--- Day changed Thu Sep 04 2008 01:22 -!- pk [i=pk@cash-3b-130.res.umass.edu] has quit [] 01:37 -!- biopunk [n=p@h91n3c1o261.bredband.skanova.com] has quit [] 01:39 -!- ybit [n=h@unaffiliated/ybit] has quit [Read error: 104 (Connection reset by peer)] 02:28 -!- Nofaris_ [n=Nofaris@adsl-75-42-77-105.dsl.scrm01.sbcglobal.net] has quit [Read error: 110 (Connection timed out)] 02:36 * nsh reads buffer 02:41 < kanzure> :) 02:41 * nsh is rather dubious of measures like PoPS and RiPS 02:41 < kanzure> read on 02:41 < kanzure> I solve it 02:41 < kanzure> or at least that part 02:42 * nsh reads on 02:45 < nsh> so 02:45 < kanzure> there's also stuff in #bioinformatics where faceface got whiff of what I'm doing 02:46 < kanzure> *caught scent 02:47 < nsh> you want to systematically optimise a simulated gene regulatory network by training a stochastic model, such as HMM or ANN against a simulation of the GRN itself, such that the model becomes successful at predicting whether a certain change in the configuration of the GRN variables will lead to greater or lower efficiency, according to some metric? 02:47 < kanzure> correct 02:48 < nsh> but rather than replicate the computational complexity redundantly in two models, you feel it would be possible to feed the stochastic efficiency predicter only the uniqueness-factor of the GRN model and have its implicit modeling recreate it somehow? 02:48 < kanzure> but, not only is it (1) either a HMM or ANN or something -- this is variablized, and not only is (2) the GRN design circuit variablized, and not only is (3) the simulation results (obviously) variablized, but I'd also like to (4) variablize the fact that it's a GRN ... 02:48 < kanzure> uhm 02:48 < kanzure> hm 02:49 < kanzure> No, what I'm interested in seeing is if the stochastic model predictor thingy applies across different problems 02:49 < nsh> oh 02:49 < kanzure> and the ranking of these 'stochastic models' 02:49 < kanzure> particularly rules ("graph grammars") used to do some internal fixups on the 'stochastic models' 02:49 < kanzure> which makes it even weirder 02:49 < nsh> that sounds dangerously like an attempt to produce a general purpose solver 02:49 < nsh> which is almost certain to not work well 02:49 < kanzure> because then you have to ask, how many iterations of rule applications are you willing to go through as well? or will you just "lock it in" at some point 02:50 < kanzure> right 02:50 < kanzure> but it's not intended to do general purpose solving really 02:50 < kanzure> I think i'd be good if it works for just a given problem 02:50 * nsh nods 02:50 < kanzure> problem space, something with context. i.e., a particular GRN that you're making 02:50 < nsh> right 02:51 < kanzure> faceface was happy that it sounded like it could possibly automatically generate all possible GRNs / biochem pathways from some unmodeled organism, and then the stochastic models could help fish out which one of those has the correct 'gaps' filled in that nobody knows about yet from the bioinformatics 02:51 < kanzure> :-) which is one possibility :-) 02:51 < nsh> mmm 02:52 < kanzure> there's probably a simpler way to do it than a bruteforce tree search + dynamic programming / backpropagation 02:52 < kanzure> but anyway. 02:52 < nsh> bear in mind combinatorial explosion 02:52 < kanzure> right right 02:52 < nsh> complex networks scale terribly 02:52 < kanzure> bear in mind we have supercomputer access 02:52 < nsh> oh, we do? :-) 02:52 * nsh notes this 02:52 * kanzure grins 02:53 < kanzure> so, 02:53 < kanzure> you understand what it is that I'm doing, but 02:53 < kanzure> what the hell do I feed into the ANN/HMM/stochastic-model-thingy ? 02:53 < kanzure> something that is a function of a given generated/possible model, I know this much. 02:53 < kanzure> that's not much of a start :) 02:53 < nsh> aye 02:54 * nsh was thinking the same 02:54 < kanzure> so I was going to use biobricks as some seed content 02:55 < nsh> you can convolute a set of data in many different ways and still retain its uniqueness. some forms will be more facile to predict from than others; the question is, why? 02:55 < kanzure> and there's some ways I can get kinetic reaction or gene expression models for the biobrick parts, so that's a start 02:55 < kanzure> right, you could conceivably take the ID number of the generated design 02:55 < kanzure> or the md5 02:55 < nsh> well 02:55 < kanzure> but that's not necessarily going to produce something interesting with a limited ANN 02:55 < nsh> md5 is lossy 02:55 * nsh nods 02:56 < nsh> and the id number would only contain the data of the final model when coupled with the complexity of the GRN generating algorithm 02:56 < kanzure> the idea is to somehow model the simulator in less steps, right? 02:56 * nsh nods 02:56 < kanzure> i.e., hidden variables that aren't really variables in the simulation, but rather represent some 'computational complexity wetdream of a mix of variables' 02:57 < kanzure> computational complexity in the sense of complexity science, not the real deal 02:57 * nsh smiles 02:57 < nsh> but there's no free lunch, of course 02:57 < nsh> you can't arbitrarily reduce the complexity of the model 02:57 < nsh> only whatever is redundant in it 02:57 < nsh> and if you can't tell for certain whether you're optimising or approximating 02:58 < nsh> you run the risk of inaccuracy for certain marginal cases 02:58 < kanzure> right. 02:59 < nsh> so, maybe you can think of it like image compression, or mp3 02:59 < kanzure> it = ? 02:59 < nsh> it, the optimisation of the GRN simulation 03:00 < nsh> what features are extraneous to the final output variable 03:00 < nsh> which is, for example, some measure of optimality 03:00 < nsh> so, in testing efficiency, you're taking a large data set and reducing it finally to a very small one 03:00 < kanzure> not optimizing the simulator, btw 03:01 < nsh> mmm 03:01 < kanzure> okay, so that's possible, 03:01 < kanzure> maybe the stochastic models can figure out the 'interesting stuff' on the fringes 03:01 * nsh bbl 03:01 < kanzure> that don't "compress well" in the typical patterns of other generated designs (GRNs, or stoichiometrical chem, ..) 03:01 < kanzure> hrm 03:04 < kanzure> http://www.me.utexas.edu/~adl/graphsynth/images/flowchart.png <-- harder than this claims it to be 03:05 < kanzure> wait 03:05 < kanzure> isn't that the semantic search facilitator? 03:05 * kanzure scratches his head 03:09 < kanzure> Oh, maybe this is better described as a runtime/online interpreter to 'dynamic programming' 03:35 < kanzure> do I need the stochastic model? if I can autogenerate all possible designs using a set of 'rules' for the manipulation of the circuits/designs, then surely some rules are going to be more useful in a certian problem domain than in others. These rules, that when applied, if they consistently make 'better stuff' then that's what should be used more often, no? 03:37 < kanzure> hypothesis: there are hidden variables between genotype and phenotype 03:40 < bkero> Perhaps you could adapt acovea for what you're doing? :) 03:40 < kanzure> http://www.coyotegulch.com/products/acovea/ 03:40 < kanzure> 'ACOVEA (Analysis of Compiler Options via Evolutionary Algorithm)' 03:40 < kanzure> oh god 03:40 < kanzure> just what I need :) 03:40 < bkero> Bwahaha 03:41 < bkero> btw the guy who runs it is insane 03:41 < bkero> clinically 03:41 < kanzure> How serious are you? 03:42 < kanzure> insane as in insane in the sense that you'd have to be to do that code, 03:42 < bkero> Heh 03:42 < bkero> He does some weird shit 03:42 < kanzure> or in an institution for the insane? 03:42 < bkero> former 03:42 < kanzure> neat 03:42 < kanzure> wait, this is just for compiler flags? 03:42 < bkero> Exactly 03:42 < kanzure> Nothing more? 03:42 < bkero> Hm? 03:43 < kanzure> Compiler optimization usually looks at code and then tries to find a better way and fixes your mistakes 03:43 < kanzure> evolutionary algorithms and backpropagation and such are what I've been thinking of 03:43 < kanzure> and really this /is/ a compiler/optimization problem 03:43 < bkero> This basically takes algorithms, and figures out what combinations of compiler options yield the fastest runtimes. 03:43 < kanzure> if it's just "flags" in general, and not analyzing algorithms (loops?) in the code, then blah 03:44 < kanzure> why not just implement the code from within the compiler in the first place ? 03:44 < bkero> But it's built in such a modular way you could adapt it to running any dataset through a filter to see which performs better(less entropy, faster runtime, faster compile time, etc) 03:44 < bkero> It's testing combinations of flags 03:45 < bkero> based on evolution 03:45 < kanzure> yes, I know 03:45 < kanzure> it says that. 03:45 < bkero> There might be a plugin for gcc that does this, but I've never heard of it. 03:47 < fenn> "AMD’s products are not designed, intended, authorized or warranted for use as components in systems intended for surgical implant into the body" well shucks, so much for that idea 03:48 < bkero> Fuck AMD 03:48 < bkero> go arm 03:48 < kanzure> from their legal mumbo jumbo? 03:48 < kanzure> bkero: Atmel meeting later today :) 03:48 < bkero> lol 03:48 < kanzure> they're showing up for the local IEEE chapter 03:48 < bkero> Gotta love those 8-bit microcontrollers 03:48 < kanzure> methinks they've quadrupled that by now 03:49 < bkero> Seriously though, ARM chips are where it's at if you need any serious cruning done in an embedded environment. 03:49 -!- faceface [n=dbolser@bioinformatics.org] has joined #hplusroadmap 03:49 < fenn> there are 32 bit AVR's but most people use the 8 bit chips because they dont come in ridiculous BGA packages 03:49 < fenn> and the free compiler yadda yadda 03:49 < kanzure> I'm nearly convinced that the Semantic Search Facilitator project is equivalent to the Automated Design Lab project. 03:50 < fenn> you've rediscovered the do what i mean interface 03:50 < bkero> Heh 03:50 < kanzure> does this mean I have to become Larry Wall ? 03:50 < fenn> there can only be one 03:50 < bkero> You're doomed to a terrible fate. 03:50 < bkero> Dude, have you SEEN larry wall? 03:50 < kanzure> No? 03:50 < kanzure> Have YOU? 03:51 < kanzure> srsly 03:51 < bkero> yup 03:51 < bkero> He was at oscon this year 03:51 < kanzure> Oh. 03:52 < kanzure> me was going to use GAs to mutate the search strings 03:53 < kanzure> and then find those GA / rulesets that prove most lucrative in certain search domains 03:53 < kanzure> that's basically the same thing here 03:53 < kanzure> except not natural language 03:53 < kanzure> and not the web 03:54 < bkero> lol 03:56 < bkero> Sounds like some bioperl 03:56 < kanzure> Yes, I've toyed with bioperl.org before. 03:57 < kanzure> the wikipedia article on ANNs vaguely references them as "systems that learn in an /optimal/ sense" 03:58 < kanzure> all these darned things are really just ways of otherwise saying 'optimization' 03:58 < bkero> lol 03:58 < kanzure> life is a compiler, I am the optimizer and you sir are a very, very Big O notational headache 03:58 < kanzure> hrm 03:58 < kanzure> I'm sure that line could be improved 03:59 < bkero> I'm actually bytecode. 03:59 < bkero> Already interpreted and ready to be run 03:59 < kanzure> Where's your virtual machine? 03:59 < bkero> bvm 03:59 < bkero> It has yet to be written. 03:59 < bkero> So I'm actually useless bytecode, and very hard to reverse back into higher-level. 04:04 < kanzure> http://heybryan.org/~bbishop/docs/2008-07-31/ 04:06 < kanzure> http://www.cs.jyu.fi/ai/OntoGroup/InBCT_May_2004.html <-- Semantic Search Facilitator 04:06 < kanzure> "Semantic Search Assistant hides from users the complexity of query language of concrete search engine and performs routine actions that most of users do in order to achieve better performance and get more relevant results." 04:19 < faceface> kanzure, I was working with a guy who worked on that field 04:20 < kanzure> 'that' ? 04:20 < kanzure> bkero: I've found myself wanting a compiler optimization database of tricks and such. gcc is more of a hack IIRC. llvm looks more structured, maybe. 04:20 < faceface> oh, semantic search facilitator type stuff 04:21 < kanzure> Neat. 04:21 < kanzure> What'd you come up with? 04:21 < faceface> cant find his website 04:21 < kanzure> btw, don't care about the semantic aspect .. just the facilitator part ;-) 04:22 < faceface> actually we were working on protein-protein interactions 04:22 < kanzure> excellent 04:22 < kanzure> so, how did that work? 04:22 < kanzure> helping researchers to find the right proteins to fit into the puzzle? 04:22 < faceface> i.e. nothing to do with semantics 04:22 < faceface> his idea was that ontologies should help you lean a new field 04:23 < kanzure> damn right 04:23 < bkero> science 8) 04:25 < faceface> found it at last... http://www.vincent-wolowski.net/ 04:25 < faceface> yup, science it is :-) 04:25 < kanzure> polish 04:25 < faceface> its just most tools don't let you use an ontology in that kind of didactic / explorative way 04:26 < kanzure> right 04:26 < kanzure> explorative = realtime/online debuggng, profiling or compiler optimization 04:26 < kanzure> meaning some really funky code 04:26 < faceface> http://www.vincent-wolowski.net/informatics.html 04:26 < kanzure> or a giant virtual machine 04:26 < faceface> well.. time to work 04:27 < kanzure> what's your latest? 04:28 < faceface> boring 04:28 -!- willPow3r [n=will@cpe-66-75-6-181.san.res.rr.com] has joined #hplusroadmap 04:28 < faceface> people hear are boring 04:28 < kanzure> where? 04:28 < willPow3r> hear 04:28 * kanzure gets ready to mark a place off his map 04:28 < kanzure> hehe 04:28 < faceface> guess what... disks are getting larger... data is getting more... lets spend the morning talking about how much dna exists... 04:29 < faceface> dundee, scotland 04:29 < kanzure> eh? 04:29 < kanzure> they want to calculate the amount of existing dna? 04:29 < faceface> oh well... (I'm only jelous because they are having this conversation without me) 04:29 < faceface> no that would be interesting 04:29 < kanzure> oh 04:29 < kanzure> :) 04:29 < kanzure> then what are they talking about? 04:29 < faceface> they can only see what people who do sequencing present to them 04:30 < kanzure> they have no sequencer? 04:30 < kanzure> wtf 04:30 < faceface> sequening = faster, data = more 04:30 < faceface> not here 04:30 < faceface> oh if data = more then ... OMFG! 04:30 * kanzure wonders about an 'ecological search facilitator' to find unknown niches/links 04:30 < kanzure> in thermodynamic systems in general. hrm. 04:30 < faceface> well.. have fun 04:30 < kanzure> heh 04:30 < kanzure> okay 04:32 -!- marainein [n=marainei@220.253-204-179.VIC.netspace.net.au] has joined #hplusroadmap 04:40 < kanzure> oh 04:40 < kanzure> it occurs to me that in the semantic search facilitator, or my 'autogoogler', if I had the 20 different queries up, I would only obviously select a few (or maybe all (eventually)), 04:41 < marainein> kanzure: what thing are you talking about? 04:41 < kanzure> from that selection information, the 'GA, rules, HMM, or ANN aspect'' that were used to generate those queries would be reinforced 04:41 < kanzure> marainein: hold on 04:41 < kanzure> as nsh said it: 04:41 < kanzure> "you want to systematically optimise a simulated gene regulatory network by training a stochastic model, such as HMM or ANN against a simulation of the GRN itself, such that the model becomes successful at predicting whether a certain change in the configuration of the GRN variables will lead to greater or lower efficiency, according to some metric?" 04:42 < kanzure> "but rather than replicate the computational complexity redundantly in two models, you feel it would be possible to feed the stochastic efficiency predicter only the uniqueness-factor of the GRN model and have its implicit modeling recreate it somehow?" 04:46 < kanzure> applications include automatically generating genetic regulatory networks, factories, search queries, and filling in the gaps in our knowledge of genotype/phenotype relations. 04:49 < kanzure> you know what I like, the simulator allows numeric output 04:49 < kanzure> instead of flipping through google search results 04:50 < kanzure> although google search results give semantic feedback 04:50 < kanzure> hm, tradeoffs :) 04:50 * marainein nods 04:57 < kanzure> Suppose you took the opposite approach of limiting the search results and instead expanded it; perhaps you applied the probabilistic web crawling papers to the issue of design. Then, you have weird, strange variables that relate various models (out of the 10,000 generated models) together. 04:57 < kanzure> By laws of connectivity, clicking from one to the next should only take a certain amount of time to find The Good One -- some upper-bound on the number of clicks/hops that it would take. But it has to be more links than the number of nodes overall, obviously. 05:02 < kanzure> So, I'm interested in taking the 'Autogoogler' idea in terms of design. There's no "web" that is pre-generated. There's "not much" to search. The user input is, what? The results are ranked by simulator results, obviously -- whatever parameters the user is ultimately optimizing for. 05:02 < kanzure> What's the user input? It has to be constrained - it's not going to be 'words' since there's no 'words' attached to the generated designs. I suppose the user input is the selection of different 'rules' to apply to rewrite the design-graphs (possible manipulations between, i.e., biobricks), by clicking on the equivalent of the 'list of 20 search queries'. 05:41 -!- nsh [n=nsh@eduroam-52.uta.fi] has quit [Read error: 104 (Connection reset by peer)] 05:41 -!- nsh [n=nsh@eduroam-52.uta.fi] has joined #hplusroadmap 05:59 < kanzure> http://www.sustainabledave.org/ 06:00 < kanzure> http://www.youtube.com/watch?v=fdJjyoHdnIA 06:00 < kanzure> http://www.youtube.com/watch?v=opaTDxWUyZc 06:00 < kanzure> the npr link might be beter 06:00 < kanzure> *better 06:26 * nsh wonders how kanzure has time to find all the links he spouts, when he doesn't even have time to read them all 06:26 < nsh> he->I 06:27 -!- willPow3r [n=will@cpe-66-75-6-181.san.res.rr.com] has quit ["Leaving"] 06:27 < kanzure> nsh: the confused mumblings of /me, ah yes :-) 06:27 < kanzure> silly mortal, there is no /me 06:27 < kanzure> Anyway, the trick is to be not human. 06:30 -!- willPow3r [n=will@cpe-66-75-6-181.san.res.rr.com] has joined #hplusroadmap 06:31 -!- willPow3r [n=will@cpe-66-75-6-181.san.res.rr.com] has quit [Client Quit] 06:35 -!- willPow3r [n=will@cpe-66-75-6-181.san.res.rr.com] has joined #hplusroadmap 06:38 -!- faceface [n=dbolser@bioinformatics.org] has quit [Remote closed the connection] 06:49 * nsh guesses as much 06:56 < kanzure> http://partsregistry.org/cgi/partsdb/pgroup.cgi?pgroup=measurement&show=1 06:56 < kanzure> grumble grumble, why does this have html output 06:59 -!- Overand [i=overand@pdpc/supporter/active/Overand] has quit [No route to host] 07:00 < kanzure> http://www.biodas.org/wiki/Main_Page 07:01 < kanzure> spec: http://www.biodas.org/documents/spec.html 07:13 * nsh reads http://www.nature.com/nrg/journal/v9/n9/full/nrg2414.html 07:13 < nsh> , cringes at capitalised wiki 07:15 < kanzure> hrm 07:17 < nsh> heh, type "nucleur" 07:17 < nsh> Nature should hire a proofreader 07:17 * nsh will do it for 5 figures 07:18 < nsh> two typos "compute" 07:20 < nsh> this is pretty kintergarten stuff :-/ 07:20 < nsh> good that they have a concrete example problem though 07:25 < kanzure> I got stupid and turned the televisional news on 07:26 < nsh> :-/ 07:26 -!- faceface [n=dbolser@bioinformatics.org] has joined #hplusroadmap 07:26 < nsh> usually it happens the other way around 07:26 < nsh> hey dan 07:26 < nsh> (people watch the television news and then become stupid) 07:26 < kanzure> and there's this silly stuff saying "mmr vaccine doesn't cause autism - it's the nail in the coffin to the debate .. so stop talking about it." but they'll just say it's a conspiracy 07:26 -!- jimjim [n=dbolser@bioinformatics.org] has joined #hplusroadmap 07:28 * nsh tries to ignore the crazies 07:30 < kanzure> yay, let's treat these there darned children with these thingies called brainers or somethin' 07:30 < kanzure> that'll learn 'em 07:30 * kanzure fetches a pitchfork 07:31 * nsh smiles 07:31 < kanzure> http://www.usautism.org/ 07:31 < nsh> [[[ 07:31 < nsh> A mature biology cyberinfrastructure should make this type of analysis much more straightforward (Fig. 2b). Ideally, the researcher would be able to design the experiment at a high level by describing the data sets he wishes to work with and the relationships he wishes to traverse (protein to gene to exon to conservation score) by using a graphical tool or a high-level description language. 07:31 < nsh> The infrastructure would then do the hard work of finding databases, analysis services and compute resources that can satisfy the request, thereby transforming and integrating the data and returning the results. If the researcher desired, he could then easily share the method and results with the research community by pushing a 'publish' button; this information would then become a discoverable service that could be re-used by others. 07:31 < nsh> Over time, other members of the research community could add value to this work by commenting on it, linking it to related work, contributing modifications to the method, and submitting new raw and analysed data sets that enrich it. 07:31 < kanzure> I'm going to go yell at them 07:32 < nsh> ]]] --http://www.nature.com/nrg/journal/v9/n9/full/nrg2414.html#B4 07:32 < kanzure> I think they're down the street 07:32 < kanzure> screw them 07:32 < kanzure> look at their wiki 07:32 < nsh> (above is essentially saying: wouldn't it be cool if there was MAGIC!?!?!?!?!) 07:32 < kanzure> they cite SKDB for a reason 07:32 < kanzure> that's not quite magic 07:32 < kanzure> just throw it into git repositories 07:32 < kanzure> we'll do the rest, m'am 07:33 * nsh smiles 07:34 * nsh actually believes all the above is quite possible 07:34 < kanzure> http://www.youtube.com/user/USAutismAndAsperger 07:34 < kanzure> "how do we fix these kids" 07:34 < nsh> do i really want to watch that? 07:34 < kanzure> no 07:34 < nsh> IGNORE THE CRAZIES 07:34 < kanzure> but it's next door :-( 07:35 < nsh> then leave your rubbish bags outside their porch 07:35 < nsh> but don't give them brainspace 07:35 < nsh> that's valuable finding-cool-links-for-nsh time 07:35 < nsh> and coming-up-with-interesting-ideas time 07:43 < kanzure> now wonder these kids "withdraw" (whatever that means) 07:44 < kanzure> *no 07:46 < nsh> mm 07:52 < kanzure> find me my variables for my GRN optimizer 07:52 < kanzure> heh 08:01 -!- marainein [n=marainei@220.253-204-179.VIC.netspace.net.au] has quit [Read error: 113 (No route to host)] 08:08 -!- kikl [n=h@c-71-207-240-143.hsd1.al.comcast.net] has joined #hplusroadmap 08:09 < kanzure> Hey kikl 08:09 < kanzure> oh, probably ybit 08:09 < kanzure> no fun. 08:09 < kanzure> What's with the random nickjoins, anyway? 08:14 -!- Phreedom_ [n=freedom@ip-194-50-167-184.mir.dn.ua] has quit [Read error: 110 (Connection timed out)] 08:36 < faceface> ignore me 08:37 < jimjim> I thought xchat had crashed 08:37 -!- jimjim [n=dbolser@bioinformatics.org] has quit ["leaving"] 08:40 < faceface> nsh, thanks for the link - we were just talking about that at lunch 08:40 < faceface> I saw it on nature networks forum 08:41 < faceface> nsh, kanzure, are you on nature networks? 08:41 < faceface> not that its good for anything really... 08:42 < kanzure> No, is it a social community thingy? 08:42 * kanzure stole nature.com 08:42 < kanzure> All of it ... 08:42 < faceface> sure 08:42 < faceface> kanzure, you did that with a log in somewhere? 08:43 < faceface> cus a guy over on #bioinformatics would be interested in dumping those PDF's into a P2p 08:43 < kanzure> faceface: I need somebody god damned serious about it. 08:44 < faceface> kanzure, yeah... 08:44 < kanzure> I'm willing to invest some money into it for a machine to put behind a coffee shop. 08:44 < kanzure> 40 GB is a *huge* torrent. 08:44 < faceface> coffee shop? 08:44 < faceface> kanzure, dont torrent the whole thing man! 08:44 < kanzure> Right, just leave it on a hdd uploading or torrenting. 08:44 < kanzure> Hm? 08:44 < kanzure> Why not ? 08:44 < faceface> cus its not what ppl want to download... a few yes 08:44 < kanzure> you and me both 08:44 < kanzure> that's about it :-) 08:44 < kanzure> No, I'm sure some foreign universities would love to meet me 08:45 < faceface> ppl want papers 08:45 < faceface> oh yeah... them too 08:45 < faceface> forgot about the poor folk 08:46 < kanzure> cd3wd comes to mind 08:46 < kanzure> Although CD3WD is only 700 MB. 08:46 < kanzure> This would substantially increase their collection. Oops. 08:46 < kanzure> von Neumann probe project here we come? 08:47 < kanzure> (not with natural language) 08:50 < kikl> kanzure, good guess 08:50 < fenn> CD3WD is many GB 08:50 -!- kikl is now known as ybit 08:51 < fenn> about 14GB i think 08:51 < ybit> do you guys ever sleep o.O 08:51 < fenn> it's getting close to my bedtime 08:52 < faceface> this week 08:56 < kanzure> sleep? 08:57 < kanzure> Sleep is for the weak! Food is for the weak! err .. 08:57 < kanzure> (2008-08-26 11:21:12) kanzure: nsh: Food is for the weak. Sleep is for the weak. Fight entropy: start doing nothing today! 09:02 < ybit> :) 09:53 -!- ybit [n=h@unaffiliated/ybit] has quit [Read error: 54 (Connection reset by peer)] 10:04 < faceface> bot? 10:10 -!- pk [i=pk@cash-3b-130.res.umass.edu] has joined #hplusroadmap 10:21 < nsh> []_+````{ pk: STATE PURPOSE. }''''+_[] 10:22 < pk> lawl what? 10:23 * nsh smiles 10:39 -!- Overand [i=overand@pdpc/supporter/active/Overand] has joined #hplusroadmap 11:24 -!- Nade [n=lixasd@cpc2-clif5-0-0-cust516.nott.cable.ntl.com] has joined #hplusroadmap 13:39 -!- Phreedom_ [n=freedom@ip-194-50-167-184.mir.dn.ua] has joined #hplusroadmap 13:44 -!- Phreedom_ [n=freedom@ip-194-50-167-184.mir.dn.ua] has quit [Remote closed the connection] 13:45 < kanzure> And now I know the Great Evils of CAD. 13:46 < kanzure> while the interface isn't as bad as blender, it doesn't promote the cleanest thinking 13:46 -!- Phreedom [n=freedom@ip-194-50-167-184.mir.dn.ua] has joined #hplusroadmap 13:46 < kanzure> specifically solidworks 13:47 < bkero> How about CAM? 13:50 < kanzure> Haven't tried yet. 13:50 < kanzure> 'There is a plate of dark chocolate cookies with tiny chocolate chips in the middle office.  It has been brought to our attention that, if these cookies are not consumed by the end of the business day, certain peril might doom every PCR ever carried out in this building henceforth.  13:50 < kanzure> It is our suspicion that super intelligent bears have laid plans to dope DNAses into every tiny 200 uL eppendorf in every lab, and they are using the chocolate chips within the cookies as a homing signal to know where their target lies.  In order to safeguard the success of the work carried out within the MBB, we highly encourage you, the brave citizens of Ellingtonia, to consume these cookies with all due haste.  ' 13:50 < kanzure> Ah yes, interlab politics 13:50 < kanzure> I'll (never?) miss it 13:51 < bkero> Mmm cookies. 13:52 < bkero> kanzure: fscking awesome 13:52 < kanzure> :) 13:53 < kanzure> oh the terror http://heybryan.org/~bbishop/docs/me302/ 13:53 < bkero> I just have faceless cube. 13:53 < kanzure> cube? 13:53 < bkero> Cubicle 13:53 < kanzure> there were one or two cubes in Andy's lab 13:53 < Phreedom> kanzure, fenn, nsh: I wonder if any of you use jabber? 13:53 < bkero> Coworkers are made out of cardboard. 13:53 < kanzure> they had the Holy Post Docs that No One Must Ever Disturb. 13:53 < kanzure> Phreedom: yes 13:53 < kanzure> Phreedom: kanzure@gmail.com 13:54 * kanzure also knows jer, the guy responsible for xmpp in the first place .. *cough* 13:54 < kanzure> bkero: Ironically, these cubes were right next to Andy's (the PI's) office 13:54 < kanzure> you'd think they might get some of the most freedom 13:54 < bkero> Cubes are a sterile and uninteresting environment. 13:54 < kanzure> i.e., furthest desks maybe 13:55 < kanzure> Oh no, 13:55 < kanzure> there is no sterility and uninterest in a molecular bio lab 13:55 < kanzure> far from it :) 13:55 < bkero> There is at intel. :P 13:55 < bkero> My cube is J14-JF4-104 13:56 < kanzure> Isn't that a number from Star Wars? 13:56 < bkero> Not that I can remember. 13:56 < bkero> 3720:1? 13:56 < kanzure> Intel has stormtroopers? 13:56 < kanzure> Are those the DRM goons ? 13:57 < bkero> More or less 13:57 < bkero> Mostly they're overfed downs kids. 13:57 < kanzure> That's unfortunate. 14:17 < kanzure> use case for the automated design prototype with biobricks 14:18 < kanzure> consider Charlie/newgenome 14:18 < kanzure> who has been showing up with his various ideas to use various thingies to do things 14:18 < kanzure> one of his interests is synthetic biology 14:18 < kanzure> How would he, somebody who could learn a techical spec if necessary, address his ooh's and aah's in the right way to let the computer hash out the possible designs via the permutations and scoring and evaluators and so on? 14:19 -!- willPow3r_ [n=will@cpe-66-75-6-181.san.res.rr.com] has joined #hplusroadmap 14:19 * kanzure doesn't like assuming only one type of variable worth dealing with in these contexts ("gene expression levels, yay we're done") 14:19 -!- willPow3r_ [n=will@cpe-66-75-6-181.san.res.rr.com] has quit [Client Quit] 14:19 < kanzure> guess it's dependent on the information available more than anything 14:19 < kanzure> so given a dataset that he selects from, there's certain amount of pre-encoded functionality that could be combined together 14:57 -!- Netsplit niven.freenode.net <-> irc.freenode.net quits: procto 14:58 -!- Netsplit over, joins: procto 14:58 < kanzure> Daily wtf by way of Damien Broderick: 'David Duchovny has entered a rehabilitation center for sex addiction, The Associated Press reported. Mr. Duchovny, who plays a sex-obsessed character on the Showtime show "Californication," did so voluntarily, according to a statement on Thursday from ... ' 15:02 < kanzure> http://www.neuro.gatech.edu/groups/potter/papers/DagstuhlAIBakkumpreprint.pdf Removing the 'a' from 'ai' 15:11 < kanzure> http://machineslikeus.com/news/thinking-causes-weight-gain <-- studying students and their calorie intake /after/ intellectually daunting tasks 15:12 < kanzure> this sounds like an amazingly awesome excuse for me to head downstairs for a bite to eat right nwo 15:29 -!- Nofaris_ [n=Nofaris@adsl-75-42-77-105.dsl.scrm01.sbcglobal.net] has joined #hplusroadmap 16:06 < kanzure> http://austinbrains.org/austin.ics <-- 1.4 MB of events. Not that it's interesting to any of you, of course. 16:06 < kanzure> just had to throw it up there :) 16:44 < kanzure> anybody surprised that Google Calendar can't handle the calendar? 16:56 * kanzure is downloading 15 GB of data from the UT Solar Vehicle Team server 16:58 < kanzure> http://utsvt.ece.utexas.edu/wiki/index.php/IT:Quick_Start_Guide 16:58 < kanzure> wtf, somebody on campus uses a bug tracker 16:58 < kanzure> hurray 17:07 < kanzure> argh 17:07 < kanzure> can't escape the bullshit 17:07 < kanzure> just trying to get some cartoons in the background ... 17:08 < kanzure> "A man was dragged over a mile today ... investigation at 10 ... the Autism Conference started today including this ectostatic chamber " -- note that that this was looking like a giant body bag, the thing the Star Trek vessels shoot you out of when you're as dead as you're ever going to be (you have to really worry when even they, the Trekkies, cannot save you) -- "but what did they think about the new vaccine research? Coming up." do'h 17:08 < kanzure> when you're as dead as you're ever going to be (you have to really worry when even they, the Trekkies, cannot save you) -- "but what did they think about the new vaccine research? Coming up." do'h 17:24 < kanzure> 'hyperbaric chamber' 17:25 < kanzure> oh 17:25 < kanzure> sensory deprivation 17:25 < kanzure> ' Portable Hyperbaric Chamber|Hyperbaric Oxygen Chamber|Portable ...Manufacturer of low pressure portable chambers for use in clinics and offices. Includes specifications and distributor locations.!' 17:29 -!- jk4930 [n=jk@p57B758C8.dip.t-dialin.net] has joined #hplusroadmap 17:42 -!- jk4930 [n=jk@p57B758C8.dip.t-dialin.net] has left #hplusroadmap ["Konversation terminated!"] 18:20 -!- ybit [n=h@unaffiliated/ybit] has joined #hplusroadmap 18:32 -!- ybit [n=h@unaffiliated/ybit] has quit [Remote closed the connection] 18:32 -!- ybit [n=h@unaffiliated/ybit] has joined #hplusroadmap 19:08 -!- Phreedom [n=freedom@ip-194-50-167-184.mir.dn.ua] has quit [Read error: 110 (Connection timed out)] 20:10 -!- ybit [n=h@unaffiliated/ybit] has quit [Remote closed the connection] 20:28 < bkero> Heh 20:28 < bkero> I need myself a hyperbaric chamber. 20:28 < bkero> Hang out at around 6 BAR all day. 20:57 -!- Phreedom [n=freedom@ip-194-50-167-184.mir.dn.ua] has joined #hplusroadmap 21:26 -!- ybit [n=h@unaffiliated/ybit] has joined #hplusroadmap 21:39 < fenn> three years later and this thing still reaches out and grabs me: http://www.positron.org/projects/A51/ 22:10 < kanzure> augh 22:10 < kanzure> so 22:10 < kanzure> of all people 22:10 < kanzure> does anyone remember that I was once dating a blind woman? 22:11 < kanzure> before she knew me she was with someone who turns out to be the 'head sys admin' of the robotics group here 22:11 < kanzure> which is awkward because I think after she knew me she went back to him 22:11 < kanzure> what happened to "there can only be one [programmer per high school]" anyway? 22:12 < fenn> i think that went out of fashion after they stopped playing the Highlander series 22:15 < kanzure> is CentOS worth getting? 22:26 < fenn> no 22:29 -!- Nofaris___ [n=Nofaris@adsl-75-42-77-105.dsl.scrm01.sbcglobal.net] has joined #hplusroadmap 22:30 < ybit> anyone speak mandarin? 22:30 < ybit> meh 22:32 < kanzure> it's on my todo list 22:37 < ybit> mine too 22:37 < ybit> after spanish and portuguese 22:38 < ybit> if you learn spanish, you essentially know portuguese 22:38 < ybit> guessing it will take half the time or less to learn compared to spanish 22:46 -!- Nofaris_ [n=Nofaris@adsl-75-42-77-105.dsl.scrm01.sbcglobal.net] has quit [Read error: 110 (Connection timed out)] 22:50 < fenn> i would learn mandarin first 22:50 < fenn> spanish is too easy 22:51 < kanzure> agreed 22:51 < kanzure> also, mandarin = more people 22:58 < ybit> spanish = latin america 23:00 < ybit> also, countries with less control if that appeals to anyone 23:01 < ybit> spanish will only take about 2 years to learn anyway, and since i've been studying for about a month now, i'll continue with it :) 23:12 -!- ybit [n=h@unaffiliated/ybit] has quit [] 23:14 -!- ybit [n=h@unaffiliated/ybit] has joined #hplusroadmap 23:25 -!- Nofaris___ [n=Nofaris@adsl-75-42-77-105.dsl.scrm01.sbcglobal.net] has quit [Read error: 110 (Connection timed out)] 23:27 -!- ybit [n=h@unaffiliated/ybit] has quit [] 23:29 -!- ybit [n=h@unaffiliated/ybit] has joined #hplusroadmap 23:33 -!- Nofaris___ [n=Nofaris@adsl-75-42-77-105.dsl.scrm01.sbcglobal.net] has joined #hplusroadmap 23:34 -!- Nofaris___ is now known as Nofaris 23:57 -!- Nade is now known as Nade|away