints has joined #aichallenge 2011-11-20T08:36:39 *** Clex has joined #aichallenge 2011-11-20T08:36:44 *** bysin has joined #aichallenge 2011-11-20T08:37:29 *** Harpyon has joined #aichallenge 2011-11-20T08:37:56 *** Harpyon_ has joined #aichallenge 2011-11-20T08:43:19 *** dlila has joined #aichallenge 2011-11-20T08:45:16 *** welterde has joined #aichallenge 2011-11-20T08:45:39 *** Ttech has joined #aichallenge 2011-11-20T08:46:40 *** retybok_ has quit IRC (Ping timeout: 240 seconds) 2011-11-20T08:54:14 [14,15,13,140,0] 2011-11-20T08:54:29 my new bot beats 3 old one in food collection 2011-11-20T08:54:33 and exploration 2011-11-20T08:55:17 antimatroid: how much better is your new bot than your old one? 2011-11-20T08:55:34 haven't played them 2011-11-20T08:55:40 haven't got old copies of the current 2011-11-20T08:55:54 i'm still working on something that plays well within the time limits 2011-11-20T08:56:06 and i haven't told the bot about hills yet 2011-11-20T08:56:11 except for in battle resolution 2011-11-20T08:56:15 oh 2011-11-20T08:56:55 so why dont you fight them? 2011-11-20T08:57:11 id be interested 2011-11-20T08:57:50 i'm hoping to have something to play on tcp in a few hours or the next day or so 2011-11-20T08:58:21 people will hate playing me 2011-11-20T08:58:25 but i will use my turn limit aha 2011-11-20T08:58:57 the 5 secs? 2011-11-20T08:59:09 is it 5 seconds on tcp? 2011-11-20T08:59:19 does it give the bots a smaller one then cushion it? 2011-11-20T08:59:28 or expect you to cushion it yourself for network overhead? 2011-11-20T08:59:47 it used to just tells bots the normal one i think then add like 4 seconds 2011-11-20T09:00:12 i think i remember seein 5000ms when i debugged 2011-11-20T09:00:28 my bot will be trying to use near 5000ms then :P 2011-11-20T09:00:53 *** Antvolution has joined #aichallenge 2011-11-20T09:01:34 you will time out 2011-11-20T09:01:49 my bot used <100ms i think and still times out frequently on tcp 2011-11-20T09:06:39 *** hacklash has quit IRC (Ping timeout: 248 seconds) 2011-11-20T09:08:29 my bot uses min(500, settings["turn_time']) ms per turn 2011-11-20T09:08:49 haven't seen any timeouts on tcp (other than real ones where it really used over 5s due to bugs) 2011-11-20T09:13:19 *** mj41 has joined #aichallenge 2011-11-20T09:26:21 *** retybok_ has joined #aichallenge 2011-11-20T09:29:24 *** praveen has joined #aichallenge 2011-11-20T09:39:11 *** retybok_ has quit IRC (Ping timeout: 248 seconds) 2011-11-20T09:43:09 yeah the tcp server should really just tell the clients the timeout is 500ms and have the actual timeout be longer 2011-11-20T09:43:35 *** twymer has joined #aichallenge 2011-11-20T09:46:47 *** sigh has quit IRC (Remote host closed the connection) 2011-11-20T09:53:48 why dont you set your bot for 500 then? 2011-11-20T09:55:02 *** rajanaresh has quit IRC (Ping timeout: 258 seconds) 2011-11-20T09:55:56 this sucks 2011-11-20T09:56:03 i cant properly manage versions of my bot 2011-11-20T09:56:12 i do 2011-11-20T09:56:20 but it's dumb that you have to 2011-11-20T09:56:20 how? 2011-11-20T09:56:31 (set my bot for 500ms max) 2011-11-20T09:56:34 k 2011-11-20T09:56:42 versions of my bot? i tag them in git 2011-11-20T09:56:46 my editor(kdevelop) uses sessions 2011-11-20T09:57:07 and if i copy a version to another directory i have to make a new session for it 2011-11-20T09:57:15 i just discovered i completely broke my bot sometime a couple days ago 2011-11-20T09:57:16 i like emacs's desktop-load much better 2011-11-20T09:57:23 but it still works pretty well 2011-11-20T09:57:34 completely? 2011-11-20T09:57:48 i'm not even sure wtf is wrong, the output just looks wrong 2011-11-20T09:58:07 is your bot that long? 2011-11-20T09:58:17 i think i can still keep it in mind 2011-11-20T09:58:21 mine 2011-11-20T09:59:27 however im not satisfied with their level of inteligence 2011-11-20T09:59:37 im gonna try to make use of my ant class 2011-11-20T09:59:56 make them more autonomus 2011-11-20T10:02:18 Watch out for ant revolt 2011-11-20T10:03:27 oh, they will rebel against their god? 2011-11-20T10:03:54 i have ^c the ultimate weapon 2011-11-20T10:05:41 *** jasox has joined #aichallenge 2011-11-20T10:14:37 *** raemde_ has joined #aichallenge 2011-11-20T10:18:19 *** raemde has quit IRC (Ping timeout: 244 seconds) 2011-11-20T10:18:25 *** raemde_ is now known as raemde 2011-11-20T10:23:15 *** cyphase has quit IRC (Ping timeout: 260 seconds) 2011-11-20T10:27:43 *** bergmark has joined #aichallenge 2011-11-20T10:35:32 *** cyphase has joined #aichallenge 2011-11-20T10:38:02 *** Blkt has joined #aichallenge 2011-11-20T10:38:04 you mean ~Ant :p 2011-11-20T10:39:43 *** Hexren has joined #aichallenge 2011-11-20T10:39:48 *** Hexren has left #aichallenge 2011-11-20T10:42:54 *** mleise has joined #aichallenge 2011-11-20T10:43:02 *** ALplus has joined #aichallenge 2011-11-20T10:43:37 *** retybok_ has joined #aichallenge 2011-11-20T10:44:38 *** Antimony has joined #aichallenge 2011-11-20T10:47:42 *** Harpyon_ has quit IRC (Quit: Computer has gone to sleep.) 2011-11-20T10:47:52 *** u_ has joined #aichallenge 2011-11-20T10:48:50 *** hacklash has joined #aichallenge 2011-11-20T10:50:26 *** Blkt has quit IRC (Read error: Connection reset by peer) 2011-11-20T10:50:45 *** Palmik has joined #aichallenge 2011-11-20T10:57:18 how can i run locally my 5 versions of bots in many games, randomly, with random maps, and get ranking at the end? 2011-11-20T11:02:21 i was contemplating the same 2011-11-20T11:04:14 set up a local tcp server and connect with all 5 bots? 2011-11-20T11:05:14 too cpu intensive for my lappy 2011-11-20T11:06:23 modify the server to run all clients one by one instead of simultaniously? 2011-11-20T11:06:38 scripting... 2011-11-20T11:06:48 Yexo: duh, option with local tcp is not bad 2011-11-20T11:10:53 Fluxid: you are probably aware of this, i'll say it anyway: ranking your bot against other versions of itself could lead to some pretty heavy bias. you're far from guaranteed to actually find the "best" bot (as in: the one that would do best on the official server) this way. 2011-11-20T11:11:28 Yeah, you would need some other bots in the mix 2011-11-20T11:11:59 But you are already running the tcp server, soo... 2011-11-20T11:14:42 bqf: i want to see how small changes in my bot change its behaviour in different situations. just running it locally manually via playgame.py is tiring 2011-11-20T11:25:54 *** Harpyon_ has joined #aichallenge 2011-11-20T11:32:52 *** chridi has joined #aichallenge 2011-11-20T11:32:52 *** chridi has joined #aichallenge 2011-11-20T11:36:14 *** Palmik has quit IRC (Ping timeout: 244 seconds) 2011-11-20T11:39:37 *** Palmik has joined #aichallenge 2011-11-20T11:44:33 *** amstan has joined #aichallenge 2011-11-20T11:44:33 *** ChanServ sets mode: +o amstan 2011-11-20T11:54:12 *** praveen has quit IRC (Ping timeout: 265 seconds) 2011-11-20T11:54:55 *** flag has joined #aichallenge 2011-11-20T12:01:27 *** yoden has joined #aichallenge 2011-11-20T12:07:43 *** raemde_ has joined #aichallenge 2011-11-20T12:07:45 *** bergmark has quit IRC (Ping timeout: 260 seconds) 2011-11-20T12:09:00 *** NightExcessive has joined #aichallenge 2011-11-20T12:11:17 *** raemde has quit IRC (Ping timeout: 245 seconds) 2011-11-20T12:11:28 *** retybok_ has quit IRC (Ping timeout: 244 seconds) 2011-11-20T12:18:10 *** praveen has joined #aichallenge 2011-11-20T12:20:40 *** jrogge_ has joined #aichallenge 2011-11-20T12:25:43 *** jasox has left #aichallenge ("Leaving") 2011-11-20T12:26:24 *** okan_ has joined #aichallenge 2011-11-20T12:28:28 *** treeform has joined #aichallenge 2011-11-20T12:32:22 *** Antvolution has quit IRC (Quit: Leaving.) 2011-11-20T12:35:15 *** u__ has joined #aichallenge 2011-11-20T12:37:57 *** u_ has quit IRC (Ping timeout: 245 seconds) 2011-11-20T12:37:57 *** u__ is now known as u_ 2011-11-20T12:49:06 *** Antimony has quit IRC (Ping timeout: 276 seconds) 2011-11-20T12:51:02 *** retybok_ has joined #aichallenge 2011-11-20T12:56:05 *** roflmao has joined #aichallenge 2011-11-20T12:56:21 *** JorgeB has joined #aichallenge 2011-11-20T13:00:01 *** cyphase has quit IRC (Read error: Connection reset by peer) 2011-11-20T13:04:40 *** retybok_ has quit IRC (Ping timeout: 240 seconds) 2011-11-20T13:07:03 *** danielharan has joined #aichallenge 2011-11-20T13:08:20 *** Antimony has joined #aichallenge 2011-11-20T13:08:52 *** danielharan has quit IRC (Remote host closed the connection) 2011-11-20T13:13:00 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-20T13:13:22 *** g0llum has joined #aichallenge 2011-11-20T13:15:27 *** ficode has joined #aichallenge 2011-11-20T13:16:25 *** cyphase has joined #aichallenge 2011-11-20T13:20:21 *** Anilm3 has joined #aichallenge 2011-11-20T13:20:39 Hi 2011-11-20T13:22:53 *** Mazer1010 has joined #aichallenge 2011-11-20T13:23:22 *** roflmao has left #aichallenge 2011-11-20T13:23:58 hi 2011-11-20T13:26:33 *** jstrong has joined #aichallenge 2011-11-20T13:26:53 screen 2011-11-20T13:28:20 *** Tombana has joined #aichallenge 2011-11-20T13:28:36 hi 2011-11-20T13:28:40 *** Antvolution has joined #aichallenge 2011-11-20T13:31:51 any suggestions on how one could debug a mono bot from monodevelop after launching in playgame.py ? 2011-11-20T13:32:09 or at least some best practice for that 2011-11-20T13:40:25 *** retybok_ has joined #aichallenge 2011-11-20T13:41:26 *** roflmao has joined #aichallenge 2011-11-20T13:41:31 hey 2011-11-20T13:41:53 hi 2011-11-20T13:41:58 whatup :P 2011-11-20T13:42:15 reading through the forums and pondering when i should do that aiclass midterm 2011-11-20T13:42:22 fun 2011-11-20T13:42:27 i woke up at 6:30 this morning with the sudden realization that i had a major bug in my combat resolution code 2011-11-20T13:42:53 hah, nice a1k0n 2011-11-20T13:42:56 what was it? 2011-11-20T13:43:26 I do my best work when asleep 2011-11-20T13:43:35 say you have a provisional next state of a..b.a and the left a moves right. i would update b to have two enemies but i wouldn't update the a on the right to be alive again 2011-11-20T13:44:14 ah 2011-11-20T13:44:16 which led to a huge number of mistakes which would be sampled out eventually so it wouldn't matter *that* much but it still is pretty bad 2011-11-20T13:44:26 yeah those can be really hard to catch 2011-11-20T13:44:48 I really need to code in a battle simulation function 2011-11-20T13:44:51 don't have one yet 2011-11-20T13:45:27 Hey everyone, I wrote a script to allow C++ players to use Boost! 2011-11-20T13:45:31 You can find it on the forums 2011-11-20T13:46:00 a script? whoa 2011-11-20T13:46:30 oic 2011-11-20T13:48:16 *** okan_ has quit IRC (Ping timeout: 265 seconds) 2011-11-20T13:53:46 *** yoR has joined #aichallenge 2011-11-20T13:57:20 *** speakeasy has joined #aichallenge 2011-11-20T13:58:11 *** choas has joined #aichallenge 2011-11-20T14:03:29 *** roflmao has quit IRC (Ping timeout: 244 seconds) 2011-11-20T14:03:30 *** jstrong has quit IRC (Ping timeout: 260 seconds) 2011-11-20T14:03:53 hello 2011-11-20T14:03:53 *** raemde_ is now known as raemde 2011-11-20T14:04:10 *** iglo has joined #aichallenge 2011-11-20T14:05:09 I've seen people show pictures of their ants with all kinds of extra visual info, like color maps that show distance and stuff. Did they customize the visualiser to show all that info or did they take a screenshot and modify it afterwards? 2011-11-20T14:05:27 *** roflmao has joined #aichallenge 2011-11-20T14:05:30 Tombana: http://forums.aichallenge.org/viewtopic.php?f=25&t=1633 2011-11-20T14:05:31 would such a special visualiser (that can show debug info) be available? 2011-11-20T14:05:33 oh ok 2011-11-20T14:05:50 thnx 2011-11-20T14:05:59 it's a modified version of the visualizer, you cann add info and all the viz stuff in it easily 2011-11-20T14:06:51 awesomeness all over the place 2011-11-20T14:06:57 this is gonna help so much 2011-11-20T14:07:30 yup, it's very helpful 2011-11-20T14:08:09 i think in the last post, some guy summarized the available commands. there is really no documentation besides the forum as i see it 2011-11-20T14:08:22 *** retybok_ has quit IRC (Ping timeout: 245 seconds) 2011-11-20T14:08:27 ok. and where's the latest version? 2011-11-20T14:08:31 cuz i see all kinds of links to the git repo 2011-11-20T14:08:37 but not sure which one i should get 2011-11-20T14:10:39 Pick the vis_overlay branch 2011-11-20T14:10:53 and then the files that j-h-a has modified 2011-11-20T14:11:03 https://github.com/j-h-a/aichallenge/tree/vis_overlay/ants 2011-11-20T14:11:39 just replace your tools/ants.py and tools/visualizer/* with the files from there 2011-11-20T14:14:41 my new code is no good, damn 2011-11-20T14:15:28 *** speakeasy has quit IRC (Ping timeout: 240 seconds) 2011-11-20T14:16:17 is it evil? 2011-11-20T14:16:56 ] 2011-11-20T14:19:48 Thanks HaraKiri 2011-11-20T14:20:23 *** choas has quit IRC (Ping timeout: 252 seconds) 2011-11-20T14:20:47 *** yoR has quit IRC (Quit: Page closed) 2011-11-20T14:25:45 *** Surya has joined #aichallenge 2011-11-20T14:31:39 *** Mazer1010 has quit IRC (Quit: Page closed) 2011-11-20T14:31:59 *** Mazer1010 has joined #aichallenge 2011-11-20T14:35:42 *** Antimony has quit IRC (Ping timeout: 276 seconds) 2011-11-20T14:37:45 *** retybok_ has joined #aichallenge 2011-11-20T14:40:31 *** Conorach_ has joined #aichallenge 2011-11-20T14:42:31 *** Conorach has quit IRC (Ping timeout: 252 seconds) 2011-11-20T14:43:30 *** retybok_ has quit IRC (Ping timeout: 276 seconds) 2011-11-20T14:45:04 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-11-20T14:45:34 *** xathis has quit IRC () 2011-11-20T14:45:49 *** smiley1983 has quit IRC (Ping timeout: 252 seconds) 2011-11-20T14:47:53 *** smiley1983 has joined #aichallenge 2011-11-20T14:50:04 *** sapy has joined #aichallenge 2011-11-20T14:50:13 *** retybok_ has joined #aichallenge 2011-11-20T14:52:21 *** sapy has left #aichallenge 2011-11-20T14:55:25 *** Accoun has quit IRC () 2011-11-20T15:02:55 *** NoxiaZ^ has joined #aichallenge 2011-11-20T15:07:52 *** willem has quit IRC (Quit: leaving) 2011-11-20T15:09:39 *** Accoun has joined #aichallenge 2011-11-20T15:13:49 *** Surya has quit IRC (Ping timeout: 265 seconds) 2011-11-20T15:14:39 *** dom7b5 has quit IRC (Quit: Ex-Chat) 2011-11-20T15:18:47 *** cbad has quit IRC (Read error: Connection reset by peer) 2011-11-20T15:21:26 *** Anilm3 has quit IRC (Quit: leaving) 2011-11-20T15:22:30 *** okan_ has joined #aichallenge 2011-11-20T15:23:51 *** cbad has joined #aichallenge 2011-11-20T15:23:51 *** cbad has joined #aichallenge 2011-11-20T15:26:35 *** maria has joined #aichallenge 2011-11-20T15:28:45 *** Anilm3 has joined #aichallenge 2011-11-20T15:29:07 Would manhattan distance work well on Astar? 2011-11-20T15:29:25 as the heuristic? yes, just make sure to wrap it properly 2011-11-20T15:29:47 pathfinding for each ant to their target will probably be much too slow though 2011-11-20T15:30:22 I have a working version that works quite fast but I just wanted to know if manhattan was ok 2011-11-20T15:31:09 you just can't overestimate the distance in the heuristic if you want the shortest path 2011-11-20T15:31:38 I'll have to see if it's wrapped properly 2011-11-20T15:32:30 *** retybok_ has quit IRC (Ping timeout: 260 seconds) 2011-11-20T15:40:21 *** NightExcessive has quit IRC (Quit: Leaving) 2011-11-20T15:41:23 *** NightExcessive has joined #aichallenge 2011-11-20T15:48:16 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-11-20T15:49:06 *** Mazer1010 has quit IRC (Ping timeout: 265 seconds) 2011-11-20T15:51:10 *** besh has joined #aichallenge 2011-11-20T15:56:25 *** mcstar has left #aichallenge ("WeeChat 0.3.6") 2011-11-20T16:01:07 *** delt0r_ has joined #aichallenge 2011-11-20T16:02:23 *** Palmik has quit IRC (Remote host closed the connection) 2011-11-20T16:03:13 *** dom7b5 has joined #aichallenge 2011-11-20T16:03:47 *** praveen has quit IRC (Quit: Page closed) 2011-11-20T16:06:25 *** xar0l has joined #aichallenge 2011-11-20T16:09:50 *** helkafen has joined #aichallenge 2011-11-20T16:10:12 I finally settled on how to collect food by maximizing vision. 2011-11-20T16:10:21 Here is a video of it http://www.youtube.com/watch?v=YmIlhMSGX_c 2011-11-20T16:10:47 *** Conorach has joined #aichallenge 2011-11-20T16:11:28 That is against a dummy opponent, but you get the idea. Let me know what you think 2011-11-20T16:12:10 besh: thats great ;) 2011-11-20T16:12:47 really. I struggled a lot on how to do it. That is the best I can come up with :) 2011-11-20T16:13:17 yeah, that seems to work really well 2011-11-20T16:13:18 good idea but they are defenceless i guess 2011-11-20T16:13:37 *** mleise has quit IRC (Read error: Connection reset by peer) 2011-11-20T16:13:48 I try to put my ants vision radius away. 2011-11-20T16:13:51 *** Conorach_ has quit IRC (Ping timeout: 276 seconds) 2011-11-20T16:15:05 I think I've finally fixed the bugs in my defense branch 2011-11-20T16:15:25 http://ants.fluxid.pl/replay.7836 2011-11-20T16:15:35 maybe sort your ants just like groups for example 4 of them stay together :P 2011-11-20T16:16:01 strcat_homeland has my new defense code, and it seems to work a lot better than my complete lack of defense before 2011-11-20T16:16:26 my other attempts at defense actually kept ants near my hill so my bot wasn't as good at foraging/exploration 2011-11-20T16:22:28 thestinger, I've been wondering if it would be better to prevent enemy ants from even finding your hill in the first place. 2011-11-20T16:22:34 even if it means suiciding one ant. 2011-11-20T16:22:49 most people will have some offensive code to attack enemy hills in force. 2011-11-20T16:22:54 Conorach: that might be very hard in maze maps 2011-11-20T16:23:22 Conorach: yeah, once the hill is found it's usually screwed 2011-11-20T16:24:00 Enemies explore so they will eventually find your hill. 2011-11-20T16:24:10 this code relies on a constant supply of new ants for defense (if I'm losing at foraging defense isn't going to help) and just tries to keep away the enemy as long as possible 2011-11-20T16:24:12 but in that fight it's amazing how you keep them out for a lon time :) 2011-11-20T16:24:13 so I can keep getting new ants 2011-11-20T16:25:26 besh, but it's different if they find your hill right? they will send their ants to attack it. Of course when there is only one place in the map unexplored, they will come in force as well, but that's usually a lot better than around the 15 ant mark, when two enemies see your hill and go all-in ;) 2011-11-20T16:25:27 hehe 2011-11-20T16:27:03 I always try to keep twice the number of enemy ants around each of my hills. 2011-11-20T16:27:38 It works most of the time unless some crazy bot sends all of its ants before I have time to regroup. 2011-11-20T16:27:55 that's pretty much what my defense does now 2011-11-20T16:28:39 Also always keep one ant close to your hill just to have vision. 2011-11-20T16:29:00 yeah, I need to do that 2011-11-20T16:29:08 *** SparrowG has quit IRC (Quit: Connection closed for inactivity) 2011-11-20T16:29:33 they can gather food in the area around the hill so I don't see any disadvantage to 1 "guard" per hill 2011-11-20T16:29:51 *** chridi has quit IRC (Ping timeout: 248 seconds) 2011-11-20T16:30:48 well in the early stages it could cost some amount of exploration 2011-11-20T16:30:57 *** retybok_ has joined #aichallenge 2011-11-20T16:31:03 you don't have to do it in the very early game though 2011-11-20T16:31:33 yes, it is good also in those maps where hills are placed close to one another. 2011-11-20T16:32:11 yeah, the game specification has a rule about minimum hill distance for map generation but the current maps don't seem the follow it 2011-11-20T16:32:20 "Hills must be between 100 and 250 steps away from other hills (details still being worked on)" 2011-11-20T16:33:44 I think those maps need to go for the final games 2011-11-20T16:33:54 http://aichallenge.org/specification.php#map_generators 2011-11-20T16:33:59 yeah, that makes sense 2011-11-20T16:34:28 the maps with hills really close together are way too random/luck-based imo - if 1 or 2 of the bots next to you go in your direction right away, you're screwed 2011-11-20T16:34:31 for testing the bot in starterpackage in c++ with play_one_game_live.sh ; is it correct to load it with "cpp mybotfolder/mybot.cc" ? 2011-11-20T16:34:37 maybe you can fend them off, and then you're both screwed 2011-11-20T16:35:20 yes, especially with multip-player games. 2011-11-20T16:35:23 iglo, no, first compile that, the run the executable with playgame.py 2011-11-20T16:35:46 I mean if it was just 2 you would probably give the same damage as you recieve. 2011-11-20T16:36:08 ok 2011-11-20T16:36:10 yeah, the 2-player maps are really nice 2011-11-20T16:36:11 ill try that :) 2011-11-20T16:36:24 @ g0llum 2011-11-20T16:36:25 iglo: An error occured while trying to show the previous error. 2011-11-20T16:36:51 i heard that anyway. 2011-11-20T16:40:16 *** retybok_ has quit IRC (Ping timeout: 240 seconds) 2011-11-20T16:43:58 nice, the tcp client just leaked a thousand or so sockets in about a minute 2011-11-20T16:44:05 and all of my bots timed out 2011-11-20T16:44:35 OSError: [Errno 24] Too many open files 2011-11-20T16:45:05 I should just wrap the script with another script so it doesn't keep running for so long 2011-11-20T16:45:52 *** mj41 has quit IRC (Ping timeout: 245 seconds) 2011-11-20T16:47:21 *** thestinger has quit IRC (Quit: WeeChat 0.3.6) 2011-11-20T16:47:54 thestinger, doing the looping from another script, thus starting a fresh python every time, will probably help 2011-11-20T16:48:22 ok 2011-11-20T16:48:23 done 2011-11-20T16:48:32 but how to test it? 2011-11-20T16:48:57 but finding the leak, and send an issue, would be even more awesome. 2011-11-20T16:49:31 *** xar0l has quit IRC (Ping timeout: 265 seconds) 2011-11-20T16:49:33 Id try to call playgame.py and the map is shown but the game never start 2011-11-20T16:50:25 iglo, paste your cmdline, someone'll find the flaw 2011-11-20T16:51:15 #!/usr/bin/env sh 2011-11-20T16:51:15 ./playgame.py -So --player_seed 42 --end_wait=0.25 --verbose --log_dir game_logs --turns 1000 --map_file maps/maze/maze_02p_01.map "$@" \ 2011-11-20T16:51:15 "python mybot/MyBot.py" \ 2011-11-20T16:51:15 "python sample_bots/python/LeftyBot.py" | 2011-11-20T16:51:15 java -jar visualizer.jar 2011-11-20T16:51:55 I don't think the $@ belongs there 2011-11-20T16:51:56 *** nle85 has quit IRC (Ping timeout: 265 seconds) 2011-11-20T16:52:16 iglo, maze_02p_01.map needs 2 players, append them at the end 2011-11-20T16:52:52 1 of them your bot, of course 2011-11-20T16:53:00 i copied that line from play_one_game_live.sh 2011-11-20T16:53:08 *** needsch has joined #aichallenge 2011-11-20T16:53:25 what is the output on the commandline? 2011-11-20T16:53:36 it's all right , just add the players 2011-11-20T16:53:46 Reading live stream from stdin... 2011-11-20T16:53:46 Loading visualizer... 2011-11-20T16:53:46 Creating canvas... 2011-11-20T16:53:51 g0llum: he did, MyBot.py / LeftyBot.py 2011-11-20T16:53:59 the visualizier starts but not the game 2011-11-20T16:54:14 I see the amze and 2 ants 2011-11-20T16:54:48 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-20T16:55:13 *** g0llum has joined #aichallenge 2011-11-20T16:56:34 and if you run it without the visualizer? 2011-11-20T16:56:38 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-20T16:56:49 *** g0llum has joined #aichallenge 2011-11-20T16:57:12 wait 2011-11-20T16:59:01 #!/usr/bin/env sh 2011-11-20T16:59:01 ./playgame.py -So --player_seed 42 --end_wait=0.25 --verbose --log_dir game_logs --turns 1000 --map_file maps/maze/maze_02p_01.map "$@" \ 2011-11-20T16:59:01 "python mybot/MyBot.py" \ 2011-11-20T16:59:01 "python sample_bots/python/LeftyBot.py" 2011-11-20T16:59:03 like that? 2011-11-20T16:59:08 yes 2011-11-20T16:59:12 i start it in a sh script 2011-11-20T16:59:25 that's fine 2011-11-20T16:59:38 what is the output on the commandline like now? 2011-11-20T16:59:39 # score 0 3 2011-11-20T16:59:39 # status crashed survived 2011-11-20T16:59:39 # playerturns 0 0 2011-11-20T16:59:39 # waiting 0.25 seconds for bots to process end turn 2011-11-20T16:59:50 there you ahve it, your first player crashed 2011-11-20T16:59:54 its long output and that are the last lines 2011-11-20T16:59:57 I still think it's becaseu of that "$@" 2011-11-20T17:00:13 if there was output before it might have worked 2011-11-20T17:00:19 check game_logs/0.replay.html 2011-11-20T17:01:04 *** Fandekasp has quit IRC (Ping timeout: 240 seconds) 2011-11-20T17:01:32 hmm 2011-11-20T17:01:39 its the old replay 2011-11-20T17:04:34 did you try removing the "$@" already? 2011-11-20T17:05:08 sure 2011-11-20T17:05:12 *** ltriant has joined #aichallenge 2011-11-20T17:05:30 # turn 0 stats: [1,1,0] 0 [0,0] - 32 0 None [0,1] [1,1] [0,1] 0 None 2011-11-20T17:05:30 end 2011-11-20T17:05:36 #!/usr/bin/env sh 2011-11-20T17:05:36 ./playgame.py -So --player_seed 42 --end_wait=0.25 --verbose --log_dir game_logs --turns 1000 --map_file maps/maze/maze_02p_01.map\ 2011-11-20T17:05:36 "python mybot/MyBot.py" \ 2011-11-20T17:05:36 "python sample_bots/python/LeftyBot.py" 2011-11-20T17:06:09 maybe I didnt compiled it the right way 2011-11-20T17:06:37 used the makefile from c++ startepack 2011-11-20T17:06:39 from the output you gave it said that the first bot (your one) crashed 2011-11-20T17:06:43 are you sure it didn't? 2011-11-20T17:07:09 id crashed obviosly there is no round played 2011-11-20T17:07:31 so debug your own bot to see why it crashed 2011-11-20T17:07:38 that's got nothing to do with playgame.py perse 2011-11-20T17:08:56 *** Alexer has quit IRC (Read error: Operation timed out) 2011-11-20T17:09:53 *** xar0l has joined #aichallenge 2011-11-20T17:10:46 there is no MyBot.py file only an executable MyBot 2011-11-20T17:11:39 *** NoxiaZ^ has quit IRC (Read error: Connection reset by peer) 2011-11-20T17:13:18 are you using C++ or python? 2011-11-20T17:13:35 if you have an executable named MyBot replace "python mybot/MyBot.py" by "./MyBot" 2011-11-20T17:13:37 c++ 2011-11-20T17:13:48 oh.. ok :D 2011-11-20T17:13:55 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-20T17:14:12 *** g0llum has joined #aichallenge 2011-11-20T17:15:37 *** okan_ has quit IRC (Ping timeout: 265 seconds) 2011-11-20T17:16:00 iglo, just try to tun MyBot straight from the console, without playgame.py. just input: turn 0\n rows 10\n cols 10\n ready\n if it still crashes, you're doing the parsing wrong there. 2011-11-20T17:16:12 *** rajanaresh has joined #aichallenge 2011-11-20T17:17:33 where to enter that input 2011-11-20T17:17:58 *** mj41 has joined #aichallenge 2011-11-20T17:18:07 in other words, if you emulate, what playgame/engine is giving you, you can even debug it witth gdb or similar 2011-11-20T17:18:46 just start our bot and type it 2011-11-20T17:19:19 *** Euphony has joined #aichallenge 2011-11-20T17:19:28 *** besh has quit IRC (Quit: Page closed) 2011-11-20T17:19:46 hm 2011-11-20T17:20:13 turn 0\n rows 10\n cols 10\n ready\n 2011-11-20T17:20:18 ups 2011-11-20T17:20:55 ai_challenge/tools/mybot$ ./MyBot 2011-11-20T17:21:00 after that ? 2011-11-20T17:21:24 yes 2011-11-20T17:21:33 nothing happens 2011-11-20T17:21:46 also look at test_bot.sh 2011-11-20T17:22:02 did you literarily type \n or a newline instead? 2011-11-20T17:22:14 replace all those \n by enters if you didn't do so before 2011-11-20T17:23:38 he said "go" 2011-11-20T17:23:47 that's your bot 2011-11-20T17:23:55 after the go it expects the input for the first turn 2011-11-20T17:24:33 ok 2011-11-20T17:24:36 *** Alexer has joined #aichallenge 2011-11-20T17:24:59 the engine says ready on turn 0, you say go 2011-11-20T17:25:29 so than it should start 2011-11-20T17:25:53 it has already started, after that "go" the bots wait for input from the engein again 2011-11-20T17:26:06 which, if you started your bot from the commandline, is you 2011-11-20T17:27:36 ok got it... I just copied the bot into the tools folder and ran "./myBot" 2011-11-20T17:27:48 now it works... sorry for all that trouble 2011-11-20T17:28:02 *** needsch has left #aichallenge 2011-11-20T17:29:04 welcome to the club! 2011-11-20T17:29:09 *** helkafen has quit IRC (Ping timeout: 265 seconds) 2011-11-20T17:29:29 tomorrow I start the work# 2011-11-20T17:29:35 thank u guys 2011-11-20T17:33:24 guess I should begin to edit the bot.cc file 2011-11-20T17:34:05 contestbot: seen seletskiy 2011-11-20T17:34:05 amstan: seletskiy was last seen in #aichallenge 1 day, 10 hours, 17 minutes, and 36 seconds ago: Minthos: Higher than 50ms? 2011-11-20T17:36:03 *** pqpq has joined #aichallenge 2011-11-20T17:37:40 *** Euphony has quit IRC (Quit: Page closed) 2011-11-20T17:38:28 *** _Lone_Wolf_ <_Lone_Wolf_!LW@bl10-238-206.dsl.telepac.pt> has joined #aichallenge 2011-11-20T17:38:42 hey, my bot has said "0 players are ahead." and "Next game should be any time now." all day now. is it because its deactivated? i thought it was still chosen as an enemy even then though (when my last version was deactivated it still played games at least) 2011-11-20T17:41:55 *** avdg1 has quit IRC (Quit: Leaving.) 2011-11-20T17:41:58 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-11-20T17:42:40 *** avdg has joined #aichallenge 2011-11-20T17:43:22 hi all. I wanted to ask if A* is a good way to collect food if an ant sees one, considering I have all the knowledge of the environment. Is it going to timeout in certain cases? 2011-11-20T17:43:54 rajanaresh: i've heard a lot of people saying it's slow 2011-11-20T17:44:01 A* is good if you care about 1 or a few target only (since its good in concentrating its extension in 1 direction) 2011-11-20T17:44:16 rajanaresh: collaborative diffusion seems like a pupular idea 2011-11-20T17:44:17 if you want more, you probably want bfs 2011-11-20T17:44:23 see forums 2011-11-20T17:45:24 amstan: collaborative diffusion didn't work so well for me 2011-11-20T17:45:29 aww 2011-11-20T17:45:32 ok, bfs then 2011-11-20T17:46:03 collaborative diffusion works pretty well 2011-11-20T17:46:05 its fine for finding food moving around... but you still end up with a mess when you want to add any real strategy 2011-11-20T17:46:23 of course not for the actual fights 2011-11-20T17:46:26 I don't understand how can bfs be any faster than uniform cost search or A* 2011-11-20T17:46:28 not just that 2011-11-20T17:47:02 rajanaresh: A* is fast if you go from one (or a few) source to a single target 2011-11-20T17:47:16 its things like sticking around to defend for example... sometimes enemy ants should attract other times they need to repeal etc 2011-11-20T17:47:56 so change the evaluation function depending on location / turn / etc. 2011-11-20T17:48:11 in which case its not longer really the right tool 2011-11-20T17:48:23 since your hacking in more than it does... 2011-11-20T17:49:26 so what would be the right tool there? 2011-11-20T17:49:47 *** iglo has quit IRC (Remote host closed the connection) 2011-11-20T17:50:55 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-11-20T17:52:29 *** _Lone_Wolf_ <_Lone_Wolf_!LW@bl10-238-206.dsl.telepac.pt> has quit IRC () 2011-11-20T17:52:49 *** HaraKiri has quit IRC () 2011-11-20T17:53:28 *** Harpyon_ has quit IRC (Quit: Computer has gone to sleep.) 2011-11-20T17:53:59 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-11-20T17:54:01 *** mj41 has quit IRC (Ping timeout: 244 seconds) 2011-11-20T18:05:49 *** pqpq has quit IRC (Quit: Page closed) 2011-11-20T18:07:39 aichallenge: Alexandru Stan epsilon * ra1a938a / manager/worker_ssh.py : added sshping, but the timeout thing likes to keep threads running(up to 2 minutes) is ssh server is not present - http://git.io/-59Rvw 2011-11-20T18:07:39 aichallenge: Alexandru Stan epsilon * r7ada95d / (manager/sql.py manager/worker_ssh.py): worker_ssh: removed limit for the sql, discovered ssh timeout(code is now much simpler), ssh pings won't remember ssh keys for the hosts pinged - http://git.io/7epNRw 2011-11-20T18:08:16 *** smiley1983 has quit IRC (Ping timeout: 240 seconds) 2011-11-20T18:09:35 *** smiley1983 has joined #aichallenge 2011-11-20T18:10:52 aichallenge: Alexandru Stan epsilon * re895a0b / manager/worker_ssh.py : worker_ssh: sorted the worker list - http://git.io/61lx0A 2011-11-20T18:11:10 janzert: it's all good now 2011-11-20T18:11:28 exactly as the website list reports 2011-11-20T18:11:37 nice, stanford is doing an online graphical model class: http://www.pgm-class.org/ 2011-11-20T18:13:05 *** besh has joined #aichallenge 2011-11-20T18:13:08 a1k0n: i don't even know which course to pick now anymore 2011-11-20T18:13:10 a1k0n: they're doing a game theory one too although it looks a tad easy :( 2011-11-20T18:13:20 yeah. and ML, and NLP 2011-11-20T18:13:49 amstan: that depends on your interest and background. can you do more than one? 2011-11-20T18:14:12 a1k0n: i'm not sure, i tried to only do one this term because i have other school too 2011-11-20T18:14:18 but next term i have coop 2011-11-20T18:14:25 *** Antimony has joined #aichallenge 2011-11-20T18:21:18 That course looks fairly interesting 2011-11-20T18:21:29 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-11-20T18:23:21 She talks really really fast 2011-11-20T18:26:22 *** Tombana has quit IRC (Quit: Page closed) 2011-11-20T18:31:18 amstan: great 2011-11-20T18:37:06 *** mleyen has quit IRC (Quit: ~ Trillian Astra - www.trillian.im ~) 2011-11-20T18:38:19 *** Jak_o_Shadows has joined #aichallenge 2011-11-20T18:42:07 *** smiley1983 has quit IRC (Ping timeout: 245 seconds) 2011-11-20T18:45:47 *** devrand has joined #aichallenge 2011-11-20T18:46:47 Guys... anybody know how the server compiles C/C++ entries? Does it use the makefile you supply it? 2011-11-20T18:48:02 It doesn't use a makefile 2011-11-20T18:49:12 https://github.com/aichallenge/aichallenge/blob/epsilon/worker/compiler.py 2011-11-20T18:49:15 thanks... any idea what compiler settings it's using then? I was playing around w/ different optimizations and found that the unroll loops command greatly hurts 2011-11-20T18:49:19 *** mikewintermute has quit IRC (Quit: mikewintermute) 2011-11-20T18:49:20 thanks 2011-11-20T18:49:34 look aroundline 308 2011-11-20T18:49:39 *** smiley1983 has joined #aichallenge 2011-11-20T18:50:04 doh... yeah, that funroll loops flag really hurts :/ 2011-11-20T18:50:14 k, thanks for the help though 2011-11-20T18:50:20 it uses the file extension to decide language and then just compiles all the code it finds 2011-11-20T18:50:51 *** Redgis has quit IRC (Ping timeout: 244 seconds) 2011-11-20T18:51:41 I guess it makes sense... actually really shocked they didn't use the march flag 2011-11-20T18:53:25 well I think the idea is to be conservative and break the fewest things. 2011-11-20T18:53:50 yeah... I don't know if I should bring up the loops (as well as -03 actually) 2011-11-20T18:53:51 although I guess -O3 and is not that really 2011-11-20T18:54:04 gcc documentation recommends not using it, loll 2011-11-20T18:54:21 actually the unroll loops thing too 2011-11-20T18:54:29 well thankfully the compiler for my language has no optimisation 2011-11-20T18:54:33 it cuts my (local) performance almost in half on a subroutine 2011-11-20T18:54:52 can you disable it with a pragma or something 2011-11-20T18:55:05 hmm... maybe 2011-11-20T18:55:14 not sure where priority would go... worth taking a look into 2011-11-20T18:55:22 *** pairofdice has quit IRC (Quit: leaving) 2011-11-20T18:55:35 you using python? 2011-11-20T18:55:40 Go 2011-11-20T18:55:47 awesome 2011-11-20T18:55:57 thought about that, but it seemed the whole point of go was multi-threading 2011-11-20T18:56:05 kinda' shooting yourself in the foot using it, aren't you? 2011-11-20T18:56:21 well.. .tying a hand behind your back - wrong expression 2011-11-20T18:56:21 It just really pleasant to work in. 2011-11-20T18:56:47 eh - that's good I guess 2011-11-20T18:58:01 only toyed with it myself 2011-11-20T18:58:01 and seems to be plenty fast enough. I did have to look at the emitted assembly code a few times for inner loop things 2011-11-20T18:58:01 seems more pleasant than C, but not as familiar with it either 2011-11-20T18:58:01 lol - way too much work on an x86 arch to do that 2011-11-20T19:00:18 damn - compiler options specified on the command line can override compiler options used in #pragma options 2011-11-20T19:01:20 devrand: why do you need march? 2011-11-20T19:01:35 don't need march... just want to get rid of funroll loops really 2011-11-20T19:01:51 testing a new bot locally, the funroll loops totally pwns the script 2011-11-20T19:02:06 g++ documentation recommends not using it (or -03 for that matter) 2011-11-20T19:02:07 who added them anyway? 2011-11-20T19:02:29 beats me... just a lowly player here :) 2011-11-20T19:03:32 *** Antimony has quit IRC (Ping timeout: 260 seconds) 2011-11-20T19:03:59 if it gets put up to a vote though... I'd recommend -02 and -march=native for gcc/g++ compiling... definitely the best you can do while being generic and conservative 2011-11-20T19:04:21 *** ChrisH_ has joined #aichallenge 2011-11-20T19:04:46 janzert: ping 2011-11-20T19:05:37 actually you might get it to not unroll loops with a limit declared volatile 2011-11-20T19:06:22 maybe... but with -03 all bets are probably off anyway 2011-11-20T19:06:41 will compiler respect a volatile then? 2011-11-20T19:06:48 I expect it will. 2011-11-20T19:07:03 @later tell janzert game #110490 and #110491 were exact duplicates 2011-11-20T19:07:03 ChrisH_: OK 2011-11-20T19:07:03 its pretty important for hardware stuff 2011-11-20T19:09:19 helps, but still about 175% running time compared to not having funroll loops at all 2011-11-20T19:09:28 better than the 210% ish 2011-11-20T19:10:11 bleh, thanks for the help... I'll be afkish. Hate compilers 2011-11-20T19:10:17 ChrisH_ not exactly the same - the player seed was not the same . 2011-11-20T19:10:31 *** delt0r_ has quit IRC (Read error: Connection reset by peer) 2011-11-20T19:10:45 btw ChrisH they covered that before... supposedly some small percent of games are duplicates 2011-11-20T19:10:52 bugnuts: no, but the scheduler isn't supposed to pick the same two players on the same map in consecutive games. 2011-11-20T19:11:08 an error that's too hard to debug.. happens due to concurrency issues (multiple servers) 2011-11-20T19:11:10 fair enough. 2011-11-20T19:11:17 devrand: yes, I am aware of that as well, but thought they would want to know when it happens 2011-11-20T19:11:32 kk 2011-11-20T19:12:23 thanks for the comments though, both of you, it's good that others help when the admins aren't online. :) 2011-11-20T19:15:25 *** besh has quit IRC (Quit: Page closed) 2011-11-20T19:21:35 *** replore has joined #aichallenge 2011-11-20T19:31:11 *** ficode has quit IRC (Ping timeout: 248 seconds) 2011-11-20T19:35:10 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-11-20T19:40:53 *** Antimony has joined #aichallenge 2011-11-20T19:42:03 *** NightExcessive has quit IRC (Quit: Leaving) 2011-11-20T19:43:47 *** Antimony has quit IRC (Client Quit) 2011-11-20T19:44:07 *** devrand has quit IRC (Quit: Page closed) 2011-11-20T19:44:38 *** CowTipperVirus has joined #aichallenge 2011-11-20T19:45:03 o/ any organizers on atm? 2011-11-20T19:46:42 something "interesting" happened here... looks like a server glitch, but I want them to look anyways: http://aichallenge.org/profile.php?user=8633 2011-11-20T19:46:45 doh 2011-11-20T19:46:54 http://aichallenge.org/visualizer.php?game=110604 2011-11-20T19:48:43 @later Amstan: http://aichallenge.org/visualizer.php?game=110604 would you mind taking a look at this? Server glitch? 2011-11-20T19:48:44 CowTipperVirus: User error -- Replace user. 2011-11-20T19:48:48 the servers have been really laggy 2011-11-20T19:49:00 doh... what's Amstan's handle? 2011-11-20T19:49:00 the amount of CPU you get for setup is really uneven 2011-11-20T19:49:13 CowTipperVirus: @later tell 2011-11-20T19:49:24 lol... umm hey 2011-11-20T19:49:55 CowTipperVirus: submit an issue please 2011-11-20T19:50:12 trhough git? 2011-11-20T19:50:25 CowTipperVirus: github 2011-11-20T19:50:38 ok. thanks 2011-11-20T19:54:24 *** NightExcessive has joined #aichallenge 2011-11-20T19:56:15 *** NightExcessive has quit IRC (Client Quit) 2011-11-20T19:56:41 *** NightExcessive has joined #aichallenge 2011-11-20T19:58:55 *** Anilm3 has joined #aichallenge 2011-11-20T19:59:20 which tcp server is the people using more? 2011-11-20T19:59:45 @tcp 2011-11-20T19:59:45 amstan: tcp could be http://ants.fluxid.pl/howto. 2011-11-20T19:59:50 that's pretty popular ^ 2011-11-20T19:59:53 or tcPants.com 2011-11-20T19:59:59 ok 2011-11-20T20:00:03 thanks 2011-11-20T20:02:59 *** NightExcessive has quit IRC (Changing host) 2011-11-20T20:02:59 *** NightExcessive has joined #aichallenge 2011-11-20T20:03:02 *** fpcfan has quit IRC (Ping timeout: 252 seconds) 2011-11-20T20:06:17 *** b0rder has joined #aichallenge 2011-11-20T20:14:04 *** fpcfan has joined #aichallenge 2011-11-20T20:17:56 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-11-20T20:25:01 *** JorgeB has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-11-20T20:28:41 *** smiley1983 has quit IRC (Ping timeout: 258 seconds) 2011-11-20T20:29:24 *** smiley1983 has joined #aichallenge 2011-11-20T20:35:21 *** roflmao has quit IRC (Quit: Lost terminal) 2011-11-20T20:46:40 *** fpcfan has quit IRC (Ping timeout: 260 seconds) 2011-11-20T20:52:05 *** JorgeB has joined #aichallenge 2011-11-20T20:57:03 *** besh has joined #aichallenge 2011-11-20T21:04:26 *** fpcfan has joined #aichallenge 2011-11-20T21:05:30 *** besh has quit IRC (Quit: Page closed) 2011-11-20T21:08:48 *** fpcfan has quit IRC (Ping timeout: 244 seconds) 2011-11-20T21:09:36 *** fpcfan has joined #aichallenge 2011-11-20T21:13:23 *** bmh has joined #aichallenge 2011-11-20T21:14:17 *** smiley1983 has quit IRC (Quit: leaving) 2011-11-20T21:46:10 *** bmh has quit IRC (Quit: Page closed) 2011-11-20T21:46:20 *** yoden has quit IRC (Quit: Leaving.) 2011-11-20T21:53:09 *** heathkit has quit IRC (Remote host closed the connection) 2011-11-20T21:55:49 *** ibdknox has joined #aichallenge 2011-11-20T22:19:17 *** Mazer1010 has joined #aichallenge 2011-11-20T22:20:03 *** ibdknox has quit IRC (Remote host closed the connection) 2011-11-20T22:27:03 *** dlila has quit IRC (Quit: Leaving) 2011-11-20T22:28:11 *** snthaoeu has joined #aichallenge 2011-11-20T22:40:49 *** Antvolution1 has joined #aichallenge 2011-11-20T22:43:11 *** Antvolution has quit IRC (Ping timeout: 248 seconds) 2011-11-20T22:50:52 *** Mazer1010 has quit IRC (Quit: Page closed) 2011-11-20T23:13:21 *** ibdknox has joined #aichallenge 2011-11-20T23:29:39 *** thestinger has joined #aichallenge 2011-11-20T23:38:06 *** Harpyon has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-11-20T23:42:36 *** jstrong has joined #aichallenge 2011-11-20T23:42:40 *** jstrong is now known as roflmao 2011-11-20T23:46:52 *** ibdknox has quit IRC (Remote host closed the connection) 2011-11-20T23:48:29 *** ibdknox has joined #aichallenge 2011-11-20T23:48:47 *** ibdknox has quit IRC (Remote host closed the connection) 2011-11-20T23:55:01 *** Antvolution1 has left #aichallenge 2011-11-20T23:55:32 *** TheLinker has quit IRC (Remote host closed the connection)