--- Log opened Sun Aug 04 00:00:00 2013 | ||
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has quit [Ping timeout: 264 seconds] | 00:01 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 00:16 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 00:16 | |
@sonney2k | gsomix_, I am thinking about using C_ORDER per default | 00:16 |
---|---|---|
@sonney2k | gsomix_, with csv data it is most likely that one has one vector per line | 00:17 |
@sonney2k | (C_ORDER) | 00:17 |
@sonney2k | IIRC there is the .tcsv format | 00:17 |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has joined #shogun | 00:17 | |
@sonney2k | that does one vector per column | 00:18 |
-!- travis-ci [~travis-ci@ec2-54-224-79-102.compute-1.amazonaws.com] has joined #shogun | 00:26 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg'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/9816903 | 00:26 |
-!- travis-ci [~travis-ci@ec2-54-224-79-102.compute-1.amazonaws.com] has left #shogun [] | 00:26 | |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 7da4bc1 / src/shogun/io/CSVFile.cpp: https://github.com/shogun-toolbox/shogun/commit/7da4bc1f0bffea270703ed2fd3622e3275a261f9 | 00:26 |
shogun-notifier- | shogun: switch to C_ORDER per default | 00:26 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 68834f7 / src/shogun/features/DenseFeatures.cpp: https://github.com/shogun-toolbox/shogun/commit/68834f729478f5690a06de1efbffd6f738b3a3c2 | 00:26 |
shogun-notifier- | shogun: don't call DotFeatures(fname) to fix loading from within constructor | 00:26 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * f8e2b9a / examples/undocumented/python_modular/kernel_anova_modular.py: https://github.com/shogun-toolbox/shogun/commit/f8e2b9a04a737dda42620ef83308fdfebd1a7d6c | 00:26 |
shogun-notifier- | shogun: move anova kernel to use csvfile class | 00:26 |
@sonney2k | gsomix_, ok changed that now please fix the vector reader asap please | 00:29 |
@sonney2k | then I can finally get things working | 00:29 |
@sonney2k | alrighty | 00:29 |
* sonney2k ZZZzzzz | 00:29 | |
shogun-buildbot | build #1150 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1150 blamelist: Soeren Sonnenburg <sonne@debian.org> | 00:34 |
shogun-notifier- | shogun: Sergey Lisitsyn :develop * f6fbd64 / tests/unit/transfer/MALSAR_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/f6fbd649d11d45bc48a27c90c84ca31aa13e8cc1 | 00:36 |
shogun-notifier- | shogun: Added simple tests for algorithms from the MALSAR | 00:36 |
shogun-buildbot | build #1151 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1151 blamelist: Soeren Sonnenburg <sonne@debian.org> | 00:48 |
shogun-buildbot | build #1152 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1152 blamelist: Soeren Sonnenburg <sonne@debian.org> | 00:52 |
-!- travis-ci [~travis-ci@ec2-54-224-79-102.compute-1.amazonaws.com] has joined #shogun | 00:53 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg'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/9817515 | 00:53 |
-!- travis-ci [~travis-ci@ec2-54-224-79-102.compute-1.amazonaws.com] has left #shogun [] | 00:53 | |
shogun-buildbot | build #1459 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/1459 blamelist: Soeren Sonnenburg <sonne@debian.org> | 01:16 |
-!- travis-ci [~travis-ci@ec2-54-227-68-127.compute-1.amazonaws.com] has joined #shogun | 01:17 | |
travis-ci | [travis-ci] it's Sergey Lisitsyn'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/9817783 | 01:17 |
-!- travis-ci [~travis-ci@ec2-54-227-68-127.compute-1.amazonaws.com] has left #shogun [] | 01:17 | |
@iglesiasg | lisitsyn, sonney2k: did you guys see https://github.com/shogun-toolbox/shogun/issues/1351? | 01:28 |
@iglesiasg | I think there is something weird with get_distance_matrix | 01:28 |
shogun-buildbot | build #1153 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1153 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 01:31 |
shogun-buildbot | build #1460 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/1460 blamelist: Soeren Sonnenburg <sonne@debian.org>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 01:59 |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has quit [Remote host closed the connection] | 02:50 | |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has joined #shogun | 02:56 | |
-!- pickle27 [~Kevin@67.193.243.174] has quit [Quit: Leaving] | 03:33 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 03:36 | |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has quit [Ping timeout: 245 seconds] | 03:43 | |
shogun-buildbot | build #478 of nightly_default is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/478 | 03:49 |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has joined #shogun | 03:56 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 04:20 | |
-!- lambday [67157d4c@gateway/web/freenode/ip.103.21.125.76] has joined #shogun | 06:57 | |
-!- lambday [67157d4c@gateway/web/freenode/ip.103.21.125.76] has quit [Ping timeout: 250 seconds] | 09:56 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 11:06 | |
shogun-notifier- | shogun: lambday :develop * 09b5bf2 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/09b5bf2cc0049d10c314ce1d4fda58594610a6ef | 11:06 |
shogun-notifier- | shogun: Lanczos eigensolver added | 11:06 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 3966c12 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/3966c12f65af9e07f20ec53b1765668b5c1a8dea | 11:06 |
shogun-notifier- | shogun: Merge pull request #1353 from lambday/feature/log_determinant | 11:06 |
shogun-notifier- | shogun: | 11:06 |
shogun-notifier- | shogun: Lanczos eigensolver added | 11:06 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 37bac96 / / (15 files): https://github.com/shogun-toolbox/shogun/commit/37bac96cc454612a4496e5a439a7b44cdcc79bdc | 11:09 |
shogun-notifier- | shogun: -added unbiased estimator for marginal likelihood. | 11:09 |
shogun-notifier- | shogun: -new class for probability distributions and a Gaussian for sampling and log-pdf (numerically stable) | 11:09 |
shogun-notifier- | shogun: -log-sum-exp and log-mean-exp in CMath | 11:09 |
shogun-notifier- | shogun: -CLikelihood model now is able to evaluate many likelihoods at once | 11:09 |
shogun-notifier- | shogun: -some fixes | 11:09 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 13ec4b7 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/13ec4b7f87afd704b5db853f40b4e779c496e863 | 11:09 |
shogun-notifier- | shogun: removed SVD_QR for Gaussian, too unstable, replaced by ridge on diagonal | 11:09 |
shogun-notifier- | shogun: Heiko Strathmann :develop * e49119c / src/interfaces/modular/Distribution.i/ (2 files): https://github.com/shogun-toolbox/shogun/commit/e49119c28654dd123967f221100ceab8603f0aea | 11:09 |
shogun-notifier- | shogun: added GaussianDistriution to modular interfaces | 11:09 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 7fcbcbf / src/shogun/machine/gp/InferenceMethod.cpp: https://github.com/shogun-toolbox/shogun/commit/7fcbcbfef1bdf08816d130052205700c057d9ee6 | 11:09 |
shogun-notifier- | shogun: make covariances symmetric | 11:09 |
shogun-notifier- | shogun: Heiko Strathmann :develop * e14944d / src/shogun/distributions/classical/GaussianDistribution.cpp: https://github.com/shogun-toolbox/shogun/commit/e14944d5766fa724928144f919118a76ae066518 | 11:09 |
shogun-notifier- | shogun: some security checks for psd matrices | 11:09 |
shogun-notifier- | shogun: Heiko Strathmann :develop * f4340e8 / src/shogun/distributions/classical/GaussianDistribution.cpp: https://github.com/shogun-toolbox/shogun/commit/f4340e818a7f62c7bbc71856f8a3d64c9d31ca6d | 11:09 |
shogun-notifier- | shogun: removed hacky psd checks, user has to do this. Print smallest eigenvalue instead | 11:09 |
shogun-notifier- | shogun: Heiko Strathmann :develop * f5d402e / src/shogun/kernel/LinearARDKernel.cpp,src/shogun/kernel/LinearARDKernel.h: https://github.com/shogun-toolbox/shogun/commit/f5d402e4f3d7b080ad2ee1b0f75b72e35fab01f5 | 11:09 |
shogun-notifier- | shogun: added method to set/get weights as a whole and some doc updates | 11:09 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 11:27 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 11:27 | |
-!- travis-ci [~travis-ci@ec2-54-227-68-127.compute-1.amazonaws.com] has joined #shogun | 11:38 | |
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/9825642 | 11:38 |
-!- travis-ci [~travis-ci@ec2-54-227-68-127.compute-1.amazonaws.com] has left #shogun [] | 11:38 | |
-!- travis-ci [~travis-ci@ec2-54-224-79-102.compute-1.amazonaws.com] has joined #shogun | 11:59 | |
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/9825717 | 11:59 |
-!- travis-ci [~travis-ci@ec2-54-224-79-102.compute-1.amazonaws.com] has left #shogun [] | 11:59 | |
-!- vgorbati [~vgorbati@91.216.173.29] has joined #shogun | 12:32 | |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has quit [Ping timeout: 268 seconds] | 13:03 | |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has joined #shogun | 13:04 | |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has quit [Ping timeout: 268 seconds] | 13:11 | |
-!- vgorbati [~vgorbati@91.216.173.29] has quit [Quit: vgorbati] | 13:19 | |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has joined #shogun | 13:24 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 14:10 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 16:43 | |
-!- foulwall [~user@110.17.1.73] has joined #shogun | 17:05 | |
-!- HeikoS [~heiko@nat-160-11.internal.eduroam.ucl.ac.uk] has joined #shogun | 17:34 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 17:34 | |
-!- foulwall [~user@110.17.1.73] has quit [Remote host closed the connection] | 17:49 | |
lisitsyn | hey HeikoS | 18:13 |
@HeikoS | lisitsyn: hi! | 18:14 |
lisitsyn | HeikoS: how is it going? | 18:14 |
@HeikoS | well :) | 18:14 |
@HeikoS | and you? | 18:14 |
lisitsyn | yeah fine too | 18:14 |
lisitsyn | HeikoS: I am still debugging this thing that fails in python :) | 18:14 |
@HeikoS | lisitsyn: is that a hard one? | 18:15 |
@HeikoS | lisitsyn: oh btw do the tests work fine on your machine on current git? | 18:15 |
lisitsyn | HeikoS: yeah IIRC it worked yesterday | 18:15 |
lisitsyn | HeikoS: what fails on your machine? | 18:15 |
@HeikoS | lisitsyn: ok good | 18:15 |
@HeikoS | lisitsyn: nothing, but travis seems to freeze and fernando also had a problem with my latest commit | 18:15 |
lisitsyn | HeikoS: I was surprised with the number of tests | 18:16 |
@HeikoS | lisitsyn: where? | 18:16 |
lisitsyn | HeikoS: the total number | 18:17 |
lisitsyn | like 11** | 18:17 |
@HeikoS | lisitsyn: well half of it is created automagic | 18:17 |
lisitsyn | yeah | 18:17 |
@HeikoS | lisitsyn: but still good | 18:17 |
@HeikoS | we test things on every class | 18:17 |
@HeikoS | lisitsyn: still I would like to take this a step further in the near future | 18:17 |
@HeikoS | make serialisation/clone/equals 100% reliable | 18:18 |
@HeikoS | currently its only tested on empty classes | 18:18 |
lisitsyn | HeikoS: what about migration and etc? | 18:18 |
@HeikoS | lisitsyn: I am for dropping that (and the integration tests with it) since we cannot really maintain it. But open for discussions | 18:19 |
lisitsyn | HeikoS: the only one that can maintain it is you probably | 18:20 |
@HeikoS | lisitsyn: and I fail miserably ;) | 18:20 |
@HeikoS | lisitsyn: the code is too complicated, too dirty, not tested etc | 18:20 |
lisitsyn | no you are doing other things | 18:20 |
lisitsyn | that are more important | 18:20 |
@HeikoS | lisitsyn: it is extremely hard to use | 18:20 |
@HeikoS | lisitsyn: and nobody really uses that parts that work | 18:20 |
@HeikoS | lisitsyn: rather, we always re-commit the integration tests | 18:20 |
@HeikoS | lisitsyn: well thats the same result, I dont think an open-source project focussed on ML can offer migration for its fil format | 18:21 |
lisitsyn | HeikoS: hehe that's true everybody should do their job | 18:22 |
@HeikoS | lisitsyn: how are things going with you? | 18:22 |
lisitsyn | HeikoS: I feel we are maintaining a bit more than we should | 18:22 |
@HeikoS | lisitsyn: yeah maybe | 18:23 |
@HeikoS | lisitsyn: see my issues on dropping code | 18:23 |
lisitsyn | HeikoS: what things? | 18:23 |
@HeikoS | marking stable things, etc | 18:23 |
@HeikoS | lisitsyn: work, shogun etc | 18:23 |
@HeikoS | gsoc | 18:23 |
lisitsyn | ahh | 18:23 |
lisitsyn | we had some failure at my job and I moved to a new place | 18:23 |
lisitsyn | :D | 18:23 |
@HeikoS | exciting? | 18:24 |
lisitsyn | gsoc wise everything is ok - fernando and kevin are doing pretty well | 18:24 |
lisitsyn | HeikoS: well I changed my 12 m2 to 58 m2 | 18:24 |
lisitsyn | ;) | 18:24 |
@HeikoS | I see | 18:25 |
@HeikoS | like I change my e2 7dj to e2 7ja recently :)( | 18:25 |
lisitsyn | HeikoS: ehmhmm | 18:25 |
lisitsyn | :D | 18:25 |
lisitsyn | what? | 18:25 |
lisitsyn | HeikoS: m2 = m^2 | 18:26 |
@HeikoS | ah | 18:26 |
lisitsyn | like sq.m. | 18:26 |
@HeikoS | haha :) | 18:26 |
lisitsyn | hahah | 18:26 |
@HeikoS | wow thats nice then | 18:26 |
lisitsyn | I get your joke haha | 18:26 |
@HeikoS | I though that was a postcode | 18:26 |
lisitsyn | ahh | 18:26 |
lisitsyn | it wasn't | 18:26 |
lisitsyn | :D | 18:26 |
lisitsyn | I thought you said some non-sense like I did | 18:26 |
lisitsyn | HeikoS: yeah that's area | 18:26 |
lisitsyn | HeikoS: what about you? | 18:28 |
lisitsyn | ohohoh | 18:28 |
lisitsyn | what's with doc sprint | 18:28 |
@HeikoS | ? | 18:28 |
lisitsyn | I think we're missing it | 18:28 |
@HeikoS | lisitsyn: we could apply | 18:28 |
@HeikoS | deadline is 6th | 18:28 |
lisitsyn | do we have real idea? | 18:28 |
@HeikoS | I am not really sure though | 18:28 |
lisitsyn | HeikoS: carol didn't contact me too | 18:29 |
@HeikoS | we are missing a clear goal with that | 18:29 |
lisitsyn | about gsoc meeting | 18:29 |
@HeikoS | contact? | 18:29 |
lisitsyn | well I am on the waiting list | 18:29 |
@HeikoS | ah we still need to get the PO number | 18:29 |
@HeikoS | ah man | 18:29 |
@HeikoS | so annoying with my bank | 18:29 |
lisitsyn | what's with your bank? | 18:29 |
@HeikoS | I dont have an account on my name | 18:29 |
@HeikoS | or I do have but I dont know the number | 18:29 |
lisitsyn | hmm | 18:30 |
@HeikoS | and my parents are in vaccation | 18:30 |
lisitsyn | what kind of account? | 18:30 |
@HeikoS | post lying at their place | 18:30 |
@HeikoS | for GSoC money | 18:30 |
lisitsyn | but you do have some mastercard or visa or anything? | 18:30 |
@HeikoS | I will call them tomorow again | 18:30 |
@HeikoS | not in Germany | 18:30 |
lisitsyn | ahh | 18:30 |
@HeikoS | in England I would have to pay taxes | 18:30 |
lisitsyn | complicated | 18:30 |
@HeikoS | yeah annoying | 18:31 |
@HeikoS | lisitsyn: I will get some food now, be back in 15 mins | 18:31 |
lisitsyn | HeikoS: I think I'll ask carol if I have any chances | 18:31 |
lisitsyn | ok see you then ;) | 18:31 |
@HeikoS | lisitsyn: ask sonney, I think we get a fixed amount | 18:31 |
@HeikoS | so we should book flights asap | 18:31 |
@HeikoS | so I need to sort out bank stuff asap | 18:32 |
@HeikoS | ill be back soon, catch up then | 18:32 |
lisitsyn | HeikoS: hmm but that's the flight | 18:32 |
lisitsyn | we need a hotel reservation for me etc | 18:32 |
lisitsyn | so it is all about google's decision | 18:32 |
lisitsyn | (if I get it right) | 18:32 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 18:52 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 18:52 | |
@iglesiasg | hello guys | 18:53 |
@HeikoS | re | 18:54 |
@HeikoS | iglesiasg: any updates on that freeze you had? | 19:12 |
@iglesiasg | HeikoS, no, not yet sorry. I am doing the unit test sonney2k ask for this issue https://gist.github.com/iglesias/6145782 | 19:12 |
@iglesiasg | HeikoS, I won't be able to take a further look at it until tomorrow | 19:13 |
@iglesiasg | HeikoS, Tuesday morning latest | 19:13 |
@HeikoS | iglesiasg: ah I see, unit test is good :) | 19:13 |
@iglesiasg | but I won't forget about it! I leave your mail as not read :) | 19:13 |
@HeikoS | iglesiasg: cool thanks :) | 19:16 |
@HeikoS | iglesiasg: no rush on this btw | 19:16 |
@HeikoS | as long as travis and the buildbot dont complain | 19:16 |
@iglesiasg | HeikoS, sure | 19:16 |
@iglesiasg | HeikoS, although the travis build for that PR was a bit weird, wasn't it? Lot of grey in there | 19:17 |
@HeikoS | iglesiasg: yeah it was | 19:17 |
@HeikoS | I thought it ran out of time (it does sometimes= | 19:17 |
@HeikoS | so I merged since the previous compile error were gone | 19:17 |
@iglesiasg | aham I see | 19:18 |
@iglesiasg | HeikoS, I was thinking for the multi-threaded cross-validation issue | 19:33 |
@HeikoS | iglesiasg: yes? | 19:34 |
@iglesiasg | what if the subset is not in shared memory, so you have several process running with its own subset | 19:34 |
@iglesiasg | then I think it would not be necessary to have multiple subsets | 19:34 |
@HeikoS | iglesiasg: yeah thats the goal | 19:35 |
@HeikoS | iglesiasg: but the labels should stay shared | 19:35 |
@HeikoS | or labels dont matter | 19:35 |
@HeikoS | features should be shared | 19:35 |
@iglesiasg | HeikoS, indeed | 19:35 |
@HeikoS | so they need multiple subsets | 19:35 |
@HeikoS | since all threads work on the same feature object | 19:35 |
@iglesiasg | mmm | 19:35 |
@iglesiasg | I was thinking if there is a way to have the features shared but not the subset | 19:36 |
@iglesiasg | but that would break with how it is currently done | 19:36 |
@iglesiasg | since the subset is a member of the features IIRC | 19:36 |
@HeikoS | iglesiasg: where is the difference? | 19:36 |
@HeikoS | iglesiasg: when features store multiple subsets and 1 object is shared | 19:37 |
@HeikoS | or if implementations store their own subset? | 19:37 |
@HeikoS | I think they also should be kept within the features to avoid problems | 19:37 |
@iglesiasg | HeikoS, I have the idea that it could be different. If cross-validation is run by different processes, then each process has its own subset in its own memory space | 19:38 |
@iglesiasg | but in the code there is only one subset, no multiple ones | 19:38 |
@HeikoS | iglesiasg: yes, but where is the difference or advantage to store the subset on the shared feature object? | 19:38 |
@iglesiasg | HeikoS, I think it should be easier to program if there is only one subset to take into account | 19:39 |
@HeikoS | iglesiasg: there is, you just have to pass/remember the id | 19:39 |
@HeikoS | the only difference is where it is stored | 19:39 |
@iglesiasg | HeikoS, I didn't understand that, sorry | 19:40 |
@HeikoS | iglesiasg: only difference is where the subset is stored | 19:41 |
@iglesiasg | HeikoS, yes | 19:42 |
@iglesiasg | in any case I think my idea is not correct | 19:43 |
@HeikoS | why nor? | 19:43 |
@HeikoS | not? | 19:43 |
@iglesiasg | I don't think it is possible to have several processes that share part of the memory (the features) and not other part (the subsets) | 19:43 |
gsomix_ | good evening | 19:53 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 20:02 | |
shogun-notifier- | shogun: Fernando Iglesias :develop * 1c598fd / tests/unit/distance/EuclideanDistance_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/1c598fd9180a3397058e69b1586cf489031e7435 | 20:02 |
shogun-notifier- | shogun: Add EuclideanDistance unit test | 20:02 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 646363b / tests/unit/distance/EuclideanDistance_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/646363b95f060f58afae340a769e2177ec22b55f | 20:02 |
shogun-notifier- | shogun: Merge pull request #1357 from iglesias/euclidean_unittest | 20:02 |
shogun-notifier- | shogun: | 20:02 |
shogun-notifier- | shogun: Add EuclideanDistance unit test | 20:02 |
-!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has joined #shogun | 20:03 | |
lambday | HeikoS: hi | 20:03 |
lambday | HeikoS: should we try seeing the performance of cg_m instead of individual cg? I think this would be the bottleneck! | 20:04 |
lambday | cg solvers are not trustworthy :( | 20:04 |
lambday | HeikoS: another thing, the only thing that changes in "LogRationalApproximationIndividual::submit_jobs" is the part which creates new operators for dense and sparse matrices... I don't want to write just another class for sparse... so I can check for types dynamically I think! | 20:08 |
shogun-buildbot | build #1646 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1646 blamelist: Soeren Sonnenburg <sonne@debian.org> | 20:21 |
-!- travis-ci [~travis-ci@ec2-204-236-192-191.compute-1.amazonaws.com] has joined #shogun | 20:55 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg'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/9835487 | 20:55 |
-!- travis-ci [~travis-ci@ec2-204-236-192-191.compute-1.amazonaws.com] has left #shogun [] | 20:55 | |
shogun-buildbot | build #1645 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1645 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 20:57 |
lisitsyn | HeikoS: ha it seems that my code fails because of C++11 | 21:20 |
lisitsyn | :D | 21:20 |
lisitsyn | lambday: you worked with complex a bit | 21:20 |
lisitsyn | may be you have seen something like that | 21:20 |
lisitsyn | I have the following line | 21:20 |
lisitsyn | cout << "eigenvalue " << eigensolver.eigenvalues()[i] << "=" << eigensolver.eigenvalues()[i].real() << "+i" << eigensolver.eigenvalues()[i].imag() << endl; | 21:20 |
lisitsyn | the output is | 21:20 |
lisitsyn | eigenvalue (0.666667,0)=3.22526e-319+i8.25667e-317 | 21:20 |
-!- gsomix__ [~gsomix@178.45.37.229] has joined #shogun | 21:21 | |
-!- lambday_ [67157d37@gateway/web/freenode/ip.103.21.125.55] has joined #shogun | 21:21 | |
lambday_ | lisitsyn: sorry I got dc | 21:21 |
lisitsyn | lambday | 21:21 |
lisitsyn | lambday_: | 21:21 |
lambday_ | lisitsyn: which one is failing for c++11 | 21:21 |
lisitsyn | (11:20:38 PM) lisitsyn: I have the following line | 21:21 |
lisitsyn | (11:20:43 PM) lisitsyn: cout << "eigenvalue " << eigensolver.eigenvalues()[i] << "=" << eigensolver.eigenvalues()[i].real() << "+i" << eigensolver.eigenvalues()[i].imag() << endl; | 21:21 |
lisitsyn | (11:20:56 PM) lisitsyn: the output is | 21:21 |
lisitsyn | (11:20:57 PM) lisitsyn: eigenvalue (0.666667,0)=3.22526e-319+i8.25667e-317 | 21:21 |
lisitsyn | the question is what the fuck :D | 21:22 |
lisitsyn | may be you have seen something like that? | 21:22 |
lambday_ | holy shit! | 21:22 |
lambday_ | :-o | 21:22 |
@sonney2k | lisitsyn, valgrind? | 21:22 |
lambday_ | lisitsyn: try taking eigensolver.eigenvalues() into a VectorXcd and then print may be? | 21:23 |
lisitsyn | lambday_: it is okay in vector | 21:23 |
lisitsyn | lambday_: shit happens on real() | 21:23 |
lisitsyn | sonney2k: checking | 21:23 |
lambday_ | lisitsyn: try real(val) instead of val.real() | 21:23 |
-!- gsomix_ [~gsomix@109.169.187.164] has quit [Ping timeout: 240 seconds] | 21:23 | |
lambday_ | lisitsyn: I don't remember exactly but there was some stuff in c++11 | 21:24 |
-!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has quit [Ping timeout: 250 seconds] | 21:24 | |
@sonney2k | lambday, but that shouldn't even compile | 21:24 |
lisitsyn | sonney2k: why that? | 21:24 |
lambday_ | sonney2k: that thing that we were doing for c++11, remember? | 21:25 |
lambday_ | :-/ | 21:25 |
@sonney2k | the complex.real()=x | 21:25 |
lambday_ | I am not sure! | 21:25 |
@sonney2k | yeah | 21:25 |
@sonney2k | lisitsyn, it just didn't here | 21:25 |
@sonney2k | only complex.real(x) did compile | 21:25 |
lisitsyn | sonney2k: what's the reason? | 21:25 |
lisitsyn | real(x)? | 21:25 |
lisitsyn | what does it do? | 21:25 |
lisitsyn | puts real to x? | 21:25 |
@sonney2k | lisitsyn, set the real part of a complex | 21:25 |
@sonney2k | yes | 21:25 |
@sonney2k | no x -> to real of the complex number | 21:25 |
@sonney2k | named complex | 21:25 |
lisitsyn | okay | 21:26 |
lisitsyn | how to get real part then | 21:26 |
@sonney2k | lisitsyn, same thing w/o args | 21:26 |
@sonney2k | should be correct | 21:26 |
lisitsyn | that's what I was doing | 21:26 |
lisitsyn | std::complex<MatrixXd::Scalar> eigenvalue = eigensolver.eigenvalues()[i]; | 21:29 |
lisitsyn | cout << "eigenvalue " << eigenvalue << "=" << std::real(eigenvalue) << "+i" << std::imag(eigenvalue) << endl; | 21:29 |
lisitsyn | the output is | 21:29 |
@sonney2k | lisitsyn, that is why I rather think it is some memory corruption | 21:29 |
lisitsyn | eigenvalue (0.666667,0)=3.22526e-319+i7.06327e-304 | 21:29 |
@sonney2k | lisitsyn, errm why not eigenvalue.real() ? | 21:29 |
lisitsyn | sonney2k: tried both | 21:30 |
@sonney2k | same result? | 21:30 |
lisitsyn | sonney2k: yes uninitialized memory | 21:30 |
lambday_ | lisitsyn: that's weird! | 21:31 |
@sonney2k | lisitsyn, ok so try a minimal unit test to figure out why that is happening | 21:32 |
lisitsyn | yeah.. | 21:32 |
lambday_ | lisitsyn: shouldn't it be MatrixXcd?? | 21:33 |
lambday_ | oh no sorry | 21:33 |
lisitsyn | lambday_: no that's just double | 21:33 |
lambday_ | yeah sorry I misunderstood | 21:33 |
@sonney2k | iglesiasg, btw you found a nice one - bug in threading code. it works ok with pthreads disabled | 21:34 |
lambday_ | lisitsyn: eigensolver is an instance of SelfAdjointEigenSolver? | 21:38 |
lisitsyn | lambday_: no, it is basic eigensolver | 21:39 |
lisitsyn | I can't assume it is selfadj | 21:39 |
lambday_ | okay | 21:39 |
lisitsyn | lambday_: lets see if I reproduce that in a minimal script | 21:39 |
lambday_ | alright | 21:40 |
gsomix__ | sonney2k, hello. if you have a minute. https://github.com/shogun-toolbox/shogun/pull/1356 | 21:41 |
-!- gsomix__ is now known as gsomix | 21:43 | |
-!- zxtx [~zv@c-98-207-60-193.hsd1.ca.comcast.net] has quit [Ping timeout: 240 seconds] | 21:43 | |
lambday_ | lisitsyn: in my machine its working! :-/ for random matrix and i=0 | 21:46 |
lisitsyn | lambday_: yes same for me | 21:46 |
lisitsyn | I have no idea :D | 21:49 |
shogun-notifier- | shogun: Evgeniy Andreev :develop * 1c3bc6f / src/shogun/io/CSVFile.cpp,tests/unit/io/CSVFile_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/1c3bc6f90bde2f5446fc672e4986d57c1a772de6 | 21:49 |
shogun-notifier- | shogun: changed behavior of get_vector in CSVFile | 21:49 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * b56da3c / src/shogun/io/CSVFile.cpp,tests/unit/io/CSVFile_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/b56da3cd29ca0ede02b1cc8b660ea219b72a12b3 | 21:49 |
shogun-notifier- | shogun: Merge pull request #1356 from gsomix/feature/get_vector | 21:49 |
shogun-notifier- | shogun: | 21:49 |
shogun-notifier- | shogun: Changed behavior of get_vector in CSVFile | 21:49 |
lisitsyn | I am checking if it is really related to C++11 | 21:49 |
lisitsyn | but I'd guess because it appeared once sonney2k added it | 21:49 |
lambday_ | lisitsyn: where does it fail? | 21:50 |
lambday_ | argh in 8 mins my net connection is gonna get cut off! | 21:52 |
shogun-buildbot | build #1647 of deb1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1647 | 21:53 |
shogun-buildbot | build #836 of rpm1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/836 blamelist: Evgeniy Andreev <gsomix@gmail.com> | 21:56 |
-!- lambday_ [67157d37@gateway/web/freenode/ip.103.21.125.55] has quit [Ping timeout: 250 seconds] | 22:03 | |
shogun-buildbot | build #1154 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1154 blamelist: Evgeniy Andreev <gsomix@gmail.com> | 22:05 |
shogun-buildbot | build #1461 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1461 blamelist: Evgeniy Andreev <gsomix@gmail.com> | 22:34 |
shogun-buildbot | build #1648 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1648 blamelist: Soeren Sonnenburg <sonne@debian.org> | 22:41 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 56b4f34 / src/shogun/kernel/Kernel.cpp,src/shogun/kernel/Kernel.h: https://github.com/shogun-toolbox/shogun/commit/56b4f342dafe3857a7102cdcf39617b8760cd4e7 | 22:43 |
shogun-notifier- | shogun: fix formatting / doc | 22:43 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * d474227 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/d474227dac6e804fd8376877237f5b6bf49af67a | 22:52 |
shogun-notifier- | shogun: compute distance matrix the same way kernel matrix is computed | 22:52 |
@sonney2k | iglesiasg, ^ that should fix it | 22:52 |
@sonney2k | iglesiasg, any idea for a proper name for a base class of CKernel and CDistance? | 22:52 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 240 seconds] | 22:55 | |
shogun-buildbot | build #837 of rpm1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/837 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:02 |
-!- zxtx [~zv@199-188-193-9.PUBLIC.monkeybrains.net] has joined #shogun | 23:03 | |
shogun-buildbot | build #1155 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1155 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:12 |
lisitsyn | sonney2k: I am disabling it with C++11 for now :D | 23:13 |
@sonney2k | lisitsyn, errm? so you have a small test case now? | 23:14 |
lisitsyn | sonney2k: no | 23:14 |
lisitsyn | I can't reproduce it | 23:14 |
@sonney2k | lisitsyn, so it only fails on travis? | 23:14 |
lisitsyn | sonney2k: it fails on my machine too | 23:15 |
-!- travis-ci [~travis-ci@ec2-184-73-123-75.compute-1.amazonaws.com] has joined #shogun | 23:15 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg'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/9838977 | 23:15 |
-!- travis-ci [~travis-ci@ec2-184-73-123-75.compute-1.amazonaws.com] has left #shogun [] | 23:15 | |
lisitsyn | but I can't reproduce it separately | 23:15 |
lisitsyn | just in this case and with C++11 | 23:15 |
lisitsyn | without C++11 it works ok | 23:15 |
-!- zxtx [~zv@199-188-193-9.PUBLIC.monkeybrains.net] has quit [Ping timeout: 240 seconds] | 23:15 | |
@sonney2k | lisitsyn, so it fails on your machine with c++11 too ok | 23:16 |
lisitsyn | sonney2k: it seems for me it is something quite complex with eigen3 and C++11 | 23:16 |
lisitsyn | I'll try to debug it again later | 23:17 |
lisitsyn | but for now I'll just disable it for C++11 | 23:17 |
@sonney2k | lisitsyn, and did you use g++ or clang++ ? | 23:17 |
lisitsyn | sonney2k: both | 23:17 |
@sonney2k | and it fails with both? | 23:17 |
lisitsyn | yes | 23:18 |
shogun-notifier- | shogun: Sergey Lisitsyn :develop * ba2bb88 / src/shogun/ (2 files): https://github.com/shogun-toolbox/shogun/commit/ba2bb88463c013f78bfbaf334ea31d0a61004609 | 23:19 |
shogun-notifier- | shogun: Disabled malsar clustered LR for C++11 due to wrong behaviour | 23:19 |
@sonney2k | lisitsyn, but then you could probably create a small code snippet that shows this failure? | 23:20 |
shogun-buildbot | build #1462 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1462 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:32 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 2301565 / src/shogun/ (3 files): https://github.com/shogun-toolbox/shogun/commit/2301565053670c63fc1ece2e63a95d6cba2c0489 | 23:33 |
shogun-notifier- | shogun: ifdef lapack in inference method - since gaussian is only available with it | 23:33 |
shogun-buildbot | build #1651 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1651 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 23:35 |
-!- travis-ci [~travis-ci@ec2-184-73-123-75.compute-1.amazonaws.com] has joined #shogun | 23:35 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg'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/9839311 | 23:35 |
-!- travis-ci [~travis-ci@ec2-184-73-123-75.compute-1.amazonaws.com] has left #shogun [] | 23:35 | |
shogun-buildbot | build #1649 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1649 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:35 |
shogun-buildbot | build #1652 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1652 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:38 |
lisitsyn | argh | 23:40 |
lisitsyn | okay will fix in a minute | 23:41 |
lisitsyn | sonney2k: I failed to create such a snippet :/ | 23:45 |
@sonney2k | lisitsyn, hmmhh | 23:45 |
@sonney2k | lisitsyn, well sth not too big would at least be of interest to the eigen3 people | 23:46 |
@sonney2k | I mean it must be their bug to fix | 23:46 |
lisitsyn | sonney2k: yeah I'll try to minimize it | 23:46 |
-!- iglesiasg [~iglesias@83.179.44.135] has joined #shogun | 23:47 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 23:47 | |
@iglesiasg | sonney2k, oh cool | 23:47 |
@iglesiasg | sonney2k, how is possible though? It works fine when the number of threads is equal to 1 | 23:47 |
-!- travis-ci [~travis-ci@ec2-184-73-123-75.compute-1.amazonaws.com] has joined #shogun | 23:48 | |
travis-ci | [travis-ci] it's Sergey Lisitsyn'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/9839737 | 23:48 |
-!- travis-ci [~travis-ci@ec2-184-73-123-75.compute-1.amazonaws.com] has left #shogun [] | 23:48 | |
@iglesiasg | anyway, time to rest | 23:48 |
@iglesiasg | see you tom | 23:48 |
-!- iglesiasg [~iglesias@83.179.44.135] has quit [Client Quit] | 23:48 | |
shogun-notifier- | shogun: Sergey Lisitsyn :develop * 33d16a4 / src/shogun/transfer/multitask/MultitaskClusteredLogisticRegression.cpp: https://github.com/shogun-toolbox/shogun/commit/33d16a4f3fbd6d55b7831c05b96cb25a55c499ec | 23:55 |
shogun-notifier- | shogun: Really disabled clustered LR with C++11 | 23:55 |
--- Log closed Mon Aug 05 00:00:01 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!