2011-12-11T00:01:41 McLeopold: btw, the change to make seed selection and insert one statement did help but not eliminate the problem, basically went from 17-20% down to 9-12% 2011-12-11T00:02:10 I'll try and put together an external script to keep a buffer of matches soon 2011-12-11T00:03:13 the only problem with that is we are getting close to the point where a pairing takes longer to produce than the speed we're using them 2011-12-11T00:05:21 janzert: thanks for the help, i'll check the code you emailed me to see what orders it outputs 2011-12-11T00:05:37 janzert: ok 2011-12-11T00:06:46 *** zyberkiddy has quit IRC (Remote host closed the connection) 2011-12-11T00:09:21 *** Paradoxial has quit IRC (Quit: :) 2011-12-11T00:11:56 *** Antimony has joined #aichallenge 2011-12-11T00:26:48 *** besh has quit IRC (Quit: Page closed) 2011-12-11T00:45:45 having been smacked around by bj_v5, I think it's time for combat code. Any suggestions? 2011-12-11T01:05:11 bmh: minimax :) 2011-12-11T01:06:26 antimatroid1: I'm going to stick to monte carlo techniques 2011-12-11T01:06:55 do you know how i'm doing minimax?= 2011-12-11T01:07:08 it's like taking a tiny sample of the possible moves and running minimax on those 2011-12-11T01:07:17 how about this: Consider all ants within combat radius 2. Pick an ant and tell everyone to move toward or away from that ant 2011-12-11T01:07:31 not very strategic :P 2011-12-11T01:07:41 and i don't like the combat + 2 thing 2011-12-11T01:07:48 why? 2011-12-11T01:07:56 a..%b and a%..b 2011-12-11T01:07:59 yeah yeah.. 2011-12-11T01:08:12 it isn't going to produce bad results 2011-12-11T01:08:23 it could just be more expensive than optimal 2011-12-11T01:09:22 something i thinks worthwhile doing is working out the maximum enemies each location your ants can move into could possibly be within battle range next turn 2011-12-11T01:09:30 then minimise that or minimise positive that 2011-12-11T01:09:46 that's a bit of a mouthful :P 2011-12-11T01:10:31 I really ought to have studied AI in university. 2011-12-11T01:10:44 To all the kids out there: Take a machine learning class. You'll regret it if you don't. 2011-12-11T01:10:57 the "computing school" at my university is a joke, i would have done better taking more maths classes 2011-12-11T01:11:10 i did both the ai and machine learning classes, they were a joke :P 2011-12-11T01:11:24 i'm very tempted to do that stanford ml class though 2011-12-11T01:11:40 where are you Sydney? NSW? 2011-12-11T01:11:46 tasmania 2011-12-11T01:11:47 utas 2011-12-11T01:11:52 down down under! 2011-12-11T01:12:00 aha yeah, the deep south 2011-12-11T01:15:33 I certainly talk to more Australians on the internet than when I was living in the UK 2011-12-11T01:16:28 smiley1983 and i are always up at odd hours anyway 2011-12-11T01:35:18 antimatroid1: how's your bot coming along? 2011-12-11T01:35:29 ChrisH_: not amazingly 2011-12-11T01:35:45 i'm just trying to get it functional to submit at this point :P 2011-12-11T01:36:00 and thinking about my battle stuff 2011-12-11T01:36:40 I will be interested to see how your battle code does when you get it going. 2011-12-11T01:37:11 ie. when evaluating a state, i'll add enemyDeaths-friendDeaths to the payoff, but i think something like 10*(eDeaths-fDeaths) + eDeaths/(fDeaths+1) might work better? 2011-12-11T01:37:43 ie. (eDeaths, fDeaths), (2,1) is better than (3,2) right? 2011-12-11T01:38:39 yeah, 2,1 is better than 3,2 2011-12-11T01:38:59 if you can achieve that each time there are casualties you come out with more in the end 2011-12-11T01:39:18 but 3,2 is better than a stalemate, so it's a balancing act 2011-12-11T01:39:35 between making progress and being too conservative 2011-12-11T01:41:46 yeah i know, just trying to work out how to order the outcomes properly 2011-12-11T01:41:55 i'm going to add in a third set of moves for each set of ants too 2011-12-11T01:42:05 which doesn't move into the attack radius of any current enemy locations 2011-12-11T01:42:14 *** dvladim has joined #aichallenge 2011-12-11T01:47:33 http://paste.aichallenge.org/Hf3RV/ 2011-12-11T01:48:55 *** taeke has joined #aichallenge 2011-12-11T01:49:12 Hi 2011-12-11T01:50:06 How does this work? 2011-12-11T01:50:27 Is there anybody out there? 2011-12-11T01:50:31 hi 2011-12-11T01:50:53 Hi Leopold 2011-12-11T01:51:28 I am looking for some answer about the current games on the servers 2011-12-11T01:51:40 do you have a bot submitted? 2011-12-11T01:51:42 i see strange behavier 2011-12-11T01:51:59 my bot playing two games on the same time 2011-12-11T01:52:17 Yes i have a bot up and running 2011-12-11T01:53:09 so, when scheduling games, often similar bots are chosen in a row, and you are more likely to be pulled into a games during that time 2011-12-11T01:54:13 so you can end up playing in two games at the same time? 2011-12-11T01:54:18 sure 2011-12-11T01:54:31 i don't mind but had not seen that before 2011-12-11T01:54:37 we have 25 servers now 2011-12-11T01:54:41 *** dvladim has quit IRC (Ping timeout: 252 seconds) 2011-12-11T01:54:50 each with 2-10 players 2011-12-11T01:55:12 and somehow i don't regonise the behavier of my bot 2011-12-11T01:55:16 potentially 50-250 players in a game at any one time 2011-12-11T01:55:34 great 2011-12-11T01:55:36 I always feel like that too 2011-12-11T01:55:39 ChrisH_: I posted on the forum about timeouts too 2011-12-11T01:55:44 :) yeahh 2011-12-11T01:55:54 BenJackson: better or worse with the new servers? 2011-12-11T01:55:55 one game where I "timed out" against pguillory I downloaded the input and fed it to the same version locally 2011-12-11T01:55:56 it isn't doing to bad 2011-12-11T01:56:05 earlier turns in that same game took 5x longer to compute than the one that "timed out" 2011-12-11T01:56:21 <_flag> BenJackson: This happens to me ocassionally too 2011-12-11T01:56:22 taeke: it looks like you pick random directions and just send ants in a straight line 2011-12-11T01:56:27 nothing in the last 30 or so turns took more than 200ms on my machine, which is slower than the contest machines 2011-12-11T01:56:36 <_flag> Sometimes two players timeout in the same moves 2011-12-11T01:56:41 <_flag> same move* 2011-12-11T01:56:47 <_flag> Which is awfully suspicious 2011-12-11T01:56:49 yeah but that is not the behavier i have locally 2011-12-11T01:57:13 <_flag> For example: http://aichallenge.org/visualizer.php?game=179756&user=31 2011-12-11T01:57:26 ChrisH_: Also re analyzing games, I was saying on irc earlier that my technique is basically to look at the number-of-ants graph and exmine the point where it diverges in favor of the ultimate winner 2011-12-11T01:57:30 and try to see why it happened 2011-12-11T01:57:46 yeah 2011-12-11T01:57:46 taeke: it looks like you submitted leftybot? 2011-12-11T01:58:25 in one test game at home where the map had a hill near a dead end the winning bot ignored the initial food that was deeper in the dead end and went out instead 2011-12-11T01:58:25 but that isn't the case and my result are also not matching lefty bot 2011-12-11T01:58:30 leaving that food for a future ant 2011-12-11T01:58:42 the losing bot moved backwards to the first food 2011-12-11T01:58:45 everything followed from that 2011-12-11T01:58:53 skill of 50 and up with lefty bot? 2011-12-11T01:59:05 I don't know if you saw my conversation with janzert earlier, but I had a game where my bot wasn't moving all of it's ants even early in the game. When I run it with that game input locally the orders are issued, but when janzert ran it on a worker he got the same results as the game replay showed, which is really scary to me. 2011-12-11T01:59:21 wow 2011-12-11T01:59:40 so your thing may not be timing 2011-12-11T01:59:47 is your local system 32 or 64 bit? 2011-12-11T01:59:56 ChrisH_: do you know if we was doing the ec2 instances or the new ones for that test? 2011-12-11T02:00:31 BenJackson: do you ever look at the won at turn? 2011-12-11T02:00:54 no, but I did have one timeout recently that I posted in the forums about and that I think was a server thing, because my code is pretty good about not timing out. 2011-12-11T02:01:46 McLeopold: on the aichallenge site I note it when I'm skimming over games to see how decisive they were 2011-12-11T02:01:47 McLeopold: I don't know which kind of worker janzert tested that on, and I don't know which type the particular game ran on, but it was this game if you are interested: http://aichallenge.org/visualizer.php?game=186986&user=10 2011-12-11T02:02:19 http://aichallenge.org/visualizer.php?game=188704&user=9213 is not lefty bot but http://aichallenge.org/visualizer.php?game=186506&user=9213 looks like lefty bot but is is all version 28 2011-12-11T02:02:33 ChrisH_: the worker id should be in the replay, we can trace that to the worker ip, which would probably let us know the type 2011-12-11T02:03:32 BenJackson: my local system is 32 bit, but Go should make that a non-issue, unless it is due to a difference in the 32 bit or 64 compiler, the language itself doesn't allow implicit conversions between different bit numbers, so it is all explicit. 2011-12-11T02:04:17 oh Go 2011-12-11T02:04:28 I guess a compiler issue is an order of magnitude more likely with Go than with, say, C++ 2011-12-11T02:04:35 yes 2011-12-11T02:04:47 I have to remind myself of that when I use GCC on obscure architectures 2011-12-11T02:05:58 ChrisH_: that game was worker_id 661 at ip 77.95.X.X, which is a new worker 2011-12-11T02:06:00 but the behavior of my bot in the weird game was consistent with it either thinking it had less time left in the turn than reality (but janzert's test pretty much refuted that) or that it was occassionally dropping the last few ant orders on the floor before they get sent. 2011-12-11T02:06:26 and then the next turn it would work 2011-12-11T02:06:42 ChrisH_: so, you have to send a go line after all orders 2011-12-11T02:06:43 did it emit the stale order right at the start of the NEXT turn? 2011-12-11T02:06:51 there was a stretch of about 15 turns (around 130, still early in the game) where several ants just didn't move 2011-12-11T02:07:06 no, the order just didn't happen 2011-12-11T02:07:28 unfortunately the way my bot works it isn't really possible to stop computing orders early 2011-12-11T02:07:36 and my bot doesn't cache any orders, paths or any info about ants between turns either 2011-12-11T02:07:38 how many ways can your bot skip to the go line? 2011-12-11T02:07:47 one 2011-12-11T02:07:54 and given the way I timed out one time on the official site it must have been in a way my bot didn't even run 2011-12-11T02:08:35 ChrisH_: could it be something to do with how the sandboxing code affects the runtime? 2011-12-11T02:08:45 McLeopold could you take a look at http://aichallenge.org/visualizer.php?game=188704&user=9213 and http://aichallenge.org/visualizer.php?game=186506&user=9213 same version but tottaly different behavier the latest does not look like my bot the first does but it is all version 28 2011-12-11T02:08:47 all orders are collected in one structure and held there until there is nothing left to do or time is up, then a three line function loops to send the orders and when that returns the final "go" is sent 2011-12-11T02:08:52 I don't know how feasible it is for someone to run under an equivalent sandbox 2011-12-11T02:10:52 McLeopold: have any of my other games run on server 661? 2011-12-11T02:11:21 ChrisH_: let me check... 2011-12-11T02:11:40 taeke: those are both submission 32527, there is no way we run the wrong bots 2011-12-11T02:13:21 ChrisH_: game ids 186986 and 188450 2011-12-11T02:13:27 the behavior isn't that obviously different 2011-12-11T02:13:52 for taeke, no, especially the first turn pause and bounce after hitting a wall 2011-12-11T02:15:32 McLeopold: 186986 is the weird game, 188450 is my most recent one and there are no apparent problems with that one, in fact my bot dominated that matchup 2011-12-11T02:15:38 ChrisH_: I don't think I see static ants in the second game 2011-12-11T02:15:53 no, it looks perfect to me 2011-12-11T02:16:17 maybe more ants makes you go faster :) 2011-12-11T02:16:17 in fact, I had already checked it but didn't know it was the same server as the other one. 2011-12-11T02:16:24 LOL, I wish 2011-12-11T02:16:53 the only static ants in the second game are late and it is normal for my bot in that version 2011-12-11T02:17:13 too many ants and I hadn't seen the last enemy hill yet 2011-12-11T02:17:19 so they had nothing to do 2011-12-11T02:17:44 I'm wondering if unseeable water would kill my bot 2011-12-11T02:17:59 I think they might all surround the lake... 2011-12-11T02:18:35 trying to explore the middle? 2011-12-11T02:18:49 yeah, I should run a test 2011-12-11T02:19:00 it might break my path finding 2011-12-11T02:19:06 my bot would put a goal in there, but since there is not path to it the ants would ignore it 2011-12-11T02:19:18 I've seen my current bot do stupid things when a stalemate is blocking explorable area 2011-12-11T02:19:49 I actually restructured my pathing so I could do things like avoid stalemate areas 2011-12-11T02:19:56 but never did 2011-12-11T02:19:58 Has anyone seen bots controlling ant spawning by standing on their own hills? 2011-12-11T02:20:01 I think I would need to detect no path spots and mark them water 2011-12-11T02:20:10 no 2011-12-11T02:20:21 but I've seen some maps where it would help 2011-12-11T02:20:52 ChrisH_: I thought Xathis does it 2011-12-11T02:20:55 Yeah, me too, so I've coded it up and it's mostly working, but I haven't run enough tests yet to decide if I'll keep it. 2011-12-11T02:21:05 when do you do it? 2011-12-11T02:21:20 well, that's the interesting part isn't it. 2011-12-11T02:21:20 *** epicmonkey has joined #aichallenge 2011-12-11T02:21:37 I did implement code to keep track of my hive 2011-12-11T02:21:49 but usually if I have ants in my hive I'm fine 2011-12-11T02:21:53 the goal is to spawn ants as close to targets as possible to avoid wasting the time traveling from the "wrong" hill 2011-12-11T02:21:54 I get screwed on maps with no food 2011-12-11T02:22:25 but I also don't want to leave any ants in the hive if it can be avoided 2011-12-11T02:22:39 McLeopold: Ok thanks. It must be me looking at it the wrong way. Thanks for the information. 2011-12-11T02:22:40 and the resulting code started to remind me of the future state tracking I did for PW. 2011-12-11T02:22:58 I didn't realize how bursty food was until I saw how bursty hive depth was and then looked into why 2011-12-11T02:22:59 yeah, I'm totally reusing PW code :) 2011-12-11T02:23:22 *** bmh has quit IRC (Quit: Page closed) 2011-12-11T02:23:49 I like implementing things like hive counting and symmetry detection because it's something I can solve 2011-12-11T02:23:58 whereas balancing my strategy heuristics is open ended 2011-12-11T02:24:16 and currently my tweaking has made a bot which loses 100% of the time to my last bot in some situations 2011-12-11T02:24:29 the only concern I have is that it means I have to leave an ant on each of my hills to control the spawning, but the bonus is that none of my hills can get ambushed and I can spawn defenses there faster as well 2011-12-11T02:25:02 all assuming a multi-hill map of course 2011-12-11T02:25:06 if you are that worried, then you are trying to control 1 ant spawn, probably not worth it 2011-12-11T02:25:36 after you get a bunch of food, just pick a hill closest to the good stuff, then don't move the new ants off the others 2011-12-11T02:25:51 i think it will be an advantage in some cases, but not sure how much of an advantage 2011-12-11T02:26:05 McLeopold: that's what I'm hoping will happen 2011-12-11T02:26:15 there is a cell maze that I need to find... 2011-12-11T02:26:21 my "solution" to this is to not worry about my worst hills being taken in multi-hill 2011-12-11T02:26:21 also, I think my bot does better with fewer hills 2011-12-11T02:26:38 if I have a hill near/between a strong enemy better to let it die than feed ants into a bottleneck 2011-12-11T02:26:55 that was my solution back when I had a "static" defense 2011-12-11T02:27:16 *** Kurnevsky has joined #aichallenge 2011-12-11T02:27:17 what is driving me nuts about my bot is that my "active" defense lets things by 2011-12-11T02:27:20 it doesn't die 2011-12-11T02:27:25 it just turns sideways and slips past 2011-12-11T02:27:31 and I'm willing to allow that to happen still, but we'll see how it goes, i need to get this code tested and then upload it to see how it does 2011-12-11T02:27:38 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-11T02:27:40 my attempts to stop that have been mostly futile 2011-12-11T02:28:09 BenJackson: yeah, that is tricky to get right 2011-12-11T02:29:11 btw I hope you Go guys release source 2011-12-11T02:29:20 my reaction to Go tutorials and so on has been "eww" 2011-12-11T02:30:00 not sure my code is very "pretty" 2011-12-11T02:33:20 http://aichallenge.org/map.php?map=cell_maze/cell_maze_p05_09.map 2011-12-11T02:37:15 yep, that's a good example McLeopold 2011-12-11T02:37:29 *** Palmik has joined #aichallenge 2011-12-11T02:37:32 its not the one I was looking for 2011-12-11T02:38:06 but I think the situation can come up dynamically as well when you have one hill deep in your own area and another closer to the front lines or an enemy hill. 2011-12-11T02:38:08 its a map where you have 3 hills in front of a large snaking cave and 1 hill on the other side of the map that is surrounded 2011-12-11T02:38:17 it is a cell_maps, I think at least 4 players 2011-12-11T02:39:14 at least in the map I found, you have a hill inside a cave, which can just be blocked 2011-12-11T02:41:00 this map could be like that: http://aichallenge.org/visualizer.php?game=166036 2011-12-11T02:41:07 no 2011-12-11T02:41:35 there was a main channel that had a slightly negative slope everyone had to fight over 2011-12-11T02:41:36 no not it, but one hill out in the open and other inside a horseshoe could lead to interesting spawning strategies 2011-12-11T02:43:29 that's a crazy map 2011-12-11T02:44:17 yeah, and that was a crazy game, one of the few "good games" I posted in the forums 2011-12-11T02:45:42 my pathfinding would explorer from the center of a lake (if unseen) and stop at the water, and then quit. so I think I'll lose 5ms in a game for maps that have those 2011-12-11T02:46:02 *** Antimony has joined #aichallenge 2011-12-11T02:47:45 not an issue, and it's not the total time lost during the game that matters, but the maximum time lost on any single turn that really matters 2011-12-11T02:53:17 ChrisH_: how many ants to you commit to razing a hill? 2011-12-11T02:53:26 *** amstan has quit IRC (Ping timeout: 248 seconds) 2011-12-11T02:54:19 I commit at least 3 plus any ants with nothing else to do 2011-12-11T02:54:28 why 2011-12-11T02:54:29 ? 2011-12-11T02:54:58 I'm working on a formula 2011-12-11T02:55:08 I started with 2 and added more if needed 2011-12-11T02:55:52 the thing is, my bot also attacks enemy ants, so the more there are around the enemy hill, the more ants I send, plus the 3 2011-12-11T02:56:09 that was how I was deciding to add more 2011-12-11T02:56:15 ChrisH_: so the test I did earlier was on one of the new workers 2011-12-11T02:56:29 janzert: ok 2011-12-11T02:56:29 just repeated it on one of the old workers with the same result though 2011-12-11T02:56:34 ok 2011-12-11T02:57:01 while I'm sure it's worse for you, I'm rather relieved ;) 2011-12-11T02:57:17 I haven't seen this happening a lot, so I'm not going to worry about it now, my next version has changed enough hopefully it won't happen when I resubmit 2011-12-11T02:57:35 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-11T02:58:23 *** dvladim has joined #aichallenge 2011-12-11T03:00:00 janzert: here's $5 to omit another batch of ChrisH_'s mvoes 2011-12-11T03:00:13 LOL 2011-12-11T03:00:25 shh, you're supposed to keep that to IM's :P 2011-12-11T03:01:14 oh god how did this get here I am not good with computer 2011-12-11T03:02:46 the amazing thing is that I tied for 1st in that game anyway, but my last hill was under serious attack and I lost the ant count war when the turn limit hit, so I easily could have lost if the game had gone another few turns, but even tying for 1st my mu went down because the competition all had lower mu's 2011-12-11T03:03:19 do you actually estimate scores while playing? 2011-12-11T03:03:25 so, i did ok even with a couple fingers tied behind my back 2011-12-11T03:03:41 no, i have no idea how to do that accurately. 2011-12-11T03:04:07 I don't think it's possible 2011-12-11T03:04:27 you'd have to do serious work just to know for sure how many players there were 2011-12-11T03:05:09 ChrisH_: mu went down because you tied with a lower mu 2011-12-11T03:05:24 you could play a whole game and not see all the players 2011-12-11T03:05:31 ie me 2011-12-11T03:05:35 janzert: yes, I understand, not a complaint 2011-12-11T03:05:53 recent game, I tie chrish for 4th, 2 games back (for me) he wins and I come in last 2011-12-11T03:05:59 that is basically the problem with V10 of my bot 2011-12-11T03:06:04 it's a bit... bi-modal 2011-12-11T03:07:38 i think you see that for a lot of bots when they are playing roughly equal competition 2011-12-11T03:08:27 the margin for error in a game can be pretty slim sometimes 2011-12-11T03:09:18 at some point this week I may have to knuckle down and quit playing with new ideas and just do another bugfix pass 2011-12-11T03:09:33 *** Antimony has quit IRC (Ping timeout: 268 seconds) 2011-12-11T03:23:39 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-11T03:44:31 *** ChrisH_ has joined #aichallenge 2011-12-11T03:44:42 McLeopold: http://ants.fluxid.pl/replay.25353 2011-12-11T03:47:37 McLeopold: It's looking promissing, there were some nice behaviors in that game. 2011-12-11T03:48:33 *** dvladim has quit IRC (Read error: Connection reset by peer) 2011-12-11T03:53:22 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-11T03:54:29 *** dvladim has joined #aichallenge 2011-12-11T03:54:58 *** Redgis has joined #aichallenge 2011-12-11T04:15:04 *** replore_ has quit IRC (Remote host closed the connection) 2011-12-11T04:15:59 *** foRei has joined #aichallenge 2011-12-11T04:24:43 *** jon1 has joined #aichallenge 2011-12-11T04:27:31 nice game against xathis here: http://ants.fluxid.pl/replay.25082 makes me aware of the weakness of my combat. xathis strategy is to outflank the opponent. can't figure out yet how to command my ants to do the same formation against the enemy 2011-12-11T04:36:40 *** GeorgeSebastian has joined #aichallenge 2011-12-11T04:36:47 *** GeorgeSebastian has joined #aichallenge 2011-12-11T04:40:02 *** pairofdice has joined #aichallenge 2011-12-11T04:46:02 *** Anilm3 has joined #aichallenge 2011-12-11T05:08:29 *** djstrong has joined #aichallenge 2011-12-11T05:09:15 *** GeorgeSebastian_ has joined #aichallenge 2011-12-11T05:09:40 *** GeorgeSebastian_ has left #aichallenge 2011-12-11T05:11:02 is there any good solution how to deal with timeouts in java? i found out that GC can take 500ms and it is the problem 2011-12-11T05:12:12 ...ofc except making everything "static" ;) 2011-12-11T05:31:08 *** dvladim has quit IRC (Ping timeout: 240 seconds) 2011-12-11T05:37:54 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-12-11T05:45:04 *** Akranis has joined #aichallenge 2011-12-11T05:51:28 *** taeke has quit IRC (Quit: Page closed) 2011-12-11T05:52:41 *** kire has joined #aichallenge 2011-12-11T06:02:20 *** nha has joined #aichallenge 2011-12-11T06:04:54 *** mj41 has joined #aichallenge 2011-12-11T06:08:20 *** kilae has joined #aichallenge 2011-12-11T06:12:03 in this game: http://ants.fluxid.pl/replay.25581 my ants are not moving at the end due to an absence of goals. i can see that there is one enemy hill still remaining but my ants can seem to see it because it is covered by an enemy ant 2011-12-11T06:12:53 my question now is if the hill is still being included in the input even if it is covered? 2011-12-11T06:13:04 it should 2011-12-11T06:13:05 *** heinrich5991 has quit IRC (Excess Flood) 2011-12-11T06:15:07 It's visible 2011-12-11T06:15:43 ok. it seems like a bug in my setting the hill being razed or not. so the enemy could hide it's hill to my ants by just covering it. lol 2011-12-11T06:16:10 It's a bit odd though 2011-12-11T06:16:15 *** heinrich5991 has joined #aichallenge 2011-12-11T06:16:19 :) 2011-12-11T06:17:34 i set a hill razed or not if it is visible. problem is that the map table is overwritten by an ant instead of a hill. so when i loop over the map to check for hills, it couldn't find it 2011-12-11T06:17:35 *** ikaros has joined #aichallenge 2011-12-11T06:17:52 Yeah 2011-12-11T06:19:59 I have suspected this bug for so long but it is only now that I've verified it. Thanks guys for the clarification. 2011-12-11T06:27:37 this is actually something that came up a few times in this channel; i suspect "hiding" your hills like that might turn out to be a viable option when fighting lower-ranked bots ;) 2011-12-11T06:32:14 *** codetiger has joined #aichallenge 2011-12-11T06:32:34 anybody tried submitting pypy code 2011-12-11T06:34:44 *** sigh has quit IRC (Remote host closed the connection) 2011-12-11T06:42:35 http://aichallenge.org/language_profile.php?language=PyPy 2011-12-11T06:42:35 *** Anilm3 has joined #aichallenge 2011-12-11T06:42:39 @rankings 2011-12-11T06:42:40 Anilm3: Top 10 players: xathis(92.6), pguillory(92.5), fourmidable(90.2), ChrisH(90.0), lazarant(89.9), GreenTea(89.3), delineate(88.9), Memetix(87.7), RVeerdonk(86.9), Speedy_Consoles(86.8) 2011-12-11T06:42:48 :O 2011-12-11T06:43:04 Whaat 2011-12-11T06:47:04 And my bot is at 231... ;_; 2011-12-11T06:47:19 #2 in Py3 rankings though :D 2011-12-11T06:49:13 we're near then 2011-12-11T06:51:03 My bot has been plummeting 2011-12-11T06:51:27 I guess that means others aren't standing still :) 2011-12-11T06:51:27 what's plummeting? 2011-12-11T06:52:09 collapse, crash, decrease, descend, dip, dive... 2011-12-11T06:52:12 *** foRei has quit IRC (Ping timeout: 244 seconds) 2011-12-11T06:52:18 oh I see 2011-12-11T06:52:24 thanks 2011-12-11T06:52:40 I don't know if I will have time to do anything new 2011-12-11T06:52:46 :( 2011-12-11T06:53:38 Yeah, I'm a bit demotivated to do anything to my bot 2011-12-11T06:53:41 ok, I successfully submitted my bot in pypy. Seems I should only change the extension of MyBot.pypy and not ant.py 2011-12-11T06:54:57 codetiger: indeed 2011-12-11T06:55:15 codetiger: pypy is nice, I am using it 2011-12-11T06:55:28 pairofdice: I'm not demotivated, I just have a lot of work 2011-12-11T06:55:29 So much faster than cpython 2011-12-11T06:55:58 It would be nice if the contest deadline was extended.. 2011-12-11T06:56:04 heh 2011-12-11T06:56:05 but that's not going to happen 2011-12-11T06:56:35 My friends and I will be improving our bots and competing among ourselves for a while after the competition ends, since we started into the comp late 2011-12-11T06:57:02 *** flowenol has joined #aichallenge 2011-12-11T06:57:06 too bad I don't have any friends on the contest lol 2011-12-11T06:57:23 I have two in the top 100 but they started a month before the rest of us 2011-12-11T06:57:31 most of us were too busy until last week 2011-12-11T06:57:43 :) I wish i got into the contest when it was on beta 2011-12-11T06:57:59 codetiger: me too 2011-12-11T06:57:59 We also plan on tossing any bots we can find afterwards that people open source into our pool of bots to play against 2011-12-11T06:58:01 :P 2011-12-11T06:58:05 yea ill try that next time too codetiger :) 2011-12-11T06:58:26 Wraithan: that's interesting!! 2011-12-11T06:58:57 coz, they displayed "You will regret if you upload your test bot secretly", I thought I'll only work after the beta is removed 2011-12-11T07:00:16 *** jstemmer has quit IRC (Quit: leaving) 2011-12-11T07:01:18 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-11T07:01:35 Wraithan: planning on making your minicontest open? :P 2011-12-11T07:02:18 anyone having timeout issues? 2011-12-11T07:02:31 We are still discussing it, having it closed means security is easy, as well having limit server resources 2011-12-11T07:02:54 I see 2011-12-11T07:02:59 brb 2011-12-11T07:03:00 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-12-11T07:04:18 my bot had some timeouts wich seem impossible to me 2011-12-11T07:04:29 i even ran the input on my side 2011-12-11T07:10:03 *** bretep has joined #aichallenge 2011-12-11T07:20:37 *** g0llum has joined #aichallenge 2011-12-11T07:23:39 *** Jacob_Strauss has joined #aichallenge 2011-12-11T07:27:35 *** Anilm3 has joined #aichallenge 2011-12-11T07:43:05 wow i still dont get memetix stuff 2011-12-11T07:44:04 my last commit was november 1st :( 2011-12-11T07:44:37 i was hoping to have a few days at least so i could submit something.. maybe i still can 2011-12-11T07:44:39 *** Kommander has joined #aichallenge 2011-12-11T07:55:56 *** jon1 has quit IRC (Quit: Leaving.) 2011-12-11T07:58:43 *** sigh has joined #aichallenge 2011-12-11T07:59:53 *** lilEzek has joined #aichallenge 2011-12-11T07:59:56 hi 2011-12-11T08:00:06 anyone can help me with an easy task? 2011-12-11T08:00:39 I would like to determine the direction an ant should take to get close to another point 2011-12-11T08:00:54 and I can't handle torus 2011-12-11T08:01:34 I mean, given two points, determine the direction that should the first point take to get close to second point 2011-12-11T08:02:46 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-11T08:03:39 *** Lithosphere has quit IRC (Remote host closed the connection) 2011-12-11T08:03:55 *** Lithosphere has joined #aichallenge 2011-12-11T08:03:55 *** Lithosphere has joined #aichallenge 2011-12-11T08:04:14 lilEzek: you realize that if you don't use pathfinding your ant can get stuck right? 2011-12-11T08:04:30 I know 2011-12-11T08:04:35 I use bfs 2011-12-11T08:04:50 if you use bfs you have stored the direction 2011-12-11T08:04:53 or the path 2011-12-11T08:05:01 I'm storing a stack of steps 2011-12-11T08:05:17 follow the steps then 2011-12-11T08:05:19 that is, a stack in c++ STL 2011-12-11T08:05:23 yeah 2011-12-11T08:05:32 but I need a function that 2011-12-11T08:05:35 In my case 2011-12-11T08:05:41 pointsToDirection(Location,Location) 2011-12-11T08:05:43 I use A* 2011-12-11T08:05:46 to know the direction to follow 2011-12-11T08:05:47 and get the whole path 2011-12-11T08:06:21 I have a BFS for the purpose of exploration and for that I just store the direction 2011-12-11T08:06:39 so when I find the interesting node on the BFS I just go to that direction 2011-12-11T08:07:01 yeah 2011-12-11T08:07:04 I imagine so 2011-12-11T08:07:24 the problem is I'm not storing all the locations on the stack 2011-12-11T08:07:27 in your case I think you would have to reconstruct the path or store it on each node 2011-12-11T08:07:46 imagine I need to go from 1,0 to 200,0 2011-12-11T08:07:56 if no obstacle is in the path 2011-12-11T08:08:02 the stack only stores 1,0 and 200,0 2011-12-11T08:08:10 because I can deduce the direction 2011-12-11T08:08:14 the first step of the bresenham line algorithm 2011-12-11T08:08:25 bresenham line algorithm? 2011-12-11T08:08:28 yup 2011-12-11T08:08:32 google it 2011-12-11T08:08:54 what does it gives? 2011-12-11T08:08:57 what does it give? 2011-12-11T08:09:10 it's an algorith to draw a line on a screen of pixels 2011-12-11T08:09:32 in this case you adapt it to draw a line without diagonal moves 2011-12-11T08:09:42 and since no obstacles are in the way 2011-12-11T08:09:45 the problem is I'm in a torus 2011-12-11T08:09:58 so I can't get two points and draw that line 2011-12-11T08:10:08