--- Log opened Tue May 17 00:00:07 2011 | ||
-!- blackburn [~qdrgsm@188.122.225.203] has quit [Read error: Connection reset by peer] | 00:24 | |
-!- alesis-novik [~alesis@188.74.87.84] has joined #shogun | 06:19 | |
-!- blackburn [~qdrgsm@188.168.5.178] has joined #shogun | 06:59 | |
blackburn | alesis-novik: how are your exams? | 07:01 |
---|---|---|
alesis-novik | 2 left. Done this Saturday, then the fun starts! | 07:15 |
blackburn | nice | 07:27 |
-!- blackburn [~qdrgsm@188.168.5.178] has quit [Quit: Leaving.] | 10:35 | |
-!- sploving [~root@124.16.139.196] has joined #shogun | 10:40 | |
sploving | anybody aroud? | 10:40 |
-!- meerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has joined #shogun | 10:56 | |
-!- ameerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has quit [Ping timeout: 252 seconds] | 10:59 | |
-!- meerkat is now known as ameerkat | 10:59 | |
sploving | sonney 2k, great news. The vector typemap works well! | 11:28 |
sploving | I will push it | 11:29 |
-!- ameerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has quit [Ping timeout: 246 seconds] | 13:45 | |
-!- heiko [~heiko@infole-06.uni-duisburg.de] has joined #shogun | 14:29 | |
-!- blackburn [~qdrgsm@188.168.4.130] has joined #shogun | 14:38 | |
sploving | hello blackburn | 14:41 |
@bettyboo | you here? | 14:41 |
sploving | glad to see you | 14:41 |
sploving | bettyboo, yes | 14:41 |
@bettyboo | sploving: yes? how that? | 14:41 |
blackburn | sploving: hi | 14:42 |
blackburn | almost full gsoc roster here ;) | 14:42 |
sploving | blackburn, I am writing the matrix typemap for java | 14:42 |
blackburn | how it's going? | 14:43 |
sploving | It compile well. do you know which function call the matrix. I would write a example to test it | 14:43 |
blackburn | what you mean saying call? | 14:43 |
blackburn | as arg? | 14:44 |
sploving | for vector example, it like this:Labels x = new Labels(); | 14:44 |
sploving | double y[] = {1, 2, 3, 4}; | 14:44 |
sploving | x.set_labels(y); | 14:44 |
sploving | double z[] = x.get_labels(); | 14:44 |
sploving | yeap I mean arg | 14:44 |
blackburn | may be get_feature_matrix from Features | 14:44 |
blackburn | e.g. Features feats = new Features(some_matrix) | 14:45 |
sploving | Ok thx. I will have a try | 14:45 |
sploving | how is going for your work? | 14:45 |
blackburn | I have almost done LLE, cleaning, testing and etc to go | 14:46 |
sploving | great work | 14:46 |
blackburn | maybe 15-20% of all work | 14:46 |
blackburn | I just going to get some things to work before my exams | 14:46 |
sploving | good plan | 14:46 |
blackburn | sploving: have you passed your exams already? | 14:49 |
sploving | wait for a moment. I am trying | 14:49 |
sploving | blackburn, double x[][] = {{1, 2},{3, 4}, {5, 6}}; Features x = new Features(x); | 14:50 |
sploving | you mean this? | 14:50 |
blackburn | yeap, sth like this | 14:51 |
sploving | how to get the matrix from x | 14:51 |
blackburn | x.get_feature_matrix(); | 14:51 |
blackburn | ah no | 14:52 |
blackburn | you should use some more specified class | 14:52 |
sploving | I know you mean. just like Label | 14:53 |
sploving | So many classes. I need some time to find a suitable one | 14:53 |
sploving | if sonney2k is here, he could give me quick help | 14:53 |
blackburn | in python I'd suggest to use RealFeatures (shortname for CSimpleFeatures<float64_t>) | 14:54 |
sploving | blcakburn, I wil take a look at it | 14:56 |
blackburn | I don't sure how to instance template class in your java interface, but it is the proper class for this test | 15:00 |
sploving | blackburn, I found in SGBase.i, a lot of template typemap were added | 15:03 |
sploving | why add these? before add this, my vector could work well. but after added it, I could not compile my example well | 15:04 |
@sonney2k | sploving, I will have some time this evening | 15:08 |
@sonney2k | not now | 15:08 |
sploving | sonney, OK | 15:09 |
-!- heiko [~heiko@infole-06.uni-duisburg.de] has quit [Ping timeout: 258 seconds] | 15:50 | |
-!- heiko [~heiko@infole-06.uni-duisburg.de] has joined #shogun | 17:18 | |
-!- sudarshan [~sudarshan@c-174-50-225-239.hsd1.tn.comcast.net] has joined #shogun | 17:46 | |
sudarshan | hi all | 17:46 |
-!- sudarshan [~sudarshan@c-174-50-225-239.hsd1.tn.comcast.net] has quit [Remote host closed the connection] | 18:06 | |
blackburn | sonney2k: what do you think about ScaLAPACK? | 18:42 |
blackburn | this night will try to test if atlas will be better for particular sgvectors and sgmatrices then adding, multiplying, etc | 18:47 |
blackburn | I'd really like if we have very efficient techniques encapsulated there | 18:48 |
-!- ameerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has joined #shogun | 19:34 | |
-!- heiko [~heiko@infole-06.uni-duisburg.de] has left #shogun [] | 19:35 | |
@sonney2k | blackburn, scalapack is the parallel one? | 20:29 |
@sonney2k | IIRC these are all transparent | 20:29 |
@sonney2k | so it just depends on what you link against to | 20:29 |
blackburn | sonney2k: yeap, parallel | 20:29 |
@sonney2k | gotoblas is also *very* fast | 20:29 |
@sonney2k | sploving, I am around now - you too? | 20:30 |
blackburn | gotoblas? | 20:30 |
blackburn | sonney2k: would you mind using it? | 20:30 |
blackburn | it became my idea fix :D | 20:30 |
@bettyboo | ha ha | 20:31 |
@sonney2k | blackburn, you can just link shogun against it - these are all blas/lapack wrappers... | 20:32 |
blackburn | ah, I see | 20:32 |
blackburn | sonney2k: so it is just the matter of specified library, right? | 20:34 |
blackburn | so will it be as fast as one's setup for shogun? | 20:34 |
@sonney2k | yes | 20:34 |
blackburn | sonney2k: what do you prefer for usage in SGVector, etc? | 20:37 |
blackburn | will blas be better for it? | 20:37 |
blackburn | sonney2k: ah, another one question, what is Array, Array2 and Array3? | 20:39 |
blackburn | never have seen it used anywhere.. | 20:39 |
@sonney2k | blackburn, it is a 1d 2d 3d array | 20:40 |
blackburn | hehe | 20:40 |
@sonney2k | I never used it though | 20:40 |
@sonney2k | but it is probably sth one should keep in mind to merge it with SGVector etc on the long run | 20:41 |
blackburn | sonney2k: why? does it have some good ideas to implement in SGVector? | 20:42 |
@sonney2k | blackburn, I mean a 1d array is a vector rigth? | 20:43 |
blackburn | exactly, but do we need Array.h if we have SGVector?.. | 20:43 |
blackburn | sonney2k: can you suggest me a way to test the performance of SGVectors addition.. | 20:45 |
@sonney2k | blackburn, create a few thousand vectors and measure add times | 20:46 |
blackburn | sonney2k: where? | 20:46 |
blackburn | 'technically' | 20:46 |
@sonney2k | I would even create it as a method of SGVector | 20:47 |
blackburn | what is I should call it from? | 20:48 |
blackburn | will python_modular interface calling be fair? | 20:48 |
@sonney2k | from whatever language you like then ... | 20:48 |
@sonney2k | sure | 20:48 |
blackburn | the only problem - how can I instance sgvector in python | 20:49 |
blackburn | seems it isn't described in .i? | 20:49 |
@sonney2k | x=SGRealVector() | 20:49 |
blackburn | ah. sorry, my fault | 20:49 |
blackburn | sonney2k: are you sure? there is no SGRealVector in Library.i | 20:51 |
blackburn | ah | 20:51 |
blackburn | SGBase | 20:51 |
@sonney2k | blackburn, it is in features | 20:51 |
@sonney2k | let me check | 20:52 |
blackburn | no it isn't | 20:53 |
blackburn | in SGBAse | 20:53 |
@sonney2k | blackburn, err RealVector() | 20:53 |
blackburn | ah | 20:54 |
blackburn | seems Features.i is related to SGBase.i | 20:54 |
@sonney2k | blackburn, any module will include SGBase | 20:55 |
blackburn | okay | 20:55 |
blackburn | sonney2k: how are Jan? :) | 20:59 |
@sonney2k | blackburn, growing gained some weight at least but otherwise sleeping most of the time | 21:00 |
blackburn | sonney2k: what was his weight 'initially'? | 21:01 |
@sonney2k | 3585g now it is >3700 | 21:01 |
blackburn | hehe | 21:02 |
blackburn | sonney2k: i'd say he is богатырь :) | 21:03 |
@bettyboo | ;> | 21:03 |
blackburn | oh damn allergy | 21:08 |
CIA-31 | shogun: Heiko Strathmann master * r8d63904 / src/python_modular/swig_typemaps.i : Merge remote branch 'upstream/master' - http://bit.ly/iOjSLk | 21:18 |
CIA-31 | shogun: Heiko Strathmann master * r19e72c1 / (2 files): renamed new SGVector/SGMatirx based add_matrix(...) and add_vector(...) methods to add(...) - http://bit.ly/lpgA6f | 21:18 |
@sonney2k | blackburn, what kind of allergy? | 21:24 |
@sonney2k | sploving, if vectors work - for matrices you should ask mikiobraun again | 21:24 |
@sonney2k | sploving, supporting jblas or ujmp later on | 21:25 |
blackburn | sonney2k: related to flowers, etc I think, sneezing and so on | 21:25 |
@sonney2k | blackburn, you should aim for working in some clean room environment after your studies :D | 21:25 |
@bettyboo | *g* | 21:25 |
blackburn | sonney2k: I have it as long as I remember.. :) | 21:26 |
@bettyboo | not bad | 21:26 |
blackburn | bettyboo: but it is bad for me! | 21:28 |
@sonney2k | or not so good | 21:28 |
@bettyboo | blackburn: but we don't need save / load right? just for setting / getting | 21:28 |
@sonney2k | heh | 21:28 |
@sonney2k | betty is always right | 21:28 |
@bettyboo | sonney2k: it is compiling still.. | 21:28 |
blackburn | bettyboo: are you right? | 21:29 |
@bettyboo | blackburn: which is what you care about, right? | 21:29 |
@bettyboo | never mind | 21:29 |
@sonney2k | nice - sploving managed to get the java in / out typemaps to work for vectors | 21:44 |
@sonney2k | so the SGVector stufff is now working for python and java already for vectors at least | 21:45 |
blackburn | yeap | 21:47 |
@sonney2k | all that is mising are lists of strings / sparse matrices / matrices | 21:48 |
@sonney2k | of arbitrary kind though | 21:48 |
@sonney2k | but yeah it will be more difficult to use ujmp / jblas for the matrix stuff later | 21:49 |
CIA-31 | shogun: Baozeng Ding master * r9881d16 / src/java_modular/swig_typemaps.i : add SGMatrix support in java typemaps - http://bit.ly/ihL9lh | 21:50 |
CIA-31 | shogun: Soeren Sonnenburg master * r5145468 / (2 files in 2 dirs): Merge branch 'master' of git://github.com/sploving/shogun - http://bit.ly/m7vByF | 21:50 |
CIA-31 | shogun: Soeren Sonnenburg master * rdffdda0 / src/java_modular/swig_typemaps.i : | 21:50 |
CIA-31 | shogun: disable matrix typemaps for now - these should be based on jblas/ujmp | 21:50 |
CIA-31 | shogun: later on. - http://bit.ly/mBGveE | 21:50 |
@sonney2k | blackburn, have you seen this http://www.h-online.com/open/news/item/Linux-PC-in-a-browser-1244723.html | 21:51 |
@sonney2k | linux running in .js in a browser? | 21:51 |
@sonney2k | pretty impressive... | 21:52 |
blackburn | sonney2k: yeap, it is gorgeous | 21:57 |
-!- alesis-novik [~alesis@188.74.87.84] has quit [Quit: I'll be Bach] | 22:15 | |
* blackburn is late night plumber | 23:07 | |
--- Log closed Wed May 18 00:00:09 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!