Open in new window / Try shogun cloud
--- Log opened Wed Jun 26 00:00:05 2013
-!- lisitsyn [~lisitsyn@46.20.65.164] has quit [Ping timeout: 248 seconds]00:56
shogun-notifier-shogun: iglesias :develop * ad6fed6 / src/shogun/distance/ (3 files): https://github.com/shogun-toolbox/shogun/commit/ad6fed6bb703099f6a8b845c7cbde2bf34eb87d601:03
shogun-notifier-shogun: Add CustomMahalanobisDistance01:03
shogun-notifier-shogun: iglesias :develop * ccf16f7 / tests/unit/distance/Distance_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/ccf16f7629f1f7107658bdc27d773466ab2939d401:03
shogun-notifier-shogun: Very simple unit test for CustomMahalanobisDistance01:03
shogun-notifier-shogun: iglesias :develop * 70d9fca / src/shogun/metric/LMNN.cpp,src/shogun/metric/LMNN.h: https://github.com/shogun-toolbox/shogun/commit/70d9fca923a8f5822f00de50500af98d324f3d7401:03
shogun-notifier-shogun: [WIP] Add LMNN without train method01:03
shogun-notifier-shogun: Fernando Iglesias :develop * 384185f / / (6 files): https://github.com/shogun-toolbox/shogun/commit/384185f9c1aeb1f8213e63173f9407e27ea1099d01:03
shogun-notifier-shogun: Merge pull request #1188 from iglesias/feature/metric01:03
shogun-notifier-shogun:01:03
shogun-notifier-shogun: Feature/metric01:03
shogun-buildbotbuild #974 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/974  blamelist: iglesias <fernando.iglesiasg@gmail.com>01:10
shogun-buildbotbuild #975 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/975  blamelist: iglesias <fernando.iglesiasg@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>01:12
shogun-buildbotbuild #1142 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1142  blamelist: iglesias <fernando.iglesiasg@gmail.com>01:18
shogun-buildbotbuild #976 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/976  blamelist: iglesias <fernando.iglesiasg@gmail.com>01:19
shogun-buildbotbuild #1143 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1143  blamelist: iglesias <fernando.iglesiasg@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>01:26
-!- travis-ci [~travis-ci@ec2-54-226-51-241.compute-1.amazonaws.com] has joined #shogun01:31
travis-ci[travis-ci] it's Fernando Iglesias's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/844440401:31
-!- travis-ci [~travis-ci@ec2-54-226-51-241.compute-1.amazonaws.com] has left #shogun []01:31
shogun-buildbotbuild #1144 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1144  blamelist: iglesias <fernando.iglesiasg@gmail.com>01:34
-!- FSCV [~FSCV@187.210.54.166] has quit [Quit: Leaving]01:36
-!- nube [~rho@49.244.126.8] has quit [Ping timeout: 240 seconds]01:52
shogun-buildbotbuild #1265 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1265  blamelist: iglesias <fernando.iglesiasg@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>02:01
-!- nube [~rho@49.244.118.165] has joined #shogun02:05
shogun-buildbotbuild #1266 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/126602:14
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun02:28
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]02:59
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]04:03
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving]04:17
-!- nube [~rho@49.244.118.165] has quit [Ping timeout: 260 seconds]04:21
-!- hushell [~hushell@c-67-160-139-116.hsd1.or.comcast.net] has joined #shogun04:32
-!- hushell [~hushell@c-67-160-139-116.hsd1.or.comcast.net] has quit [Ping timeout: 240 seconds]05:46
-!- hushell [~hushell@c-67-160-139-116.hsd1.or.comcast.net] has joined #shogun05:46
-!- nube [~rho@116.90.239.3] has joined #shogun06:32
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 252 seconds]07:06
-!- nube [~rho@116.90.239.3] has joined #shogun07:20
-!- gsomix_ [~Miranda@46.20.65.164] has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]07:52
-!- lisitsyn [~lisitsyn@46.20.65.164] has joined #shogun08:18
-!- lisitsyn [~lisitsyn@46.20.65.164] has quit [Quit: Leaving.]08:32
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun09:11
-!- mode/#shogun [+o iglesiasg] by ChanServ09:11
-!- nube1 [~rho@116.90.239.13] has joined #shogun09:13
-!- vgorbati [c3ee5cb1@gateway/web/freenode/ip.195.238.92.177] has joined #shogun09:13
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 276 seconds]09:13
-!- nube [~rho@116.90.239.13] has joined #shogun09:16
-!- nube1 [~rho@116.90.239.13] has quit [Read error: No route to host]09:16
-!- nube [~rho@116.90.239.13] has quit [Quit: Leaving.]09:25
-!- nube [~rho@116.90.239.13] has joined #shogun09:25
-!- nube [~rho@116.90.239.13] has quit [Quit: Leaving.]09:35
-!- nube [~rho@116.90.239.13] has joined #shogun09:36
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun09:37
thoralfHello *.*09:58
lisitsynthoralf: hey09:59
-!- lambday [67157e4d@gateway/web/cgi-irc/kiwiirc.com/ip.103.21.126.77] has joined #shogun10:14
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]10:15
-!- hushell [~hushell@c-67-160-139-116.hsd1.or.comcast.net] has quit [Quit: WeeChat 0.3.7]10:15
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun10:18
-!- mode/#shogun [+o iglesiasg] by ChanServ10:18
-!- nube [~rho@116.90.239.13] has quit [Remote host closed the connection]10:43
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun11:04
van51hello11:04
van51sonne|work: about your comment on the Tokenizer classes. I get what you;re saying, but I wanted to be able to use a custom Tokenizer and an n-gram Tokenizer interchangeably in the same workflow11:10
sonne|workyes but that would work right?11:11
sonne|workNGramTokenizer & DelimiterTokenizer?11:11
sonne|workvan51: ^11:12
van51sonne|work: yeah it would work I believe11:13
van51sonne|work: although now I'm confused as to whether you meant to maintain just one class of Tokenizers (what I understood) or to have the WhiteSpaceTokenizer I made into a DelimiterTokenizer11:13
sonne|workyes whitespacetokenizer -> delimitertokenizer11:14
van51sonne|work: ok then :)11:15
van51going to make some coffee and start coding :D11:15
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving]11:19
sonne|workgo van51 go!11:40
van51heh I'm on it ;)11:46
-!- nube [~rho@116.90.239.13] has joined #shogun11:55
-!- nube [~rho@116.90.239.13] has quit [Quit: Leaving.]12:10
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun12:10
shogun-notifier-shogun: lambday :develop * 5bcdd54 / / (25 files): https://github.com/shogun-toolbox/shogun/commit/5bcdd547fab1c0ba3cad5d915d9b67a27e62f29c12:10
shogun-notifier-shogun: params registered, sample modified (log-det), gitignore modified12:10
shogun-notifier-shogun: Heiko Strathmann :develop * 2aa7804 / / (25 files): https://github.com/shogun-toolbox/shogun/commit/2aa78041e8717f39419c83f6ad8a52b50c0bd92812:10
shogun-notifier-shogun: Merge pull request #1189 from lambday/feature/log_determinant12:10
shogun-notifier-shogun:12:10
shogun-notifier-shogun: params registered, sample modified (log-det), gitignore modified12:10
shogun-buildbotbuild #977 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/977  blamelist: lambday <heavensdevil6909@gmail.com>12:19
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has quit [Quit: Leaving.]12:23
shogun-buildbotbuild #1145 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1145  blamelist: lambday <heavensdevil6909@gmail.com>12:27
-!- HeikoS [~heiko@nat-163-105.internal.eduroam.ucl.ac.uk] has joined #shogun12:30
-!- mode/#shogun [+o HeikoS] by ChanServ12:30
lisitsynHeikoS is back!12:35
@HeikoSlisitsyn: yes! :)12:35
shogun-buildbotbuild #1146 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1146  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>12:37
@HeikoSserialization borken again :)12:37
-!- travis-ci [~travis-ci@ec2-23-22-249-119.compute-1.amazonaws.com] has joined #shogun12:39
travis-ci[travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/845860612:39
-!- travis-ci [~travis-ci@ec2-23-22-249-119.compute-1.amazonaws.com] has left #shogun []12:39
shogun-buildbotbuild #978 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/978  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>12:41
-!- nube [~rho@49.244.79.97] has joined #shogun12:54
lambdayHeikoS: heyyyy! :D13:02
@HeikoSlambday: heya! :)13:03
lambdayhow are you??13:03
@HeikoSlambday: pretty good :)13:03
@HeikoSand yourself?13:03
lambdayam good too :)13:03
lambdayand the trip?13:03
lambdayhow many kms did you travel total?13:03
@HeikoSlambday: was very nice13:04
@HeikoSwe did about 75km a day on average13:05
@HeikoS5 days13:05
lambdaygreat!!! :-o13:05
lambdayin england?13:05
@HeikoSno Germany13:05
@HeikoS"M?nsterland"13:05
@HeikoSnice area13:05
@HeikoSlots of lakes, rivers and woods13:05
@HeikoSand it rained every day13:05
lambdayah..classic! may be we'll get to see some photos :)13:06
lambdaylol @10^6 emails :P13:06
@HeikoSlambday: no photos made13:07
@HeikoSjust pure cycling13:07
@HeikoSlambday: yeah emails are quite a bit13:07
lambdayhmm...13:07
@HeikoSspending the early morning on working thorugh them :)13:07
@HeikoSliked your patches13:07
@HeikoSlooks like the general framework is well set up13:07
lambdaythanks... by the way, I thought of making it uint64_t (m_dimension) but then changed it, what do you think?13:07
lambdayyeah!13:07
lambdaycan't wait to see it working for approximation case13:08
@HeikoSlambday: I dont know, maybe leave it for now13:08
lambdayalright13:08
@HeikoS2^31 -1 is quite a bit already13:08
lambdayhmm13:08
lambdayI coded up CLinearSolver and CDirectLinearSolverComplex...13:08
lambdayshall I send the PR now?13:09
@HeikoSlambday: nice, yeah show me what you got13:09
lambdayokay... sending13:09
lambdayHeikoS: sent13:11
@HeikoSlambday: what solver is this?13:12
lambdayllt13:12
@HeikoSllt13:12
@HeikoSso only PSD matrices13:12
lambdayyes13:12
@HeikoSwe should somehow reflect this in the same13:12
@HeikoSor even a different approiach:13:12
@HeikoSthere should be an enum to select how to solve the system13:12
lambdayand we initialize it in the constructor13:13
@HeikoSI suggest: Cholesky, QR-no permutation, QR-col-permutation, QR-full-permutation, and SVD13:13
@HeikoSits only for dense matrices right?13:13
lambdayyes13:13
@HeikoShttp://eigen.tuxfamily.org/dox/TopicLinearAlgebraDecompositions.html13:14
lambdayokay..13:14
@HeikoSlambday: sorry to make this more complicated13:14
@HeikoSbut then we are on the safe side13:14
@HeikoSSVD being ultra stable13:14
@HeikoSQR being tradeoff between fast and accurate13:15
@HeikoSand for PSD LLT is both13:15
lambdayHeikoS: yeah... okay.. so, solve will have switch cases, and depending on some enum EDirectSolverType, it will use corresponding solvers13:16
@HeikoSlambday: yeah, that should be minor code changes since its all in eigen313:16
@HeikoSjust the enum13:16
@HeikoSand init it to LLT13:16
@HeikoScholesky13:16
lambdayHeikoS: alright..13:17
@HeikoSlambday: oh and make sure you use the rank-revealing features of eigen13:17
@HeikoSe.g. llt fails if matrix is not psd13:17
@HeikoSlambday: gotta make a phone call, be back in a few minutes13:17
lambdayHeikoS: okay, I'm checking13:18
thoralfHeikoS: Is your eclipse confused by "#ifndef _XXX__H__" and greys the whole class?13:18
@HeikoSthoralf: which file?13:18
thoralfHeikoS: _BINARY_LABELS__H__13:18
thoralfHeikoS: BinaryLabels.h13:18
@HeikoSthoralf: checking ...13:18
thoralfHeikoS: But it also affects other files.13:18
@HeikoSthoralf: yes13:19
thoralfHeikoS: Adding another underscore to the define fixes this, but I don't know why.13:19
@HeikoSthoralf: wow I always completely ignored/not noticed that :)13:19
thoralfHeikoS: This also affects auto completion.13:19
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun13:19
@HeikoSthoralf: there should be no underscore at the beginning of defines anyway13:19
@HeikoSc++ style standard13:20
@HeikoSremoving also makes it non-gray13:20
@HeikoSthoralf: how is eclipse going apart from that?13:20
thoralfHeikoS: Sorry, don't understand the question.13:20
lisitsynHeikoS: what's the standard stating that ;)13:21
thoralfhttps://bugs.eclipse.org/bugs/show_bug.cgi?id=31393313:22
thoralfSeems related.13:23
sonne|workHeikoS, lisitsyn can we finalize the workshop program this evening?13:24
lisitsynsonne|work: I am available13:24
sonne|workif you can please already prepare the schedule13:25
sonne|worklisitsyn: ^13:25
shogun-buildbotbuild #1268 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1268  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>13:26
lisitsynsonne|work: I am a bit lost - any template?13:27
sonne|worklisitsyn: just when who gives which talk13:27
-!- Rasmi [~Rasmi@117.197.255.93] has joined #shogun13:29
Rasmiwhile compiling one example ie. classifier_minimal_svm.cpp it shows error ie. undefined reference to SGMatrix<float 64_t>.... etc13:30
-!- Rasmi [~Rasmi@117.197.255.93] has quit [Ping timeout: 246 seconds]13:39
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun13:43
-!- mode/#shogun [+o iglesiasg] by ChanServ13:43
-!- Rasmi [~Rasmi@117.197.251.58] has joined #shogun14:02
-!- gsomix [~Miranda@37.61.181.106] has joined #shogun14:05
-!- Rasmi [~Rasmi@117.197.251.58] has quit [Quit: irc2go]14:11
-!- foulwall` [~user@2001:da8:215:6100:2c22:84c9:e288:c210] has joined #shogun14:11
-!- foulwall` [~user@2001:da8:215:6100:2c22:84c9:e288:c210] has left #shogun ["ERC Version 5.3 (IRC client for Emacs)"]14:25
@HeikoSlambday, thoralf, lisitsyn back14:27
lambdayHeikoS: wb14:27
@HeikoSsonne|work: yes can do, which time?14:27
@HeikoSlambday: how are the solvers going?14:28
sonne|workHeikoS: evening 20hrs14:28
lambdayHeikoS: was away, just checking.. will update the PR soon :)14:28
@HeikoSsonne|work: thats 19h here, fine for me, although earlier would be better14:29
@HeikoSsonne|work: BTW i will be in Berlin from Wednesday, so if there is any need of organising things locally before Friday, I could help14:29
sonne|workHeikoS: ahh btw we have a room at the TU for the weekend!14:30
@HeikoSsonne|work: nice!14:31
-!- foulwall [~user@2001:da8:215:6100:2c22:84c9:e288:c210] has joined #shogun14:39
-!- zxtx [~zv@cpe-66-68-190-37.austin.res.rr.com] has quit [Ping timeout: 248 seconds]14:42
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has quit [Quit: Leaving.]14:49
-!- foulwall` [~user@2001:da8:215:6100:2c22:84c9:e288:c210] has joined #shogun14:51
-!- foulwall` [~user@2001:da8:215:6100:2c22:84c9:e288:c210] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]14:51
-!- nube [~rho@49.244.79.97] has quit [Ping timeout: 252 seconds]14:52
thoralfI don't understand MulticlassMachine::get_submachine_outputs() - how is it possible to call (CMachine*)machine->apply_binary() without arguments?14:54
thoralfI can only find implementations with signature "CBinaryLabels* apply_binary(CFeatures *)"14:55
thoralfAm I missing something?14:55
@iglesiasgthoralf, a machine has a CFeatures* member15:06
@iglesiasgif apply is called without arguments, then these ones are used15:06
@iglesiasgcheck the signature in the headers, they should be with a default argument15:07
@iglesiasgsomething like  apply_binary(CFeatures* features = NULL)15:07
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]15:10
thoralfiglesiasg: Yes, thanks.  The "magic" happens in apply_get_outputs()15:39
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun15:56
-!- zxtx [~zv@12.231.120.253] has joined #shogun16:04
pickle27hey everyone16:07
-!- zxtx [~zv@12.231.120.253] has quit [Ping timeout: 240 seconds]16:08
pickle27guys Im have a heck of time trying to read a csv matrix into an SGMatrix properly16:08
pickle27I keep getting some really weird orderings in the variable16:08
pickle27Im using CAsciiFile to load the matrix16:09
pickle27It loads my matrix as a transpose which makes sense according to the documentation but for some reason I can't untranspose it properly16:10
-!- nube [~rho@49.244.65.164] has joined #shogun16:16
-!- zxtx [~zv@12.231.120.253] has joined #shogun16:16
pickle27I've got it working fine now but it seems like their might be a better way16:20
-!- zxtx [~zv@12.231.120.253] has quit [Ping timeout: 256 seconds]16:20
pickle27for the record here is my solution http://pastebin.com/hNNNf4nc16:21
pickle27I'm loading a csv file created from R, its a 10x300 csv and I need to load it and re shape it into a 10x10x30 ndarray16:21
lambdayHeikoS: there??16:22
pickle27if anyone has an idea for a better solution I'd love to see it16:22
@HeikoSlambday: yep, hi!16:22
lambdayHeikoS: except FullPivHouseholderQR, everything else seems to fail for even a small matrix for complex16:23
lambdaythe matrix I am testing with is16:23
lambday2, 1+2i16:23
lisitsynpickle27: so you are loading 3d matrix?16:23
lambday1+2i, 316:23
-!- zxtx [~zv@12.231.120.253] has joined #shogun16:23
lambdayb=[1,1]16:23
@HeikoSlambday: do real matrices work?16:23
pickle27lisitsyn: R flattens it for you though16:23
@HeikoSdid you check the condition number?16:23
pickle27lisitsyn: so I need to re form it in 3D16:23
@HeikoSoh well thats small16:24
lambday1.770516:24
@HeikoSI mean small matrix16:24
@HeikoSreal matrices?16:24
lambdayfor real, llt works16:24
lambday(I mean, just keeping the complex part 0)16:24
lambdayrest, I wanted it to work with complex16:24
lambdayoh, and we'll have complex in the diagonal16:24
@HeikoSis there some documentation for eigen's complex solvers?16:24
lambdayso no use of using LLT at all16:25
lambdayHeikoS: not that I ran into.. specific complex cases are there for eigensolvers16:25
lambdaym\b gives16:26
lambday0.26804 - 0.10309i16:26
lambday   0.17526 - 0.14433i16:26
@HeikoSoctave I assume?16:26
lambdayyep16:26
lambdayI didn't make this class template cause we'll only use it for complex purpose16:27
lambdayis that okay?16:27
pickle27I realize that CAsciiFile was done writing a column to a row for efficiency but its sometimes kind of inconvenient no?16:28
@HeikoSlambday: dont know16:29
@HeikoSI think yes16:29
@HeikoSthere is a talk here now, will be back in 1 hr, sorry16:29
lambdayHeikoS: no problem :)16:29
@HeikoSlambday: I was wrong, no talk now, but in 30 mins16:31
lambdayHeikoS: okay :)16:31
lambdayso what shall we do? :-/16:31
@HeikoSwhat does eigen say?16:32
lambdayI didn't see anything specific for complex for these16:33
-!- FSCV [~FSCV@50.7.50.60] has joined #shogun16:33
lambdayI'm checking16:33
-!- foulwall [~user@2001:da8:215:6100:2c22:84c9:e288:c210] has quit [Ping timeout: 245 seconds]16:33
-!- zxtx [~zv@12.231.120.253] has quit [Ping timeout: 276 seconds]16:35
-!- thoralf [~thoralf@enki.zib.de] has quit [Quit: Konversation terminated!]16:36
lambdayHeikoS: theoretically, non-hermitan matrices shouldn't be used with LLT, right? cholesky assumes it to be hermitan spd16:38
lambdayhermitian16:38
@HeikoSlambday: yep16:38
lambdayso, we'll have complex shifts... which will be in the diagonal16:38
lambdaythat will make it non-hermitian16:39
lambday:-/16:39
@HeikoSlambday: yep, but so what?16:40
@HeikoSlambday: wait16:40
lambdayokay16:40
@HeikoSI solved them with Cholesky sometimes IIRC16:40
@HeikoSlambday: no my mind tricks me16:41
@HeikoScannot be16:41
lambdayI was trying on octave...16:41
lambdayn =16:41
lambday   5 + 2i   1 + 2i16:41
lambday   1 + 2i   3 + 4i16:41
lambdayn =16:41
lambday   5 + 2i   1 + 2i16:41
lambday   1 + 2i   3 + 4i16:41
lambdayoops! sorry16:42
lambdayhol(n,'lower')*chol(n)16:42
lambdayans =16:42
lambday   5.0000 + 0.0000i   1.0000 + 2.0000i16:42
lambday   1.0000 + 2.0000i   1.4000 + 0.8000i16:42
lambdayargh!16:42
@HeikoSah oray, I was using backslash sorry16:42
lambdayhmmm16:43
@HeikoSjust wondering what matlab uses with backslash16:43
@HeikoSI think Gaussian elemination16:43
lambdayhmmm16:44
-!- gsomix [~Miranda@37.61.181.106] has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]16:45
@HeikoSlambday: the direct solvers are more or less only for debuggin right?16:46
lambdayumm... yeah16:46
-!- gsomix [~Miranda@37.61.181.106] has joined #shogun16:46
lambdaysort of...16:46
lambdayI mean, they will almost never be used16:46
lambdaywe'll use preconditioned CG, so16:47
gsomixgood evening?16:47
@HeikoSlambday: then just go for svd16:47
lambdaythat doesn't work :(16:47
lambdayQR-full does16:47
@HeikoSdo it via the enum (so it can be extended, maybe well add QR)16:48
@HeikoSwhat does svd tell you?16:48
lambdaysvd gives an answer though..16:48
lambdaybut that's not correct16:48
@HeikoSlambday: which answer?16:48
lambdaysolution vec16:48
@HeikoSand that is wrong???16:48
lambday(Ax-b).norm/b.norm = 0.7 something16:49
lambdayJacobiSVD16:49
@HeikoSwhich dimension?16:49
@HeikoSwow16:49
@HeikoSthat sucks16:49
lambday2x216:49
@HeikoSno that cannot be16:49
@HeikoSmaybe an issue with complex number operators?16:50
lambdayours16:50
lambday?16:50
@HeikoSyes, how does eigen store complex matrices?16:50
lambdayusing std::complex only16:50
lambdayand full-perm QR works just fine16:50
@HeikoSlambday: its weird, SVD has the best accuracy of all algorithms16:50
@HeikoSoptimal one16:51
@HeikoSwhat does the full QR give?16:51
lambdaymay be the complex version isn't working yet?16:51
lambdayumm16:51
lambdayrel error 0.016:51
lambdayexactly what I get from octave16:51
lambday(0.268041,-0.103093)16:51
lambday(0.175258,-0.14433)16:51
@HeikoShttp://eigen.tuxfamily.org/dox/Eigen2ToEigen3.html16:52
@HeikoSmaybe this?16:52
lambday0.26804 - 0.10309i16:52
lambday   0.17526 - 0.14433i16:52
lambdaychekcing16:52
@HeikoSlambday: do you use solve of the SVD class or do you do that by hand?16:53
@HeikoSbased on the factors?16:53
lambdaysolve of the class..16:54
lambdayI am using JacobiSVD only, that's in Eigen316:54
@HeikoSlambday: ok, since its not really important, just use the full QR then16:55
lambdaydrop all the rest?16:55
@HeikoSbut with an enum that can be changed and switch in the solve method16:55
lambdayalright16:55
@HeikoSleave them in but add a warning or so16:55
@HeikoSwe want this to work, but its not important now16:55
lambdayokay... so removing these testcases except full QR..16:56
@HeikoSlambday: make sure to try some other (larger) matrices16:56
lambdayokay16:56
lambdayI'm checking16:56
@HeikoSlambday: there must be an error somewhere since the SVD in eigen is stable16:57
@HeikoSlambday: maybe produce a small reproducable example and post it to the eigen3 forum16:57
lambdayHeikoS: okay..16:58
lisitsynHeikoS: btw eigen guys are sooo unhappy with the current svd implementation16:58
@HeikoSdevelopers or users?16:59
-!- vgorbati [c3ee5cb1@gateway/web/freenode/ip.195.238.92.177] has quit [Quit: Page closed]17:02
-!- foulwall [~user@2001:da8:215:c252:5940:b422:1b94:2c7] has joined #shogun17:06
lisitsynHeikoS: devs17:06
@HeikoSlisitsyn: did not know that17:08
@HeikoSlisitsyn: well QR is also nice :17:08
@HeikoS :)17:08
-!- gsomix [~Miranda@37.61.181.106] has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]17:09
lisitsynHeikoS: some students working on divide and conquer SVD17:09
lisitsynso they are expecting to improve some day17:09
@HeikoSlisitsyn: nice!17:09
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Ping timeout: 252 seconds]17:14
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving]17:20
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.]17:34
-!- foulwall [~user@2001:da8:215:c252:5940:b422:1b94:2c7] has quit [Ping timeout: 264 seconds]17:36
lambdayHeikoS: something is weird.. I tested again and SVD, QR (all 3) works.. LLT fails however..18:09
@HeikoSlambday: which did you change?18:09
@HeikoSand what type of systems?18:09
lambdaypd.. same matrix18:10
lambdaycomplex18:10
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun18:10
lambdayI'm adding separate tests for all these.. I must have done something wrong earlier18:12
@HeikoSlambday: well, that sounds good then :)18:12
@HeikoSsince LLT cannot work anyways18:12
lambdayyeah!18:12
lambdayyep18:12
lambdayso, should we change the default to something else, svd18:12
lambday?18:12
@HeikoSyeah SVD, or the half-pivot QR since eigen people dont like their svd, you can choose :)18:13
lambdaywell, svd is slow18:13
lambdayso I am thinking qr18:13
@HeikoSlambday: okay then18:15
@HeikoSif its possible to choose, we dont hurt ourselves anyway18:15
lambdayyes18:15
lambday:)18:15
lambdayHeikoS: updated18:25
lambdaycould you please check it?18:25
-!- lisitsyn [~lisitsyn@37.61.181.106] has joined #shogun18:33
lambdayoh wait.. I forgot to remove something18:33
lambdayHeikoS: updated! tested with old eigen3 too...18:43
lambdayHeikoS: please check.. I'll be back in an hour18:43
-!- pickle27 [~kevin@rcv3-lab-pc.ee.queensu.ca] has joined #shogun18:45
-!- pickle27 [~kevin@rcv3-lab-pc.ee.queensu.ca] has quit [Quit: Leaving]18:56
van51why does every SG_PRINT show tha name of the function?19:06
van51i remember it being asked again but I don't remember seeing the answer19:06
@HeikoSvan51: because we made it (for some reason)19:08
@HeikoSonly debug is supposed to do this19:08
@HeikoSvan51: feel free to change, its all in SGIO.h19:08
van51HeikoS: okie :)19:08
@HeikoSlambday: there is no PR19:08
@HeikoSvan51: seems weird to me btw :) only makes sense for error/warning/debug19:09
@sonney2kHeikoS, well about the print stuff I asked you if this is what you had in mind ...20:01
@sonney2kanyways HeikoS, lisitsyn around?20:01
lisitsynsonney2k: yes20:01
@sonney2ktime to finalize the program20:01
@HeikoSsonney2k: then we got each other wrong20:01
@HeikoSprint should stay unchanged in my eyes20:01
@HeikoSbut the others should not20:01
@HeikoSanyways20:01
@HeikoSlets talk about the workshop20:01
@sonney2khttp://shogun-toolbox.org/page/Events/workshop2013_program20:01
@sonney2kthat was what we had20:01
@HeikoSwe dont have posters20:02
lambdayHeikoS: back20:02
lambdayHeikoS: I updated the same PR https://github.com/shogun-toolbox/shogun/pull/119020:02
@HeikoSand we dont have user contributed talks20:02
@HeikoSso I suggest we remove them20:02
@sonney2klambday, please postpone the discussion for 30 mins - thanks20:02
@HeikoSlambday: will have a look after this meeting20:03
lambdaysonney2k: okay.. no problem :)20:03
lambdayHeikoS: yes :)20:03
lisitsynsonney2k: alright let me put it into gdoc20:03
@sonney2klisitsyn, IIRC we had that in gdoc20:03
lisitsynsonney2k: like spreadsheet?20:03
@sonney2kiglesiasg transferred this to the website20:03
@HeikoSsonney2k, lisitsyn how many talks will there be?20:03
@HeikoSone from each of us?20:03
@sonney2kHeikoS, well how many speakers do we have?20:03
@HeikoSwe have a google doc for that too20:03
@HeikoSwait20:03
@HeikoShttps://docs.google.com/spreadsheet/ccc?key=0Amig4CX9mHQbdG1lQ1lNcDFsYUluZnA5ZDBiT0N1VHc20:04
lisitsynooh these gdocs are unmanageable20:04
@HeikoSwe have 5 people20:04
@HeikoSif gunnar comes20:04
lisitsyndoes gunnar come?20:04
@HeikoSand svm/mkl were two topics in the list that people voted for20:04
@HeikoSand GPS/ kernel-stats both were in there two20:05
@sonney2ksure20:05
@HeikoSso there is the potential for 5+2 talks20:05
@HeikoSwhere one is the intro from gunnar20:05
@HeikoSquestion is whether thats too much developer stuff20:05
@sonney2klisitsyn, no I meant when we did the planning we came up with some schedule IIRC20:05
@sonney2kbut no idea where that is20:05
lisitsynsonney2k: yeah I am trying to find it20:06
@sonney2kno idea it is gone I guess20:07
lisitsynstrange20:07
@HeikoSdoesnt matter too much does it?20:07
@HeikoSthe one on the weppage is goog enough to do it now20:07
@sonney2kHeikoS, btw after this we need to do the PO number for GSoC money stuff btw20:07
@sonney2kHeikoS, true20:07
@HeikoSsonney sure20:07
@HeikoSso how many talks?20:07
@HeikoSwe are mostly on our own anyways ;)20:07
@sonney2kHeikoS, now we have 29 rows btw20:07
@HeikoSnice20:08
lisitsynheh two from me20:08
@HeikoSthe votes should not have changed too much though20:08
@sonney2kheh20:08
lisitsynreferals :D20:08
lisitsynyou may notice their russian names20:08
lisitsynsonney2k: the last one is vojtech's student!20:08
lisitsynoriginally from my city what a coincidence :D20:09
@HeikoSso again for the talks, no user talks, no posters, what to do instead?20:09
@HeikoSjust have us talking and provide some space for discussions?20:09
@HeikoSand maybe hacking sessions?20:09
lisitsynHeikoS: yes I think so20:09
@sonney2kHeikoS, it is actually not so bad we can have a more relaxed program20:09
@sonney2klisitsyn, please post the current schedule in some gdoc20:10
@HeikoSMaybe the talks should end with some tasks that people could solve, to get their hands on some of the things20:10
lisitsynsonney2k: doing20:10
@HeikoSsonney2k: agreed20:10
@sonney2kHeikoS, yeah we can do some hands-on during / after the talks20:10
@HeikoSsonney2k: thats pretty nice then20:10
@HeikoShow long the talks?20:11
@HeikoSwhat to talk about?20:11
@sonney2kI would say about 40 mins then per talk20:11
@HeikoSsonney2k: thats long, we had 15 mins in the old schedule20:11
@sonney2k5 talks right?20:11
@sonney2kyes we squeezed it into 1520:11
@HeikoSwhat about some 30 mins and then another 30 mins for people to try things out?20:11
@sonney2kbut now we can relax it20:12
@HeikoS(including breaks and setting up the next one)20:12
@sonney2kgunnar can do a 1 hour keynote20:12
lisitsynlets put a hour then20:12
@HeikoSsonney2k: question also is whether we combine some talk topics20:12
@HeikoSSVM/MKL can be combined right?20:12
@HeikoSbut GPs/Kernel tests cannot :)20:13
lisitsynwhat is poster sessions?20:13
@sonney2kthen each talk can contain the theory and already in the talk practical sessions20:13
@sonney2klisitsyn, dead20:13
lisitsynright20:13
lisitsynsonney2k: what do we do when having food?20:13
@sonney2kHeikoS, yes I can do all kinds of SVMs & MKL20:13
@HeikoSok20:13
@sonney2klisitsyn, eat !20:13
lisitsynjust talk to each other I guess20:13
lisitsynwell no poster sessions so just food20:14
@sonney2kand eat :D20:14
@HeikoSso 5hrs of talks/session20:14
@HeikoSincluding the keynote20:14
@HeikoSthats enough for a day20:14
lisitsynHeikoS: intensive shit20:14
lisitsyn:D20:14
lisitsynwe will get out dying20:14
@HeikoSI suggest to do things relaxed20:14
@HeikoStake time in the talks etc20:14
lisitsynHeikoS: yes and turn on some bob marley songs20:14
@sonney2kyes not NIPS conference level for sure20:14
@sonney2kbut really slow if possible20:15
@HeikoSdont know  that :D20:15
@sonney2kHeikoS, slow?20:15
@sonney2kalways rushing heh?20:15
@HeikoSNIPS20:15
@HeikoShavent been there yet ;)20:15
@sonney2kthis year right?20:15
@HeikoSnono, I am never rushing20:15
@HeikoShopefully, we'll see20:15
@sonney2klisitsyn, where is the doc?20:15
lisitsynsonney2k: will share in a minute20:16
@HeikoSok then with the intro/speeddating and the final discussion and lunch, thats a lot of stuff20:16
@sonney2kI woudl suggest we do 1 hour gunnar for the ML tutorial as a start20:16
@HeikoSmaybe we should even go for 45 mins instead20:16
lisitsynsonney2k: HeikoS shared20:16
@sonney2kthen directly lunch20:16
@HeikoSyep agreed20:17
@HeikoSwhat about this shogun in bioinformatics ?20:17
@HeikoSsonney2k you could talk about that during MKL/SVM maybe?20:17
@HeikoSthe genome stuff you did20:17
@sonney2kthen I we somehow have to do a basic shogun tutorial. basic as in what for / how to install / let people install20:18
@sonney2kmaybe extend this with SVMs & MKL20:18
@HeikoSsonney2k: ok20:18
@HeikoSno a seperate thing would be cool to have20:18
@HeikoSyou probably have some talks about shogun on disc ?20:18
@HeikoSgeneral overview etc20:19
@sonney2kno good ones20:19
@HeikoSstill you are suited best for this one ;)20:19
@sonney2kmaybe we can do this quickly in 15-30 mins20:19
@HeikoSbtw in the SVM one you should mention some x-validation stuff since that is not talked about otherwise20:20
@HeikoSyep agreed20:20
@sonney2kthen I am not sure what to do with the 4 tutorials20:20
@HeikoS30 mins then everyone can install20:20
@HeikoSwhat about having two-coffee-two?20:20
@sonney2kand we really need to have some discussions20:20
@HeikoSI suggest to do the talks not too long20:21
@HeikoSthen afterwards there is room for discussion and to try20:21
@HeikoStrying out should only be reproducing things20:21
@HeikoSlike one slide that is left on20:21
@sonney2kwe have to know what we want to discuss about20:21
@HeikoSif the talk is only 35-40 mins, there is plenty of time20:21
@sonney2kas I've seen future of shogun is a topic20:22
@HeikoSjust give some room after each talk I would say, and a guided one later on20:22
@HeikoSafter the talks20:22
@sonney2kshogun development process the other20:22
@HeikoSsonney2k, yeah20:22
@HeikoSbut that should be seperate afterwards dont you think?20:22
@HeikoSlisitsyn:  are you typing the times?20:22
@sonney2kand shogun in new environments20:23
lisitsynHeikoS: yes20:23
lisitsynHeikoS: well just draft to estimate20:23
@HeikoSI would do 1 hr slots for each talk/questions/session20:23
@sonney2kHeikoS, we have to mix this in20:23
lisitsynHeikoS: 1hr per talk?20:23
@sonney2kotherwise it is too much talks and people just fall asleep20:23
lisitsynsounds too much20:23
@HeikoSlisitsyn: not talking time but everything20:23
@HeikoSshort break, try out the things maybe, ask some questions20:24
@HeikoSsonney2k: I agree, so what about two discussion sessions then?20:24
lisitsynI see no place for sessions then HeikoS?20:24
@sonney2kHeikoS, yes makes sense20:24
@HeikoSthere should be some headroom for spontaneous discussions in my eyes20:25
@HeikoSso the 1 hr x 4 seems good20:26
@HeikoS+ 2x30 mins for diccussions20:26
@HeikoSmaybe we set a red line for each of them20:26
@HeikoSfirst one should be more user-related, second one can be more dev-related20:26
-!- iglesiasg [53b32c87@gateway/web/freenode/ip.83.179.44.135] has joined #shogun20:26
@HeikoSand one coffee break can be combined with discussion?20:27
@sonney2kHeikoS, I would say there are mostly users there20:27
lisitsyniglesiasg: join our discussion20:27
@sonney2kHeikoS, maybe we have a coffee break just before the discussion20:27
@HeikoSsonney2k: my idea was just to put the stuff that mostly interests us to the end (if people want to go)20:27
@sonney2kother way round20:27
iglesiasglisitsyn: that's why I am here :)20:27
@sonney2kHeikoS, well we have saturday & sunday for that too20:27
@HeikoSsonney2k: ok20:28
@sonney2kHeikoS, we should figure out what users want /expect20:28
@HeikoSthen its pretty simple now20:28
@HeikoSsonney2k: well thats what the form was for right?20:28
@sonney2kHeikoS, open the gdoc and do the changes20:28
@sonney2kHeikoS, no not helpful20:28
@sonney2kthey only clicked points they kew20:28
@HeikoSok20:28
@sonney2kwe gave them I mean20:28
* sonney2k tired20:29
@HeikoSsonney2k: so then there should be a free slot maybe20:29
@HeikoSwhich we can fill by demand (optional)20:29
@sonney2kHeikoS, I would do that in the coffee / lunch break / discussions20:30
@HeikoSwhat about putting the shogun intro before the lunch?20:31
iglesiasgthere won't be poster sessions at the end then?20:31
iglesiasgnone applied?20:31
@HeikoSiglesiasg: nope ;)20:31
iglesiasgtoo sad20:31
lisitsyniglesiasg: are you up to show us some SO?20:32
@HeikoSwe really should make sure that the tutorials are nice with lots of applications and pictures20:32
@HeikoSotherwise its too much20:32
lisitsynsure20:32
iglesiasglisitsyn: I could talk about how to do SO in Shogun20:32
iglesiasgrather than SO in general since many people there will actually know more about it than me :D20:32
@HeikoSiglesiasg: I think it should be a mix20:33
lisitsynwell same holds for dimension reduction :D20:33
@HeikoSsame holds for all gsoc projects :D20:33
@HeikoSbut thats not bad20:33
@HeikoSI dont know any SO so happy to get an intro20:34
@HeikoSsame with dimension reduction actually20:34
@HeikoSso does everyone agree with the schedule?20:34
lisitsynlooks ok20:34
@HeikoSthen we need to set the order of the talks20:34
@sonney2kHeikoS, we need to put names to the talks20:34
@HeikoSyep20:34
@sonney2ktimes are ok with everyone?20:35
@HeikoSI take the second :)20:35
@sonney2kI made tutorial 3 & 4 45 mins only20:35
@sonney2kthe other 1 hour20:35
@HeikoSah thats good then we can fit in the double ones in there20:35
iglesiasgI think times are ok20:36
@HeikoSsonney2k: what do you think about this mixture of GPs and kernel mmd20:36
@HeikoSGPs are somehow very relatedto SVMs20:36
@HeikoSso might be a bit redundant20:36
lisitsynHeikoS: no no20:37
@HeikoS?20:37
lisitsynnobody understands GPs20:37
lisitsynso please keep it20:37
lisitsynyou and votjakovr may be20:37
lisitsynbut not me20:37
lisitsyn:D20:37
@HeikoSok then20:37
@HeikoSlisitsyn: what do you think on the kernel MMD stuff?20:37
@sonney2kHeikoS, what was the prio list again20:37
iglesiasgI will be able to re-use the slides from my final degree project presentation for the tutorial :)20:38
@sonney2kfor some reason I cannot find the doc20:38
@sonney2kHeikoS, but I would love to learn more about MMD too :D20:38
lisitsynHeikoS: I don't know20:38
@HeikoSI would like to talk aobut it since it is something new20:38
@HeikoSGPs and SVMs are quite mature already20:38
lisitsynsonney2k: would be nice if you take a look at most methods as MKL cases20:39
@sonney2kHeikoS, the mature stuff is of course the one practitioners are interested in and the researchers want the new stuff20:39
lisitsynI mean lasso is MKL20:39
@HeikoSsonney2k: sure20:39
lisitsyndeep learning is mkl20:39
lisitsynetc20:39
lisitsyn:D20:39
@sonney2klisitsyn, no deep learning is not (not convex like mkl and certainly more layers)20:40
iglesiasghttps://www.dropbox.com/s/a0bolzzxgr669sg/fjig_pfc_slides.pdf20:40
@HeikoSso any suggestions for this clash?20:40
lisitsynsonney2k: yeah sure, jk20:40
iglesiasgtake a look guys in case you want to use any of the figures I have there ^20:40
iglesiasgit is in Spanish but figures do not talk any language :)20:40
@sonney2kiglesiasg, you will give the talk20:41
lisitsyniglesiasg: yes :D20:41
lisitsyndid you know you are going to give a talk?20:41
@sonney2kiglesiasg, so just do it in english and add more shogun howto in there then you are good20:41
iglesiasgsonney2k, lisitsyn yes20:41
iglesiasgbut for instance at the beginning I am talking there about SVMs20:41
@sonney2kHeikoS, well you could do 2 tutorials then20:41
iglesiasgmaybe sonney2k wants to use any of the figures there20:41
lisitsyniglesiasg: ahh that's what you mean20:42
@HeikoStoo much20:42
@HeikoSor two short ones20:42
iglesiasglisitsyn: :)20:42
@HeikoSIll try that20:42
lisitsynHeikoS: what if you do20:42
@sonney2kiglesiasg, I will use shogun demos :D20:42
lisitsynSVM, GPs, MKL, MMD20:42
iglesiasgsonney2k: all right ;)20:42
@sonney2kHeikoS, or GPs and an a brief MMD20:42
-!- pickle27 [~kevin@rcv3-lab-pc.ee.queensu.ca] has joined #shogun20:42
iglesiasgI have generated many there with Shogun anyway20:42
@HeikoSsonney2k: yeah thats fine20:43
@sonney2kanyway we have to finalize this now20:43
@sonney2kiglesiasg, could you put it live on the website again?20:43
lisitsyniglesiasg: maquina de vectores soporte!20:43
@HeikoSor maybe lets shift it to 90 mins for that one20:43
@HeikoSthat its more relaxed20:43
iglesiasgsonney2k: the new schedule you mean?20:43
@sonney2kiglesiasg, yes20:43
iglesiasgupdate workshop program, yes sure20:43
@sonney2kwe should also make some announcement like program finalized20:43
@HeikoS1930 should be fine still20:44
@sonney2kand also instructions20:44
@sonney2kHeikoS, no20:44
iglesiasglisitsyn: maquina de vectores soporte rule them all :P20:44
@sonney2kwe don't have the main hall that long20:44
@sonney2kthere might be other events20:44
lisitsynwhere are we heading after 19.00?20:44
@sonney2klisitsyn, stay in c-base20:44
@sonney2kbut grill outside20:44
@HeikoSok then20:45
lisitsynsonney2k: do they have some yard?20:45
lisitsynor what?20:45
@sonney2khttps://encrypted.google.com/search?q=c-base&safe=active&source=lnms&tbm=isch&sa=X&ei=yzbLUYesEcmctAbs64C4Dw&ved=0CAkQ_AUoAQ&biw=1745&bih=83020:45
@sonney2klisitsyn, http://upload.wikimedia.org/wikipedia/de/7/7e/C-base-29.jpg20:46
@sonney2kthat is actually directly in front of c-base20:46
lisitsynsonney2k: ok some place outside20:46
@sonney2khttp://blog.diediplomandin.de/wp-content/uploads/cbase051.jpg20:46
@sonney2kand that is the plan!20:46
iglesiasgI think there is a c-base in budapest a bit diffirent than the one we are going to20:46
@HeikoSsonney2k, what about all the stuff we need (brill, meat, drinks, projector (?), cables, bla)20:46
lisitsynrussian vodka?20:47
lisitsynhah20:47
@HeikoSlisitsyn: yeah of course :) but how to get it there and who and when20:47
@sonney2kHeikoS, well they have lots of stuff there20:48
@sonney2kbut sure I have to buy quite a bit20:48
@sonney2kvideo recording is still hmm...20:48
@HeikoSsonney2k: as mentioned, I am in Berlin a few days earlier if you need help buying things20:48
lisitsynI am too :D20:48
lisitsynbut what can I do is to get lost20:48
lisitsynthat's the only thing I can do in berlin :D20:48
@sonney2kyou are an experienced drinker too20:49
lisitsynsonney2k: HeikoS: anyone is up to meet *before* the event btw?20:50
@sonney2kHeikoS, wait you will in berlin from 10th of july?20:50
@HeikoSyes 1020:50
@HeikoSmeeting some friends etc20:51
@sonney2kso what is missing in the schedule is a name to what we want to discuss in the discussion sessions20:51
lisitsynsonney2k: I suggest the second one is the future20:52
lisitsynof shogun20:52
@sonney2klisitsyn, agreed20:52
@HeikoSfirst one dev-proces20:52
lisitsynand the first one may be usability20:52
lisitsynha20:52
@HeikoSand usability yes20:52
@sonney2kok change it then!20:52
@HeikoSthat should be fine20:53
@HeikoSnote I changed the times slightly for discussion 1(if you agree)20:54
@sonney2kHeikoS, no20:54
@sonney2kDiscussion should be extra20:54
@sonney2knot in coffee break20:54
@HeikoSchanged20:55
@sonney2kHeikoS, if we are running late we just shorten the discussion/break20:55
@HeikoSyeah thats fine20:55
@HeikoSso all fine then=?20:56
lisitsynyes fine for me20:56
@HeikoSshould we maybe introduce the devs in the shogun intro?20:56
@HeikoS5 mins each or so20:56
lisitsyntoo much!20:56
@HeikoSor 220:56
@HeikoSjust to let people know who we are20:57
lisitsyncan we have badges?20:57
@HeikoSand maybe also say how we work together20:57
@HeikoSsince thats quite special - we havent ever met each other20:57
lisitsynhaha20:57
lisitsynHeikoS: haven't you met Soeren?20:57
@HeikoSexplains a lot of things in shogun ;)20:57
@HeikoSnope20:57
lisitsynoh so nobody ever met20:57
lisitsynthat's what we call virtual20:58
@sonney2kok schedule fixed right?20:58
@HeikoSok20:58
@sonney2kthen iglesiasg please put this online!20:59
@sonney2kwe should write some announcement for that then too20:59
@sonney2kand instructions how to get to c-base20:59
@sonney2kvolunteers?20:59
@HeikoSwho wrote the last ones?21:00
lisitsynsonney2k: what kind of announcement?21:00
lisitsynI wrote some email while ago you didn't like it may be we can re-use it?21:00
@sonney2kOnce that is done we should send an email to the participants21:00
@sonney2knot really21:00
@HeikoSI can do this then21:02
@sonney2kthat is where the hands on will be https://maps.google.de/maps?q=Marchstra%C3%9Fe+23&hl=en&ie=UTF8&t=h&hnear=Marchstra%C3%9Fe+23,+10587+Berlin&z=1621:02
@HeikoSsonney2k: you should do the way description since you know the place better21:02
@sonney2kroom MAR 0.00121:02
@sonney2kHeikoS, just use google maps21:02
@sonney2kit really is enough21:02
@HeikoSwell then no need to do that since people can do that on their own21:03
@sonney2kthere are train / underground stations 2 mins away from everything21:03
@HeikoSI thought you meant description of way21:03
lisitsynoh close to me!21:03
@HeikoSlike go there, turn left and jump up and down three times21:03
@HeikoSthen a magical door opens21:03
@HeikoSand three dwarves let you in :D21:03
@sonney2kHeikoS, heh21:03
lisitsynis that open for anyone?21:04
@sonney2kHeikoS, btw we should do the found shogun e.V. at 10 am then21:04
@HeikoSok21:04
@sonney2kdo you think 1 hour is enough?21:04
@HeikoSsonney2k: yes, we should distribute the satzung before21:04
@sonney2kHeikoS, yeah it should be finalized21:05
@HeikoSso that we just meet, quickly say what we do and then everyone signs21:05
@sonney2kHeikoS, did you get a reply from the KDE guys?21:05
@HeikoSwill do that in the next weeks21:05
@HeikoSsonney2k: not yet21:05
@HeikoSwaiting21:05
@sonney2khmmhh21:05
@sonney2kping them21:05
@HeikoSwill be fine21:05
@HeikoSwill do that21:05
@sonney2kbut politely21:05
@HeikoSjust returned from holidays21:05
@HeikoSbtw what about this unconference stuff? saturday?21:06
@sonney2kHeikoS, I think you have to work it out - but IMHO everyone brings his notebook21:06
@sonney2kand you start hacking/ discussing21:06
@HeikoSsonney2k, lisitsyn: I will send you the announcement tomorrow and if you agree Ill send it to all participants21:07
@sonney2kyou could of course solve 'real' toy problems21:07
@HeikoSsonney2k: ah yeah you are not around saturday right?21:07
@HeikoSok then21:07
@sonney2kI am going to be shot for organizing this anyways21:07
@HeikoSsonney2k: I would like to go home now, could we do the PO tomorrow same time?21:07
@HeikoSsonney2k: really? haha, TUB nice21:08
@sonney2kHeikoS, or during the day from work21:08
@HeikoSok then21:08
@sonney2kHeikoS, I actually only wanted to ask if we can do ti21:08
@sonney2kit21:08
@sonney2kvia you again21:08
@HeikoSsonney2k: yeah its fine, no income in Germany21:08
@sonney2kbut then it is easy21:08
@sonney2kjust this one form to fill out21:09
@HeikoSand then Ill donate it after the workshop21:09
@HeikoSgoing home, see you guys tomorrow!21:09
-!- vgorbati [~vgorbati@37.19.133.36] has joined #shogun21:09
lisitsynsee you21:09
lisitsynsonney2k: what time you expect us to be at that TU room?21:10
@sonney2klisitsyn, it is reserved from 8-21hrs21:10
lisitsynuh21:10
lisitsynI am definitely won't stay that long!21:11
@sonney2kso I guess 9-17 or 10-1821:11
@sonney2kdo whatever you want :D21:11
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun21:11
shogun-notifier-shogun: lambday :develop * 84258f4 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/84258f49454fe6d5bc5d26d52d00863cc2740b6321:11
shogun-notifier-shogun: LinearSolver base, direct solver implementation added (log-det)21:11
shogun-notifier-shogun: Heiko Strathmann :develop * 2534fbe / / (4 files): https://github.com/shogun-toolbox/shogun/commit/2534fbeace514a0165cb9928fdcb42ae8a05e0d721:11
shogun-notifier-shogun: Merge pull request #1190 from lambday/feature/log_determinant21:11
shogun-notifier-shogun:21:11
shogun-notifier-shogun: CLinearSolver base, direct solver implementation added (log-det)21:11
-!- HeikoS [~heiko@nat-163-105.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]21:11
lisitsynsonney2k: sure I just mean I will be first time in that city :)21:12
@sonney2klisitsyn, how long will you be staying?21:14
@sonney2khttps://docs.google.com/forms/d/1zDCB8aozZOQJjqDm5rLsOso2xAbIn7OBcYGdLhL7OTs/viewform21:14
lisitsynsonney2k: 10,11,12,13,1421:14
@sonney2kahh heiko just left21:14
@sonney2klisitsyn, ohh short21:14
@sonney2kI thought a week21:14
lisitsynsonney2k: yes that's why I am confused with staying in room whole day :)21:14
lisitsynsonney2k: any idea when can they decide on me?21:15
@sonney2ksorry what?21:20
lisitsynsonney2k: summit21:21
lisitsynsonney2k: my participance21:21
lisitsynwaiting list21:21
lisitsyn#waitinglist21:21
lisitsyn:D21:21
@sonney2kno idea21:22
shogun-buildbotbuild #1147 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1147  blamelist: lambday <heavensdevil6909@gmail.com>21:26
shogun-buildbotbuild #979 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/979  blamelist: lambday <heavensdevil6909@gmail.com>21:28
-!- zxtx [~zv@12.231.120.253] has joined #shogun21:28
van51sonney2k: can I ask a couple things?21:29
van51or someone else actually21:33
lisitsynvan51: what's up?21:33
van51lisitsyn: first of all, if I change the display of the histogram to that : http://pastebin.com/cbReK2Hn in Alphabet.cpp21:34
van51is it good?21:34
lisitsynvan51: what is changed?21:34
lisitsynCharacter is not in alphabet?21:35
van51lisitsyn: instead of the char it used to print the index of the character21:35
van51and what you sai21:35
van51said*21:35
lisitsynlike 5921:35
lisitsyninstead of some letter?21:35
van51yeap21:35
lisitsynlike that indeed21:35
lisitsynI do I mean21:35
van51so I leave it like this?21:36
lisitsynyes I don't mind such improvement21:36
van51ok21:36
van51one more thing21:36
van51I changed the default behavior of SGIO that always printed the function name21:37
van51like we discussed with HeikoS above21:37
shogun-buildbotbuild #1148 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1148  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:37
van51should I have it to set this function_name display automatically if the loglevel is set to >= warn?21:37
lisitsynvan51: I don't know but I tend to think it is better to show these names only when debuggin21:38
van51lisitsyn: actually now there are 3 possibilities, show nothing, show function name, show file and line21:39
shogun-buildbotbuild #980 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/980  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:41
-!- travis-ci [~travis-ci@ec2-174-129-182-140.compute-1.amazonaws.com] has joined #shogun21:42
travis-ci[travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/847684321:42
-!- travis-ci [~travis-ci@ec2-174-129-182-140.compute-1.amazonaws.com] has left #shogun []21:42
-!- vgorbati [~vgorbati@37.19.133.36] has quit [Ping timeout: 256 seconds]21:42
-!- zxtx [~zv@12.231.120.253] has quit [Read error: Connection reset by peer]21:47
lisitsynvan51: may be it makes the sense to put an option like show line21:47
@wikingyo21:47
lisitsynwiking: hey21:48
van51lisitsyn: maybe I'll make a PR with these changes only and that would mean that the user could/should also specify what to display as well21:49
van51and we can update it afterwards21:49
lisitsynalright21:51
@sonney2kvan51, the hist['E'] is not a good idea21:57
@sonney2kthink of '\n' as the char21:57
@sonney2kor other control characters21:58
@sonney2kvan51, best is you try it with the full range 0..25521:58
* sonney2k off21:58
van51sonney2k: hmm idd21:59
lisitsynvan51: I'd handle alphanumerical this way21:59
lisitsynnumber otherwise21:59
lisitsynand special like EOF and EOL22:00
lisitsynmarked some way too22:00
lisitsynbut that's me ;)22:00
van51lisitsyn: makes sense actually22:00
shogun-buildbotbuild #1269 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/126922:01
-!- pickle27 [~kevin@rcv3-lab-pc.ee.queensu.ca] has quit [Quit: Leaving]22:05
shogun-buildbotbuild #1270 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1270  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>22:27
@wikinglisitsyn: http://osrc.dfm.io/lisitsyn22:32
@wiking:DDD22:33
lisitsynwiking: ha!22:34
@wikingsonney2k: http://osrc.dfm.io/sonney2k22:35
iglesiasgoh that is super cool man!22:35
iglesiasglisitsyn: man you are a top 6%22:35
iglesiasgyou are like famous22:35
lisitsyniglesiasg: haha22:36
lisitsynyou are a trend setting C++ expert iglesiasg22:36
iglesiasg:D22:37
* lambday yawns22:38
lambdaygood night guys22:38
-!- lambday [67157e4d@gateway/web/cgi-irc/kiwiirc.com/ip.103.21.126.77] has quit [Quit: lambday]22:39
lisitsynhttp://osrc.dfm.io/karlnapf22:39
@wikinglol man this shit says work best in the morning.22:41
lisitsynwiking: I am dead in the morning :D22:44
van51made a PR22:48
van51lisitsyn: if you have time now, please have a look. it's rather small22:48
lisitsynvan51: yes sure22:48
van51thanks :)22:48
lisitsynvan51: ahh so there is a function for that22:48
lisitsyngut22:49
van51lisitsyn: yeah thankfully22:49
van51I had a long if before22:49
lisitsynvan51: let sonney2k merge it, I am totally happy with it22:50
lisitsynvan51: ahh one more thing may be22:50
lisitsynvan51: what about EOF and EOL? makes sense to state them too22:50
van51lisitsyn: tell me22:50
lisitsynI don't remember any code of \n22:51
lisitsynso may be it would be more convenient for users too22:51
van51lisitsyn: yeah you're right22:51
van51i'll do it in a while then22:51
-!- iglesiasg [53b32c87@gateway/web/freenode/ip.83.179.44.135] has quit [Quit: Page closed]22:51
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun22:55
-!- mode/#shogun [+o iglesiasg] by ChanServ22:55
@iglesiasgsonney2k: around?22:57
@iglesiasgsonney2k: for some reason the webpage crashes in http://www.shogun-toolbox.org/admin/22:59
-!- zxtx [~zv@12.231.120.253] has joined #shogun23:05
@iglesiasgsonney2k: maybe I remember wrong that was the page to change stuff23:06
@iglesiasglisitsyn do you happen to remember if it was that one?23:06
lisitsynuhmm I don't remember23:07
lisitsynbut admin is not working yeah23:07
@iglesiasgBernard just told me it should be that one too23:08
shogun-notifier-shogun: van51 :develop * 9132e15 / src/shogun/features/Alphabet.cpp,src/shogun/io/SGIO.cpp: https://github.com/shogun-toolbox/shogun/commit/9132e156eefe183e91d5071414ce0388c05057b123:23
shogun-notifier-shogun: Changed display messages23:23
shogun-notifier-shogun: Heiko Strathmann :develop * f6bdbf1 / src/shogun/features/Alphabet.cpp,src/shogun/io/SGIO.cpp: https://github.com/shogun-toolbox/shogun/commit/f6bdbf1f1e889232c88292cd6e12562f4388881c23:23
shogun-notifier-shogun: Merge pull request #1191 from van51/develop23:23
shogun-notifier-shogun:23:23
shogun-notifier-shogun: Changed display messages23:23
van51I think just a sec after I updated it, it was merged :p23:24
-!- zxtx [~zv@12.231.120.253] has quit [Read error: Connection reset by peer]23:28
shogun-buildbotbuild #981 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/981  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>23:28
shogun-buildbotbuild #1149 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1149  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>23:36
-!- lisitsyn [~lisitsyn@37.61.181.106] has quit [Quit: Leaving.]23:36
shogun-buildbotbuild #982 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/982  blamelist: van51 <vangelis_51@hotmail.com>23:39
shogun-buildbotbuild #1150 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1150  blamelist: van51 <vangelis_51@hotmail.com>23:47
--- Log closed Thu Jun 27 00:00:06 2013