2010-12-10T00:03:41 *** medrimonia has joined #aichallenge 2010-12-10T00:05:17 *** medrimonia1 has quit IRC (Ping timeout: 265 seconds) 2010-12-10T00:06:44 *** bhasker has joined #aichallenge 2010-12-10T00:09:39 *** Rubicon-|-Cross has quit IRC (Ping timeout: 265 seconds) 2010-12-10T00:36:50 *** delt0r___ has quit IRC (Read error: Operation timed out) 2010-12-10T00:37:09 *** delt0r___ has joined #aichallenge 2010-12-10T00:39:14 *** withate has joined #aichallenge 2010-12-10T00:39:46 *** yasith has quit IRC (Ping timeout: 245 seconds) 2010-12-10T00:48:21 *** bhasker has quit IRC (Quit: bhasker) 2010-12-10T00:50:19 *** bhasker has joined #aichallenge 2010-12-10T00:53:28 *** bhasker has quit IRC (Client Quit) 2010-12-10T00:55:35 *** medrimonia has quit IRC (*.net *.split) 2010-12-10T00:55:36 *** wh1teside has quit IRC (*.net *.split) 2010-12-10T00:55:36 *** delt0r has quit IRC (*.net *.split) 2010-12-10T00:55:36 *** politik has quit IRC (*.net *.split) 2010-12-10T00:56:00 *** Mathnerd314 has quit IRC (Ping timeout: 245 seconds) 2010-12-10T00:58:16 *** pyro- has joined #aichallenge 2010-12-10T00:59:44 *** medrimonia has joined #aichallenge 2010-12-10T00:59:44 *** wh1teside has joined #aichallenge 2010-12-10T00:59:44 *** delt0r has joined #aichallenge 2010-12-10T00:59:44 *** politik has joined #aichallenge 2010-12-10T01:00:42 *** Rubicon-|-Cross has joined #aichallenge 2010-12-10T01:03:57 antimatroid1 2010-12-10T01:04:00 you there? 2010-12-10T01:04:45 is anyone up 2010-12-10T01:08:09 *** sigh has left #aichallenge ("Leaving") 2010-12-10T01:11:28 *** jamesdotnet has joined #aichallenge 2010-12-10T01:23:17 *** Appleman1234 has joined #aichallenge 2010-12-10T01:27:55 *** Zoo is now known as Guest2611 2010-12-10T01:30:09 epona: hey 2010-12-10T01:30:17 of course i'm up, it's 5 30pm :) 2010-12-10T01:30:48 ahh 2010-12-10T01:31:17 someone brought up a problem in #ai, was wondering if you might be able to help them out a little 2010-12-10T01:31:41 it looks like a horribly dumbed down version of planetwars 2010-12-10T01:35:09 if your not up for it its whatever... 2010-12-10T01:35:16 what's the problem? 2010-12-10T01:35:38 http://compeng.hud.ac.uk/scomhro/Courses/CIS2344/Assignments/Assignment2010-2011-Formative/ 2010-12-10T01:38:23 hey, withate 2010-12-10T01:39:08 * epona pokes withate 2010-12-10T01:39:15 * withate *shy* 2010-12-10T01:39:23 when they say the bots resources increaase by 20%, if they're currently at 50%, does that mean it becomes 60 or 70%? 2010-12-10T01:39:41 it means 70 2010-12-10T01:39:54 that specification is wrong 2010-12-10T01:40:23 why is it wrong? 2010-12-10T01:40:29 *** pyro- has left #aichallenge 2010-12-10T01:40:58 "Ie. a visit to A or F will increase a bot's resources by 20%" 2010-12-10T01:41:09 it should say percentage points 2010-12-10T01:41:12 if i'm being pedantic 2010-12-10T01:42:02 later on it says (of the bot's maximum capacity) 2010-12-10T01:42:14 which is 100%, so i guess it makes sense, to some point 2010-12-10T01:42:58 *** Palmik has joined #aichallenge 2010-12-10T01:43:01 it makes sense if you ignore what it says :P 2010-12-10T01:43:23 fair enough, not going to argue ^_^ 2010-12-10T01:43:31 this actually looks like a really interesting assignment 2010-12-10T01:43:55 and really stressfull as well :o not like looking for help on IRC is the best 2010-12-10T01:43:58 a student could do 2010-12-10T01:44:00 ;o 2010-12-10T01:45:13 yeah, i wish i got assignemnts this awesome 2010-12-10T01:45:36 i take it this is an algorithms assignemnt? not ai? 2010-12-10T01:47:29 yeh 2010-12-10T01:47:41 but it looks like AI to me 2010-12-10T01:48:13 hmmm, i think they want you do actually solve the maps, not sort of use ai techniques, but it's still a cool assignemnt 2010-12-10T01:48:27 if i was doing the quick one without requirements, i'd probably try dynamic programming 2010-12-10T01:48:46 well what ive got so far is 2010-12-10T01:48:53 track back from the end to teach node and work out how much you need when entering to be able to complete it and in how many steps 2010-12-10T01:48:55 depth-first traversal search exploring the 2010-12-10T01:48:56 any ongoing contests around? 2010-12-10T01:49:05 graph 2010-12-10T01:49:21 rabidus: a few actually, mit battlecode people seem most interested in, but it's java only 2010-12-10T01:49:36 the problem is, that youve only got entry node at the start 2010-12-10T01:49:38 arimaa is anohter, then there was some other sequential game, but i forget what it was called 2010-12-10T01:49:58 and in order to get to exit node (the goal) 2010-12-10T01:49:59 rabidus, not for ai challenge 2010-12-10T01:50:00 withate: you don't know where exit nodes are? 2010-12-10T01:50:04 is there a framework for arimaa ? 2010-12-10T01:50:10 no, i just know the entry node 2010-12-10T01:50:15 ahhh 2010-12-10T01:50:17 and the only way 2010-12-10T01:50:24 to "communicate" is to leave the messages 2010-12-10T01:50:27 at the nodes 2010-12-10T01:50:47 so basically you need to search out the entire graph and communicate that? then calculate the shortest path on that graph once you know it? 2010-12-10T01:50:48 thats kind of my problem right now, im not sure what should i be storing in there 2010-12-10T01:50:57 yeh, true 2010-12-10T01:51:03 shortest path should not be a problem i guess 2010-12-10T01:51:07 loads of algorithms for it 2010-12-10T01:51:14 i think it's slightly tricker 2010-12-10T01:51:36 and you only get access to adjacency list of the node 2010-12-10T01:51:36 because of the requirements that the overall cost is >0 at all times 2010-12-10T01:51:46 when you are actually in the node 2010-12-10T01:51:49 yup 2010-12-10T01:52:20 have you got greedy bot sorted? 2010-12-10T01:52:42 no, not really. I do not know what should I be storing in messages 2010-12-10T01:52:48 since at some point, bot is going to die 2010-12-10T01:53:06 and bot can leave a message at the current node he is 2010-12-10T01:53:16 but if say ive got 10% resources, i leave A, i get to the B 2010-12-10T01:53:25 i die, and I can only leave the message at B 2010-12-10T01:53:40 i would write your own graph class that can add nodes and edges from "messages", then just have it filter added edges/nodes, and use all messages to build yourself a gaph 2010-12-10T01:53:47 so if lets say i run another bot, how do I know, that I am going to die in B, if ive got 10% before visiting it 2010-12-10T01:53:52 yeh, Ive got that 2010-12-10T01:54:00 i mean ive got my own graph class 2010-12-10T01:54:12 but i get stuck at building it 2010-12-10T01:54:16 because as soon as the bot dies 2010-12-10T01:54:36 im lost of how should i know that it is going to die 2010-12-10T01:54:48 at that node, if it enter having resources level below X 2010-12-10T01:55:23 since you can't really access the % modiffier either 2010-12-10T01:55:32 *node % modiffier 2010-12-10T01:55:53 unless doing something like ping-pong messages? 2010-12-10T01:56:22 if left with 10%, went there, and came back with 30% 2010-12-10T01:56:50 it's given me a cool idea for an ai game actually 2010-12-10T01:56:50 30%-10%= value of node's resources modiffier 2010-12-10T01:56:53 am i making sense? 2010-12-10T01:57:00 whats the idea? 2010-12-10T01:58:58 epona: maps are random connected graphs, you don't see where their bots are (except maybe at adjacent nodes) then at each turn, you can send any number of your bots at a node along any edge or leave it there, any any combination of these, then if your bots end up in the same square as theirs then you battle like planet wars, and then i'm not sure how to do "growth" 2010-12-10T01:59:19 or any* combination 2010-12-10T01:59:36 withate: sort of, i'm not sure i full understand what's happening where, let me read it more 2010-12-10T01:59:40 :) 2010-12-10T02:00:00 so chess + planetwars? 2010-12-10T02:00:03 right, thanks 2010-12-10T02:00:50 not chess? 2010-12-10T02:00:56 it's simultaneous, quite like planet wars 2010-12-10T02:01:03 but you have more private information 2010-12-10T02:01:11 and the graph isn't fully connected 2010-12-10T02:01:25 you could even have uni-directional edges 2010-12-10T02:01:37 oh i see 2010-12-10T02:01:41 and have bots have to work out where the sink holes are 2010-12-10T02:01:45 that'd be cool actually 2010-12-10T02:02:23 what if you played planet wars 2010-12-10T02:02:25 in a cube 2010-12-10T02:02:32 3d? 2010-12-10T02:02:36 yea 2010-12-10T02:02:44 it's not much different, other than more planets can be "closer" to each other 2010-12-10T02:03:16 I guess thats true, doesn't really mess up the distance formulas 2010-12-10T02:03:19 same goes for any number of dimensions, treating it like a graph you just sort of change the laws of how many planets can be at the same distance to each other 2010-12-10T02:03:29 *** tapwater has quit IRC (Quit: tapwater) 2010-12-10T02:05:37 withate: what if you do a sort of go forward, go back leave message, go forward, go forward, go back 2010-12-10T02:06:13 you may even say on the second round 2010-12-10T02:06:14 go forward, leave message of node you're trying, if you end up getting to the node again and have that message without a return node, you know you can't go there, unless you have more than you did before 2010-12-10T02:06:26 return message* 2010-12-10T02:06:49 imo they went and took an awesome problem and made it stupidly messy :C 2010-12-10T02:06:58 lol 2010-12-10T02:07:16 that solves the negative problem though 2010-12-10T02:07:40 *** narnach has joined #aichallenge 2010-12-10T02:07:40 *** narnach is now known as narnach_mb 2010-12-10T02:08:14 i'm going to post that game idea in the forums and see if people like or hate 2010-12-10T02:08:37 do it! ill support it 2010-12-10T02:08:40 antimatroid1, i was thinking about it 2010-12-10T02:09:09 but lets say 2010-12-10T02:09:40 weve got A+20 (entry node) 2010-12-10T02:10:05 B+0, D-10, C-10 2010-12-10T02:10:31 and then A: B->D; B: A->D 2010-12-10T02:10:42 D: A->B->C 2010-12-10T02:10:53 so if you start off at A, you are at 20% 2010-12-10T02:11:05 you go to B - still 20% 2010-12-10T02:11:17 you go to D - youve got 10% left 2010-12-10T02:11:25 then you go to C - ur dead 2010-12-10T02:11:26 thats when you go back 2010-12-10T02:11:44 why would you move foward three time 2010-12-10T02:11:45 s 2010-12-10T02:11:56 you ve got the point. 2010-12-10T02:12:50 so i start of at A (20), leave message (resources=20, traversal=A,backtrack=A, going=B, returned=false). 2010-12-10T02:13:06 *** sigh has joined #aichallenge 2010-12-10T02:13:28 if the node you arrive at is +20 +10 or +0 then you can use it as a staging place 2010-12-10T02:13:43 meaning, drop an object that doesn't make you go back 2010-12-10T02:14:07 i go to B (20), leavemessage(resources=20,traversal=A,B; backtrack=A,B; going=A, returned=false) 2010-12-10T02:14:48 when i return to A, i get last element of backtrack, and realise, that its resAdjustment = 0, and that it is safe to go there from A with 20%? 2010-12-10T02:15:13 but the thing is, i dont really know, what + or - it is at 2010-12-10T02:15:15 withate: leave a mesage at D saying, "I'm going to C, I have 10% left", then when you return say leave a message saying you got back 2010-12-10T02:15:30 then if you're ever at D with 10% left and no return message was sent, don't follow that path again 2010-12-10T02:16:01 but the thing is 2010-12-10T02:16:08 i cant really send messages 2010-12-10T02:16:15 i can just drop it where it is 2010-12-10T02:16:27 so how would i simulate sending? 2010-12-10T02:16:40 getting the last element of backtrack? 2010-12-10T02:16:52 that's my point, leave a mesage at the node saying where you're about to go and what resources you have 2010-12-10T02:17:05 then ammend that if you return from that node saying how much is there 2010-12-10T02:17:14 ^ 2010-12-10T02:17:52 why they ruined such a nice problem i have no idea :\ 2010-12-10T02:18:12 exactly, giving it to stupid students like me :)) 2010-12-10T02:18:30 well all this extra work isn't really teaching you anything algorithms wise, it's just making it messy 2010-12-10T02:18:58 thanks, I will try that. btw, since we got these ping-pong messages, should I still be creating my own graph class and storing info in it? 2010-12-10T02:19:09 oh yes, thats Hughes, hes always messy 2010-12-10T02:19:30 (as well as his assignments) 2010-12-10T02:21:06 one more question, before I try implementing that, lets say I realise I can not go from B to C if im below X resources, what do I do? or should i be storing positive cycles as well? 2010-12-10T23:23:40 *** contestbot has joined #aichallenge 2010-12-10T23:23:41 *** ChanServ sets mode: +o contestbot 2010-12-10T23:23:49 @topic 2010-12-10T23:23:49 amstan: Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || Congratulations to bocsimacko for first place! || Next Contest sometime in February 2010-12-10T23:23:57 woot 2010-12-10T23:33:33 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-12-10T23:34:39 :) 2010-12-10T23:43:19 evening 2010-12-10T23:43:45 afternoon 2010-12-10T23:44:14 how goes it 2010-12-10T23:44:39 not bad 2010-12-10T23:44:44 what about yourself? 2010-12-10T23:47:12 *** amriedle has quit IRC (Quit: Lost terminal) 2010-12-10T23:48:10 doing okay ...long day at work 2010-12-10T23:55:42 sigh: rofl, what the heck? So this Goldman Sachs programmer used open source modules in his programs and reuses them at another company and Goldman Sachs is pissed? 2010-12-10T23:57:45 Frontier_: no idea, I don't trust anything that article says... for obvious reasons