[06:43] MSG: Ping timeout: 255 seconds [07:56] Join: multimode joined #corewars [07:57] MSG: Client Quit [07:58] Join: multimode joined #corewars [08:07] Join: fiveop joined #corewars [08:28] MSG: [08:30] Join: multimode joined #corewars [15:47] Join: asw joined #corewars [15:48] Join: Nenad joined #corewars [15:48] hi N [15:49] Hi [15:50] I'm surprised to see that Barkley isn't here the last couple of days [15:51] he was usually monitoring the channel, even when there was little or no activity [15:51] have you been working on that warrior from 94b since yesterday? [15:51] nope [15:53] though i might today [15:54] I've been finishing today the seminar paper about some cw-related stuff for one of my univ. courses. [16:00] Join: inversed joined #corewars [16:01] hi Inversed [16:01] Hi [16:01] hmm, Kline said that i should boot the scanner-part, and while i agree with that, what should i do with the part that boots the clear [16:01] hi [16:01] should i just leave it and pray.. [16:02] it's always a bit complicated to find the proper booting strategy when your code is large [16:03] I didn't have a good look at it, really, yesterday. I'll do so now and tell you how I feel about it [16:03] * Nenad goes to look [16:03] btw, inversed, what do you think about a tourney in November? Would you play if I organized one? [16:05] Of course yes [16:05] :) [16:07] though I was thinking about organizing my own tournament [16:07] :) it seems that there will be more tournaments than people playing... lol [16:08] * asw waves [16:08] * Nenad waves back to asw [16:08] hi asw [16:09] hey. [16:10] Miz: it would take too long to boot everything away, I think. If you add too much booting lines in your code, its size will increase enough to remaove the advantage that you had there over some qscans - small size... and you will waste enough time booting so that your attack will result in fewer quick wins [16:10] asw: how is qcoreworld going? [16:10] booting both paper and scan is 14 lines [16:10] any news, new software versions, etc.? [16:17] btw, offtopic, I think that 94b is a little bit too weak now... no real challenge for the players. it's also not good that there has been so little activity here, so, basically, if some newbies came, they didn't have anyone to ask for advice... any ideas how to change that? ok. some CE soon would be nice, that's my fault, I've been too busy for too long. maybe some fresh tutorials with a little bit more about current trends, or at least, some advice as ho [16:18] cut at "some advice as ho--" [16:18] oh :) [16:19] or at least, some advice as how to make decent scanners? [16:19] I finally found one person who is interested in CW, so expect some -b traffic soon [16:19] :) I've been trying for the past two years, but to no avail [16:20] even if it sounded interesting (and it did) to some people, they just said: I've got no time to learn, or smt. like that [16:20] I am going to write some scanner tutorials soon: "Making of Pro.Verka" and "Self-checking in Scanners" [16:20] :) [16:20] yay [16:21] darnit, i need to continue on my stone-article [16:21] I will finish that oneshot tutorial (which is getting quite huge) that I started a long time ago [16:22] Stone-article? [16:22] http://koti.mbnet.fi/mizcu/stonearticle.txt [16:23] That is a little bit older than the one i have on my own 'puter [16:23] Just happens that i am not on it right now [16:24] one more thing -> I've gotten no reply at rgc to that post a few days ago. would any of you be interested in some sort of mini-meeting sometime next year (in the summer) [16:25] Have little problems myself getting anywhere [16:26] Maybe we'd make some sort of CW-megameeting here? Try to break the 20-people limit and get some real talk going? [16:26] well, nothing can replace the physical presence... [16:26] I know [16:27] How much will it cost? [16:28] no idea, but shouldn't be too much [16:30] I, mean, things are cheaper here than in most of the Europe, and there should also be no problem in finding cheap accomodation. If there is only a few people, a few of us could probably make arrangements so that you sleep at our places, so that should be no problem [16:31] the trip would probably be the most expensive bit, but it all depends... [16:31] And I suspect that for me it'd be quite hard to talk to smb. else in English [16:31] ? [16:31] not sure I understand what you mean [16:32] heavy accent? [16:32] people here? [16:33] well, the older generations haven't learned it, but for the most of the young people, ... we all speak english quite well, I'd say :) [16:33] well, there are always those that don't want to learn, but still... [16:33] communication should be no problem [16:33] OMG [16:34] ? [16:34] Do you think so? I never had a conversation in English with a foreignier. [16:34] naughty inversed dropped SotBS! [16:34] :) [16:34] inversed: where do you live? [16:34] Miz sorrrrryyy!!! Last time Silverblade failed to enter 94nop [16:35] Russia, town near Moscow [16:35] Guess i have to release its source now.. [16:36] inversed: when some Russians came to PSC (never mind) two years ago, they talked in russian, we talked in serbian, and we understood each other [16:36] so, even that is not impossible [16:36] although probably not the best solution [16:39] well, think about it:) [16:40] o_O [16:40] new suddenly came slow ar darn [16:40] *net [16:40] I'd just like to know in time if there are enough people interested, and, if so, I'll do my best to organize the whole thing [16:41] it might be possible for me to convince my professors to allow us to use the computer center there (since it's not used for lectures in the summer) [16:42] I was looking forward to it being organized this summer, when Roy suggested it, but nothing came out of it [16:43] btw, if you're interested, I could send you the paper I just finished about some corewar categorization [16:43] the work is still in progress, but the first part of it was finished [16:44] Categorization of warriors? [16:44] nothing special, though. haven't had time for more [16:44] what's you email? [16:44] *your [16:45] [16:45] no wonder.. my firewall is blocking hotmail O_o [16:46] ... and [16:47] ok. sent. (to the first one) [16:48] Again that KOTH time-travel bug: sometimes when I send a dat, hill status rolls back one challenge [16:48] :) haven't noticed that [16:49] or is it simply something special about your DATs? :) [16:49] probably some new phenomenon... haven't DATed a hill for a long time.. [16:50] I've got to go now, have some more stuff to do tonight. see you soon! btw, come if you can in sunday on an irc minitourn. (if enough of us comes (hopefully)) [16:51] * Nenad waves [16:51] wait a moment [16:51] http://koti.mbnet.fi/mizcu/sotbs.red [16:52] i dont bother to send it to RGC [16:52] ? [16:52] there, now you can go [16:52] :) [16:52] bye [16:53] MSG: [16:55] It is what I thougth: vain-like stone and a- and b- imps [16:55] i did mention a long time ago that it was surprisingly simple for what it was [16:55] actually that is an emerald-stone [16:56] And btw my monsters are completely different [16:56] O_o [16:57] well, i guess my score-predicition is a bit rusted [16:58] *analyzation [17:00] Yay! q^i outperforms q^4 ! [17:01] (look at 94nop) [17:01] barely [17:02] Anyway it's quite a feat, because it's totally different [17:03] Got to go, bye [17:04] byby [17:04] MSG: [17:57] MSG: Quit: Lost terminal [18:07] Join: willvarfa joined #corewars [18:08] Join: Nenad joined #corewars [18:09] * Nenad hops back in for a while [18:09] evening all [18:10] :) hi Will [18:10] how has it been? [18:10] all going well, I hope [18:12] oh yes; yourself? [18:13] yep. well, actually - almost no sleep, exam headaches, a lot of coffee, and a bit of everything else :) [18:13] and a shitty weather at the moment [18:13] but, otherwise, fine :) [18:14] i've been wanting to ask you about something for a while now, btw. [18:15] If I understood correctly, you did some automatic warrior categorization by looking at how many times an instruction is executed in melee mode, right? Or was it someone else? [18:15] I did [18:15] "fingerprinting" [18:16] :) [18:16] did you get nice results? [18:16] what I did was "clustering" by score [18:16] I then visually compared the results of the op-code fingerprint [18:16] warriors with visually similiar graphs scored similiarly [18:16] it was strange - I wasn't expecting such a perfect correlation at all [18:17] heh. I also did benchmarking, and clustered (but it's not over yet), but also tried a pure syntax analysis clustering [18:17] I worked out how to improve my measures, [18:17] the results are similar so far [18:17] and then started an evolver based upon the principle - Species Blossom [18:17] but still a lot of testing to be done [18:17] Species Blossom? did it produce some good or interesting warriors? [18:18] actually, I'm asking because I thought about comparing my own results (once they are complete) with yours, if possible [18:18] http://redcoder.sourceforge.net/?p=kepler [18:18] have you kept the logs somewhere? [18:19] and my test results: http://redcoder.sourceforge.net/?p=kepler-wilfiz [18:19] I thought about syntax analysis, because it should be a lot quicker than running the simulation [18:19] the actual execution logs are gone - they were taking up about 40 GB on one of bvowk's boxes, and that was zipped [18:20] I was doing this a couple of years back [18:20] before I completely lost my free time ;-/ [18:20] did you, by any chance write your findings into some sort of a paper or smt like that? [18:20] (for me to be able to reference to) [18:21] http://redcoder.sourceforge.net/kepler-wilfiz/34/ [18:21] for example - you'll see the correlation in opcodes/time [18:21] but remember I am clustering all warriors by score [18:22] and using a simple add to nearest-group algorithm, nothing fancy [18:22] I never "wrote it up" because I moved straight on to wondering how to use the stats [18:22] basically, scoring similarity == instruction similarity [18:23] therefore, scoring similarity suggests a better chance of mating successfully in an evolver [18:23] yes, my results should be similar to those... [18:23] I used some sort of "modified bag of instructions" [18:23] looking at instructions, instruciton pairs, and imps [18:23] (not all instruction pairs, only the characteristic ones) [18:23] therefore, Species Blossom uses scoring similarity (Euler distance against a paper, scissors and a stone benchmark) to select mates, rather than an overall fitness [18:24] ah I tried a while back to work on syntax analysis of "genes" in warriors [18:24] no success for me though :-( [18:24] Species can do it [18:24] treating runs of self-referencing code as a gene [18:25] but it rather hurt the evolver, which is probably down to my dodgy interpretation rather than genes being a bad idea [18:26] hrm. I'm still not sure if I'll include an evolver in my next project. (not wanting to bite more than I can chew) [18:26] I'd like to make a good optimizer utilizing several tecniques first [18:26] I've never spotted the difference ;-) [18:26] :) [18:26] so where have you got to in your research? [18:27] I can send you the current paper draft, if you'd like [18:27] what's your mail? [18:29] varfar [18:29] at [18:29] yahoo [18:29] co [18:29] uk [18:31] the score table clustering is not over yet, so those results weren't included in the draft, at all. will be some time soon, though [18:31] we intend to test the syntax clustering by comparing it to the clustering by score which sgould be more accurate [18:31] *should [18:32] hmm [18:32] well, message sent [18:32] I wonder what the maths for correlating clustering from n different score-functions is [18:33] there are some [18:33] received [18:33] but I haven't looked into it yet. [18:33] we had another idea in mind, but you'll read it in the "future plans" section [18:34] we? [18:35] well, I'm working on it with a colleague [18:37] * willvarfa is reading as fast as he can [18:38] hi will [18:39] hi Miz [18:43] will you take the analysis a step further? [18:43] you say that is a draft [18:43] is that a draft of the write-up, or is that just the first step? [18:43] yes, I will continue with the analysis [18:43] but we had a deadline for this [18:44] * willvarfa is eager to see this for, say, Koenigstuhl [18:44] so that we could get a grade [18:44] it is very good [18:44] :) I'm glad you think so. [18:44] and the English is a very high level too [18:44] well, it's ok., I guess... we did are best [18:45] *our [18:45] (sigh) [18:45] but, as you can see, I still don't know how to type fast :) [18:45] I still don't know how to read fast [18:45] anyway, we'll continue with more clustering, comparisons, etc. [18:45] amateurs [18:46] :) [18:46] did you decide on the grouping i.e. MOVSUB or did the clustering find that relationship? [18:46] the clustering algorithm found it [18:46] I only determined the pairs to go into the representation [18:46] the EM took care for the rest [18:47] so, hmm, this is like looking for markovs? [18:47] well, do you know how EM works? [18:47] no [18:47] it is basically similar to incremental clustering algorithms, but then again, totally different :) [18:47] basically, [18:47] I guess this result is saying that MOVSUB is a very common combination? [18:48] not really [18:48] it says that by dividing on that node in the decision tree, you gain a lot of info about a class/cluster [18:48] that's why it's called info gain, in the first place :) [18:49] it means that there is a significant difference between the clusters when you look at that pair [18:49] ah right [18:50] understood [18:50] so you've clustered [18:50] and then you look for the largest dissimilarities? [18:50] yes [18:50] to see what distinguishes one cluster from another [18:51] the clustering in the first place was by? [18:52] define: by [18:52] what was the clustering using to determine similarity in the first place? [18:53] by the instructions? [18:53] well, some sort of distance measure, I suppose. it works similar to incremental clustering methods, but it tries to guess the parameters of normal distribution within clusters, then recalculates the clusters and does it again, etc. until it reaches a local/global extrema [18:53] yes, by the bag of instructions [18:54] that's why it is called: expectation maximization [18:54] ah right [18:54] ok [18:54] way beyond my comprehension :-) [18:54] :) [18:55] anyway, since you're here: how do you feel about a cw meeting next year? [18:55] one bad thing as come thing switching to Irssi [18:55] now whenever Nenad mentions "maximization", it catches "Miz" out of it [18:56] or optimization.. [18:56] heh [18:58] I am in favour of a meeting next year [18:58] :) [18:58] I've been meaning to go to the previous one, and these days I more have the means [18:58] I am a corewar clinger, though - not actually productive myself [18:58] now I really have to go unfortunately [18:59] :( [18:59] * Nenad waves [18:59] I do hope to continue your progress report next time we bump into each other here, or you write to r.g.c [18:59] ok. [18:59] and I will try to be in #corewars a bit more often too in the future [18:59] * willvarfa waves [18:59] see you sson, then [18:59] MSG: Remote host closed the connection [18:59] *soon [19:00] ... [19:00] well, time for me to go, as well [19:00] * Nenad waves to Miz [19:00] MSG: [20:14] Join: Core_old joined #corewars [21:19] MSG: Read error: Connection reset by peer