verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
awhitesong has left #mlpack []
stephentu has quit [Ping timeout: 246 seconds]
stephentu has joined #mlpack
curiousguy13 has quit [Ping timeout: 252 seconds]
jbc_ has quit [Quit: jbc_]
curiousguy13 has joined #mlpack
stephentu has quit [Ping timeout: 250 seconds]
stephentu has joined #mlpack
nssssandeep has joined #mlpack
nssssandeep has quit [Remote host closed the connection]
jbc_ has joined #mlpack
jbc_ has quit [Quit: jbc_]
jbc_ has joined #mlpack
jbc_ has quit [Client Quit]
kshitijk has joined #mlpack
kshitijk has quit [Ping timeout: 252 seconds]
kshitijk has joined #mlpack
kshitijk has quit [Ping timeout: 256 seconds]
awhitesong has joined #mlpack
awhitesong has left #mlpack []
curiousguy13 has quit [Ping timeout: 250 seconds]
kshitijk has joined #mlpack
kshitijk has quit [Ping timeout: 250 seconds]
curiousguy13 has joined #mlpack
kshitijk has joined #mlpack
sahmed95 has joined #mlpack
stephentu has quit [Ping timeout: 265 seconds]
sahmed95 has quit [Ping timeout: 252 seconds]
sahmed95 has joined #mlpack
kshitijk has quit [Ping timeout: 244 seconds]
kshitijk has joined #mlpack
dhfromko_ has joined #mlpack
dhfromko_ has quit [Remote host closed the connection]
kshitijk has quit [Ping timeout: 245 seconds]
dhfromko_ has joined #mlpack
sahmed95 has quit [Ping timeout: 246 seconds]
dhfromk__ has joined #mlpack
dhfromko_ has quit [Ping timeout: 245 seconds]
dhfromk__ has quit [Read error: Connection reset by peer]
dhfromko_ has joined #mlpack
dhfromko_ has quit [Read error: Connection reset by peer]
dhfromko_ has joined #mlpack
sahmed95 has joined #mlpack
kshitijk has joined #mlpack
jbc_ has joined #mlpack
dhfromko_ has quit [Remote host closed the connection]
dhfromko_ has joined #mlpack
curiousguy13 has quit [Ping timeout: 264 seconds]
dhfromko_ has quit [Remote host closed the connection]
curiousguy13 has joined #mlpack
dhfromko_ has joined #mlpack
kshitijk has quit [Ping timeout: 255 seconds]
sahmed95 has quit [Ping timeout: 240 seconds]
dhfromko_ has quit [Remote host closed the connection]
dhfromko_ has joined #mlpack
sahmed95 has joined #mlpack
sahmed95 has quit [Ping timeout: 250 seconds]
dhfromko_ has quit [Remote host closed the connection]
dhfromko_ has joined #mlpack
sahmed95 has joined #mlpack
stephent1 has joined #mlpack
dhfromko_ has quit [Read error: Connection reset by peer]
sahmed95 has quit [Ping timeout: 246 seconds]
< naywhayare>
stephent1: non-exciting news: to properly provide an eigs_sym_pair() function so you can get a few eigenvectors/eigenvalues of the system Ax = Bx\lambda, I need solve(SpMat)
< naywhayare>
which is bad news, because it means I now need to spend some time wrapping SuperLU or something like that, but it's good news, because afterwards solve() and lu() become available for sparse objects
sahmed_95_ has joined #mlpack
stephent2 has joined #mlpack
awhitesong has joined #mlpack
stephent2 has quit [Ping timeout: 245 seconds]
< naywhayare>
awhitesong: you asked yesterday about the generalized eigenvalue solver
< naywhayare>
I'm in the process of wrapping ARPACK to do this for Armadillo, but it's somewhat complex, and I need to wrap SuperLU first so that I can have solve() and lu() for sparse matrices
< naywhayare>
you're welcome to help out, but I'm not actually sure what wrapping SuperLU will look like yet...
dhfromko_ has joined #mlpack
dhfromko_ has quit [Remote host closed the connection]
stephent2 has joined #mlpack
sumedhghaisas has joined #mlpack
dhfromko_ has joined #mlpack
curiousguy13 has quit [Ping timeout: 264 seconds]
sahmed_95_ has quit [Ping timeout: 265 seconds]
curiousguy13 has joined #mlpack
sumedhghaisas has quit [Ping timeout: 246 seconds]
< stephent2>
naywhayare: ill try to get to mlpack bugs thsi weekend
< stephent2>
sorry i've been busy lately
< awhitesong>
naywhayare: Hi. sure, will try to dive into SuperLU's source for lu decomposition function and see about wrapping it. Thanks.
stephent2 is now known as stephentu_
< stephentu_>
naywhayare: oh wow i'm glad I didn't attempt to do this
< stephentu_>
but its cool i think this would be a good gsoc project so we should keep it
< stephentu_>
unimplmeented for now
awhitesong has left #mlpack []
< naywhayare>
awhitesong:
< naywhayare>
ah, too slow
< stephentu_>
haha
< naywhayare>
stephentu_: wrapping SuperLU and ARPACK for Armadillo is going to be something that probably most GSoC students would find too difficult
< stephentu_>
we need like a mailing list
awhitesong has joined #mlpack
< stephentu_>
naywhayare: well definitely, i mean once you do it, the manifold project will be feasible :)
< naywhayare>
I hope to get it done by the start of GSoC, and I think that's possible
< naywhayare>
awhitesong: wrapping SuperLU is going to be somewhat complex. if I remember right, the really hard part is going to be shimming SuperLU correctly so that we can use all four variants (float, double, std::complex<float>, std::complex<double>)
< naywhayare>
I think that when you include SuperLU, you #include <superlu_<type>.h> where type is one of those four types above
< naywhayare>
and then that .h file typedefs some symbol which appears in each of those files... so I can't include all four at once
< stephentu_>
oh that sounds like some macro fun
< naywhayare>
if by "fun" you mean "nightmare disaster" :)
< stephentu_>
yes
< stephentu_>
QEMU does this all over the place
< stephentu_>
ifdef COND_!
< stephentu_>
#ifdef cond_1
< stephentu_>
#include "file".h
< stephentu_>
#undef cond_1
< naywhayare>
I actually met the SuperLU maintainer... she works at LBL, right up the hill from you. I think I should probably send a patch so that it doesn't do stupid crap like that
< stephentu_>
#ifdef cond_2
< stephentu_>
#include "file.h"
< stephentu_>
etc
< naywhayare>
ah, this is worse, something like this
< naywhayare>
in superlu_double.h (or whatever): "typedef double superlu_type;"
< naywhayare>
and so forth
< naywhayare>
so you can't actually include all four
< stephentu_>
oh god
< naywhayare>
meaning that what will probably have to be built is a wrapper in four separate compilation units that hides that type inside of only that compilation unit
< stephentu_>
is it written in pure C?
< stephentu_>
or did they just not learn about templates
< naywhayare>
I think so, yeah
< naywhayare>
if it is C++, then it's basically C with classes instead of structs, or something along those lines
< naywhayare>
I think the folks that wrote it are more comfortable in FORTRAN...
< stephentu_>
that is scary
< naywhayare>
yeah, which is why I'm glad the scariness will be hidden in Armadillo (which is, in its own way, scary) and not directly in mlpack...
< stephentu_>
ya i dont touch armadillo w/ a long stick
< naywhayare>
it's a very cool design internally, but it's a one-man show (and the maintainer likes it that way), so the internal code makes complete sense to him but isn't documented very well to the outside world
< naywhayare>
I wish that he would focus more on documenting the project and growing the developer community around it, but that project is very much his own personal project, so whatever, it's his choice. as long as he continues to maintain it, there's not too much compelling reason to switch away
< naywhayare>
considering how much work that would take...
< awhitesong>
naywhayare: ohk. i will try to see the code . If possible, we can also consider using Eigen c++ library, i guess its implementation is simpler. guess its using sparseLU class that is taking SparseMatrix<typename> in all the four formats.
dhfromko_ has quit [Remote host closed the connection]
< naywhayare>
awhitesong: unfortunately I don't think the Armadillo maintainer would be okay with wrapping Eigen, but I think maybe we can use the same techniques they are for wrapping
< naywhayare>
I think they use SuperLU internally
dhfromko_ has joined #mlpack
< naywhayare>
the first problem to deal with will be how to provide functions that call the correct SuperLU overload depending on the type of the matrix; I hope to be able to take a little time tomorrow to try some things out
dhfromko_ has quit [Remote host closed the connection]
< awhitesong>
ok.i will try to go through its code. Also will try to see if and how eigen is using superLU. Thanks.
awhitesong has left #mlpack []
stephentu_ has quit [Ping timeout: 250 seconds]
dhfromkorea has joined #mlpack
dhfromkorea has quit [Read error: Connection reset by peer]
dhfromkorea has joined #mlpack
dhfromkorea has quit [Remote host closed the connection]