2011-12-13T00:04:51 looks serious, try some antibiotics 2011-12-13T00:16:11 my condition is more like >_< 2011-12-13T00:18:01 argh 2011-12-13T00:18:24 I got sidetracked on this idea of trying to make a bfs step in the "long" direction rather than the short one 2011-12-13T00:18:28 harder than it sounds 2011-12-13T00:20:45 <_flag> Well I beat xathis in a single game, so I think that's enough work for today 2011-12-13T00:21:31 <_flag> http://ants.fluxid.pl/replay.29611 2011-12-13T00:22:13 *** Areks has joined #aichallenge 2011-12-13T00:22:22 hey you're #2 on aichallenge, grats 2011-12-13T00:22:35 <_flag> Thanks :) 2011-12-13T00:23:12 *** bluegaspode has quit IRC (Ping timeout: 258 seconds) 2011-12-13T00:28:35 with the current game rate I'm tempted to upload some fairly untested code 2011-12-13T00:29:22 <_flag> I figure I might as well just let myself sit near the top of the rankings for exposure until my final entry, with xathis playing tcp is good enough testing anyway 2011-12-13T00:29:51 I'm only finally finding strategies that can reliably beat my last entry 2011-12-13T00:31:06 *** b0rder has joined #aichallenge 2011-12-13T00:36:01 welp, kicking off 20 games on fluxid will give me time to play skyrim 2011-12-13T00:51:11 *** keith__ has quit IRC (Ping timeout: 258 seconds) 2011-12-13T00:53:03 *** meatroaf has joined #aichallenge 2011-12-13T00:53:06 *** pguillory has quit IRC (Ping timeout: 258 seconds) 2011-12-13T01:16:30 wait... BenJackson, isn't what you described called a "depth first search"? ^_^ 2011-12-13T01:16:56 searching in the "long direction" rather than the "short one" sounds exactly like dfs vs bfs.... 2011-12-13T01:20:42 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-12-13T01:22:25 *** memolicious has joined #aichallenge 2011-12-13T01:22:44 i have a completely noobish question regarding distance calculation. 2011-12-13T01:23:08 memolicious: k 2011-12-13T01:23:15 I'm on, let's see if I can answer 2011-12-13T01:23:44 when i compare the result of the included distance function against a provided radius, say attackradius, do I use the squarert of the attackradius? 2011-12-13T01:23:55 or is the distance function squaring the result and I compare directly? 2011-12-13T01:24:08 and memolicious, for future reference (IDK if it applies to this room) there's a standing IRC meme. "Don't ask if you can ask a question, or say you're going to ask a question. Just ask the question" ^_^ 2011-12-13T01:24:30 At dragonfyre13 yeah I realized I totally blew that haha. 2011-12-13T01:24:38 yes, sqrt of attackradius 2011-12-13T01:24:58 waiit. 2011-12-13T01:25:01 hangon 2011-12-13T01:25:03 hehe 2011-12-13T01:25:08 waiting 2011-12-13T01:25:10 yeah, that's right 2011-12-13T01:25:14 wanted to make sure 2011-12-13T01:25:41 thanks :) 2011-12-13T01:25:42 Though I think that only works if it's manhatten distance 2011-12-13T01:25:55 eh? 2011-12-13T01:26:06 travel 2011-12-13T01:26:36 basically, if you're moving in a straight line 2011-12-13T01:26:39 the distance function i am referring to is the one documented on the site, the Euclidean metric I believe. 2011-12-13T01:26:46 k, hangon 2011-12-13T01:26:53 in the python starter pack, or no? 2011-12-13T01:26:53 right - which is what you need to determine something like attack range 2011-12-13T01:26:57 that's the one I was looking at 2011-12-13T01:27:02 mine is the JavaScript start pack actually =) 2011-12-13T01:27:05 ah 2011-12-13T01:27:07 hangon 2011-12-13T01:27:10 k 2011-12-13T01:27:30 it matches the euclidean metric function described on the site so i assume its the same (well - should be) 2011-12-13T01:28:13 ah yea 2011-12-13T01:28:17 k 2011-12-13T01:28:44 so, if ants.distance <= sqrt(attackradius2) then its w/in attack range, yeah? 2011-12-13T01:29:05 so you need to compare sqrt(attackradius2) >= distance(loc1,loc2) 2011-12-13T01:29:14 yep 2011-12-13T01:29:16 yeah 2011-12-13T01:29:17 cool 2011-12-13T01:29:17 that's the ticket 2011-12-13T01:29:21 thanks for your help. 2011-12-13T01:29:24 bp 2011-12-13T01:29:25 np 2011-12-13T01:29:52 *** McLeopold has joined #aichallenge 2011-12-13T01:53:18 memolicious: you can also just compare euclidean_distance2 with attackradius2 2011-12-13T01:54:03 instead of taking sqrt() for each euclidean calculation and for attackradius2 2011-12-13T01:54:32 yeah i was wondering about that - good call. thanks! 2011-12-13T01:57:53 *** Garf has joined #aichallenge 2011-12-13T01:58:06 *** kiv has joined #aichallenge 2011-12-13T02:05:42 amstan: you around? 2011-12-13T02:06:12 amstan: we just got the most interesting markov chain spam I've ever seen 2011-12-13T02:06:21 *** GeorgeSebastian has joined #aichallenge 2011-12-13T02:08:33 *** Palmik has joined #aichallenge 2011-12-13T02:14:14 *** kiv has quit IRC (Quit: Leaving) 2011-12-13T02:16:21 McLeopold: hey I meant to ask you 2011-12-13T02:16:29 ... 2011-12-13T02:16:31 you posted a visualization that had a bullseye sort of ring around your hill 2011-12-13T02:16:38 did that mean something? 2011-12-13T02:16:43 ants in hive 2011-12-13T02:16:53 heh 2011-12-13T02:16:57 I count the rings to make sure my calc is correct 2011-12-13T02:17:48 growth rings 2011-12-13T02:18:02 wow I just watched a replay where I'm not orange or white 2011-12-13T02:18:10 I'm staring at orange thinking WTF was my bot doing 2011-12-13T02:18:25 orange is reserved for *good* players :P 2011-12-13T02:20:46 *** sigh has joined #aichallenge 2011-12-13T02:23:07 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-13T02:24:34 argh I'm going to have to fix my 2-thick-water bug 2011-12-13T02:26:15 *** sigh has quit IRC (Remote host closed the connection) 2011-12-13T02:32:27 *** sigh has joined #aichallenge 2011-12-13T02:33:18 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-13T02:41:28 < dragonfyre13> yes, sqrt of attackradius ← NO, WHY 2011-12-13T02:41:49 why would you need to sqrt 2011-12-13T02:42:12 x^2 + y^2 <> attackradius 2011-12-13T02:42:17 no need to sqrt anything 2011-12-13T02:43:51 ok, "included distance function', disregard my answer, i just woke up 2011-12-13T02:44:24 i wish we never used euclidean for anything 2011-12-13T02:44:34 it makes life so complicated :P 2011-12-13T02:44:51 http://ants.fluxid.pl/replay.29718 2011-12-13T02:44:54 that map is ... intimate! 2011-12-13T02:44:59 *** sigh has quit IRC (Remote host closed the connection) 2011-12-13T02:45:00 not sure I've ever seen it 2011-12-13T02:45:20 herds of sheep 2011-12-13T02:46:14 is that randomwalk 202 or something? :P 2011-12-13T02:47:05 02/12 2011-12-13T02:48:46 arg. Time for bed. Reimpliment my attack logic tomorrow night. 2011-12-13T02:48:53 it sucks right now. 2011-12-13T02:52:09 xathis didn't like being in second place then? 2011-12-13T02:54:05 http://ants.fluxid.pl/replay.29745 2011-12-13T02:54:13 haha that demonstrates my "anti timeout" code 2011-12-13T02:54:32 got 250 ants? ok, time for the WWI style charges across no-mans-land 2011-12-13T02:55:14 you can see in the graph where I hit 25o and blue takes a nosedive 2011-12-13T02:56:33 BenJackson: that's a great way to deal with timeouts xD 2011-12-13T02:56:48 if I have 500 ants, I'll just skip doing combat completely 2011-12-13T02:57:02 and then it will kick back in when a few die :P 2011-12-13T02:58:28 http://ants.fluxid.pl/replay.29837 2011-12-13T02:58:33 that one is interesting if you center on blue 2011-12-13T02:58:39 initially I do well to keep orange clear 2011-12-13T02:58:47 then suddenly it's ant soup as I fail to hold any lines 2011-12-13T02:58:54 then I clear it out again 2011-12-13T02:58:55 i like that symmetry detection hasn't been very worthwhile 2011-12-13T02:59:29 thestinger: you could at least have them all just make their offensive or defensive moves 2011-12-13T02:59:38 ie. move the ants out of the most dangerous squares at least 2011-12-13T02:59:45 well, if I disable combat they're just going to charge at the enemy 2011-12-13T03:00:04 yeah, but you could be a little smarter about it imo 2011-12-13T03:00:06 I do all my movement and end up with an orders hash table, and then run combat 2011-12-13T03:00:19 well, I could handle way more than 500 ants now :P 2011-12-13T03:00:30 my last bot had a single timeout on the server, with around 550 ants 2011-12-13T03:00:38 and it did combat globally (simulation for the whole map) 2011-12-13T03:00:50 hmmm 2011-12-13T03:00:58 i tell mine to stop when it runs out of time 2011-12-13T03:01:10 and tell it to use up to 50% of the time on combat 2011-12-13T03:01:18 I used to do that, but the calls to gettimeofday() were the bottleneck 2011-12-13T03:01:22 and do that first before anything else 2011-12-13T03:01:28 so I would have to put it in the outer loop where it wouldn't really catch anything 2011-12-13T03:01:30 yeah i have considered that 2011-12-13T03:01:36 gettimeofday is actually very cheap 2011-12-13T03:01:38 on modern linux 2011-12-13T03:01:39 but you can always just check after each battle partition has run 2011-12-13T03:01:47 BenJackson: not in an inner loop 2011-12-13T03:01:53 could be -pg distorting your results 2011-12-13T03:01:58 thestinger: it's not even a system call 2011-12-13T03:02:05 oh 2011-12-13T03:02:13 well, the profiler screwed up I guess :\ 2011-12-13T03:02:39 really tiny inner-loop functions are greatly distorted on -pg 2011-12-13T03:04:16 neat, gcc has some special C++ profiling thing now 2011-12-13T03:04:40 *** epicmonkey has joined #aichallenge 2011-12-13T03:04:40 http://gcc.gnu.org/onlinedocs/libstdc++/manual/profile_mode.html#manual.ext.profile_mode.using 2011-12-13T03:05:39 wow, this makes my code run like 100x slower 2011-12-13T03:07:48 it's amazing how much better xathis's bot is than mine 2011-12-13T03:07:57 and even my older version can hover around 10th 2011-12-13T03:08:41 my bot looks like xathis's ants stupid drunk 2011-12-13T03:08:57 BenJackson: last contest bocsimacko was leaps and bounds ahead of anyone else 2011-12-13T03:09:37 wow xathis only *tied* for 1st in the last game he played and his mu went DOWN nearly 0.5 2011-12-13T03:15:16 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-12-13T03:26:00 *** ikaros has joined #aichallenge 2011-12-13T03:35:56 food spawning adjacent to ants is a pita to handle 2011-12-13T03:36:09 well, other than just first moving ants adjacent to food 2011-12-13T03:36:14 but that's not necessarily optimal 2011-12-13T03:36:20 *** mj41 has joined #aichallenge 2011-12-13T03:37:43 *** Blkt has joined #aichallenge 2011-12-13T03:41:49 *** Bluedgis has quit IRC (Quit: leaving) 2011-12-13T03:42:01 *** Bluedgis has joined #aichallenge 2011-12-13T03:45:36 antimatroid1: how is it a pita? 2011-12-13T03:46:04 McLeopold: consider 2011-12-13T03:46:04 a.* 2011-12-13T03:46:04 **. 2011-12-13T03:46:08 you want to go west 2011-12-13T03:46:13 east sorry 2011-12-13T03:46:34 when you search from food it's hard to do that 2011-12-13T03:47:03 a..* 2011-12-13T03:47:04 i order search items by the number of food items the path is going to collect for longer paths, but that first one is hard to do 2011-12-13T03:47:06 *.*. 2011-12-13T03:47:28 i would just collect the 1 then 2011-12-13T03:47:51 but for equal length paths i want to collect the maximal number of food first 2011-12-13T03:48:11 oh, so "hold" is a length of 1 2011-12-13T03:48:30 yeah, the ant will be available to move again in 1 turn 2011-12-13T03:48:38 so it's the same as moving into an adjacent square 2011-12-13T03:53:53 McLeopold: any ideas on how to handle that edge case? 2011-12-13T03:54:56 antimatroid1: I'm not sure how you score path, so no 2011-12-13T03:55:46 i just realised i'm unecessarily using a priority queue for the entire search 2011-12-13T03:55:49 antimatroid1: do you search from food to ants? 2011-12-13T03:56:08 i just need to use a priority queue when i add the squares adjacent to food, then i can whack it all into a queue 2011-12-13T03:56:49 yeah, if you search from food, both squares should be hit at the same time 2011-12-13T03:57:11 then the square with 2 foods should be the higher priority 2011-12-13T03:57:12 McLeopold: yeah, iteratively (ordered) bfs from uncollected food to moveable ants, and move the best found move once it's pathlength + time ant is available is less than the current search item pulled off the queue's current path length 2011-12-13T03:57:44 so it would find the food adjacent to the ant and move it before it got on to searching from squares adjacent to food 2011-12-13T03:57:52 ant adjacent to food* 2011-12-13T03:58:19 otherwise i can call off the search when i do because i order the search items right up so it searches from locations with the most adjacent food first 2011-12-13T03:59:09 my bfs would have produced both moves at the same time in a list 2011-12-13T03:59:38 but I don't track if a square has more than one food it will collect 2011-12-13T04:00:52 antimatroid1: I want to make my bfs faster, but I can't think how 2011-12-13T04:01:15 my bfs's are all slightly different 2011-12-13T04:01:19 some are ordered, some aren't etc. 2011-12-13T04:01:20 if I were to search from food to ant, I could just create a matrix of distances and pick a move that minimizes it 2011-12-13T04:01:34 but then once you collect a food item the information changes 2011-12-13T04:01:44 but if I search from ant to food, I have to propogate information on the path 2011-12-13T04:02:07 i think you want to iteratively collect food at the very least 2011-12-13T04:02:15 I can create a matrix of ant to all food, but it doesn't tell me where to go 2011-12-13T04:02:17 like others have said, there's usually not that much food to collect 2011-12-13T04:02:54 any ideas for that? 2011-12-13T04:03:26 *** jon1 has joined #aichallenge 2011-12-13T04:05:55 you can trace from food back to ants though? 2011-12-13T04:06:05 you just have to retrace the shortest paths 2011-12-13T04:06:23 not sure about implications with reusing ants 2011-12-13T04:06:38 *** Harpyon has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-12-13T04:06:59 same problem as before, the information changes every time you move an ant 2011-12-13T04:07:20 I just do a bfs from each food individually 2011-12-13T04:07:51 actually, I can't think why I would ever start a search from an ant? 2011-12-13T04:08:00 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-13T04:08:40 aw carp, i started my bot on tcp but forgot to disable debugging... it takes over 1s now, ff 2011-12-13T04:08:43 don't have a set of targets, but can check a location to see if it's a target? 2011-12-13T04:09:39 i don't think i'm doing that anymore though 2011-12-13T04:09:44 antimatroid1: does that answer why I would search from an ant? 2011-12-13T04:10:07 McLeopold: well otherwise you have to go through every square and add the targets to a set/vector 2011-12-13T04:10:26 otherwise how else would you search from the targets? 2011-12-13T04:10:57 that's kinda what I do 2011-12-13T04:10:58 fixed, sorry 2011-12-13T04:11:08 This is all in python 2011-12-13T04:11:17 I'm not really going for speed, persay 2011-12-13T04:11:33 my target sets consist of things like uncollectedFood, unseenBorder, potentialEnemyBorder, notRecentlySeenBorder, enemyAnys, enemyHills, etc. 2011-12-13T04:11:59 i only reuse ants for collecting food though 2011-12-13T04:12:11 and maybe for exploration, but i'm not atm 2011-12-13T04:12:22 *** vjacob has joined #aichallenge 2011-12-13T04:12:34 and only if the ant can't possibly be battled on it's path to the food 2011-12-13T04:13:27 hmm, I'll try the matrix thing and see how it goes... 2011-12-13T04:13:41 how are you going to update the information? 2011-12-13T04:14:10 you have to use a bfs to generate the initial information 2011-12-13T04:14:20 and a bfs every time you collect food/move an ant to update it 2011-12-13T04:14:30 i think it's just easier to bfs from the food to ants and move them 2011-12-13T04:14:48 I do a separate bfs from every food item 2011-12-13T04:15:04 do it from all food 2011-12-13T04:15:14 and just do a new search every time you collect a food item 2011-12-13T04:15:21 right, 1 whole matrix for each food 2011-12-13T04:15:32 but why? 2011-12-13T04:15:42 why not? 2011-12-13T04:15:53 you at least only need to do each matrix until F ants have been seen, where F is the number of food items 2011-12-13T04:16:15 because then you have sufficient information even if each food is collected by a unique ant 2011-12-13T04:16:38 but if you're doing that, why not just do simultaneous bfs's? 2011-12-13T04:16:59 ie. have 1 search queue but a vector of searchedMatrices 2011-12-13T04:17:27 the search items will need to mark which food item they're collecting 2011-12-13T04:17:49 if I searched from all food at once, every ant would go to the closest food 2011-12-13T04:17:59 it's a pita to reuse ants that way fyi, you have to start assigning ant ids to locations 2011-12-13T04:18:13 no, do a simultaneous bfs 2011-12-13T04:18:36 which is what? 2011-12-13T04:18:36 so do all bfs's at once, and if you pull a search item off where the "food target" has already been collected, just keep searching and ignore that 2011-12-13T04:19:21 like have, queue searchQueue and vector > searched(food.size(), Matrix(rows, cols, 0)) 2011-12-13T04:21:13 * _ _ a _ _ b _|_ _ _ _ _ _ _ * (| is the halfway mark) 2011-12-13T04:21:28 both a and b are closer to the left 2011-12-13T04:21:54 if I add both targets to the bfs, b will find the left 1 first 2011-12-13T04:22:27 yes, but once you collect the left * with a, once you pull any more search items off with target as left * you don't search further, and the search queue has search items for the right * over the left * because you have multiple searched matrices 2011-12-13T04:23:31 so basically once you do searchItem = searchQueue.front(), if(!uncollectedFood.count(searchItem.fLoc)) continue; 2011-12-13T04:23:34 so then, remove found food and add multiple targest to a new bfs? 2011-12-13T04:23:51 no, your search queue already has the other searches all in it 2011-12-13T04:25:25 the two options i got down to were this simultaneous bfs from food to ants vs. iterative bfs from all food to ants, i decided on the later 2011-12-13T04:25:30 *** aerique has joined #aichallenge 2011-12-13T04:26:11 I'm doing iterative bfs from all food to ants 2011-12-13T04:26:24 i don't think you're going to get much better 2011-12-13T04:26:53 one thing i do is only collect food if there's a known path to it (which i know from nearest friend hill information i maintain) 2011-12-13T04:27:19 *** jon1 has quit IRC (Quit: Leaving.) 2011-12-13T04:27:28 I go ahead and do it and my ant then naturally explores 2011-12-13T04:27:31 and i'm also only going to collect food in this manner if it's not at least contestable with an enemy, otherwise i'll just add it to later searches 2011-12-13T04:27:55 i did that but i decided it's taking to long when i'm sending single ants to targets 2011-12-13T04:28:03 so i'm letting exploration happen elsewhere 2011-12-13T04:28:20 basically i'm trying to collect the food that needs to be collected optimally and just that with the expensive searches 2011-12-13T04:28:25 well, he'll get distracted with other targets on the way 2011-12-13T04:28:47 okay, back to code... 2011-12-13T04:28:58 yes, but with maze maps you could be ages away from a food item that actually looks close 2011-12-13T04:29:06 i don't like it exploring based off that 2011-12-13T04:30:09 so you like to expand evenly around the hills? 2011-12-13T04:33:18 yes 2011-12-13T04:33:43 basically i'm just moving ants to their closest unseen border location (amongst other targets) 2011-12-13T04:33:57 so new ants will go towards the closest unseen location to the hill for example 2011-12-13T04:34:09 what i really like is the potential enemy border 2011-12-13T04:34:15 that's my perimeter idea 2011-12-13T04:34:22 (and many others i imagine) 2011-12-13T04:40:19 *** ikaros has joined #aichallenge 2011-12-13T04:43:56 *** sigh has joined #aichallenge 2011-12-13T04:50:26 *** Bluedgis_ has joined #aichallenge 2011-12-13T04:55:13 *** nha has joined #aichallenge 2011-12-13T05:03:23 *** Areks|2 has joined #aichallenge 2011-12-13T05:06:34 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-13T05:08:25 *** Areks|2 has quit IRC (Ping timeout: 268 seconds) 2011-12-13T05:12:38 *** Cyndre_ has joined #aichallenge 2011-12-13T05:13:48 *** Cyndre__ has joined #aichallenge 2011-12-13T05:15:24 *** vjacob has quit IRC (Quit: Leaving) 2011-12-13T05:15:31 *** Cyndre has quit IRC (Ping timeout: 252 seconds) 2011-12-13T05:17:16 *** Cyndre_ has quit IRC (Ping timeout: 240 seconds) 2011-12-13T05:18:17 *** vjacob has joined #aichallenge 2011-12-13T05:21:06 *** GarfTop has joined #aichallenge 2011-12-13T05:22:03 *** GarfTop has quit IRC (Read error: Connection reset by peer) 2011-12-13T05:22:27 *** GarfTop has joined #aichallenge 2011-12-13T05:24:05 *** Garf_ has joined #aichallenge 2011-12-13T05:24:15 *** Garf has quit IRC (Ping timeout: 240 seconds) 2011-12-13T05:25:01 *** Areks has joined #aichallenge 2011-12-13T05:27:15 *** GarfTop has quit IRC (Ping timeout: 240 seconds) 2011-12-13T05:31:11 http://ants.fluxid.pl/replay.30066 wtf, GYG 2011-12-13T05:31:44 he can raze 3 hills, and additional 2 if he kills few ants 2011-12-13T05:39:02 *** memolicious has quit IRC (Quit: Page closed) 2011-12-13T05:40:01 @rankings 2011-12-13T05:40:02 BenJackson: Top 10 players: pguillory(93.7), FlagCapper(91.4), GreenTea(90.5), Memetix(90.3), delineate(89.7), a1k0n(89.5), fourmidable(89.2), Hammerok(88.1), ThisIsNotABug(88.0), xathis(87.9) 2011-12-13T05:40:09 xathis reenters the top 10 2011-12-13T05:40:16 *** liberforce has joined #aichallenge 2011-12-13T05:41:40 *** Garf_ has quit IRC (Quit: Make a new plan, Stan!) 2011-12-13T05:41:48 Fluxid: looks like hos bot jammed 2011-12-13T05:41:52 almost no ants moving 2011-12-13T05:41:52 *** Garf has joined #aichallenge 2011-12-13T05:42:19 *** replore_ has quit IRC (Remote host closed the connection) 2011-12-13T05:42:46 oh oh :P 2011-12-13T05:44:26 *** TL__ has quit IRC (Ping timeout: 258 seconds) 2011-12-13T05:45:07 *** pairofdice has joined #aichallenge 2011-12-13T05:46:35 *** ikaros has quit IRC (Ping timeout: 252 seconds) 2011-12-13T05:48:11 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-12-13T05:49:00 *** Conorach has joined #aichallenge 2011-12-13T05:49:42 man, it's a pity i couldn't participate this challenge. just watched some top games and they're very cool 2011-12-13T05:55:10 Nooooo, last week of ai class 2011-12-13T05:59:07 Also, xathis is back in top 10 2011-12-13T06:00:42 *** McLeopold has quit IRC (Read error: Connection reset by peer) 2011-12-13T06:02:24 *** g0llum has joined #aichallenge 2011-12-13T06:02:29 *** McLeopold has joined #aichallenge 2011-12-13T06:10:34 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-13T06:14:53 *** Kurnevsky has joined #aichallenge 2011-12-13T06:39:57 *** GeorgeSebastian has joined #aichallenge 2011-12-13T06:51:07 *** ikaros has joined #aichallenge 2011-12-13T06:53:33 *** SMJ has joined #aichallenge 2011-12-13T07:06:58 *** mleise has joined #aichallenge 2011-12-13T07:07:23 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-13T07:09:58 *** ikaros has joined #aichallenge 2011-12-13T07:25:30 *** nfn has quit IRC (Quit: leaving) 2011-12-13T07:31:37 *** travelan has joined #aichallenge 2011-12-13T07:31:49 hi 2011-12-13T07:32:16 i was wondering... when the challenge ends this sunday.. what will be next? :) 2011-12-13T07:32:53 Despair 2011-12-13T07:33:58 :P 2011-12-13T07:34:04 i mean, will there be a new challenge soon? 2011-12-13T07:34:05 Finals 2011-12-13T07:35:33 Either spring or fall of next year sounds probable 2011-12-13T07:36:09 but are there any ideas what kind of challenge will it be? 2011-12-13T07:37:19 *** nfn has joined #aichallenge 2011-12-13T07:37:50 worms! 2011-12-13T07:39:00 oh, are we evolving? Like, by 2050, we'll have humans? 2011-12-13T07:39:06 bacteria 2011-12-13T07:39:26 *** flowenol has joined #aichallenge 2011-12-13T07:39:37 *** Cyndre_ has joined #aichallenge 2011-12-13T07:40:54 Mmm, Tanks! 2011-12-13T07:42:19 >physics simulation 2011-12-13T07:43:29 *** Cyndre__ has quit IRC (Ping timeout: 252 seconds) 2011-12-13T07:46:10 *** QuirionPT has joined #aichallenge 2011-12-13T07:46:18 because I just found out about this place, and I really want to participate, but there only are a few days left.. 2011-12-13T07:46:29 Yeah, it's a bit late now 2011-12-13T07:46:59 But you can still play around with the starter packages 2011-12-13T07:47:35 yeah... I know, but then the competitive element is gone 2011-12-13T07:47:52 Take it as preparation for the next round ;) 2011-12-13T07:48:56 hehe, I will ;) 2011-12-13T07:50:49 *** u_ has joined #aichallenge 2011-12-13T07:51:28 Oh, top-down counter-strike sounds good ;p 2011-12-13T07:52:14 *** goffrie_ has joined #aichallenge 2011-12-13T07:52:14 *** goffrie has quit IRC (Disconnected by services) 2011-12-13T07:52:15 *** goffrie_ is now known as goffrie 2011-12-13T07:53:00 *** vjacob has quit IRC (Quit: Leaving) 2011-12-13T07:57:39 *** monjo has joined #aichallenge 2011-12-13T08:02:03 *** monjo has quit IRC (Ping timeout: 258 seconds) 2011-12-13T08:08:11 *** zyberkiddy has joined #aichallenge 2011-12-13T08:23:31 *** meatroaf has quit IRC (Ping timeout: 258 seconds) 2011-12-13T08:29:57 *** TheLinker has joined #aichallenge 2011-12-13T08:30:39 *** GeorgeSebastian has quit IRC (Ping timeout: 252 seconds) 2011-12-13T08:37:58 *** travelan has quit IRC (Quit: Page closed) 2011-12-13T08:38:14 *** travelan has joined #aichallenge 2011-12-13T08:44:42 *** xathis has joined #aichallenge 2011-12-13T08:51:05 *** TheLinker has quit IRC (Quit: Bye) 2011-12-13T09:18:50 Hurray, my bot manages to kill the sample bots! now off to win the competition! xD 2011-12-13T09:19:17 :) 2011-12-13T09:19:41 *** sigh has quit IRC (Remote host closed the connection) 2011-12-13T09:24:02 f5f5f5 2011-12-13T09:24:05 in game right now 2011-12-13T09:24:25 will it get to first 20? 2011-12-13T09:25:23 Fluxid: loss! 2011-12-13T09:25:35 stfu 2011-12-13T09:25:42 not yet! 2011-12-13T09:26:26 haha 2011-12-13T09:31:05 awww, third 2011-12-13T09:31:24 5... 2011-12-13T09:31:52 same game ;) 2011-12-13T09:32:18 more energy went to defence than offence 2011-12-13T09:33:29 I keep seeding games with people 300 ranks below me 2011-12-13T09:33:29 pguillory timeout :) 2011-12-13T09:33:34 kinda worthless games 2011-12-13T09:35:20 Kurnevsky: if not that timeout i'd be dead for sure 2011-12-13T09:35:39 and 1000 turns too, he'd slam me with huge force 2011-12-13T09:35:45 buuuuut 2011-12-13T09:35:55 26 place after another game 2011-12-13T09:35:56 yaaaay 2011-12-13T09:36:38 *** Palmik has quit IRC (Ping timeout: 240 seconds) 2011-12-13T09:37:00 *** bergmark has joined #aichallenge 2011-12-13T09:49:19 *** UncleVasya has joined #aichallenge 2011-12-13T09:49:38 how do people find target sets like not recently seen squares? 2011-12-13T09:49:46 do you bfs from all ants to find the border? 2011-12-13T09:49:52 then search from the border back in? 2011-12-13T09:50:22 BenJackson, thestinger: ^? :P 2011-12-13T09:50:42 I do not understand 2011-12-13T09:50:51 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-13T09:50:53 why people do bfs from ants? 2011-12-13T09:51:11 well say you want to search from not recently seen locations to ants, how are you going to find the set of not recently seen locations? 2011-12-13T09:51:27 just bfs from ALL not recently seen locations? or bfs from ants to find the border of not recently seen locations? 2011-12-13T09:51:36 why I will want it? 2011-12-13T09:51:50 so you can revisit not recently seen squares for food 2011-12-13T09:52:16 Are there any cases when search from target (hill, food, etc) to my ants is not enough? 2011-12-13T09:52:37 my problem is when you don't easily already have the set of targets 2011-12-13T09:52:45 ie. not recently seen locations 2011-12-13T09:52:54 I make search from target and when ant is found he steps on the father node in search 2011-12-13T09:53:01 yes i do that too 2011-12-13T09:53:07 ah, ok 2011-12-13T09:53:07 but i need the set of targets to start the search 2011-12-13T09:53:31 I make this set by myself 2011-12-13T09:53:42 well obviously, but how 2011-12-13T09:54:19 track how many turns since you last saw each square? 2011-12-13T09:54:26 i am doing that 2011-12-13T09:54:38 but should i search from all not recently seen squares or the border of not recently seen squares? 2011-12-13T09:54:57 I don't search so I dunno from there hah 2011-12-13T09:55:03 I check every square in map. If this square is visible and one or more of his neighbours is invisible for n turns - this is internal boundary. 2011-12-13T09:56:30 I've decided do not try any optimizations here because chck once every square isn't time-consuming if compare to other staff I do in bot. 2011-12-13T09:57:13 how fast are the server computers that run our bots? Compared to a regular laptop core 2 duo 2011-12-13T09:57:32 should I expect a major slow down? 2011-12-13T09:57:57 speed up, but I believe that shouldn't matter much for timeouts since it is wall time 2011-12-13T09:58:17 cool, thanks :) 2011-12-13T10:02:06 antimatroid1: So when I have a list of internal boundary squares I do bfs from them simultaniously. 2011-12-13T10:03:30 UncleVasya, I do exactly the same ;P 2011-12-13T10:04:06 UncleVasya, except I use a slightly different method for deciding which squares are sources for the bfs 2011-12-13T10:05:22 but still depending only on the number of turns each square was last seen 2011-12-13T10:16:59 *** Anilm3 has joined #aichallenge 2011-12-13T10:17:19 @rankings 2011-12-13T10:17:20 Anilm3: Top 10 players: pguillory(92.6), FlagCapper(92.4), Memetix(90.7), xathis(90.6), a1k0n(90.5), fourmidable(89.6), delineate(89.0), ThisIsNotABug(88.7), jacob_strauss(88.1), Hammerok(87.7) 2011-12-13T10:17:27 Anilm3: hi 2011-12-13T10:17:36 Hi Uncle! 2011-12-13T10:17:53 huh, Hammerok got into top10 with Python bot 2011-12-13T10:17:59 i'm kind of sad 2011-12-13T10:20:04 Fluxid: you still have 5 days! 2011-12-13T10:23:19 *** pguillory has joined #aichallenge 2011-12-13T10:24:13 Anilm3: nah, i think i'm done, gonna finally think about other things than bot 2011-12-13T10:25:03 start thinking about the next challenge :P 2011-12-13T10:25:40 Anilm3: i have no ideas... 2011-12-13T10:26:17 *** bhasker has joined #aichallenge 2011-12-13T10:26:17 ants are awesome, can't think of anything better 2011-12-13T10:26:38 Fluxid: for what it's worth, top30 is great 2011-12-13T10:27:00 Fluxid: yeah, ants are great 2011-12-13T10:28:03 *** avdg has joined #aichallenge 2011-12-13T10:28:26 well then, see you later 2011-12-13T10:28:31 nap time 2011-12-13T10:28:33 *** Anilm3 has quit IRC (Quit: Lost terminal) 2011-12-13T10:29:14 *** treeform has joined #aichallenge 2011-12-13T10:31:13 *** mviel_ has joined #aichallenge 2011-12-13T10:31:58 McLeopold, janzert: are there any ideas for the next challenge? ;) 2011-12-13T10:34:34 *** mviel has quit IRC (Ping timeout: 252 seconds) 2011-12-13T10:38:52 I am glad I know about this and can do the next one from the beginning rather than less than a month before it starts 2011-12-13T10:46:56 *** Areks has joined #aichallenge 2011-12-13T10:48:34 *** iglo has joined #aichallenge 2011-12-13T10:49:12 http://ants.fluxid.pl/replay.30399 what a lame game 2011-12-13T10:49:31 also lol, "antsinyourpants" 2011-12-13T10:50:27 http://ants.fluxid.pl/replay.30471 AAARGH BUG 2011-12-13T10:50:53 points for going to food < penalty points for getting onto hill x( 2011-12-13T10:51:24 Oh, that's embarrasing :) 2011-12-13T10:51:27 *** zyberkiddy has quit IRC (Ping timeout: 252 seconds) 2011-12-13T10:53:07 *** Antimony has joined #aichallenge 2011-12-13T10:56:43 *** mleise has quit IRC (Quit: Leaving.) 2011-12-13T10:57:22 You could pretend your won hills are water 2011-12-13T10:57:25 own 2011-12-13T10:58:28 So instead of doing that, it would route around 2011-12-13T11:05:04 pairofdice: there is one map where hill is right by the wall 2011-12-13T11:05:26 so there tile blocked by 3 water tiles and hill 2011-12-13T11:05:36 and in this tile sometimes food appear first 2011-12-13T11:05:41 so ant could go in there 2011-12-13T11:05:49 and could not route way out ;) 2011-12-13T11:05:53 trap sprung 2011-12-13T11:06:02 but yeah, there are options 2011-12-13T11:06:44 just put a high penalty on top of it, but don't block movements to hill (in case an ant gets stuck) 2011-12-13T11:06:57 I thought all hills are now inside a water-free 3x3 box 2011-12-13T11:07:31 it wasn't the case in the first maps 2011-12-13T11:07:33 clearly not 2011-12-13T11:13:55 personally, I only block movement onto hills if my bot thinks it might have stored food 2011-12-13T11:15:56 *** travelan has quit IRC (Quit: Page closed) 2011-12-13T11:21:06 @rankings 2011-12-13T11:21:07 pairofdice: Top 10 players: pguillory(92.6), FlagCapper(92.4), xathis(91.9), Memetix(90.7), a1k0n(90.2), fourmidable(89.6), delineate(89.1), ThisIsNotABug(88.6), jacob_strauss(88.1), Hammerok(87.9) 2011-12-13T11:21:35 this is more like it 2011-12-13T11:21:38 competitive top 10 2011-12-13T11:21:47 well, with top place not a given 2011-12-13T11:21:50 i'll get you, Memetix, if it's the last thing i doooooooooooo 2011-12-13T11:22:22 a1k0n: haven't lost interest then? :P 2011-12-13T11:22:38 heh. i don't think i'll touch my bot, i'm just rooting for it. 2011-12-13T11:23:14 there's no rational explanation for it competing so well, so i'm running with it 2011-12-13T11:24:40 i have no idea where i'll end up 2011-12-13T11:24:52 have you submitted yet? 2011-12-13T11:24:53 that's kind of cool i think, i'll get to see how it does in the finals 2011-12-13T11:24:55 nope 2011-12-13T11:24:57 still working on it 2011-12-13T11:25:01 haha 2011-12-13T11:25:05 on tcp at least? 2011-12-13T11:25:12 nope :P 2011-12-13T11:25:12 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-13T11:25:18 doomed 2011-12-13T11:25:26 i don't expect to win 2011-12-13T11:25:31 anywhere in the top 50 and i'll laugh 2011-12-13T11:25:32 it would be really awesome if you do, though 2011-12-13T11:25:40 *** JorgeB has joined #aichallenge 2011-12-13T11:26:23 i have no idea what the competition is actually like 2011-12-13T11:26:36 *** aerique has quit IRC (Quit: ...) 2011-12-13T11:26:46 antimatroid1: you should at least do one upload before to make sure there isn't some really lame problem, e.g. it doesn't compile because of different system settings, or it times out too easily, or... 2011-12-13T11:26:57 nha: i know 2011-12-13T11:26:57 you'd hate yourself if you fail because of some silly thing like that 2011-12-13T11:27:09 trust me, i know better than anyone on that last point :P 2011-12-13T11:27:31 although i don't really mind so much with my bot for this, i've done it for fun rather than seriously competing 2011-12-13T11:29:42 i do at least want to get 1st in australia 2011-12-13T11:29:55 atm that doesn't look very hard 2011-12-13T11:30:13 oh, except solifugid 2011-12-13T11:32:15 a1k0n: i don't like the amount of java in the top 5, i'm partially blaming you for that 2011-12-13T11:32:16 me to is hoping to beat solifugid. 2011-12-13T11:32:48 *too 2011-12-13T11:40:33 *** SMJ has quit IRC (Quit: Konversation terminated!) 2011-12-13T11:44:00 antimatroid1: I am sad because you do except only top 50 because I suppose my rank is about [your rank + 40] and I thought you are gonna to hit top 15. 2011-12-13T11:46:56 *** nha has quit IRC (Ping timeout: 240 seconds) 2011-12-13T11:53:02 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-12-13T11:57:14 antimatroid1: me!? 2011-12-13T11:58:46 damn, i was in 5th for a little while 2011-12-13T12:01:56 UncleVasya: i have no idea where my bot will end up 2011-12-13T12:01:58 i haven't tested it against other bots 2011-12-13T12:02:23 it could finish anywhere from top 10 to top 500 for all i know 2011-12-13T12:02:41 or fail on the contest servers 2011-12-13T12:03:51 *** amstan has joined #aichallenge 2011-12-13T12:03:51 *** ChanServ sets mode: +o amstan 2011-12-13T12:04:54 no top500. I am sure even my bot will end up in top150 at least, maybe even in top100. So I believe top 70 for you guaranteed. 2011-12-13T12:08:09 *** Antimony has quit IRC (Ping timeout: 255 seconds) 2011-12-13T12:16:15 *** Darhuuk has joined #aichallenge 2011-12-13T12:20:34 *** TheLinker has joined #aichallenge 2011-12-13T12:23:09 *** Blkt has quit IRC (Remote host closed the connection) 2011-12-13T12:28:20 McLeopold: did you ban that bot at least? 2011-12-13T12:31:02 *** mviel_ has quit IRC (Ping timeout: 248 seconds) 2011-12-13T12:31:28 *** liberforce1 has joined #aichallenge 2011-12-13T12:31:56 *** liberforce has quit IRC (Ping timeout: 252 seconds) 2011-12-13T12:35:01 amstan: wow, there is some interesting story about violating rules? Could you be more specific? :) 2011-12-13T12:35:23 UncleVasya: nah, it's this thread: http://forums.aichallenge.org/viewtopic.php?f=21&t=1117 2011-12-13T12:35:28 see last post 2011-12-13T12:35:49 will read, thanks. 2011-12-13T12:37:11 UncleVasya: you don't need to read the thread, just the last post. 2011-12-13T12:38:27 heh, just spam bot. I thought about something more interesting 2011-12-13T12:38:46 UncleVasya: first time i read it, i tought it was some guy that didn't know english very well 2011-12-13T12:38:50 it partially made sense 2011-12-13T12:39:07 "Galcon was a rattling goodness selection" 2011-12-13T12:39:54 xathis got assasinated this game: http://aichallenge.org/visualizer.php?game=213756&user=4513 2011-12-13T12:41:21 can't believe it was generated by a program without human help. Maybe a human added such words as Galcon, Planet, Map etc and program genereted text with them? 2011-12-13T12:41:56 I wonder where these words come from 2011-12-13T12:42:11 from the thread above 2011-12-13T12:42:26 *** amstan__ has joined #aichallenge 2011-12-13T12:42:26 *** ChanServ sets mode: +o amstan__ 2011-12-13T12:42:27 * I can't believe bot decided to put these words in the text without human help 2011-12-13T12:43:15 UncleVasya: the spam bot probably just read all forums posts to generate its sample text 2011-12-13T12:43:47 amstan: I don't know how to bad bots... 2011-12-13T12:43:51 ban* 2011-12-13T12:44:01 It was very ergodic. 2011-12-13T12:44:08 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-13T12:44:12 heh, scary what computers can do nowadays :) 2011-12-13T12:46:20 *** liberforce1 has left #aichallenge 2011-12-13T12:48:59 hmm, cool homework at ai-class :-) 2011-12-13T12:49:08 *** Antimony has joined #aichallenge 2011-12-13T12:50:00 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-13T12:50:08 *** bhasker has quit IRC (Ping timeout: 240 seconds) 2011-12-13T12:51:15 (well, optional homework) 2011-12-13T12:51:31 ROT13 2011-12-13T12:51:57 Or something, haven't actually done it yet 2011-12-13T12:52:27 not rot13, its still not readable 2011-12-13T12:52:41 Ahh 2011-12-13T12:52:47 you just need to create a program that can predict the best combination 2011-12-13T12:52:52 kinda machine learning on text 2011-12-13T12:52:58 but I didn't watch the lectures yet 2011-12-13T12:53:25 Looks interesting 2011-12-13T12:54:19 my current view on it is to determine the probability of each character that comes after a certain character 2011-12-13T12:54:40 so you have a grid of 24 cols and 24 rows 2011-12-13T12:55:05 than calculate the penalty after processing some sample text, lowest win (depending on model) 2011-12-13T12:56:13 s/24/26/ 2011-12-13T13:00:53 *** Israfel has quit IRC (Read error: Operation timed out) 2011-12-13T13:03:05 *** Anilm3 has joined #aichallenge 2011-12-13T13:05:03 *** McLeopold has left #aichallenge 2011-12-13T13:06:32 *** Dlayne has joined #aichallenge 2011-12-13T13:06:52 Hi guys 2011-12-13T13:09:13 memetix resubmitted? nooooooooo 2011-12-13T13:09:42 You're doomed 2011-12-13T13:09:59 Also, you're 5th 2011-12-13T13:10:40 *** amstan__ has quit IRC (Quit: Konversation terminated!) 2011-12-13T13:10:50 *** amstan__ has joined #aichallenge 2011-12-13T13:10:50 *** ChanServ sets mode: +o amstan__ 2011-12-13T13:10:56 hmm. i wonder if i can do anything to improve my bot, then. 2011-12-13T13:12:09 holy shit, flagcapper's in 1st! 2011-12-13T13:12:29 Did not expect that 2011-12-13T13:12:47 i did 2011-12-13T13:13:04 Why the holy, shit then :) 2011-12-13T13:13:14 well, only partially :) 2011-12-13T13:13:16 *** Israfel has joined #aichallenge 2011-12-13T13:14:14 his bot has always been really good, and getting better with every resubmission 2011-12-13T13:15:55 Okay, "decrypted" the first sentence. 2011-12-13T13:17:07 *** amstan__ has quit IRC (Ping timeout: 240 seconds) 2011-12-13T13:20:08 *** jstrong has joined #aichallenge 2011-12-13T13:20:38 antimatroid1: did you figure out that thing about exploration/last seen tiles? 2011-12-13T13:20:48 xathis is going up quick 2011-12-13T13:21:28 antimatroid1: I have two types of "exploration" (food searching and looking for enemy hills), but yeah - I just do a BFS from all targets 2011-12-13T13:21:52 but I prevent ants from going to the same target, and mark all targets in an area as taken when an ant is going to one 2011-12-13T13:22:53 and there's another trick to get a nice even spread for foraging by preventing them from going to a target behind one that is already taken 2011-12-13T13:23:02 ug, it looks like there's a big tradeoff between speed and code prettyness 2011-12-13T13:23:04 *** nfn has left #aichallenge 2011-12-13T13:24:15 Antimony: what language? C++? :P 2011-12-13T13:24:21 Who needs pretty code? *twitches* 2011-12-13T13:24:25 C++ is already ugly as hell, it's hard to make it uglier 2011-12-13T13:24:38 wnna bet? 2011-12-13T13:24:44 use C++11 :P 2011-12-13T13:24:52 lots of stuff to make it faster/prettier 2011-12-13T13:25:00 I do, or at least the parts that my compiler supports 2011-12-13T13:25:06 are you on windows? 2011-12-13T13:25:10 no lambdas or foreach :( 2011-12-13T13:25:21 I mostly just use auto 2011-12-13T13:25:25 mingw has gcc 4.6 on windows 2011-12-13T13:25:30 oh 2011-12-13T13:25:51 I've thought about trying to upgrade before, but I'm always too lazy or scared 2011-12-13T13:25:57 I just used it when I was messing around with a java zero-day and shellcode crap 2011-12-13T13:27:36 lambdas are really nice :) 2011-12-13T13:27:39 antimatroid1: I have a "frontier" funtion that takes a predicate and produce the boundary between the yes/no areas of the predicate 2011-12-13T13:27:49 they don't seem to really hurt speed at all either, because you end up making the code a lot smaller 2011-12-13T13:27:56 initially 'ever seen' and now something like 'seen recently' 2011-12-13T13:28:09 if all goes well, my new version will soon be almost as good as my old one 2011-12-13T13:28:17 xD 2011-12-13T13:28:20 that's how I feel 2011-12-13T13:28:31 my bot _finally_ beats my pre-combat bot 2011-12-13T13:28:32 so I ran v11 overnight on fluxid 2011-12-13T13:28:42 I'm not sure it played strong enough competition to really know if it is improved 2011-12-13T13:28:55 fluxid must be really busy 2011-12-13T13:29:11 BenJackson: yeah, it has tons on people on it 2011-12-13T13:29:17 there are 10-20 other games between my games 2011-12-13T13:29:23 proper matchmaking would be really nice but it's a bit too late into the contest for that to happen 2011-12-13T13:30:00 My bot circles my pre-combat bot and gives him funny looks 2011-12-13T13:30:12 and then lets it live 2011-12-13T13:30:25 first time my combat bot met my pre combat bot the combat bot just went from 1 wide to 2 wide at the front of the snake 2011-12-13T13:30:28 ate pre-combat alived 2011-12-13T13:30:30 -d 2011-12-13T13:30:37 my code would be like half the size if I had foreach 2011-12-13T13:30:56 BenJackson: I like your bots snake, it piles into walls 2011-12-13T13:30:58 v4 that is 2011-12-13T13:31:11 I should do a regex search for "for (auto x = y.begin()..." and see how many hits I get sometime 2011-12-13T13:31:14 Antimony: the gcc on the site does not have for(a: b) in c++11 2011-12-13T13:31:21 I know 2011-12-13T13:31:23 neither do I 2011-12-13T13:31:36 If I really cared, I could include boost foreach, but it's annoying 2011-12-13T13:32:00 I could practically get c++11 behavior with "typedef vector::iterator auto" 2011-12-13T13:32:01 I probably should do some refactoring with boost when I have time 2011-12-13T13:32:07 lol 2011-12-13T13:32:16 I iterate over all kinds of containers 2011-12-13T13:32:23 *** ikaros has joined #aichallenge 2011-12-13T13:33:36 the sad part is that all of this would be one line in Python 2011-12-13T13:35:44 Now that I've seen how Python does containers, C++ STL seems so clunky and verbose 2011-12-13T13:35:56 and lacking in common features 2011-12-13T13:38:41 your not the first saying that 2011-12-13T13:38:47 *you're 2011-12-13T13:39:29 man i'm going to be lucky to stay in the top 10 at this rate 2011-12-13T13:39:48 http://aichallenge.org/visualizer.php?game=220187&user=2255 this game made me lose mu 2011-12-13T13:39:57 frickin' 16 year old kids ;) 2011-12-13T13:40:28 pguillory: it's worth it to see xathis play 2011-12-13T13:40:53 oh well... still higher than my last upload but sigma is still 2.57 2011-12-13T13:42:17 BenJackson: yep, the man's game is tight 2011-12-13T13:43:12 @rankings 2011-12-13T13:43:13 Anilm3: Top 10 players: FlagCapper(92.6), pguillory(92.5), xathis(92.4), a1k0n(90.3), fourmidable(90.2), ThisIsNotABug(89.0), delineate(88.7), jacob_strauss(88.0), Hammerok(87.8), JG.WAS(87.6) 2011-12-13T13:43:41 heh 2011-12-13T13:43:52 the top 10 is starting to look like fluxid's tcp server now :) 2011-12-13T13:43:59 whoa i'm in 4th now? 2011-12-13T13:44:17 (a distant 4th, but still) 2011-12-13T13:44:22 a1k0n: GreenTea resubmited 2011-12-13T13:44:27 these scores are really tightly clustered 2011-12-13T13:44:29 ah, no wonder 2011-12-13T13:44:47 well shit, i haven't touched my bot in weeks 2011-12-13T13:44:53 one week ago top 100 looked attainable...1 week later...hrrrmpff! 2011-12-13T13:45:24 yeah 2011-12-13T13:45:55 I've fallen from 56th to 76th, despite increasing skill slightly 2011-12-13T13:46:48 I hover just out of the top 10, getting in when people resubmit :) 2011-12-13T13:47:08 so my plan is to be 0.5 better and solidly inside :) 2011-12-13T13:47:13 i thought your bot would be in the top 10 2011-12-13T13:47:19 *** Palmik has joined #aichallenge 2011-12-13T13:47:37 i don't understand why it isn't better than mine, really 2011-12-13T13:47:46 my next version has some radical changes so I'm not sure what will happen 2011-12-13T13:51:24 a1k0n: his bot just loves being an underachiever on the official site 2011-12-13T13:51:52 *** Rinum has joined #aichallenge 2011-12-13T13:52:27 haven't been around in a while... so much has changed! 2011-12-13T13:52:30 My bot was uploaded at November 20 and for a long time had skill near 83.5. But from last week it jumped in skill to about 85 and sits there. Have no idea, why. The reason could be the grow of total amount of the bots but it doesn't seem to grow fast in last week. 2011-12-13T13:52:43 yeah, skill is all relative to everyone else 2011-12-13T13:52:46 new maps! a disruption in the rankings! 2011-12-13T13:53:27 I'm pretty much done at this point, not sure what else to code :P 2011-12-13T13:53:31 i was about to give up on my bot but if i have a chance at 4th... 2011-12-13T13:53:33 need to merge in 11 branches though... 2011-12-13T13:53:47 i hvae a bunch of numbers pulled out of my butt i could optimize i guess 2011-12-13T13:54:00 a1k0n: yeah, but that's really hard to measure 2011-12-13T13:54:07 no kidding 2011-12-13T13:54:11 you really need a significant improve to measure a difference on tcp 2011-12-13T13:54:12 I don't know if I should bother with this competition now x.x ... I'd like to maybe break into the top 100 2011-12-13T13:54:19 which is why I have so much trouble deciding if I should merge stuff 2011-12-13T13:54:29 my bot can go from rank 8 to rank 44 for no reason 2011-12-13T13:54:35 heh 2011-12-13T13:54:39 and then back up 2011-12-13T13:54:46 I think my skill *number* jumped up because of better competition in the 20..50 range 2011-12-13T13:54:50 So the growth of skiil without resubmitting means that other bots became worst? I understand when bot growth in rankings because top guys resubmit often, but does it affect the skill? 2011-12-13T13:55:09 which made room in the skill ranks for improving bots and pushed the mid 80s up 2011-12-13T13:55:15 I have now got combat sort of working again in the most horribly unoptimized way possible 2011-12-13T13:55:33 or because other bots got better but you're still not losing too badly to them 2011-12-13T13:55:41 BenJackson: yeah, mu is getting higher for the top 50 2011-12-13T13:55:52 Another explanation is that current version feels that the next one is ready and tries to do its best to not be replaced :) 2011-12-13T13:56:01 haha 2011-12-13T13:56:08 or, the new servers have more CPU power 2011-12-13T13:56:14 I went from around 87 before to a bit over 90, but my bot always end up at rank 15 to 30 2011-12-13T13:56:17 or, the new maps changed dynamics slightly 2011-12-13T13:56:18 Or, none of the above. 2011-12-13T13:56:38 *** Redgis has joined #aichallenge 2011-12-13T13:56:39 Slight inlfation is to be expected 2011-12-13T13:56:40 so who is Q on ants.fluxid.pl? 2011-12-13T13:56:43 _flag 2011-12-13T13:56:44 flagcapper 2011-12-13T13:56:45 flag 2011-12-13T13:56:46 thought so 2011-12-13T13:56:59 would you guys be interested in continuing the contest after the contest? 2011-12-13T13:57:09 it looks like xathis might get back into 1st then 2011-12-13T13:57:16 his sigma is still >2 2011-12-13T13:57:21 Anilm3: meh, I've wasted too much time on this already :P 2011-12-13T13:57:31 anyway, the top bots will release their code and it won't really be fun anymore 2011-12-13T13:57:32 99.9 mu 2011-12-13T13:57:33 heh 2011-12-13T13:58:21 Anilm3: such a question was asked during previous one too. People said 'sure' and one guy created a TCP. But no one did really use it :) 2011-12-13T13:58:30 i hope xathis, pguillory, and flag are good blog writers 2011-12-13T13:58:34 thestinger: that's true 2011-12-13T13:58:41 do they have existing blogs? 2011-12-13T13:58:42 haha 2011-12-13T13:58:48 flag does: flagcapper.com 2011-12-13T13:58:57 And greentea 2011-12-13T13:58:57 xathis has a flickr xD 2011-12-13T13:59:11 oh, he has twitter 2011-12-13T13:59:15 i don't have a blog but i'll write about the bot 2011-12-13T13:59:23 he can make 200 tweets explaining his bot 2011-12-13T13:59:26 cool. 2011-12-13T13:59:27 haha 2011-12-13T14:00:32 http://ants.fluxid.pl/replay.30676 2011-12-13T14:00:39 when I look at that game I mostly like what I did there 2011-12-13T14:00:43 but M24 did it better 2011-12-13T14:01:03 xathis: if this is really your flickr, then you sir, are beyond awesome! http://www.flickr.com/photos/xathis/5066027237/in/photostream 2011-12-13T14:01:17 yaaaay, i'm 20th! 2011-12-13T14:01:40 I'm tempted to submit my current bot with no optimization or island decomposition and see how well it does 2011-12-13T14:01:52 bqf: hahaha i took the photo but i'm not one of those guys 2011-12-13T14:02:33 xathis: aw, too bad. that lowers you down to "awesome" for knowing those people. 2011-12-13T14:02:48 silly tcp rankings 2011-12-13T14:02:57 haha der hammer 2011-12-13T14:03:04 my branches other than _stupid all dropped down a bunch 2011-12-13T14:03:29 you should reconsider your naming conventions then 2011-12-13T14:03:48 now all your bots are worst than stupid 2011-12-13T14:03:51 for _stupid I intentionally broke several things to see what would happen... 2011-12-13T14:04:18 didn't your organization say darmstadt before? am i crazy? 2011-12-13T14:04:57 BenJackson: know what really impresses me about M24 in that game? the way at turn 295 he has an enemy near his hill but he sends ants past it toward the frontier. really good judgement 2011-12-13T14:05:03 oh that's olex.s 2011-12-13T14:05:11 aww man, I get a compile error on the server even though it compiles on my own computer 2011-12-13T14:05:36 are you using gcc 4.6 now? 2011-12-13T14:05:42 xathis: when's your next game? 2011-12-13T14:05:42 pguillory: or just lucky? 2011-12-13T14:06:15 Anilm3: "Next game should be within 67 minutes" 2011-12-13T14:06:40 that's too long.. 2011-12-13T14:06:59 he had enough ants around the hill, shrug 2011-12-13T14:07:06 I'm using GCC 4.4.1 2011-12-13T14:07:15 pguillory: believe it or not this is the first version where I've assigned individual goals to ants other than for food 2011-12-13T14:07:45 I've known I wanted to do it 2011-12-13T14:07:51 but got hung up on fancy ways 2011-12-13T14:07:56 and finally just did something more brute force 2011-12-13T14:08:51 Fluxid: http://ants.fluxid.pl/replay.30803 your ants are still too scared :P 2011-12-13T14:08:53 a1k0n: i don't think so. he has about 7 ants nearby to deal with it, and he keeps 2 around and sends 5 away. very smooth 2011-12-13T14:09:13 oh, true. i'd probably end up swarming the one ant 2011-12-13T14:09:40 thestinger: yeah, "newcombat" sucks cock a bit more than "oldcombat" 2011-12-13T14:10:30 my defense fully kicks in at 15 distance from my hills, so only one ant was actually trying to defend against you 2011-12-13T14:11:14 I finally graduated my defense 2011-12-13T14:11:18 *** Antimony has quit IRC (Ping timeout: 245 seconds) 2011-12-13T14:11:37 *** Dlayne has left #aichallenge 2011-12-13T14:11:58 I think I should just ignore skill on tcp now xD - my _newdefend branch is ranked lower, but I can't find a game where it actually failed due to defense 2011-12-13T14:12:02 BenJackson: how's your performance? i tried doing that but couldn't get consistent performance. i relied on a lot of shortcuts like grouping clusters of enemy ants together and treating them as one goal, but if they just decided to all spread out one turn... timeout :) 2011-12-13T14:12:59 pguillory: couldn't that be just go? :P 2011-12-13T14:13:13 *** Garf has quit IRC (Read error: Connection reset by peer) 2011-12-13T14:13:29 out of the 57 games I played on fluxid last night the worst turn was 377ms 2011-12-13T14:13:32 *** Garf has joined #aichallenge 2011-12-13T14:13:35 strikes me as a big-O type of problem 2011-12-13T14:13:46 Anilm3: Go is fine, the 6g/6l compilers are just meant to compile fast and not make fast code though 2011-12-13T14:13:47 nice 2011-12-13T14:14:00 I actually know how to better assign goals to ants... but an obvious heuristic is distance from goals to ants 2011-12-13T14:14:13 if there was more time I'd play with those things 2011-12-13T14:14:16 but I waffle too much 2011-12-13T14:14:34 I'm cycling back around to "ok forget elegant, let's pick up a few more skill points before it's too late" 2011-12-13T14:14:40 *** Rinum has quit IRC (Quit: Page closed) 2011-12-13T14:15:04 *** jstrong has quit IRC (Quit: Leaving.) 2011-12-13T14:15:07 Time to get the gloves off, eh? Maybe put on some brass knuckles? 2011-12-13T14:15:22 *** QuirionPT has quit IRC (Quit: Leaving) 2011-12-13T14:15:53 *** bluegaspode has joined #aichallenge 2011-12-13T14:17:43 *** epicmonkey has joined #aichallenge 2011-12-13T14:17:45 i gave up on tweaking a couple of weeks ago, recognizing that my core approach wasn't going to get me to where flag and xathis (and probably others) are now. since then i've been going for one hail mary after another, total rewrite after total rewrite, looking for a big fundamental leap 2011-12-13T14:17:48 and failing to find it 2011-12-13T14:18:18 i just can't figure out how to do much better than what i've already got out there, and it's really frustrating 2011-12-13T14:18:31 *** Scryer has joined #aichallenge 2011-12-13T14:19:02 pguillory: just scheduling "defender" ants individually instead of as a mass movement is what made a bot that could consistently beat my previous version 2011-12-13T14:19:25 I did it in a stupidish brute-force kind of way but it works 2011-12-13T14:19:43 I was inspired by watching xathis's great economy with ants 2011-12-13T14:19:46 that's the difference between my old version and _newdefend 2011-12-13T14:19:58 like I said here last night, now my bot looks like a drunk version of his bot :) 2011-12-13T14:20:00 I used to just make them all move on a gradient to enemies (weighted by distance to my hill) 2011-12-13T14:20:38 huh I got the "info:" line from fluxid but now it's sitting there 2011-12-13T14:21:58 stupid kernel upgrades... now I can't load modules until I reboot 2011-12-13T14:22:58 pguillory: is your combat method similar to what memetix posted? 2011-12-13T14:25:42 a1k0n: similar 2011-12-13T14:26:05 if that combat works then I could cut a ton of time out of my bot 2011-12-13T14:26:12 and spend all my time "strategizing" 2011-12-13T14:26:20 Are there another bots that use minimax and happy wth it? 2011-12-13T14:26:49 it needs careful consideration, but it's probably the best meta-approach. i don't even compute influence but i prolly should. 2011-12-13T14:26:49 *** Antimony has joined #aichallenge 2011-12-13T14:28:29 for each potential move, i check the worst case scenario if all nearby enemies converged toward it 2011-12-13T14:28:34 if my bot stopped randomly suiciding ants who knows how good it would be 2011-12-13T14:28:40 and i do this for enemies as well 2011-12-13T14:29:06 yeah. minimax from both sides 2011-12-13T14:29:33 my combat problems are more meta now. during non-critical parts of combt my ants don't do things that would improve their chances in n moves 2011-12-13T14:29:53 then i can exclude any friendly moves that *might* be outnumbered 2011-12-13T14:30:12 how do you ever attack, then? 2011-12-13T14:30:46 *** Paradoxial has joined #aichallenge 2011-12-13T14:31:22 I guess he sets up score for killing enemy higher then penalty for losing his own 2011-12-13T14:31:40 and i repeat that process until no more moves are excluded. the repetition is necessary because if one friendly ant decides not to attack, it reduces the potential focus of nearby enemies, which can cause other friendly ants to decide not to attack. chain reaction 2011-12-13T14:31:51 oh, brilliant 2011-12-13T14:31:58 eliminating dominated moves from both sides 2011-12-13T14:32:48 well, thanks! i wonder if i have time to implement that now, haha 2011-12-13T14:33:54 any move that is not excluded above but which may attack another ant is guaranteed to win, and so i give it a strong preference 2011-12-13T14:34:49 oops I'm probably going to time otu my fluxid bot running a 10p game locally 2011-12-13T14:35:11 god damn it. i wish i had thought of that weeks ago. i was contemplating the whole eliminating dominated moves thing but didn't work it out 2011-12-13T14:35:56 and like memetix, in certain situations i accept an even trade or worse. as my ants get backed up in a choke point, i start accepting different kinds of even trades, then eventually when they get really backed up i just charge, damn the torpedoes 2011-12-13T14:36:29 heh. you're good at building up walls without backing up too much. 2011-12-13T14:36:38 so I added a --pigpile option to playgame 2011-12-13T14:36:50 where a specified player is revealed to all bots 2011-12-13T14:37:13 just did a 10 player game of prev submit vs 1 new guy 2011-12-13T14:37:26 the area near his hill after he was wiped out is highly...multicultural 2011-12-13T14:37:31 HELL FROZE 2011-12-13T14:37:36 xathis reuploaded AGAIN 2011-12-13T14:37:39 lol 2011-12-13T14:37:44 =) 2011-12-13T14:37:55 uh wait 2011-12-13T14:37:59 Fluxid: uhm, what? o.0 2011-12-13T14:38:17 wtf 2011-12-13T14:38:19 Only 2 versions 2011-12-13T14:38:23 Something odd is happening 2011-12-13T14:38:28 i didn't see your on rankings 2011-12-13T14:38:42 i refreshed and the ranking was.. different 2011-12-13T14:38:50 fourmidable was third 2011-12-13T14:39:07 and i also memorize patterns in enemy behavior and use those to make predictions about their moves in the future, but i disabled that on the last upload to keep something up my sleeve :) 2011-12-13T14:39:17 mu 99.90? seriously? 2011-12-13T14:39:43 pguillory: haha. i saw your comment to that effect in your "about me" section 2011-12-13T14:39:56 i hope my random behavior defeats that 2011-12-13T14:40:21 detecting timed out bots tho, that would be useful 2011-12-13T14:41:06 unfortunately it also detects non-deterministic behavior as such and declines to make predictions :/ 2011-12-13T14:43:00 *** bluegaspode has quit IRC (Quit: Page closed) 2011-12-13T14:43:16 which is cool but amounts to jack when i have 25% less ants because i can't figure out how to spread out properly in the expansion phase of the game 2011-12-13T14:44:16 pguillory: if your bot can calculate what my will do so maybe you can just tell it to him so he doesn't need to think by itself? :D 2011-12-13T14:44:37 hahaha 2011-12-13T14:44:42 :p 2011-12-13T14:44:46 *** Attis has joined #aichallenge 2011-12-13T14:45:11 there should be an "order" to emit suggested moves for enemies 2011-12-13T14:45:23 "no, no! don't suicide into the... awww" 2011-12-13T14:46:05 it would be interesting if future competitions allowed communication between competitors 2011-12-13T14:46:26 really put the game theory to work, prisoner's dilemma and whatnot 2011-12-13T14:46:43 planet wars with alliances 2011-12-13T14:49:09 me to 2011-12-13T14:49:55 * avdg wonders what would happen if there would be a thing like teamwork in bots 2011-12-13T14:49:58 yea, in a way, this one is not a real multiplayergame, just more than 2 competing individuals here 2011-12-13T14:50:01 how is decided what topic an aichallenge has? is there a vote on the forum based on suggestions? 2011-12-13T14:50:02 I hope for something with 'believe or not'. 2011-12-13T14:50:47 Or some game where bot can suggest moves to others and after each round all bot get some info like: 2011-12-13T14:51:03 yeah 2011-12-13T14:51:11 whoops wrong channel 2011-12-13T14:51:20 *** nfn has joined #aichallenge 2011-12-13T14:51:34 i made a big mistake in this challenge: not reading forums 2011-12-13T14:51:34 Player 1 has made his own move; Player 2 has made a move suggested to him by Player1 2011-12-13T14:55:48 *** Accoun has quit IRC () 2011-12-13T14:56:33 *** Attis has left #aichallenge 2011-12-13T15:00:53 *** thestinger has quit IRC (Quit: brb) 2011-12-13T15:04:39 *** foRei has joined #aichallenge 2011-12-13T15:05:26 *** treeform has quit IRC (Remote host closed the connection) 2011-12-13T15:12:57 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-13T15:16:14 *** Accoun has joined #aichallenge 2011-12-13T15:18:05 *** thestinger has joined #aichallenge 2011-12-13T15:20:14 @rankings 2011-12-13T15:20:16 Anilm3: Top 10 players: xathis(93.2), FlagCapper(93.1), pguillory(92.7), a1k0n(90.3), fourmidable(90.1), ThisIsNotABug(89.2), delineate(88.7), Hammerok(88.0), jacob_strauss(87.7), JG.WAS(87.4) 2011-12-13T15:20:24 hohoho 2011-12-13T15:20:27 congrats xathis 2011-12-13T15:20:38 Woohoo :) 2011-12-13T15:20:48 my mu exceeded 100 2011-12-13T15:20:49 I love how quickly the rankings shift now that the game rate is really high 2011-12-13T15:21:41 I reloaded the page but I don't see that result 2011-12-13T15:21:48 oh there it goes 2011-12-13T15:21:55 janzert: caching of rankings page wonky? 2011-12-13T15:21:58 xathis: you have now another tough rival with flag 2011-12-13T15:22:22 I think if I resubmit I might get into the 90s skill wise 2011-12-13T15:22:29 but not mu>100 :) 2011-12-13T15:26:08 *** Rinum has joined #aichallenge 2011-12-13T15:26:30 for battles... should we use ceil(attack radius)? 2011-12-13T15:27:05 <_flag> Well, first was fun while it lasted :( 2011-12-13T15:27:37 :( 2011-12-13T15:27:52 the current attack radius is 5... so we take the sqrt of that... ~2.23 .... so if an ant is 3 squares away it won't be attacked? 2011-12-13T15:28:14 well, when xathis finally loses a game you might be in 1st again 2011-12-13T15:28:16 it's 2.23 so that an ant can attack light a knight in chess 2011-12-13T15:28:20 or is euclidean distance not used for battles? 2011-12-13T15:28:21 like a knight* 2011-12-13T15:28:22 _flag: "A" will always be first on fluxid in my heart 2011-12-13T15:28:26 <_flag> a1k0n: if* 2011-12-13T15:28:39 <_flag> BenJackson: Haha, thanks 2011-12-13T15:28:47 if he manages to be 1st in every game that would be truly astonishing 2011-12-13T15:28:53 thestinger: that's a great analogy! thanks! 2011-12-13T15:28:56 he's lost games already 2011-12-13T15:29:12 a1k0n: i already lost a few games with v2 2011-12-13T15:29:26 http://aichallenge.org/visualizer.php?game=213756&user=4513 2011-12-13T15:29:28 oh yeah 2011-12-13T15:29:39 a couple stalemates with GreenTea too 2011-12-13T15:30:00 my rank: 7584! 2011-12-13T15:30:02 wow, memetix has nice movement 2011-12-13T15:30:18 prev game: 3rd to xathis, resubmit, 1st vs 4 bots in the 5000s :) 2011-12-13T15:30:20 BenJackson: beyond the occasional problem it has had? 2011-12-13T15:30:41 it seems to be fine for me atm 2011-12-13T15:30:44 BenJackson: that's a moral boost isn't it? 2011-12-13T15:30:45 janzert: someone did @rankings here and I hit reload and saw very stale data 2011-12-13T15:31:01 @rankings 2011-12-13T15:31:02 Rinum: Top 10 players: xathis(93.2), FlagCapper(93.1), pguillory(92.7), a1k0n(90.5), fourmidable(90.1), ThisIsNotABug(89.2), delineate(88.7), Hammerok(88.0), jacob_strauss(87.7), JG.WAS(87.6) 2011-12-13T15:31:19 Anilm3: well not so much when it turns out the setup was one of those starter-kit-killer things 2011-12-13T15:31:30 haha 2011-12-13T15:31:35 I had 11 ants and 8 in the hive, all my opponents suicided on move 27 2011-12-13T15:31:37 sometimes it seems to show completely incorrect data, e.g. from a specific organization or language 2011-12-13T15:31:54 but hey, the game happened so fast I hadn't even checked to see if my submit built 2011-12-13T15:31:58 so yay more servers 2011-12-13T15:32:10 janzert: earlier someone didn't see xathis on top at all 2011-12-13T15:32:36 pguillory had a rough morning 2011-12-13T15:32:44 yeah 2011-12-13T15:33:05 BenJackson: right, it'll sometimes not show a whole bunch of the players 2011-12-13T15:33:11 yup 2011-12-13T15:33:30 pretty sure it's probably something with the memcache code but I've never looked into it 2011-12-13T15:33:39 i've frequently seen just the players with 'Other' as their organization 2011-12-13T15:34:07 maybe the memcache key is funky 2011-12-13T15:36:35 just owned by a bot who sends all his ants on 1 hill :p 2011-12-13T15:39:20 *** treeform has joined #aichallenge 2011-12-13T15:41:26 the next aichallenge will be a darpa urban challenge 2011-12-13T15:41:31 http://aichallenge.org/visualizer.php?game=217973&user=235 hahaha timeouts enerywhere 2011-12-13T15:41:34 *** jstrong has joined #aichallenge 2011-12-13T15:42:17 I'm tempted to set that up as a scenario 2011-12-13T15:42:21 what's that Fluxid? 2011-12-13T15:42:29 I think my bot would have snuck out of that corner and fed 2011-12-13T15:42:31 :p I'm already laughing without seeing the game 2011-12-13T15:42:40 It's a shame that the last DARPA challenge was in 2007 2011-12-13T15:43:07 hmm, what about flag cap 2011-12-13T15:44:33 @rankings 2011-12-13T15:44:34 Anilm3: Top 10 players: xathis(93.2), FlagCapper(93.1), pguillory(92.7), a1k0n(90.6), fourmidable(90.4), ThisIsNotABug(89.2), delineate(88.7), Hammerok(88.0), jacob_strauss(87.7), JG.WAS(87.6) 2011-12-13T15:44:35 Benjackson: did you check who wins ;-) 2011-12-13T15:45:11 Anilm3: autonymous cars driven by ai in the streets 2011-12-13T15:45:27 avdg: the one live ant wins 2011-12-13T15:45:38 BenJackson: the starter bot :p 2011-12-13T15:45:46 but I think he also could have suck out, gotten that food, suicided with the attacker near his base, then gotten more food and broken out 2011-12-13T15:45:52 oh, it was a starter? haha 2011-12-13T15:45:55 Google's driverless car has driven public streets for hundreds of thousands of kms 2011-12-13T15:46:05 Fluxid: that's cool 2011-12-13T15:46:18 Fluxid: I wonder if the main sponser would be an insurance company by then ;-) 2011-12-13T15:46:21 Pretty impressive 2011-12-13T15:47:07 I think I figured out why my bot doesn't compile on the server 2011-12-13T15:47:09 pairofdice: what? 2011-12-13T15:47:17 it assumes unsigned int and size_t are convertable 2011-12-13T15:47:17 seriously? 2011-12-13T15:47:56 http://en.wikipedia.org/wiki/Google_driverless_car 2011-12-13T15:48:09 "(230,000 km) with occasional human intervention" 2011-12-13T15:48:40 *** jerez_z has joined #aichallenge 2011-12-13T15:48:49 Antimony: are you compiling with warnings on? 2011-12-13T15:49:04 I am but I'm on a 32bit machine 2011-12-13T15:49:22 how is narrowing conversion an error anyway? Is the server using -werror? 2011-12-13T15:49:32 Also, Sebastian Thrun rocks 2011-12-13T15:50:24 whoaaaaaaaaa I'm rank 34 on offical! 2011-12-13T15:50:24 :D 2011-12-13T15:50:29 In August 2011, a humanly controlled Google driverless car was involved in the project's first crash near Google headquarters in Mountain View, CA. Google has stated that the car was being driven manually at the time of the accident. 2011-12-13T15:50:33 hahaha 2011-12-13T15:50:50 heh 2011-12-13T15:51:18 meh, they should enable safety features if the car was driven manually :p 2011-12-13T15:51:25 *** jerez_z has quit IRC (Client Quit) 2011-12-13T15:51:27 *** Palmik has quit IRC (Remote host closed the connection) 2011-12-13T15:52:51 yeah, that was funny 2011-12-13T15:53:01 what, there were no darpa challenges after 2007 at all? 2011-12-13T15:53:07 sure 2011-12-13T15:53:09 that's sad! 2011-12-13T15:53:11 are there any contest admins here? 2011-12-13T15:53:16 there is (or just was) a challenge to "un shred" paper 2011-12-13T15:53:36 I don't know if there were any *driving* challenges 2011-12-13T15:53:57 i mean, right, driving challenges 2011-12-13T15:54:19 Antimony: what do you need? 2011-12-13T15:54:50 ah, there was VisLab Intercontinental Autonomous Challenge in 2010 2011-12-13T15:54:56 (whether there are any contest admins available depends on what you need) 2011-12-13T15:54:57 does the server use werror when compiling? 2011-12-13T15:55:01 oh 2011-12-13T15:55:12 werror? 2011-12-13T15:55:13 it's not really important 2011-12-13T15:55:32 yeah, if it's just a question of how something is done you can look at the repo 2011-12-13T16:06:10 *** Kurnevsky has left #aichallenge 2011-12-13T16:12:13 ok, so now first three bots are the gods fighting at the top of the olymp 2011-12-13T16:12:18 *** bluegaspode has joined #aichallenge 2011-12-13T16:15:18 *** Rinum has quit IRC (Quit: Page closed) 2011-12-13T16:16:16 *** Antimony has quit IRC (Ping timeout: 268 seconds) 2011-12-13T16:16:42 *** mj41 has joined #aichallenge 2011-12-13T16:21:52 fluxid: they are beautiful to watch 2011-12-13T16:23:15 hmm check http://aichallenge.org/games.php 2011-12-13T16:23:40 Dec 13th 9:22pm with map http://aichallenge.org/map.php?map=cell_maze/cell_maze_p10_01.map 2011-12-13T16:23:55 (bleh, I coppied the link instead of the text) 2011-12-13T16:24:06 I wish bocsimacko to come and kick the asses of this trinity. 2011-12-13T16:24:35 his wife forbid. 2011-12-13T16:24:50 forbid? 2011-12-13T16:25:14 did not allow 2011-12-13T16:25:36 already found this word. 2011-12-13T16:25:54 heheh 2011-12-13T16:26:17 (hmm, its fixed by now) 2011-12-13T16:29:54 *** ltriant has joined #aichallenge 2011-12-13T16:34:16 *** UncleVasya has quit IRC (Ping timeout: 252 seconds) 2011-12-13T16:36:44 *** Antimony has joined #aichallenge 2011-12-13T16:37:25 *** Scryer has quit IRC (Quit: Page closed) 2011-12-13T16:39:12 *** amstan has joined #aichallenge 2011-12-13T16:39:12 *** ChanServ sets mode: +o amstan 2011-12-13T16:40:34 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-13T16:49:56 Is TCP-Server bhickey down (the web-interface) ? 2011-12-13T16:55:45 *** kapoc has joined #aichallenge 2011-12-13T16:56:16 @rankins 2011-12-13T16:56:17 Anilm3: I have no idea what you mean. 2011-12-13T16:56:17 @rankings 2011-12-13T16:56:18 Anilm3: Top 10 players: xathis(93.2), FlagCapper(93.0), pguillory(92.7), a1k0n(90.6), fourmidable(90.4), ThisIsNotABug(89.2), delineate(88.7), jacob_strauss(87.7), JG.WAS(87.6), bucash(87.1) 2011-12-13T16:56:41 i'm playing a game right now 2011-12-13T16:57:04 yeah, I was thinking it could be any time :p 2011-12-13T16:57:10 haha 2011-12-13T16:57:21 The last time I just did it when you finished your game 2011-12-13T16:57:24 now during a game 2011-12-13T16:57:38 I have powers 2011-12-13T16:57:44 * avdg is doing some dirty changes to his old and slow bot, so he might have at least some chances 2011-12-13T16:57:47 yeah last time you did before i knew i had played already 2011-12-13T16:57:51 xathis: once this comp is over 2011-12-13T16:57:54 you better release the source 2011-12-13T16:57:58 hahaha 2011-12-13T16:58:00 ..... or .... nothing will actually happen to you 2011-12-13T16:58:00 i won :) 2011-12-13T16:58:03 I'm just doing time calculations :p 2011-12-13T16:58:07 we're all hoping he does 2011-12-13T16:58:10 @rankings 2011-12-13T16:58:11 Anilm3: Top 10 players: xathis(93.2), FlagCapper(93.0), pguillory(92.7), a1k0n(90.6), fourmidable(90.4), ThisIsNotABug(89.2), delineate(88.7), jacob_strauss(87.7), JG.WAS(87.6), bucash(87.1) 2011-12-13T16:58:12 well, there goes my secret 2011-12-13T16:58:33 @rankings 2011-12-13T16:58:34 Anilm3: Top 10 players: xathis(93.6), FlagCapper(93.0), pguillory(92.7), a1k0n(90.6), fourmidable(90.4), ThisIsNotABug(89.2), delineate(88.7), jacob_strauss(87.7), JG.WAS(87.6), bucash(87.1) 2011-12-13T16:58:39 haha that's it 2011-12-13T16:58:50 i think xathis skill slowly hits a limit when he can't go any hi... oh, hey, he did 2011-12-13T16:59:05 you can get to 100! 2011-12-13T16:59:07 i just know it! 2011-12-13T16:59:28 I think xathis bot could be a little bit better and not put 10 ants on 1 enemy ;-) 2011-12-13T16:59:29 oh man.. look at that top 4 2011-12-13T16:59:35 so there is some room for improvements 2011-12-13T16:59:38 in the beta i had a skill higher than 100, it's possible 2011-12-13T16:59:56 amstan: the next beta don't forget to call me 2011-12-13T17:00:08 Anilm3: that's not how it works 2011-12-13T17:00:49 avdg: i was thinking about setting a limit on how many ants should follow a single enemy, never did it though 2011-12-13T17:01:03 xathis: just curious .. is your bot an "emergent" solution 2011-12-13T17:01:06 btw amstan, how are the topics for the aichallenges chosen? 2011-12-13T17:01:08 or is it very very finetuned 2011-12-13T17:01:09 amstan: anyone can enter the beta or is it restricted? 2011-12-13T17:01:18 in other words did you just implement XYZ principles and let them work 2011-12-13T17:01:23 i hope it will be open :) 2011-12-13T17:01:26 Anilm3: anyone can, but don't expect us to advertise it 2011-12-13T17:01:27 or were you sticking your hand in there coding in special cases and that type of thing 2011-12-13T17:01:30 oh and jstrong of course i will release the source, but as i've said a few times it's a mess i'm not really proud of :D 2011-12-13T17:01:34 Anilm3: if you want to help, then stick around and help 2011-12-13T17:01:43 I will :) 2011-12-13T17:01:45 ikaros: what topics? 2011-12-13T17:01:53 well the actual challenge.. 2011-12-13T17:01:59 ikaros: suggestions forum 2011-12-13T17:02:03 ah nice 2011-12-13T17:02:08 ill have to look that up 2011-12-13T17:02:22 or is it closed / for devs only 2011-12-13T17:02:30 it's not 2011-12-13T17:02:39 there's no such things as devs, unless you mean comitters 2011-12-13T17:02:48 well ok 2011-12-13T17:02:50 jstrong: something in between... most of the intelligent behaviour is not hardcoded but i've implemented some special cases 2011-12-13T17:02:53 perfect :) 2011-12-13T17:02:54 xathis: I actually hoped that the best bots actually calculate the possible moves of the locked enemies and the required numbers of ants to kill it (in case its not in the edge of the fog of war) 2011-12-13T17:03:09 xathis: okay :), just curious 2011-12-13T17:03:47 *** kapoc has left #aichallenge 2011-12-13T17:05:24 avdg: completely calculating how to kill it in multiple turns would take too long in most cases i think 2011-12-13T17:05:26 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-12-13T17:05:40 true 2011-12-13T17:05:52 but it could work for small "islands" 2011-12-13T17:06:09 avdg: but i sometimes send 10 of my ants to kill a single enemy, that's pretty stupid 2011-12-13T17:06:32 how agressively do you try to split combat up into islands/ 2011-12-13T17:07:01 (well, "islands" in sense of islands with enemies) 2011-12-13T17:07:31 I need to find a better word for it :p 2011-12-13T17:10:42 *** besh has joined #aichallenge 2011-12-13T17:13:24 the bot I just played who was #2354 was not a starter kit but not far off 2011-12-13T17:13:33 there is such a huge bot wasteland down there 2011-12-13T17:13:37 (no offense :) 2011-12-13T17:13:55 it is BenJackson 2011-12-13T17:14:31 heh, i know. my wife chided me for arrogance when i pointed that out 2011-12-13T17:14:55 I'm kind of surprised that trueskill causes them all to have 30ish skill 2011-12-13T17:15:11 * avdg is happy that his bot is in the top 2k again 2011-12-13T17:15:19 my skill is now 34.7 and that's 7337 place 2011-12-13T17:15:51 a1k0n: that's unfair on her part!! 2011-12-13T17:15:55 haha 2011-12-13T17:16:23 one thing is being arrogant and another very different is being objective :P 2011-12-13T17:17:14 I doubt there's much ego invested in the starter kits 2011-12-13T17:17:22 but the guy in 2354 clearly wrote some code 2011-12-13T17:17:42 meh 2011-12-13T17:17:45 you can be better than over 5000 of the submissions by simply not accidentally suiciding yourself 2011-12-13T17:17:46 its all about the time you spend 2011-12-13T17:17:50 yeah rofl 2011-12-13T17:17:52 probably followed the tutorial a bit 2011-12-13T17:17:52 because in the long run the starter kits will and you'll win 2011-12-13T17:18:10 i still have remnants of tutorial code in my bot 2011-12-13T17:18:20 the question is.. who cares 2011-12-13T17:18:30 *** Zannick has quit IRC (Ping timeout: 248 seconds) 2011-12-13T17:18:45 *** Zannick has joined #aichallenge 2011-12-13T17:18:47 meh the way they handled collision was GREAT 2011-12-13T17:18:55 it gets alittle more complicated later on but 2011-12-13T17:19:02 the tutorial collision detection code is just fine 2011-12-13T17:19:03 imo 2011-12-13T17:19:07 ikaros: it's just that spending most of my time in the top 20 I lose sight of what's going on elsewhere 2011-12-13T17:19:14 I hear that the top 50 or 100 are getting stronger 2011-12-13T17:19:21 I kind of project that across the top 1000 2011-12-13T17:19:30 but there probably aren't really 1000 active participants 2011-12-13T17:19:41 hmm 2011-12-13T17:19:42 you could check what by counting bots with > 1 submission 2011-12-13T17:19:56 maybe, >1 submission and >1 day apart 2011-12-13T17:19:57 whenever I resubmit it becomes momentarily interesting to see what is going on at position 500 or whatever 2011-12-13T17:20:03 and try to figure out what the heck they did 2011-12-13T17:23:39 starter bot + entire-map-routing is about 70, which puts you around 450 2011-12-13T17:24:10 I still want to write a .replay analysis tool 2011-12-13T17:24:51 looks like there's about 3000 active participants 2011-12-13T17:25:22 rank 3000 has mu = 45, and most people there submitted 2-4 times 2011-12-13T17:30:32 *** pguillory has quit IRC (Ping timeout: 258 seconds) 2011-12-13T17:32:06 after tearing out my old island code, now I have to redo island spliiting 2011-12-13T17:32:10 lol, I submitted 30 times in a row when I was trying to fix a C++11 thing 2011-12-13T17:32:34 kept getting weird compiler errors because gcc 4.5 doesn't have some of the stuff 4.6 is letting me use 2011-12-13T17:32:50 finally figured out that std::array.size() doesn't work properly in gcc 4.5 2011-12-13T17:32:52 Use gcc 4.4 like I do! 2011-12-13T17:32:58 lambdas :( 2011-12-13T17:33:10 array size seems to be working for me 2011-12-13T17:33:10 * avdg is stuck with gcc 4.2 2011-12-13T17:33:14 also 4.4 allows stuff that 4.5 doesn't allow 2011-12-13T17:33:18 I haven't submitted yet though 2011-12-13T17:33:19 Antimony: std::array? 2011-12-13T17:33:20 oh 2011-12-13T17:33:25 not C arrays 2011-12-13T17:33:37 it's weird that it works in 4.4 and 4.6 but not 4.5 2011-12-13T17:33:44 like std::array directions = {NORTH, EAST, SOUTH, WEST}; 2011-12-13T17:33:51 directions.size() is supposed to be a constexpr 2011-12-13T17:33:51 I'm actually using 4.2.1 at home 2011-12-13T17:34:15 but 4.5 doesn't actually implement constexpr, even though it has the keyword 2011-12-13T17:34:21 it was so confusing 2011-12-13T17:34:33 so yeah, like version 11 to version 41 of my bot was fixing that xD 2011-12-13T17:35:44 I was avoiding resubmitting until I was sure I was better 2011-12-13T17:35:44 using std::array instead of std::vector really speeds up some stuff where a lot of objects are being used 2011-12-13T17:35:46 it's kind of painful seeing how bad at combt the newest version of my bot is 2011-12-13T17:35:55 but I wasn't getting that much variety of games on fluxid when I tried yesterday 2011-12-13T17:35:58 so I figured I'd resub 2011-12-13T17:36:21 currently, it solves every single ant in the map as a single island 2011-12-13T17:36:27 std::array is basically sugar for a C array, but you have .begin(), .end(), .size(), etc. - so it's a drop-in replacement for a vector 2011-12-13T17:36:29 so it always runs out of time before comign up with anthing good 2011-12-13T17:36:32 I know 2011-12-13T17:36:35 I've used boost array before 2011-12-13T17:36:57 C++11 also has std::begin() and std::end() functions 2011-12-13T17:37:05 so you can mostly use C arrays everywhere 2011-12-13T17:37:16 but then I would have to change my code to use those instead of .begin() and .end() 2011-12-13T17:37:44 I always wondered why the new foreach didn't use .begin and .end 2011-12-13T17:37:56 so it's to work with c arrays and strings? 2011-12-13T17:38:05 yeah 2011-12-13T17:38:39 for templated functions mostly 2011-12-13T17:39:01 I think it's funny how c++0x basically does have "auto" as long as you call a function 2011-12-13T17:39:07 I'm tired. I'll try making my bot non braindead later 2011-12-13T17:39:23 eg template transform (I begin, I end, ...) { for(; begin != end; ++begin) { ... }} 2011-12-13T17:39:33 I use auto heavily 2011-12-13T17:39:35 that's all very clean and you never had to write giant_template::iterator anywhere 2011-12-13T17:40:02 but if you don't indirect through a function where the template argument will be automatically applied you have to write it out 2011-12-13T17:40:17 usually I only write the type if it is 3 characters or if I need implici casting 2011-12-13T17:40:38 *** bergmark has quit IRC (Ping timeout: 240 seconds) 2011-12-13T17:41:27 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-12-13T17:41:37 BenJackson: well, it's funny that C++ had type inference for functions/classes before local type inference 2011-12-13T17:41:54 like Go has 'i := something()' which is like 'auto i = something()', but no generic functions 2011-12-13T17:42:12 well, they have polymorphism with interfaces (using virtual methods) 2011-12-13T17:45:42 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-12-13T17:51:20 *** pairofdice has quit IRC (Quit: When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl.) 2011-12-13T17:51:21 *** McLeopold has joined #aichallenge 2011-12-13T17:52:23 thestinger: examples of Go still make me cringe 2011-12-13T17:52:49 although there was a time early in my C development when I might have loved the syntax 2011-12-13T17:53:07 back when I used only the necessary spaces and parenthesis :) 2011-12-13T17:56:39 *** pguillory has joined #aichallenge 2011-12-13T17:57:55 *** iglo has quit IRC (Remote host closed the connection) 2011-12-13T17:58:14 *** Harpyon has joined #aichallenge 2011-12-13T17:58:30 *** kaemo has joined #aichallenge 2011-12-13T18:02:19 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-12-13T18:03:14 *** Jak_o_Shadows has joined #aichallenge 2011-12-13T18:03:38 *** Hexren has joined #aichallenge 2011-12-13T18:03:58 *** Hexren has left #aichallenge 2011-12-13T18:04:35 *** tdubellz_ has joined #aichallenge 2011-12-13T18:05:23 BenJackson: are you familiar with python generators? 2011-12-13T18:05:50 *** tdubellz_ is now known as tdubz 2011-12-13T18:06:25 generators? 2011-12-13T18:06:42 amstan: function that use yield 2011-12-13T18:06:48 right.. 2011-12-13T18:06:55 or a class that supports __next__ 2011-12-13T18:07:03 isn't that an iterator? 2011-12-13T18:07:15 and maybe some other magic functions, yes, same thing almost 2011-12-13T18:07:30 generators can accept data as well 2011-12-13T18:12:38 *** tdubz has quit IRC (Read error: Connection reset by peer) 2011-12-13T18:16:06 *** Yexo has joined #aichallenge 2011-12-13T18:16:23 McLeopold: yes, very 2011-12-13T18:18:55 @rankings 2011-12-13T18:19:01 Anilm3: Top 10 players: xathis(93.6), FlagCapper(93.0), pguillory(92.6), fourmidable(90.6), a1k0n(90.4), ThisIsNotABug(88.9), delineate(88.9), JG.WAS(87.7), jacob_strauss(87.6), bucash(87.0) 2011-12-13T18:20:22 BenJackson: do you know if go can do the same thing? 2011-12-13T18:20:41 I think in c you would need a switch statement and need to save state 2011-12-13T18:21:20 in C/C++ there's no way to return multiple times. You could simulate the result but your generating code would be very aware of it 2011-12-13T18:21:33 go has goroutines which probably give you something quite similar 2011-12-13T18:21:50 your generator would be a goroutine and would feed your other thread 2011-12-13T18:21:54 but I'm no Go expert 2011-12-13T18:22:50 a C++ iterator is a good example of all you can accomplish in C++. 2011-12-13T18:23:27 eg input_iterator(cin) is similar to while True: yield sys.stdin.read() 2011-12-13T18:23:44 except I just gave you the complete python implementation and merely the NAME of the c++ :) 2011-12-13T18:26:09 sorry while True: yield int(sys.stdin.read()) 2011-12-13T18:28:40 ok, thx 2011-12-13T18:28:46 http://www.chiark.greenend.org.uk/~sgtatham/coroutines.html 2011-12-13T18:30:13 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-13T18:30:37 in C++ you could do that somewhat cleaner in a functor 2011-12-13T18:33:36 BenJackson: I'm switching my python bfs over to use functors now :) 2011-12-13T18:35:58 actually, I think I'm confused... 2011-12-13T18:36:04 *** ikaros has joined #aichallenge 2011-12-13T18:37:29 *** gree has joined #aichallenge 2011-12-13T18:42:41 *** gree has quit IRC (Quit: Page closed) 2011-12-13T18:43:38 *** greghaynes has quit IRC (Ping timeout: 240 seconds) 2011-12-13T18:44:53 *** greghaynes has joined #aichallenge 2011-12-13T18:48:48 *** bluegaspode has quit IRC (Quit: Page closed) 2011-12-13T18:49:16 *** bluegaspode has joined #aichallenge 2011-12-13T18:52:56 *** Anilm3 has quit IRC (Ping timeout: 240 seconds) 2011-12-13T18:54:12 *** Anilm3 has joined #aichallenge 2011-12-13T18:54:28 *** bluegaspode_ has joined #aichallenge 2011-12-13T18:55:32 McLeopold: is there existing python code to parse replays? 2011-12-13T18:55:40 or is the js visualizer the only consumer of the data? 2011-12-13T18:56:01 *** bluegaspode has quit IRC (Ping timeout: 258 seconds) 2011-12-13T18:57:44 BenJackson: only the js 2011-12-13T18:57:56 but it is just json, so it isn't to hard to parse 2011-12-13T19:00:30 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-12-13T19:01:05 BenJackson: multiple return values is just a crappy way to make up for the lack of tuples 2011-12-13T19:01:08 which C++ has now :) 2011-12-13T19:01:10 yeah, import json, json.load() is pretty easy :) 2011-12-13T19:01:20 the hard part is reconstructing the board state from the replay 2011-12-13T19:01:32 thestinger: not multiple return VALUES, multiple return TIMES 2011-12-13T19:01:46 like for (loc in a bfs...) yield loc 2011-12-13T19:01:58 oh, so generators 2011-12-13T19:02:09 I did implement tie() once (ala boost, matched with std::pair) because I was intrigued that it was possible 2011-12-13T19:02:21 (unrelated to generators, but related to tuples :) 2011-12-13T19:02:39 *** Antimony has joined #aichallenge 2011-12-13T19:03:07 you can do generators with goroutines but it's ugly and inefficient 2011-12-13T19:03:32 is anyone else more nervous about losing the games against 1000 ranked players than the top 20? each time I'm up for such a game I'm thinking "come on, don't have some stupid bug..." 2011-12-13T19:03:32 I really like generators in Python though 2011-12-13T19:05:05 the only real problem with python generators is that they can't be copied 2011-12-13T19:06:17 mozilla's javascript also has yield for making generators (and it has list comprehensions, so it might have generator expressions too) 2011-12-13T19:06:26 I think that's going to be in ecmascript 6 2011-12-13T19:06:34 *** treeform has quit IRC (Remote host closed the connection) 2011-12-13T19:06:38 anything that make javascript less painful is welcome :) 2011-12-13T19:07:01 Antimony: there's a thing in itertools to clone them 2011-12-13T19:07:04 with the obvious storage problems 2011-12-13T19:07:35 everything in haskell can basically be a "generator" due to lazy evaluation 2011-12-13T19:07:38 which is nice :) 2011-12-13T19:07:49 sometimes... 2011-12-13T19:07:51 yeah 2011-12-13T19:08:01 arrgh I want to rewrite my bot 2011-12-13T19:08:06 so whatever you do, don't extend the contest 2011-12-13T19:08:09 thestinger: coffeescript 2011-12-13T19:08:16 there's your answer 2011-12-13T19:08:20 yeah 2011-12-13T19:08:52 list comprehensions and generators are a performance boost when the VM actually supports them though 2011-12-13T19:09:02 with coffeescript it probably becomes slower 2011-12-13T19:09:17 well coffeescript is jsut a compiler 2011-12-13T19:09:19 it compiles to native JS 2011-12-13T19:09:28 it doesn't use a VM 2011-12-13T19:09:32 https://developer.mozilla.org/en/JavaScript/Guide/Iterators_and_Generators 2011-12-13T19:09:38 I don't see it 2011-12-13T19:09:42 yeah, but JS is basically the bytecode for coffeescript :P 2011-12-13T19:10:35 I don't see any clone in itertools 2011-12-13T19:13:09 tee 2011-12-13T19:14:54 *** xathis has quit IRC () 2011-12-13T19:15:27 thestinger: c++ has tuples now? 2011-12-13T19:15:34 *** bluegaspode_ has quit IRC (Ping timeout: 258 seconds) 2011-12-13T19:16:10 * avdg is just adding a few hacks in his code so his bot performs a bit better 2011-12-13T19:16:19 antimatroid1: yeah, std::tuple 2011-12-13T19:16:44 yeah just looking 2011-12-13T19:17:03 http://en.cppreference.com/w/cpp/utility/tuple 2011-12-13T19:17:23 I think C1X has anonymous structs, so C++11 added tuples 2011-12-13T19:17:23 yay, protocolocon is back near the top 2011-12-13T19:17:43 * antimatroid1 is a fan of rufes band :P 2011-12-13T19:20:05 I've got new exploring code, who wants to download a 54mb replay file with visualizations? :D 2011-12-13T19:20:07 McLeopold: there are two ways i know of to do iteration in Go: https://gist.github.com/1474598 2011-12-13T19:21:42 but channels are about 50 times slower than callbacks 2011-12-13T19:23:29 pguillory: what about coroutines? same thing? 2011-12-13T19:23:59 McLeopold: goroutines are basically concurrent coroutines 2011-12-13T19:24:11 and they can be multiplexed over real threads 2011-12-13T19:24:39 so, would you say go could be as fast a c, or as low level? 2011-12-13T19:24:43 by default it's just concurrent (not parallel), but you can change it 2011-12-13T19:24:53 it could be as fast as idiomatic C++, if they had a good compiler 2011-12-13T19:25:19 well 2011-12-13T19:25:21 can it be functional? or do you have to do oops? 2011-12-13T19:25:25 and they need compile-time generics 2011-12-13T19:25:33 it's not really OOP in the same way as java/C++ 2011-12-13T19:25:38 there's no inheritance 2011-12-13T19:25:42 pguillory: thanks for the example 2011-12-13T19:25:45 at all? 2011-12-13T19:25:46 but there are first-class functions and closures 2011-12-13T19:25:54 pguillory: I suspect "callbacks" get inlined 2011-12-13T19:25:57 well, you use interfaces (which is their form of polymorphism) 2011-12-13T19:26:20 BenJackson: I don't think the regular Go compilers do any inlining 2011-12-13T19:26:27 abstracts? 2011-12-13T19:26:36 also Scala makes that even prettier 2011-12-13T19:26:38 can you create a class with some pre existing code? 2011-12-13T19:26:38 well, it's like manual duck typing 2011-12-13T19:26:39 negative on the inlining :( 2011-12-13T19:26:47 but I shied away from using scala because I saw comments about JVM perf problems on the forum 2011-12-13T19:26:50 gccgo does inlining 2011-12-13T19:27:17 the 6g/6l, 8g/8l, etc. compilers are just supposed to be fast - they're based on the plan9 C compilers 2011-12-13T19:27:23 they don't optimize like gcc/llvm 2011-12-13T19:27:32 *** Vaenom has quit IRC (Quit: Page closed) 2011-12-13T19:28:14 having worked on a plan9 derivative I instantly hold anything plan9 in contempt until proven otherwise 2011-12-13T19:28:33 wow people 2011-12-13T19:28:43 @rankings 2011-12-13T19:28:44 BenJackson: lots of stuff in plan9 is nice 2011-12-13T19:28:44 Anilm3: Top 10 players: xathis(94.2), FlagCapper(93.2), pguillory(92.6), fourmidable(90.5), a1k0n(90.2), ThisIsNotABug(89.0), delineate(88.8), JG.WAS(87.8), jacob_strauss(87.6), bucash(86.9) 2011-12-13T19:29:16 *** replore_ has joined #aichallenge 2011-12-13T19:29:22 BenJackson: bind mounts, /net, etc. 2011-12-13T19:29:32 bind mounts are kind of neat 2011-12-13T19:29:37 everything _really_ being a file (unlike *nix) is nice 2011-12-13T19:29:38 go a1k0n! 2011-12-13T19:29:45 Linux has absorbed most of the plan9 stuff though 2011-12-13T19:29:45 *** treeform has joined #aichallenge 2011-12-13T19:29:51 there's a kernel module in mainline for 9p, etc. 2011-12-13T19:29:59 everything really being a file is neat, but then everything (ok not everything) really being a STRING is not so neat 2011-12-13T19:30:03 eg errors 2011-12-13T19:30:06 yeah 2011-12-13T19:30:15 I've read a lot of the kernel 2011-12-13T19:30:16 *** Redgis has quit IRC (Ping timeout: 252 seconds) 2011-12-13T19:30:24 my impression is that it was interesting ideas implemented by grad students 2011-12-13T19:30:34 but not really polished like version 7 unix and its derivatives 2011-12-13T19:30:40 well, I think Ken Thompson and Rob Pike really aim for the most pragmatic solutions 2011-12-13T19:30:43 they don't like abstraction 2011-12-13T19:30:44 * avdg wonders if xathis could reach 100 2011-12-13T19:30:54 and they don't like doing things _right_, they would rather have it simple 2011-12-13T19:30:57 avdg: yes, his mu was > 100 for a while 2011-12-13T19:31:13 yeah, but that is the theoretical score 2011-12-13T19:31:18 and so far, nobody hit 100 2011-12-13T19:31:18 the only nice thign about working on a plan 9 kernel was that it was ripe for performance improvements 2011-12-13T19:31:24 I made syscalls ridiculously faster 2011-12-13T19:31:27 at least, on official 2011-12-13T19:31:44 okay, exploring replay coming up... 2011-12-13T19:34:52 http://paste.aichallenge.org/Qkpw8/ 2011-12-13T19:35:36 *** jstrong1 has joined #aichallenge 2011-12-13T19:36:07 *** goffrie has quit IRC (Quit: Konversation terminated!) 2011-12-13T19:36:50 *** Conorach has quit IRC (Ping timeout: 244 seconds) 2011-12-13T19:37:47 *** jstrong has quit IRC (Ping timeout: 240 seconds) 2011-12-13T19:39:59 *** rebelxt has joined #aichallenge 2011-12-13T19:40:34 McLeopold: what's the exploration doing? 2011-12-13T19:41:33 BenJackson: have you seen 9front? 2011-12-13T19:41:42 BenJackson: https://code.google.com/p/plan9front/ 2011-12-13T19:41:45 plan9 is still alive :P 2011-12-13T19:42:00 http://man.aiju.de/1/bullshit ever innovative 2011-12-13T19:43:22 webfs looks pretty cool 2011-12-13T19:43:49 http://man.aiju.de/4/webfs they have a torrent client and browser implemented on top of it 2011-12-13T19:45:08 anyway, it's still just people messing around with ideas 2011-12-13T19:45:22 antimatroid1: I'm doing a bfs from my hill to the edge of my explored territory, then bouncing back to ants to get a direction. It should force me to explore evenly around my hill 2011-12-13T19:46:02 I'm gonna try this technique for keeping area visible now 2011-12-13T19:46:10 thestinger: I was working for a company that used a plan 9 derivative in production 2011-12-13T19:47:26 *** treeform has quit IRC (Remote host closed the connection) 2011-12-13T19:47:37 lol, in production? 2011-12-13T19:47:45 McLeopold: hey, i just added a 2nd worker to my private test server, but all games are still getting assigned to the first worker. any quick hint how to fix? 2011-12-13T19:47:46 a web application or something? 2011-12-13T19:48:35 McLeopold: I'm now iteratively collectiong the unseen border, removing all targets in vision of a collected target 2011-12-13T19:49:11 McLeopold: I generate the unseen border when I calculate closest friend hill information 2011-12-13T19:49:13 rebelxt: make some api calls from the second worker to ensure it is authenticated 2011-12-13T19:49:53 McLeopold, the dont seem to distribute that good in your replay(paste) .. i had that same problem with this method 2011-12-13T19:50:01 antimatroid1: I've beefed up my bfs to be passed in functions that define what is a path and what blocks the search 2011-12-13T19:50:07 McLeopold: it's authenticated. heartbeats come through, but server is assigning all games to worker 1 2011-12-13T19:50:16 then I've wrapped 2 of them up in a struture that does the bounce back 2011-12-13T19:50:22 it's all python generators :) 2011-12-13T19:50:23 McLeopold: I tried to generalise path finding a while ago and decided it was a terrible mistake :) 2011-12-13T19:50:43 but i think that's just cause i do it differently all over the place 2011-12-13T19:50:45 and update things etc. 2011-12-13T19:51:01 ikaros: no, I wasn't trying to distribute yet, just go for the closest unknown to the hill 2011-12-13T19:51:21 McLeopold: have you set up potential enemy information? 2011-12-13T19:51:30 potential/possible 2011-12-13T19:51:32 ok 2011-12-13T19:52:35 ikaros: if you look at how I explored to the sw, I used 5 ants and spread out quite well covering the unknown areas 2011-12-13T19:53:32 when i tried this 2011-12-13T19:53:48 it worked well for never seen areas and the "outer ring" of ants 2011-12-13T19:54:39 antimatroid1: no, I just mark unseen locations where I thought the enemy was 2011-12-13T19:55:01 if an ant chases me at the edge of vision, I have a trail of locations I think an enemy is at 2011-12-13T19:55:14 but when all tiles are explored? do you just reset them or count the times youve seen them 2011-12-13T19:55:17 I track where enemies could possibly be 2011-12-13T19:55:26 tried so many variations and it was all crap :O 2011-12-13T19:55:28 starting with all unseen squares as possibly enemies 2011-12-13T19:55:34 so i did it totally different then 2011-12-13T19:55:49 but im still wondering why it didnt work out with simple bfs 2011-12-13T19:55:50 rebelxt: a game isn't assigned to a worker unless a worker requests one, so the second worker must not be requesting any? 2011-12-13T19:56:48 ikaros: once all tiles are seen, I'll switch from unseen to visible 2011-12-13T19:56:49 ie. at the start of a turn if a hill isn't visible then there's possibly an ant there, then at the end of each turn, each visible ant becomes a possible ant and you propogate(/whatever) the information outwards 2011-12-13T19:57:16 antimatroid1: so, an enemy influence map? 2011-12-13T19:57:29 i don't really understand what people mean by influence map? 2011-12-13T19:57:39 well thats simple :O 2011-12-13T19:57:57 antimatroid1: so, an enemy map? 2011-12-13T19:58:04 possible enemy map 2011-12-13T19:58:11 just a map where you propagate some kind of influence somehow from specific sources :) 2011-12-13T19:58:15 do you have probabilities? 2011-12-13T19:58:18 no 2011-12-13T19:58:28 if you did, it might be an influence map 2011-12-13T19:58:29 just a yes or no as to whether an enemy ant could possibly reside in that location 2011-12-13T19:59:00 ok good night. its late here 2011-12-13T19:59:04 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-13T19:59:22 antimatroid1: do you still use pidgin? 2011-12-13T19:59:23 for example that way if you expand past a dead end, you don't need to keep going back to check if enemies are there 2011-12-13T19:59:28 McLeopold: yes 2011-12-13T19:59:39 *** kevlar_work has joined #aichallenge 2011-12-13T19:59:43 or keep vision maximised around your hill 2011-12-13T19:59:43 I just noticed when ikaros left, his name went italics 2011-12-13T19:59:56 just keep a perimeter where you know there's no possibly enemies within it 2011-12-13T20:00:01 carlos.guia sure wins a lot of games 2011-12-13T20:00:05 *** Harpyon has quit IRC (Quit: Textual IRC Client: http://www.textualapp.com/) 2011-12-13T20:00:10 Does one of the servers do OOM-killing? I have a few games where I hit 100 ants and then suddenly time out, where on others I can easily go over 300 ants without a problem 2011-12-13T20:00:21 a1k0n: have you thought any more about ant formations? 2011-12-13T20:00:33 not really 2011-12-13T20:00:39 are formations really useful? 2011-12-13T20:00:45 i would have thought just get ants onto the battle border 2011-12-13T20:00:49 *** goffrie has joined #aichallenge 2011-12-13T20:00:50 pguillory's method is extremely awesome though and should help 2011-12-13T20:01:01 did he describe it somewhere? 2011-12-13T20:01:05 antimatroid1: well, it would be nice to have ants show up to the battle in a group, and not 1 by 1 2011-12-13T20:01:07 on irc here, when asked 2011-12-13T20:01:14 ooh, what time? 2011-12-13T20:01:28 thestinger: i don't see why? get ants there as soon as possible and just be defensive until the back up gets there 2011-12-13T20:01:35 at least you might be able to slow them down a bit that way 2011-12-13T20:01:47 antimatroid1: well, lets say you arrive on the front 2 by 2 2011-12-13T20:01:50 5 hours ago 2011-12-13T20:01:53 and the opponent arrives 5 by 5 2011-12-13T20:01:59 they have a big advantage and get a head start at killing you 2011-12-13T20:02:11 if you're backed into a corner 2011-12-13T20:02:11 I guess your ants would just retreat 2011-12-13T20:02:22 but yeah otherwise you can just retreat 2011-12-13T20:02:29 they can't have 5 ants turning up to every battle like that 2011-12-13T20:02:32 this guy has mad hax: http://aichallenge.org/profile.php?user=10100 2011-12-13T20:02:48 check out how many games he played from 12:45 to 12:59 2011-12-13T20:03:03 haha. seeding race condition again? 2011-12-13T20:03:17 i want to know what username account is using 2011-12-13T20:04:24 FlagCapper is awfully close to xathis 2011-12-13T20:05:23 is there any way to know the max amount of memory we're allowed to use? 2011-12-13T20:06:04 it's 1G or 1.5G or something 2011-12-13T20:06:20 hmm. 2011-12-13T20:06:27 that's heaps! 2011-12-13T20:06:28 kevlar_work: more likely is that you are running out of time. unless you are using 10MB per ant. 2011-12-13T20:06:44 antimatroid1: http://instantrimshot.com/ 2011-12-13T20:07:20 Wraithan, I can go up to ~300 ants in some games and others it "times out" at ~120. If I get close to the time, my engine punts and says "go" anyway and just doesn't move the rest of the ants 2011-12-13T20:07:57 sure, supposing everythign is wokring properly and you always catch it when it is about to time out 2011-12-13T20:08:09 as well as supposing there is no noise on the system you are playing on 2011-12-13T20:08:38 Wraithan, it lets you come back even if you time out for one move unless it loses its connection 2011-12-13T20:08:50 unless that's different online than in the dev version 2011-12-13T20:08:55 define it? 2011-12-13T20:09:00 if you timeout you are gone for the game 2011-12-13T20:09:15 and that is the same in the dev version 2011-12-13T20:09:30 hmm... 2011-12-13T20:09:35 I thought I tested that case. 2011-12-13T20:10:01 and in any case, I give myself 200ms leeway, which has been more than enough in every load scenario I've tested 2011-12-13T20:10:34 (I usually require no more than 7ms) 2011-12-13T20:11:14 oh well, I'll see what happens in the upcoming games. There's no way to save a log that you can retrieve afterward, is there? 2011-12-13T20:12:32 *** TheLinker has quit IRC (Remote host closed the connection) 2011-12-13T20:14:49 aichallenge: janzert epsilon * ra084209 / manager/matchup_buffer.py : Script to keep a buffer of matchups ready for workers - http://git.io/oi-PbQ 2011-12-13T20:16:38 *** flowenol has quit IRC (Ping timeout: 240 seconds) 2011-12-13T20:18:26 *** treeform has joined #aichallenge 2011-12-13T20:19:09 *** bmh has joined #aichallenge 2011-12-13T20:20:11 I just made my combat smarter. And now my ants run away, even when it's 3 on 1 and they're supposed to be gaurding the hill 2011-12-13T20:20:22 Antimony: "Smarter" 2011-12-13T20:20:44 I finally got around the writing the code that determines if my ants might be in combat with other ants next turn. Hooray for a 2d topographic sort. 2011-12-13T20:23:44 Antimony: that has happened to everyone so far 2011-12-13T20:24:10 but it shouldn't be possible with the wy my scoring is set up 2011-12-13T20:24:15 BenJackson: can you tell me if bhickey.net:2080 is alive? I have a really slow internet connectoin at the moment 2011-12-13T20:24:20 at the very least, they have garuteed victory 2011-12-13T20:24:27 so they shouldn't retreat at all 2011-12-13T20:24:44 bmh: if it is it's slow 2011-12-13T20:25:00 Antimony: helps to set up scenarios to test those things 2011-12-13T20:25:13 bmh: I think it replied eventually with a blank page 2011-12-13T20:27:36 *** bmh has quit IRC (Quit: Page closed) 2011-12-13T20:28:11 ok, it looks like the enemy could have made a 1v1 trade 2011-12-13T20:28:22 I guess my "accept no losses ever" strategy is a bad idea 2011-12-13T20:29:18 what can I do? 2011-12-13T20:30:25 I have a gradient of enemy value that goes up as you get near my hill 2011-12-13T20:30:43 Antimony: don't retreat near your hill, so check ant hill distance and exclude moves based on that 2011-12-13T20:30:44 if it can kill anything it will once they get close enough 2011-12-13T20:30:48 even if it's futile 2011-12-13T20:30:53 but if they can't kill anything they'll still run 2011-12-13T20:31:15 :p I like the questions on ai-class, I was thinking "no, no, no…no, euuuh… yes" 2011-12-13T20:31:35 I think I'll add a penalty for unkilled enemies within one step of the hill, so my guards will suicide to stop it 2011-12-13T20:32:16 I'm not sure if there's an easy way to work that into my current system though... 2011-12-13T20:34:24 *** Cyndre_ has quit IRC (Quit: Leaving) 2011-12-13T20:39:21 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-12-13T20:43:04 *** Darhuuk is now known as NoOne 2011-12-13T20:43:22 *** Darhuuk has joined #aichallenge 2011-12-13T20:43:34 *** NoOne is now known as Guest88313 2011-12-13T20:52:26 *** Anilm3 has quit IRC (Ping timeout: 252 seconds) 2011-12-13T20:55:27 Antimony: you could do something like if(closestEnemyToHill < 5) payoff = -a+b*closestEnemyToHill or something? 2011-12-13T20:56:23 *** pguillory has quit IRC (Ping timeout: 258 seconds) 2011-12-13T20:57:22 test 2011-12-13T20:57:34 *** Guest88313 has quit IRC (Quit: Page closed) 2011-12-13T21:02:47 *** Antimony has joined #aichallenge 2011-12-13T21:04:06 *** JorgeB has quit IRC (Ping timeout: 255 seconds) 2011-12-13T21:08:32 *** arscan has joined #aichallenge 2011-12-13T21:15:33 *** TheLinker has joined #aichallenge 2011-12-13T21:20:30 *** besh has quit IRC (Quit: Page closed) 2011-12-13T21:20:52 *** dmj111 has quit IRC (Remote host closed the connection) 2011-12-13T21:26:58 *** GeorgeSebastian has joined #aichallenge 2011-12-13T21:29:37 *** GeorgeSebastian has quit IRC (Read error: Connection reset by peer) 2011-12-13T21:30:12 *** gazpachoking has quit IRC (Ping timeout: 255 seconds) 2011-12-13T21:30:51 *** AntDroid has joined #aichallenge 2011-12-13T21:30:54 *** avdg has quit IRC (Ping timeout: 252 seconds) 2011-12-13T21:33:57 *** dmj111 has joined #aichallenge 2011-12-13T21:36:05 so how do you handle deciding when to make trades? 2011-12-13T21:36:59 *** gazpachoking has joined #aichallenge 2011-12-13T21:39:57 *** Darhuuk has quit IRC (Quit: leaving) 2011-12-13T21:41:21 *** gazpachoking has quit IRC (Ping timeout: 244 seconds) 2011-12-13T21:41:33 *** Darhuuk has joined #aichallenge 2011-12-13T21:45:46 *** GeorgeSebastian has joined #aichallenge 2011-12-13T21:46:40 *** Darhuuk has quit IRC (Quit: leaving) 2011-12-13T21:46:59 *** gazpachoking has joined #aichallenge 2011-12-13T21:48:22 *** Darhuuk has joined #aichallenge 2011-12-13T21:51:20 ug I hate debugging 2011-12-13T21:51:21 *** gazpachoking has quit IRC (Ping timeout: 255 seconds) 2011-12-13T21:56:03 *** Jak_o_Shadows has joined #aichallenge 2011-12-13T22:03:06 watching tcpclinet output is like the matrix 2011-12-13T22:03:12 I see I'm only giving one ant order 2011-12-13T22:03:19 I see the only ants I see are enemy ants 2011-12-13T22:03:23 I'm seeing new (!) water tiles 2011-12-13T22:03:29 so I surmise I'm down to 1 ant and he's on the run! 2011-12-13T22:04:04 I added some output like % of map visible, # of hills I've killed, etc. 2011-12-13T22:04:09 so it's more interesting to watch :P 2011-12-13T22:07:04 I don't even know where these games go sometimes 2011-12-13T22:07:10 I won the last two games on fluxid 2011-12-13T22:08:05 in the last 7 the worst result is 2nd and that was over an hour ago 2011-12-13T22:08:27 *** yoden-cloud has joined #aichallenge 2011-12-13T22:08:57 i just print the map state every turn 2011-12-13T22:09:01 ruins the suspense~~ 2011-12-13T22:11:37 *** avdg has joined #aichallenge 2011-12-13T22:12:20 *** Jak_o_Shadows1 has joined #aichallenge 2011-12-13T22:13:18 hmm 2011-12-13T22:13:47 commonality: the times my bot times are the games where the turntime is 500ms 2011-12-13T22:14:25 so apparently a cutoff at 90% of the turntime should be more like turntime-100ms for me... 2011-12-13T22:14:36 turntime is always 500ms 2011-12-13T22:14:45 *** Jak_o_Shadows has quit IRC (Ping timeout: 255 seconds) 2011-12-13T22:16:36 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-13T22:16:58 ah, so it is. 2011-12-13T22:18:20 I'm not sure why I thought it was 1000, lol 2011-12-13T22:21:21 interesting. Pitting a bunch of instances of my bot against one another, and it's not the one with the most ants that times out. 2011-12-13T22:23:24 meh, my game is at least taking 5 min atm 2011-12-13T22:24:43 *** Antimony has joined #aichallenge 2011-12-13T22:26:33 I'm playing in the world's slowest game on aichallenge right now 2011-12-13T22:26:53 same 2011-12-13T22:26:59 my current game has taken like 10 mins 2011-12-13T22:27:00 you too guys? 2011-12-13T22:27:04 I admit to refreshing the page 100 times 2011-12-13T22:27:14 maybe it's borked 2011-12-13T22:27:15 its still going here 2011-12-13T22:27:45 I think they are using "-serial" so games are slower 2011-12-13T22:28:34 well, my bot is fast 2011-12-13T22:28:46 it shouldn't run longer than 20ms 2011-12-13T22:28:50 each turn 2011-12-13T22:28:58 of course, there are other bots as well 2011-12-13T22:30:07 wow, weird 2011-12-13T22:30:16 my bot planned in a game a while ago and it _just_ appeared now 2011-12-13T22:30:26 and now it says I'm in another game :\ 2011-12-13T22:30:36 yay! 2011-12-13T22:30:48 still playing 2011-12-13T22:31:26 V11 has won 6 in a row to get to 1179 2011-12-13T22:31:35 didn't play strcat, but I did play advicecat 2011-12-13T22:32:00 it looks like mine won't make it in the top 2k :p 2011-12-13T22:32:11 * avdg blames slow js 2011-12-13T22:32:21 and os :p 2011-12-13T22:32:45 it's fun watching my bot vs non-combat bots 2011-12-13T22:32:52 no tedious stalemates 2011-12-13T22:32:53 meh, I don't like the climb back up 2011-12-13T22:32:57 feels like a huge waste of time 2011-12-13T22:32:59 yes 2011-12-13T22:33:10 but 6 games today 2011-12-13T22:33:14 that's way better than last time 2011-12-13T22:33:22 that's one reason I submitted a fairly untested bot 2011-12-13T22:33:25 the game rate is awesome now 2011-12-13T22:33:38 tell me all about it :/ 2011-12-13T22:33:48 yeah 2011-12-13T22:33:49 what did they do? 2011-12-13T22:34:05 new sponsor with servers 2011-12-13T22:34:20 since their previous problem was not being able to use money this solved the problem more directly 2011-12-13T22:34:21 oh, that part 2011-12-13T22:35:00 hmm, I think my game hits the 15 min mark 2011-12-13T22:35:29 is the server maybe being slow? 2011-12-13T22:35:34 i'm getting weird timeouts... 2011-12-13T22:35:41 thought maybe that's just my parent's connection -_- 2011-12-13T22:35:49 tcp? 2011-12-13T22:35:57 on ants.fluxid 2011-12-13T22:36:19 low bandwidth shouldn't be a problem, slow or unstable could be an issue 2011-12-13T22:36:44 6th game for my previous version my bot played #8 and won and got to a skill of 72.79 2011-12-13T22:36:44 meh, now I have to wait for 15 more players 2011-12-13T22:37:14 the game went probably on a server that wants to shut itself down 2011-12-13T22:37:30 ah 2011-12-13T22:37:32 this time 6th game was vs ~#1000 (times 5) and got to 60.36 2011-12-13T22:38:02 played a game in 16 sec, probably a bug :p 2011-12-13T22:38:20 or you played me 2011-12-13T22:38:21 lol 2011-12-13T22:41:04 http://aichallenge.org/visualizer.php?game=224975&user=2255 that's the game that took a REALLY long time for some reason :\ 2011-12-13T22:41:32 weird, i have good latency locally, but pretty poor to fluxid: 2011-12-13T22:41:35 --- fluxid.pl ping statistics --- 37 packets transmitted, 33 packets received, 10.8% packet loss round-trip min/avg/max/stddev = 154.007/172.095/373.270/51.119 ms 2011-12-13T22:41:47 where do you live? it's in germany 2011-12-13T22:41:48 fluxid advertises 500ms but actually gives you like 5s 2011-12-13T22:41:51 you'd have to really work to time out 2011-12-13T22:41:58 thestinger: .pl is in .de? 2011-12-13T22:42:00 it's packet loss, i think? 2011-12-13T22:42:08 BenJackson: it's hosted in germany 2011-12-13T22:42:13 i'm in ohio, but it usually works OK for me from cleveland... 2011-12-13T22:42:15 Fluxid confirmed it :) 2011-12-13T22:42:19 I guessed with tracepath though 2011-12-13T22:42:23 german web host 2011-12-13T22:43:02 my ping to there is 120ms which is pretty damn good 2011-12-13T22:43:15 since I live in toronto :P 2011-12-13T22:43:26 *** jstrong1 has quit IRC (Quit: Leaving.) 2011-12-13T22:45:21 *** Antimony has quit IRC (Ping timeout: 255 seconds) 2011-12-13T22:45:35 *** gazpachoking has joined #aichallenge 2011-12-13T22:46:05 Its probably because I'm closer to the server with a better network: 32.210/38.551/50.741/4.335 ms 2011-12-13T22:46:36 *** yoden-cloud has quit IRC (Quit: Page closed) 2011-12-13T22:46:49 *** yoden-cloud has joined #aichallenge 2011-12-13T22:48:40 lol seriously? 2011-12-13T22:48:44 I can send packets to germany 2011-12-13T22:48:47 however, my wifi 2011-12-13T22:48:50 nopee~~ 2011-12-13T22:49:33 I'm on wifi as well 2011-12-13T22:50:19 my parents have an impressive ability to pick wireless hardware that is unreliable... 2011-12-13T22:50:27 at least the ethernet part works correctly >.> 2011-12-13T22:51:21 heh, my dad just asked to install one, but in the meantime, I replaced the router 2011-12-13T22:51:28 s/just// 2011-12-13T22:52:33 *** gazpachoking has quit IRC (Ping timeout: 255 seconds) 2011-12-13T22:52:45 http://bit.ly/sQaozP (I used ip2location service) 2011-12-13T22:52:48 *** Antimony has joined #aichallenge 2011-12-13T22:53:04 dunno how the network really is :p 2011-12-13T22:54:10 (its actually crazy that signals are going all over the world multiple times in less than a second) 2011-12-13T22:56:02 hmm, ping to google (in us) seems to be faster: 17.534/23.898/32.191/3.306 ms 2011-12-13T22:56:28 google has servers everywhere 2011-12-13T22:57:02 this wait to rank up is going to give me more time to play skyrim 2011-12-13T22:57:08 yeah, but this ping didn't go to the closest server 2011-12-13T22:57:35 it went to california, at least if the ip tables are right 2011-12-13T22:57:51 BenJackson: lol, make an AI that plays skyrim for you :P 2011-12-13T22:58:05 AIs need fun too 2011-12-13T22:59:41 *** Paradoxial has quit IRC (Quit: :) 2011-12-13T23:05:29 *** gazpachoking has joined #aichallenge 2011-12-13T23:06:36 *** Antimony has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-12-13T23:06:41 *** avdg has quit IRC (Quit: Leaving.) 2011-12-13T23:08:56 *** ferd has joined #aichallenge 2011-12-13T23:14:41 *** ferd has quit IRC (Quit: Page closed) 2011-12-13T23:15:21 *** dvladim has joined #aichallenge 2011-12-13T23:20:43 *** Paradoxial has joined #aichallenge 2011-12-13T23:21:40 *** yoden-cloud has quit IRC (Ping timeout: 258 seconds) 2011-12-13T23:23:17 games are taking too long to finish now in aichallenge 2011-12-13T23:23:31 *** dvladim has quit IRC (Ping timeout: 276 seconds) 2011-12-13T23:26:34 * antimatroid1 wants a game where you're allowed to write your own scripts to run 2011-12-13T23:26:45 sort of like this, but you also get to control your bot in real time 2011-12-13T23:29:53 *** avdg has joined #aichallenge 2011-12-13T23:38:57 <_flag> Odd, the rankings aren't updating 2011-12-13T23:39:03 <_flag> @rankings 2011-12-13T23:39:05 _flag: Top 10 players: FlagCapper(93.7), xathis(93.6), pguillory(93.0), a1k0n(90.8), fourmidable(90.6), ThisIsNotABug(89.0), delineate(88.8), jacob_strauss(87.9), cumbuz(86.9), antysocial(86.7) 2011-12-13T23:39:23 <_flag> But they're updated for contestbot :P 2011-12-13T23:42:05 oshit, flag's on top 2011-12-13T23:42:23 <_flag> xathis will pass me again soon 2011-12-13T23:42:27 _flag: congrats 2011-12-13T23:42:34 <_flag> amstan: thanks :) 2011-12-13T23:42:40 apparently you're still going up, and he's going down 2011-12-13T23:42:50 <_flag> If only I could replace the version on the main server with my current version 2011-12-13T23:43:03 better version? 2011-12-13T23:43:10 <_flag> amstan: Ignore the arrows, they only give an indication of the most recent game and fluctuate a ton 2011-12-13T23:43:13 _flag: bribe amstan :) 2011-12-13T23:43:16 <_flag> mus are a better indication 2011-12-13T23:43:18 <_flag> amstan: yes 2011-12-13T23:43:43 hmm 2011-12-13T23:43:45 *** Paradoxial has quit IRC (Quit: :) 2011-12-13T23:43:49 i could replace stuff like that actually 2011-12-13T23:44:00 just replace it in the submissions folder, then restart the workers 2011-12-13T23:44:06 but you'll still have the old games 2011-12-13T23:44:24 also.. what's the point 2011-12-13T23:44:28 besides time saving? 2011-12-13T23:44:41 <_flag> There's three days left in the contest 2011-12-13T23:44:52 takes like 2-4 days to climb back up 2011-12-13T23:44:57 finals will be with reset rankings anyway 2011-12-13T23:45:45 <_flag> This is why I liked how planetwars stats kept your ranking when you resubmitted :P 2011-12-13T23:48:02 _flag: but then you never played any of the lower ranked bots :P 2011-12-13T23:48:18 <_flag> antimatroid1: does that matter? 2011-12-13T23:48:42 i think a few people had a bit of surprise in the final contest after not playing lower ranked bots in a while 2011-12-13T23:48:54 they were a lot less cautious which caused some upsets 2011-12-13T23:49:28 i think i'm one of the people you could blame for resetting the ranking every time you submit now :P 2011-12-13T23:49:49 a1k0n is up to 4th 2011-12-13T23:49:49 <_flag> Were these people top ~100 players? 2011-12-13T23:51:06 yes 2011-12-13T23:51:09 <_flag> What if you were given an option to start from the bottom or not? 2011-12-13T23:51:13 asivis or whatever is one example i think 2011-12-13T23:51:20 <_flag> Replace vs. Reset 2011-12-13T23:51:35 yeah i think it might be nice if users had a button on their profile to reset their ranking 2011-12-13T23:52:03 if there were more servers I don't think it would be a problem - maybe there will be more sponsors next time 2011-12-13T23:52:08 for zeta i suspect tcp servers will be the go but with single user accounts, just let people reset their ranking if they want 2011-12-13T23:53:08 <_flag> thestinger: I think they underestimated how many entries there would be based on the money they asked from Google 2011-12-13T23:53:16 <_flag> They wanted 5-6 servers running at all times 2011-12-13T23:53:39 <_flag> Right now there's 23 2011-12-13T23:53:50 *** Vaenom has joined #aichallenge 2011-12-13T23:53:54 are any of you going to stick around after the contest to help out with the next? :P 2011-12-13T23:54:37 <_flag> I'll probably lurk as I did for this one 2011-12-13T23:56:08 @seen mcstar 2011-12-13T23:56:08 antimatroid1: mcstar was last seen in #aichallenge 2 weeks, 2 days, 5 hours, 50 minutes, and 11 seconds ago: i dont agree 2011-12-13T23:56:20 *** Antimony has joined #aichallenge 2011-12-13T23:57:07 Wow. FlagCapper took the head pretty fast. What was it 'usual' rank range before taking the crown to Pguillory? 2011-12-13T23:58:12 I think my new combat system will actually hurt the bot 2011-12-13T23:58:19 it's too cautious 2011-12-13T23:58:37 yeah, i'm trying to work out how to make my combat less catious 2011-12-13T23:59:18 this topic is funny: http://forums.aichallenge.org/viewtopic.php?f=24&t=2115