[05:14] <HowdyDoody> I am trying to modify a desktop icon to also add an environment variable before executing the program to call.  It seems to ignore the env settting.  Do I have some bad syntax here ?  -->Exec=/usr/bin/env PYTHONPATH=/usr/lib/virtualbox/:/usr/lib/virtualbox/sdk/bindings/xpcom/python/ gksudo python "/opt/GNS3/gns3"
[10:30] <rmg51> Morning
[13:08] <teddy-dbear> Morning peoples
[18:50] <MutantTurkey> waltman: hey. Are you around? I need some math help
[18:50] <MutantTurkey> CS/math help
[18:54] <jedijf> 2
[18:54] <MutantTurkey> jedijf: 3
[18:55] <jedijf> damn; i was guessing 1+1
[18:55] <MutantTurkey> heh
[19:02] <MutantTurkey> waltman: really though...
[19:17] <waltman> MutantTurkey: yo
[19:20] <waltman> MutantTurkey: log_2(n)
[19:21] <MutantTurkey> but why?
[19:21] <waltman> it's always log_2(n)
[19:22] <waltman> ok, sometimes it's 2^n
[19:22] <MutantTurkey> waltman: so I got a job rewriting matlab into C... one of our programs uses the NNLS algorithm (it's built into matlab), but I am having trouble finding a C counterpart...
[19:23] <MutantTurkey> and our file sizes are a bit ridiculous
[19:23] <MutantTurkey> but really I am having trouble understadning the function
[19:23] <MutantTurkey> waltman: I comprehend how least squares works, at least in theory, but I don't understand the non-negative aspect of non-negative least squares
[19:24] <waltman> ok, let me take a look at it.
[19:25] <MutantTurkey> http://amath.colorado.edu/computing/Matlab/OldTechDocs/ref/nnls.html
[19:25] <MutantTurkey> what can I show you?
[19:26] <MutantTurkey> wait they're using the lsqnonneg function actually.
[19:26] <waltman> matlab obsoleted it and it's not in 2012a (which is the version I have)
[19:28] <MutantTurkey> I have 2012a, yeah it's lsqnonneg now
[19:28] <MutantTurkey> waltman: btw drexel has 2012b up as well if you wanted it
[19:28] <waltman> I need to stick with the version we run at work.
[19:29] <MutantTurkey> gotcha
[19:29] <MutantTurkey> I didn't realize there were many versioning issues
[19:29] <MutantTurkey> matlab noob here
[19:30] <waltman> best to just stick with the same version
[19:30] <MutantTurkey> gotcha
[19:30] <waltman> MutantTurkey: Well, you could try to dig up a copy of that book or some other books on least squares...
[19:31] <MutantTurkey> I found a book in temples library, maybe I'll go grab it tomorrow
[19:33] <waltman> here's an entire phd thesis on the problem: https://smartech.gatech.edu/bitstream/handle/1853/31768/?sequence=1
[19:34] <waltman> only 35 citations? wtf?
[19:34] <MutantTurkey> well fk a whole damned thesis...
[19:35] <MutantTurkey> waltman: what a sucker... only 35 citations...
[19:35] <waltman> I got as far as pseudocode for the algorithm, and it didn't look like something I'd want to write if I didn't have to
[19:35] <ChinnoDog> You are now either in over your head or loverloaded with useful information.
[19:35] <MutantTurkey> ChinnoDog: far over
[19:36] <waltman> MutantTurkey: Anyway, as you've no doubt figured out by now, I've never heard of this version of the algorithm.
[19:37] <MutantTurkey> crud
[19:37] <waltman> MutantTurkey: It seems perfectly reasonable to ask your professor if he knows of any code that might do this.
[19:37] <MutantTurkey> shr'
[19:37] <MutantTurkey> she'
[19:37] <MutantTurkey> but yes, we found some implementations that are really slow
[19:38] <MutantTurkey> aka didn't work
[19:38] <MutantTurkey> because the matrix size is like 160 millio
[19:39] <waltman> maybe there's an implementation for octave? If so, you could at least see how it's implemented.
[19:39] <waltman> do you mean 160mil x 160mil?
[19:40] <MutantTurkey> no 160mil by 1 i think
[19:40] <MutantTurkey> 160mil by 100 i mean
[19:40] <MutantTurkey> a lot of data, not a lot of other stuff
[19:41] <MutantTurkey> in that lsnonneg function there are two arguments, what is that second argument containing? I don't really get it
[19:41] <MutantTurkey> waltman: it's actually 160milx1024
[19:45] <waltman> my guess is that you're trying to minimize the solution of Cx = d
[19:45] <waltman> but again, I think you'd be much better off discussing this with your prof
[19:45] <MutantTurkey> I concur
[19:45] <waltman> bbiab
[19:45] <MutantTurkey> but I sound like an idiot
[19:46] <MutantTurkey> "durr whats a vector durr"
[19:46] <waltman> No one expects college freshmen to understand linear algebra
[19:46] <waltman> laundry &
[19:49] <MutantTurkey> ok thanks :)
[19:52] <MutantTurkey> waltman: i have a few more idiot questions...
[19:58] <waltman> I should also point out that if you're back at Drexel, Ko's students know a lot more about least squares than I do.
[19:58] <waltman> (which is not very much)
[19:59] <MutantTurkey> really you think?
[19:59] <MutantTurkey> they always impress me
[19:59] <MutantTurkey> but I am also a burden unto them :p
[19:59] <waltman> I don't know much about least squares. They use it a fair bit.
[19:59] <MutantTurkey> waltman: so I how do I load data into matlab?
[19:59] <MutantTurkey> :p
[19:59] <waltman> that I do know!
[20:00] <MutantTurkey> the format is sparse, like this:
[20:00] <MutantTurkey> SPARSE
[20:00] <MutantTurkey> row height
[20:00] <MutantTurkey> total rows
[20:00] <MutantTurkey> 1 1 .242462 (row col value)
[20:00] <MutantTurkey> for the rest
[20:00] <waltman> http://www.mathworks.com/help/matlab/ref/spconvert.html
[20:00] <MutantTurkey> does matlab understand that?
[20:00] <MutantTurkey> thanks
[20:01] <MutantTurkey> thats exactly it!
[20:01] <waltman> BTW That was the first hit on google for "load sparse matrix into matlab"
[20:02] <MutantTurkey> lol yeah
[20:02] <MutantTurkey> I am trying that as well
[20:03] <MutantTurkey> waltman: if it is reading this Gb sized file in, can i tell matlab to load another file in at the same time?
[20:06] <waltman> You could try, but it seems unlikely. You've only got the single command window, right? How are you could to do it?
[20:07] <waltman> matlab seems mainly single threaded to me. asking it to do two things at the same time, even if you could, seems like a bad idea.
[20:09] <MutantTurkey> yeah
[20:09] <waltman> ugh, my English got mangled there.
[20:10] <waltman> It seems to me that when you're in the command window, you've got to wait for one command to finish before you can run another one.
[20:10] <MutantTurkey> right
[20:12] <MutantTurkey> waltman: ugh matlab prints out the whole matrix once loaded...
[20:12] <MutantTurkey> oh god
[20:12] <MutantTurkey> oh god
[20:13] <waltman> MutantTurkey: That's because you forgot to end the statement with a ;
[20:13] <waltman> you might try Ctrc-C
[20:13] <waltman> Ctrl-C
[20:16] <MutantTurkey> woul that interpt it though?
[20:16] <MutantTurkey> the command?
[20:16] <waltman> No, I'm pretty sure it has to finish, because it's showing you the answer.
[20:16] <MutantTurkey> ok gotcha