[08:01] hrm. [08:36] Join: Nenad joined #corewars [08:41] mornin [09:20] yep, it is a good morning [09:23] why is it so good? [09:39] hes high on something? [09:40] well, dunno. :) 'cause I'm in a good mood :) [09:40] it was the first night after a while that I've actually gotten some sleep [09:41] @Miz - interesting thinking [09:41] normally i expectg you to be pretty fast to reply [09:41] alas, I'd hardly describe my current work as something that makes one high [09:41] :P [09:42] yup :) - but I was writing some report [09:42] so I wasn't paying any attention to irc [09:42] how about you? [09:43] what are u up 2 2day? [09:44] very little [09:44] maybe i should try writing that compressor [09:45] strangely i am a bit dizzy [09:45] maybe its just my sleep rythm being lost [09:48] hmm... [09:48] I know the feeling [09:55] * bvowk does too. [09:58] although, lately I usually have trouble sleeping because I'm too hungry to sleep [09:58] have to get up and eat and then I'm good to sleep again [10:03] <-- hungry every 3 hours [10:18] :P [10:19] I only have one 15 minutes break at work, so there ain't much I can do about my hunger 'til I get home about 18h [10:20] wtf? [10:36] well, yes [10:36] :P [10:36] the conditions aren't exactly ideal [10:38] also, being eve slightly late is not tolerated, so getting up early is a bit stressful, but one gets used to it. [10:38] *even [10:39] heh. also, if you stay longer at work, you don't get payed extra and you don't get any free days later on. so, yes, it could be better from that perspective [10:40] so.. you're working at a place that thinks programmers are like line workers? [10:40] but it still is one of the best local software companies to work in [10:40] yes [10:40] exactly [10:40] all work, no fun [10:41] *work in = work for [10:41] part of the problem may be that the people who own and run the company aren't programmers themselves and never were. [10:54] sounds like that would suck. [10:59] * bvowk is glad they let him eat every few hours and come in at randomish times [11:08] heh [11:08] * Nenad envies bvowk [11:10] of course you do ;) [11:14] [11:14] so, whatcha hacking on? [11:15] and if they let you IRC at work.. I really don't understand :) [11:21] well, just been reading through some documentation. waiting for the task specification process to end (a bunch of meetings and discussions), and after that I'll be doing some real coding [11:21] I've just been fiddling with some data preprocessing so far [11:22] what sort of data? [11:22] but I'll have to scrap even that if they decide to change the rdbms used or if they say that they don't want java etc [11:22] measurements of electricity consumption [11:23] nothing fancy [11:24] so you're mining that to find the hydroponics grow ops? [11:24] and the guys like me? [11:34] well :), it's all more or less related to predicting how much energy will be consumed at various levels and the trends of consumption [11:34] sounds pretty simple at first glance, I know [11:35] but the data are very low quality, so it becomes a problem. [11:36] and they also have some other ideas/demands, but that's beside the point [12:26] hrm. [12:46] * Nenad hrms back [13:41] * bvowk prods nenad with a mechanical pencil [13:41] you awake? [13:42] * Nenad says: ouch [13:42] yup [13:42] doing anything interesting? [13:42] not really :) [13:42] I just prodded my evolver [13:42] I figured every 4 months or so I better check on it [13:43] and? [13:43] meh [13:43] it's alright [13:43] I tweaked it a little because I had an idea that might keep diversity up [13:51] ? [13:51] what sort of tweak did you use? [13:52] I made the warriors maintain a list of instructions from those that spawned it [13:53] and they might be used when breeding instead of the current ones [13:53] hmm... [13:55] so, basicaly, it's almost as if some recessive properties are within each warrior, not active, latent - and can also be transferred to the offspring [13:55] does it help? [13:56] who knows.. I just did it [13:57] if the population already started converging, I fail to see how it can help, but it could, in the very start, help in preventing such convergence [13:57] * Nenad ponders [13:59] well, I'm thinking it'll prevent successful warriors too far outside a family line from interbreeding as effectively.. [13:59] so it'll force divisions in the pool and hopefully prevent single strategies from taking over [14:00] should leave a little more balance in things I hope [14:01] or it could be a total waste of time :) [14:04] list of instructions from its entire family line or just its parents? [14:04] (list = frequencies = probabilities or an actual parent code) ? [14:14] most of the family line :) [14:14] actual code [14:14] it's condensed [14:17] hmmm... [14:17] well, you'll see the results empirically. [14:17] you measure diversity along the way? [14:17] not really [14:26] then you won't really know if it worked? you can just be satisfied or not satisfied with the scores in the end - right? - and then analyse the best warriors [14:32] well, I can watch the average turnover over time.. higher turnover means more diversity [14:33] and I peek at the contents periodically.. [14:33] if they're all the same warrior.. it's a pretty safe bet things have come to an end. [14:46] :) [14:47] maybe if you just put aside some good warriors, keeping them "in reserve" and reinsert them over time with some probability and mutate them a bit before doing so [14:47] which is similar to this, but more controlled [14:47] yace did that, didnt work [14:48] well, it is artificial, of course. just a thought. [16:05] * Nenad waves [16:05] MSG: [18:12] Join: jon__ joined #corewars [18:14] MSG: Read error: Operation timed out [20:26] Join: elkauka joined #corewars [20:27] hi@all [20:38] MSG: Quit: Quit now, you'll never make it. If you disregard this advice, you'll be halfway there. [20:46] Join: jon___ joined #corewars [20:47] MSG: Read error: Operation timed out [21:11] Join: fiveop joined #corewars