IRC logs of #shogun for Tuesday, 2016-03-29

« Monday, 2016-03-28 Index Wednesday, 2016-03-30 »
--- Log opened Tue Mar 29 00:00:50 2016
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun00:16
-!- mode/#shogun [+o HeikoS] by ChanServ00:16
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun00:17
shogun-notifier-shogun: Sanuj :develop * 093ce46 / examples/meta/generator/targets/ruby.json: https://github.com/shogun-toolbox/shogun/commit/093ce46ca49dc807a785a435993419db0256f77500:17
shogun-notifier-shogun: fix enum rule for ruby00:17
shogun-notifier-shogun: Heiko Strathmann :develop * 30bbfc1 / examples/meta/generator/targets/ruby.json: https://github.com/shogun-toolbox/shogun/commit/30bbfc17eff595324334073f8d91fd27733f570b00:17
shogun-notifier-shogun: Merge pull request #3122 from sanuj/ruby_enum00:17
shogun-notifier-shogun:00:17
shogun-notifier-shogun: fix enum rule for ruby00:17
shogun-buildbotbuild #2823 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2823  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>00:20
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit]00:21
shogun-buildbotbuild #173 of FC22 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC22%20-%20libshogun/builds/173  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>00:25
-!- BurningBear [~BurningBe@121.237.190.49] has joined #shogun01:56
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]03:17
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun03:56
shogun-buildbotbuild #626 of debian wheezy - memcheck is complete: Failure [failed memory check]  Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/626  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>04:02
-!- c4goldsw [8a33fc16@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.252.22] has joined #shogun04:59
c4goldswAnyone here?  I need some quick advice on fixing a build error I'm seeing for the modular interfaces after I refactored some code.05:00
shogun-buildbotbuild #1109 of nightly_default is complete: Failure [failed notebooks]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/1109  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>05:12
-!- c4goldsw [8a33fc16@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.252.22] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]05:35
-!- BurningBear [~BurningBe@121.237.190.49] has quit [Ping timeout: 244 seconds]05:51
-!- sanuj [0e8bc409@gateway/web/freenode/ip.14.139.196.9] has joined #shogun06:08
-!- BurningBear [~BurningBe@223.3.106.36] has joined #shogun07:53
-!- sanuj [0e8bc409@gateway/web/freenode/ip.14.139.196.9] has quit [Ping timeout: 250 seconds]08:26
-!- besser82 [~besser82@fedora/besser82] has joined #shogun08:48
-!- mode/#shogun [+o besser82] by ChanServ08:48
-!- BurningBear [~BurningBe@223.3.106.36] has quit [Ping timeout: 252 seconds]09:40
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Ping timeout: 268 seconds]09:53
-!- BurningBear [~BurningBe@121.237.190.49] has joined #shogun10:25
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun12:29
-!- mode/#shogun [+o HeikoS] by ChanServ12:29
-!- BurningBear [~BurningBe@121.237.190.49] has quit [Ping timeout: 252 seconds]12:30
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]12:43
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun12:49
-!- mode/#shogun [+o HeikoS] by ChanServ12:49
-!- sanuj [0e8bc409@gateway/web/freenode/ip.14.139.196.9] has joined #shogun13:05
sanujHeikoS: i was wondering if we can use nvidia's modergpu2 for cuda backend in linalg13:12
sanujHeikoS: https://github.com/nvlabs/moderngpu13:12
@HeikoSsanuj: maybe at some point13:12
@HeikoSsanuj: I would already be happy if we had a simple GPU speedup somewhere13:12
@HeikoSthat is some algorithm that runs faster if a GPU is present13:13
@HeikoSthat is the first step13:13
@HeikoSand then shogun wide is second step13:13
@HeikoSthen we can think about tweaking backends13:13
sanuji see13:13
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]13:19
@wikingsanuj: why viennacl is not good enough?13:48
sanujwiking: viennacl is fine, lambday was talking about a separate CUDA backend in future13:49
@wikingi mean it has cuda backend support13:49
@wikingwhy would that make any sense13:49
@wikingi mean in the first place we've started to use viennacl13:49
@wikingbecause of the fact13:49
@wikingthat it should seemlessly wrap the gpu backend13:49
sanujwiking: https://github.com/shogun-toolbox/shogun/issues/310013:50
sanujwiking: i dunno much about viennacl so i can't comment13:50
@wikingok so why do we start adding more shit13:50
@wikingif we dont know what is the initial thing?13:50
@wikingi mean lets stop doing this 'add everything that is YOLO'13:51
@wikingand then things will be better attitude13:51
lisitsynYOLO13:51
sanujhaha13:51
sanujokay13:51
@wikingfucking yolo solutions13:55
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun14:02
-!- mode/#shogun [+o HeikoS] by ChanServ14:02
-!- sanuj [0e8bc409@gateway/web/freenode/ip.14.139.196.9] has quit [Ping timeout: 250 seconds]14:02
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit]14:05
-!- c4goldsw [8a337a05@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.122.5] has joined #shogun14:15
-!- c4goldsw [8a337a05@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.122.5] has quit [Client Quit]14:18
-!- c4goldsw [8a337a05@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.122.5] has joined #shogun14:19
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun14:28
-!- c4goldsw [8a337a05@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.122.5] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]14:49
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Ping timeout: 250 seconds]14:58
-!- HeikoS [~heiko@host81-152-92-78.range81-152.btcentralplus.com] has joined #shogun15:33
-!- mode/#shogun [+o HeikoS] by ChanServ15:33
shogun-buildbotbuild #137 of deb1 - libshogun - PR is complete: Failure [failed git]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/137  blamelist: ialong15:34
shogun-buildbotbuild #138 of deb1 - libshogun - PR is complete: Failure [failed git]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/138  blamelist: ialong15:40
-!- HeikoS [~heiko@host81-152-92-78.range81-152.btcentralplus.com] has quit [Quit: Leaving.]16:13
-!- HeikoS [~heiko@host81-152-92-78.range81-152.btcentralplus.com] has joined #shogun16:15
-!- mode/#shogun [+o HeikoS] by ChanServ16:15
@wikingHeikoS: yo16:17
-!- c4goldsw [8a337d98@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.125.152] has joined #shogun16:17
-!- c4goldsw [8a337d98@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.125.152] has quit [Client Quit]16:22
@HeikoSwiking: jjo16:36
@HeikoSwiking: we are almost done with the alglib code removal, which should be the most severe GPL code we had16:42
-!- c4goldsw [8a3377ce@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.119.206] has joined #shogun16:43
-!- c4goldsw [8a3377ce@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.119.206] has quit [Client Quit]16:44
-!- lambday [6a33853d@gateway/web/freenode/ip.106.51.133.61] has joined #shogun16:46
-!- mode/#shogun [+o lambday] by ChanServ16:46
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun17:06
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun17:14
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 0c84e7e / src/interfaces/modular/Preprocessor.i: https://github.com/shogun-toolbox/shogun/commit/0c84e7e3157c888cc9e4da951dd910d75f5e259317:14
shogun-notifier-shogun: temporarily comment out missing files17:14
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 4436aeb / src/interfaces/modular/Preprocessor_includes.i: https://github.com/shogun-toolbox/shogun/commit/4436aeb7c2bb7c80d3be39830f6f6a2e01f7824e17:15
shogun-notifier-shogun: remporarily remove missing files17:15
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 6ef795b / src/interfaces/modular/Preprocessor.i: https://github.com/shogun-toolbox/shogun/commit/6ef795ba6e83706b903dc84a43a36e496597c39417:21
shogun-notifier-shogun: remove more missing files17:21
-!- c4goldsw [8a33f1e4@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.241.228] has joined #shogun17:25
c4goldswHey, is anyone around right now?17:27
c4goldswI need some guidance on fixing a build error I'm receiving.17:29
shogun-buildbotbuild #139 of deb1 - libshogun - PR is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/13917:37
c4goldswSpecifically, I had four modular builds fail due to some code I refactored (an example: https://travis-ci.org/shogun-toolbox/shogun/jobs/118924063 ) - it would appears is if they all failed as I had moved CStatistics::variance into linalg::variance.  The error causing these builds to fail is "undefined symbol: _ZN6shogun11CStatistics8varianceENS_8S17:38
c4goldswGVectorIdEE".  I've replaced all usages to CStatistics::variance, so I'm not sure what the problem is.  Does anyone have any insights?17:38
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 268 seconds]17:39
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Ping timeout: 244 seconds]17:39
-!- travis-ci [~travis-ci@ec2-54-205-57-183.compute-1.amazonaws.com] has joined #shogun17:40
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11927311817:40
-!- travis-ci [~travis-ci@ec2-54-205-57-183.compute-1.amazonaws.com] has left #shogun []17:40
-!- OXPHOS [8ca3fe9e@gateway/web/freenode/ip.140.163.254.158] has joined #shogun17:41
-!- besser82 [~besser82@fedora/besser82] has joined #shogun17:41
-!- mode/#shogun [+o besser82] by ChanServ17:42
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun17:43
shogun-notifier-shogun: Esben Sorig :develop * 20db6ca / examples/meta/generator/targets/ruby.json: https://github.com/shogun-toolbox/shogun/commit/20db6cab059c1bfa44118c5db03d5ba9d9041e3c17:48
shogun-notifier-shogun: Fix ruby enum rule17:48
shogun-notifier-shogun: Heiko Strathmann :develop * 6d5e0bc / examples/meta/generator/targets/ruby.json: https://github.com/shogun-toolbox/shogun/commit/6d5e0bcaea5680bd7c4a81a8274295b9a53f271717:48
shogun-notifier-shogun: Merge pull request #3126 from sorig/enums17:48
shogun-notifier-shogun:17:48
shogun-notifier-shogun: Meta-Language: Fix ruby enum rule17:48
OXPHOSHello somebody! May I ask what IDE/editor do you recommend for Mac?17:49
@HeikoSOXPHOS: no mac user here :)17:50
@HeikoSOXPHOS: I guess you will get VIM as reply ;)17:50
c4goldswHeikoS: Have a moment?17:50
@HeikoSc4goldsw: yes, shoot :)17:50
c4goldswGreat.  I saw your message on my PR, I'm breaking things up.17:51
@HeikoSc4goldsw: that is good, these big ones are impossible to review17:51
shogun-buildbotbuild #2824 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2824  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>17:51
c4goldswStill, I have build erros I don't understand. I had four modular builds fail due to some code I refactored (an example: https://travis-ci.org/shogun-toolbox/shogun/jobs/118924063 ) - it would appears is if they all failed as I had moved CStatistics::variance into linalg::variance.  The error causing these builds to fail is "undefined symbol: _ZN6sh17:51
c4goldswogun11CStatistics8varianceENS_8SGVectorIdEE".  I've replaced all usages to CStatistics::variance, so I'm not sure what the problem is.  Does anyone have any insights?17:51
c4goldsw(copy pasted sorry)17:51
@HeikoSlambday: btw do we have not unit tests for mmd stuff?17:53
@lambdayHeikoS: not yet..17:53
@HeikoSc4goldsw: I cant really help with this little information17:54
c4goldswNot even with looking at the build report?17:54
@HeikoSI guess the best is to isolate things more17:54
c4goldswAlright, I'll try that then.17:54
@HeikoSchecking17:54
@HeikoSc4goldsw: this is caused if a method is only in header I think17:55
@HeikoSlinalg is not always available on travis17:55
@HeikoSyou have to guard things17:55
OXPHOSHeikoS: seriously..I am too noob I guess XD. I just want to find something that gives me tab when I hit tab. Other than VIM. I'll try some others. Thanks!17:55
c4goldswAh, ok, with #IFDEF ?17:56
shogun-buildbotbuild #174 of FC22 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC22%20-%20libshogun/builds/174  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>17:56
@HeikoSc4goldsw: yes17:58
c4goldswI'll get onto it yet.17:58
c4goldswHeikoS: Thanks.17:58
c4goldswOXPHOS: Eclipse works rather nicely if you ask me, with the CDT plug in (not to mention that you can also build CDT4 project files for Shogun :) )17:59
-!- c4goldsw [8a33f1e4@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.241.228] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]18:00
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 248 seconds]18:02
OXPHOSc4goldsw: Thanks! Will try.18:03
-!- travis-ci [~travis-ci@ec2-54-205-57-183.compute-1.amazonaws.com] has joined #shogun18:03
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11927339818:03
-!- travis-ci [~travis-ci@ec2-54-205-57-183.compute-1.amazonaws.com] has left #shogun []18:03
-!- HeikoS [~heiko@host81-152-92-78.range81-152.btcentralplus.com] has quit [Ping timeout: 244 seconds]18:09
-!- c4goldsw [8a33f1e4@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.241.228] has joined #shogun18:14
-!- travis-ci [~travis-ci@ec2-107-20-97-135.compute-1.amazonaws.com] has joined #shogun18:23
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11927503518:23
-!- travis-ci [~travis-ci@ec2-107-20-97-135.compute-1.amazonaws.com] has left #shogun []18:23
-!- c4goldsw [8a33f1e4@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.241.228] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]18:45
-!- OXPHOS [8ca3fe9e@gateway/web/freenode/ip.140.163.254.158] has quit [Quit: Page closed]18:46
-!- travis-ci [~travis-ci@ec2-54-167-70-137.compute-1.amazonaws.com] has joined #shogun18:53
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11928212818:53
-!- travis-ci [~travis-ci@ec2-54-167-70-137.compute-1.amazonaws.com] has left #shogun []18:53
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun18:55
-!- mode/#shogun [+o HeikoS] by ChanServ18:55
shogun-notifier-shogun: Esben Sorig :develop * 25b60c1 / examples/meta/generator/targets/lua.json: https://github.com/shogun-toolbox/shogun/commit/25b60c1825435c4fdf91b71d13326ac89f5c3d6219:00
shogun-notifier-shogun: Meta-Language: Fix Lua enum rule19:00
shogun-notifier-shogun: Heiko Strathmann :develop * 8eeda96 / examples/meta/generator/targets/lua.json: https://github.com/shogun-toolbox/shogun/commit/8eeda96169870ac93a4274b6dac8859927c6fd0719:00
shogun-notifier-shogun: Merge pull request #3127 from sorig/enums19:00
shogun-notifier-shogun:19:00
shogun-notifier-shogun: Meta-Language: Fix Lua enum rule19:00
@HeikoSlambday: how is cdflib going?19:08
@HeikoSlambday: I found a bug in mmd19:09
@lambdayHeikoS: added it.. trying with incomplete beta19:09
@lambdayHeikoS: what is the bug?19:09
@HeikoSlambday: something in permutation for null19:09
@HeikoSchecking19:09
-!- BurningBear [~BurningBe@121.237.186.161] has joined #shogun19:14
@HeikoSlambday: few bugs in fact ;)19:15
@lambdayHeikoS: what are those?19:16
@HeikoSlambday: I made a noteobook ... just re-running and sharing19:16
@HeikoSlambday: permutation I guess19:16
@HeikoSstatistical bug, not sure where it comes from in the code19:17
@lambdayHeikoS: okay.. must be quadratic-time mmd?19:17
@HeikoSyes19:17
@HeikoSlambday: but we can do the same checks with the others19:17
@HeikoSlambday: somehow these tests should be translated into unit tests actually ... but can do that later19:17
@HeikoSrandom unit tests are always annoying19:17
@lambdayHeikoS: yeah.. I'll add unit tests for these tests as well as the internal components.. I have quite a lot of them covered in flash.. gotta make those gtests19:18
@HeikoSlambday: yeah19:19
@HeikoSlambday: not sure if these "makes statistical sense" tests should be in unit19:19
@lambdayHeikoS: hehe I don't think they should be.. I'll have to take care of that via examples/notebook maybe19:20
@HeikoSlambday: yeah .... mmmh ok19:22
@HeikoSnotebooks are one option19:22
@HeikoSbut it kind of obscures them if all statistical tests are tested for their calibration19:22
@HeikoSlambday: it is a test, but not unit, more like functionality or so19:22
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun19:29
sanujHeikoS: build is passing19:30
sanujhttps://github.com/shogun-toolbox/shogun/pull/311219:30
shogun-notifier-shogun: Sanuj :develop * ef12e6b / / (8 files): https://github.com/shogun-toolbox/shogun/commit/ef12e6b1dadbb59ad08aecc431a53f351513f7f019:38
shogun-notifier-shogun: add cookbook page for linear svm19:38
shogun-notifier-shogun: Heiko Strathmann :develop * 85e677b / / (8 files): https://github.com/shogun-toolbox/shogun/commit/85e677b6ad9e57e12d7f08a26ce3908a8270221c19:38
shogun-notifier-shogun: Merge pull request #3112 from sanuj/cookbook19:38
shogun-notifier-shogun:19:38
shogun-notifier-shogun: add cookbook page for linear svm19:38
@HeikoSsanuj: nice one :)19:40
@HeikoSsanuj: did you have any further look at lisitsyn plugin stuff?19:40
sanujyes19:41
sanuji talked to besser8219:41
sanujhe told me to implement tags and string based parameters and check if it works with SWIG19:41
sanujso i'll start working on that19:42
sanujsince besser82 is working on dynaplugz19:42
@HeikoSsanuj: ah yes a prototype for swig would be good19:42
@HeikoSgood idea19:42
@HeikoSlambday: https://gist.github.com/52b1a14114d00eb545ab52796176646519:43
@HeikoSlambday: have a minute to talk about this?19:43
@lambdayHeikoS: yeah let me check19:43
@lambdaychecking your comments19:44
@HeikoSlambday: check the last two cells19:45
@HeikoSlast one first19:45
@HeikoSthere I have the case where H0 is true, check how I generate data19:45
lisitsynyes please!19:45
@HeikoSlisitsyn: pleaseplease19:46
@HeikoSlambday: so if H0 is true, then whenever I compute the statistic, it is distributed according to null19:46
@HeikoSlambday: if I permute, this doesnt change anything19:46
sanujHeikoS: besser82 told me to wait till the loading and plugin interface-part is finished in dynaplugz19:46
sanujthen i can add tags and string parameters to that19:47
@HeikoSsanuj: ok then19:47
@lambdayHeikoS: checking.. let me check whether we have to set simulate_null=true here19:47
@HeikoSsanuj: any ETA?19:47
@HeikoSlambday: no wait, doesnt matter19:47
@HeikoSpermuting or not doesnt change things19:47
@HeikoSas p and q have the same distribution already19:47
@lambdayyeah19:48
@HeikoSso I coimpute type1 error here19:48
sanujHeikoS: i don't know how much time will it take to complete19:48
@HeikoSthat is: rejected even though H0 is true19:48
sanuji'll start from tonight19:48
@HeikoSwhich should happen exactly alpha times19:49
@HeikoSalpha=0.05 here19:49
@HeikoSthis is just from the way the threshold is constructed: the 1-alpha quantile of the null19:49
@HeikoSso by construction, the statistic should lie above that threshold only 5% of the time19:49
@HeikoSbut the rejection rate is way higher19:49
@HeikoSthis case is the most simple one19:49
@lambdayyeah I see19:50
@HeikoSlambday: If you want, I can use mmd.compute_statistic to show you what I mean19:50
@HeikoSlambday: this is even true if there is a bug in compute_statistic19:50
@HeikoSany random statistic has to full fill this if the permutation works correctly19:50
@HeikoSlambday: ok this is problem 119:52
@HeikoSlambday: problem2 is related I think19:52
@HeikoSlambday: if H0 is true, then the distribution of p-values is uniform by construction19:52
@HeikoSlambday: in the second last cell, I compute p-values many times, and they are not uniform19:52
@HeikoSsimilar to the other case, this is independent of the statistic. *any* random statistic with permutation test has to satisfy this19:53
@HeikoSsame thinking19:53
@HeikoSwhether you permute or not, the statistic is computed the same way19:53
@lambdayHeikoS: yeah true.. trying to figure out where the bug might be19:54
@HeikoSlambday: the other comments are more related to performance19:54
@HeikoSlambday: it has to be in the permutation and/or threshold code19:54
-!- travis-ci [~travis-ci@ec2-107-20-97-135.compute-1.amazonaws.com] has joined #shogun20:05
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11930124020:05
-!- travis-ci [~travis-ci@ec2-107-20-97-135.compute-1.amazonaws.com] has left #shogun []20:05
@HeikoSlambday: https://gist.github.com/karlnapf/77a56528a7dbcb76498a3c16bc06593520:08
@lambdayHeikoS: rejections[i] = mmd.perform_test(), this should be rejections[i] = mmd.perform_test(alpha) I think20:08
@HeikoShere is a cell that computes p-value by hand -- this should give the correct result ... just running it20:08
@HeikoSouch right20:08
@HeikoSlambday: that might have been it for the last one :)20:09
@HeikoSlambday: ah man, that is why this thing has to be removed -- kind of type error: returning p-value or boolean...sorry if that was the problem20:09
@HeikoSlambday: maybe the others can be "fixed" that way too? :)20:10
@lambdayHeikoS: dunno :( checking.. what about mmd.compute_p_value(0.05) ? compute_p_value takes the statistic, is 0.05 the alpha here?20:10
@lambdayHeikoS: try this one.. just perform_test() for the 2nd last cell.. that will give the p-values20:11
@HeikoSyep trying20:11
@lambdayHeikoS: this is happening due to lack of documentation.. sorry about that :(20:12
@HeikoSlambday: no definitely my fault20:12
@HeikoSbut also reveals some interface issues I guess ;)20:12
@lambdayyeah definitely20:12
@HeikoSok re-running20:13
@lambdayHeikoS: regarding performance, I'll have to optimize it for permutation test and make use of the precomputed kernel matrix.. I'll add that20:15
@HeikoSlambday: ah not yet precomputing it?20:15
@lambdayHeikoS: no I did that only for stat-var computation.. not over multiple calls to compute_statistic...20:16
@HeikoSlambday: ok20:17
@lambdaynice that you pointed it out.. I'll add that support20:17
@lambdayHeikoS: really curious about whether the results make sense now20:17
@HeikoSyes same20:17
@HeikoS*waiting*20:18
-!- BurningBear [~BurningBe@121.237.186.161] has quit [Ping timeout: 248 seconds]20:18
-!- BurningBear [~BurningBe@121.237.186.161] has joined #shogun20:19
@HeikoSlove the multicore!20:19
@lambdayHeikoS: for quadratic time mmd, it's useless :D for streaming b-test/linear-time, we'll have to see whether it gives any boost20:19
@HeikoSlambday: for quadratic time the kernel matrix can be computed in parallel no?20:20
@lambdayHeikoS: well, CKernel::kernel is anyway pthreaded.. maybe for kernel selection it may make some difference20:21
@HeikoSyes, thats why I asked whether you precompute the kernel matrix?20:21
@HeikoSlambday: ok p-values are uniform now20:21
@HeikoSphew20:21
@HeikoSlambday: when you do the permutation test, and kernel  matrix is precomputed20:22
sanujlambday: viennacl 1.7.1 works with shogun right?20:22
@HeikoSthis just corresponds to summing things in a different way right?20:22
@HeikoSno additional matrix is computed or stored?20:22
@HeikoSlambday: type1 error now also is fine!20:23
@HeikoSgood20:23
-!- lambday [6a33853d@gateway/web/freenode/ip.106.51.133.61] has quit [Ping timeout: 250 seconds]20:25
-!- travis-ci [~travis-ci@ec2-54-205-57-183.compute-1.amazonaws.com] has joined #shogun20:37
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11931068920:37
-!- travis-ci [~travis-ci@ec2-54-205-57-183.compute-1.amazonaws.com] has left #shogun []20:37
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Quit: Page closed]20:41
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]20:44
-!- BurningBear [~BurningBe@121.237.186.161] has quit [Ping timeout: 244 seconds]20:46
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun21:36
-!- mode/#shogun [+o HeikoS] by ChanServ21:37
-!- lambday [6a338661@gateway/web/freenode/ip.106.51.134.97] has joined #shogun21:47
-!- mode/#shogun [+o lambday] by ChanServ21:47
@HeikoSlambday: pooooooower :)21:55
@lambdayHeikoS: hahaha21:56
@lambdayHeikoS: it's 01:25 here, and now I have to cook dinner xD21:56
@HeikoSlambday: ah man21:56
shogun-buildbotbuild #627 of debian wheezy - memcheck is complete: Failure [failed memory check]  Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/627  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>21:56
@HeikoSlambday: so how can I help best with things?21:56
@HeikoSlambday: btw one thing21:57
@lambdayHeikoS: you're already helping a lot man.. checking the sanity of the framework and suggesting api changes :)21:57
@HeikoSlambday: let's try to fix the travis build for the bigtest branch21:57
@HeikoSso that I can add examples for everything we want to be possible21:57
@HeikoSand then we can check against that on the fly21:57
@lambdayHeikoS: alright.. yeah that can be fixed if we #ifdef cxx11 the whole thing :D21:57
@HeikoSlambday: do it then :)21:58
@lambdayalrighty21:58
@lambdayHeikoS: I'll do some clean-up for bigtest today (if Bangalore state electricity board allows me, that is)..21:59
@lambdayHeikoS: BTW sending a PR for cdflib.. have a look22:00
shogun-buildbotbuild #2825 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2825  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>22:00
@HeikoSlambday: cool!22:00
shogun-buildbotbuild #175 of FC22 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC22%20-%20libshogun/builds/175  blamelist: Esben Sorig <esben@sorig.eu>, Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>22:06
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]22:12
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun22:16
-!- mode/#shogun [+o HeikoS] by ChanServ22:16
@HeikoSlisitsyn: hi22:21
lisitsynhey22:21
@HeikoSlisitsyn: all good?22:21
lisitsynHeikoS: yeah22:21
@HeikoSlisitsyn: there is a problem with the meta snippet parser22:21
lisitsynsphinx one?22:22
@HeikoSsomehow, the thing gets confused when the tag of the snippet appears in the code22:22
@HeikoSyes22:22
@HeikoSlisitsyn: know where you did the check?22:22
lisitsyngot example?22:22
@HeikoShttps://gist.github.com/karlnapf/b6746cf5a5dff068438b5fc33d49396422:23
@HeikoSlisitsyn: add this and then run make meta examples22:23
@HeikoSlisitsyn: check generated cpp22:23
lisitsynHeikoS: got syntax error22:29
lisitsynis that it?22:29
@HeikoSdoesnt compile22:29
@HeikoSthe cpp listing is incomplete22:29
lisitsynwhat's different?22:29
lisitsynit looks valid22:29
lisitsynHeikoS: it didn't generate the thing22:30
lisitsyn[100%] Generating examples from meta-language22:31
lisitsynSyntax error in input: ) on line 1422:31
lisitsynSyntax error in input: ) on line 1722:31
lisitsynSyntax error in input: ) on line 2022:31
lisitsynSyntax error in input: ) on line 2622:31
lisitsynSyntax error in input: ) on line 3222:31
lisitsynthat' what I get22:31
shogun-notifier-shogun: lambday :feature/alglib_removal * b651d92 / / (18 files): https://github.com/shogun-toolbox/shogun/commit/b651d92961061d740540e12b76af8dbcbb1cc52722:31
shogun-notifier-shogun: removed compilation errors, added libcdf22:31
shogun-notifier-shogun: Heiko Strathmann :feature/alglib_removal * b6191eb / / (18 files): https://github.com/shogun-toolbox/shogun/commit/b6191eb5d6ceb7c89d745b70b38f3e80621bb43b22:31
shogun-notifier-shogun: Merge pull request #3128 from lambday/feature/alglib_removal22:31
shogun-notifier-shogun:22:31
shogun-notifier-shogun: removed compilation errors, added libcdf22:31
@HeikoSah yes22:31
@HeikoSdamn didnt see that22:31
lisitsynHeikoS: but what's wrong?22:32
@HeikoSdont know22:32
lisitsynI am not sure I see that22:32
lisitsynlooks ok22:32
@HeikoSits the enum22:33
lisitsynI see22:33
@HeikoSlisitsyn: ok another problem22:35
@HeikoSlisitsyn: enums and ctags22:36
@HeikoSif the enum is not defined in the same header file as the class, it is not included22:36
lisitsynctags? how are they being used?22:36
@HeikoSlisitsyn: maybe I better check things properly before I ask22:37
@HeikoSlisitsyn:  ;)22:37
lisitsynHeikoS: did I miss something?22:37
@HeikoSlisitsyn: btw we have almost finished removing gpl codes from statistics22:37
lisitsynHeikoS: seen some commits22:37
@HeikoSlisitsyn: nope, just give me some time22:37
@HeikoSlisitsyn: do you know whether we have any other gpl codes?22:37
lisitsynuh22:38
lisitsynprobably22:38
lisitsynHeikoS: covertree22:38
lisitsynlets drop it is buggy anyway22:38
@HeikoSlisitsyn: can you do that in a feature branch?22:38
@HeikoScouldnt agree more if it is buggy22:39
lisitsynHeikoS: I am a bit overwhelmed with Q1 coming to the end22:39
lisitsyncan do later this week22:39
@HeikoSlisitsyn: I see22:40
@HeikoSok would be great to get this bsd release done before gsoc starts22:40
@HeikoSsince then we can use the c++1122:40
lisitsynHeikoS: got it22:40
@wikingmmmm22:40
@wikingmmmm22:40
@wikingmmmmmmmmmmmmmmmmmm22:40
@HeikoSwiking: heheh ;)22:41
@wikingHeikoS: we have a fast random backend22:41
@wikingshould we actually use that22:41
lisitsynyeahaa22:41
@wikingto generate various distribs?22:41
lisitsynt-sne is bsd-422:41
lisitsyn!22:41
lisitsynI thought it was gpl22:41
@wikingHeikoS: i mean it's a unified PRNG22:41
@HeikoSwiking: what do you mean?22:41
@HeikoSwiking: yeah up for using it22:41
@wikingit has a normal/gaussian generator22:42
@wiking*CRandom22:42
lisitsynHeikoS: SLEP is GPL22:42
@wikingthis is in response to the whole libcdf story22:42
@HeikoSwiking: ah I see, but the point of cdflib is different22:42
lisitsynoh22:42
@HeikoSwiking: we need the cdf functions22:42
@HeikoSthese are integrals22:43
lisitsynI'd have to drop my whole GSoC project :D22:43
@wikingmm yeah22:43
@HeikoSnot random numbers22:43
@HeikoSwiking: these are computed from low level stat functions, gamma etc, many of them not in closed form but have to to power/series expansions22:43
@wikingyeah22:43
@wikingbut since it follows a pdf22:43
@wikingright?22:43
@HeikoSwiking: cdflib is good since it is the exact same code as I would put in myself, but it is already used for 30 years22:44
@wikingi mean a random variable22:44
@HeikoSwiking: no randomness22:44
@HeikoSlisitsyn: any thoughts about SLEP`22:44
@HeikoS?22:44
@wikingbut i mean you use a distro for that no?22:45
lisitsynHeikoS: hard thing22:45
@HeikoSlisitsyn: we can maybe have a -DDROP_NON_BSD_CODE22:45
lisitsynwell I don't mind dropping it22:45
@wikingotherwise the cdf would make no sense22:45
lisitsynmiserable chances someone to use it22:45
@HeikoSwiking: I dont get what you mean22:46
@wikingbtw whats the deal with svmlight22:46
@wiking?22:46
lisitsynit is non-free, right?22:46
@HeikoSyes22:46
@wikingwhat's that license is like?22:46
@HeikoSnon gpl even22:46
@wikingi mean i know we cannot have it gpl22:46
lisitsynno license22:46
@wikingbut why's that22:46
@HeikoSthe guy doesnt want to22:46
@HeikoSsonney2k: asked him22:46
@wikingnono what i mean is22:47
@wikingwhat is the licensing issue there22:47
@wikingis it because they want it opensource22:47
lisitsynhe like begged him as far as I know22:47
lisitsyn:D22:47
lisitsynno free license still22:47
@wikingbut they want to keep it to themselves?22:47
@HeikoSyes22:47
@wikingso you can use the code but then again anything than academic22:47
lisitsynI don't get the point22:47
@wikingyou shoulndt use22:47
@wiking?22:47
lisitsynyes probably that22:47
@wikingbecause they could come after you and take all your bases?22:48
lisitsynraise NoCommentsException()22:48
@wikingsift (a img feature transform) is similar22:48
@wikingfree academics22:48
@wiking$$ otherwise22:48
lisitsynthankfully deep learning is not taken yet22:48
@wikinganyhow22:48
@wikingi guess we'll have then a nonBSD flag22:49
@wikinganyways22:49
@wiking(that would cover for sure svmlight)22:49
lisitsynyeah22:49
@wikingso if we are tehre22:49
@wikingwe can add some other stuff there22:49
@wikingand that's it22:49
lisitsynthen I don't have to drop my old code22:49
lisitsyn:D22:49
@wikingindeed22:49
@wikingjust flag it nonbsd22:49
@wikingand done22:49
@HeikoSwiking: +122:49
@HeikoSso all we need to do is to put a guard for those right?22:50
@wikingbasically22:50
@wikingjust transform22:50
@wikingthe GPLv3 or SVMLIGHT flag22:50
@wikingto nonbsd flag22:50
@wikingand add anything that cannot be bsd-ed22:50
@wikingUSE_SVMLIGHT22:50
@wikingthat should be22:51
@wikingNON_BSD or something22:51
@HeikoSwiking: cool, I can start a branch with that22:51
@HeikoSwiking: will do once the cdflib stuff is in22:51
@HeikoSwiking: since I already started, and its good to remove this alglib sh*** anyways22:52
lisitsynHeikoS: wiking made you rude22:52
@wiking;D22:53
@HeikoSlisitsyn: in this case, that word is very appropriate, no need for wiking to make me  use that :D22:53
lisitsynis that so bad?22:53
@wikingi think we are just following the footsteps of Linus Torvalds22:53
@wiking:))))))))))))))))22:53
lisitsynmah this guy is not even from moscow22:53
lisitsynpfff22:54
lisitsyn:D22:54
@wikingThe fact is, people need to know what my position on things are. And I can't just say "please don't do that", because people won't listen. I say "On the internet, nobody can hear you being subtle," and I mean it22:54
@wiking:)22:54
@wiking    Because if you want me to "act professional," I can tell you that I'm not interested. I'm sitting in my home office wearing a bathrobe. The same way I'm not going to start wearing ties, I'm *also* not going to buy into the fake politeness, the lying, the office politics and backstabbing, the passive aggressiveness, and the buzzwords. Because THAT is what "acting professionally" results in: people resort22:55
@HeikoShahahahaha22:55
@wiking to all kinds of really nasty things because they are forced to act out their normal urges in unnatural ways.22:55
@wikingand that's all i can say22:55
@wiking;)22:55
@wiking(c) Linus Torvalds22:55
@HeikoShahaha22:55
@HeikoSread the bio, its a funny read22:55
@HeikoSim gonna go to bed22:57
@HeikoSenjoy following linus! :)22:57
@wikinghehehe22:57
@wikingwell he has a point22:57
@wikinglook at linux .... where it came from22:57
@wikingand what it became22:57
@wiking(same goes for git22:57
shogun-notifier-shogun: lambday :feature/bigtest * 8aeaf74 / src/shogun/statistics/experimental/internals/ (34 files): https://github.com/shogun-toolbox/shogun/commit/8aeaf74dd0bbb9d7f8a296fad030f9499167448023:08
shogun-notifier-shogun: added first version of bigtesting framework23:08
shogun-notifier-shogun: lambday :feature/bigtest * b35ed96 / src/shogun/statistics/experimental/internals/ (5 files): https://github.com/shogun-toolbox/shogun/commit/b35ed96a3de784641ad604a3637286451690ddc223:08
shogun-notifier-shogun: documentation and code-style changes23:08
shogun-notifier-shogun: lambday :feature/bigtest * d427f01 / src/shogun/statistics/experimental/ (3 files): https://github.com/shogun-toolbox/shogun/commit/d427f0143ebb409754f5227a7112fdfe7055841723:08
shogun-notifier-shogun: added hypothesis test (experimental) base class23:08
shogun-notifier-shogun: lambday :feature/bigtest * 8115e62 / src/shogun/ (108 files): https://github.com/shogun-toolbox/shogun/commit/8115e62f364035561debfc018d8c41fe9c139b9623:08
shogun-notifier-shogun: removed old files23:08
shogun-notifier-shogun: lambday :feature/bigtest * 4584706 / src/shogun/hypothesistest/ (38 files): https://github.com/shogun-toolbox/shogun/commit/45847067ac91a8a428d4c45a62675fa3a41bd4ec23:08
shogun-notifier-shogun: added rest of the files23:08
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * d62994e / data,examples/meta/src/statistical_testing/quadratic_time_mmd.sg: https://github.com/shogun-toolbox/shogun/commit/d62994e4ee5a95d145106a1a7c63a70e13115e2923:08
shogun-notifier-shogun: first meta example for mmd23:08
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 05b6e61 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/05b6e61cfe633d6878de4cb6488615563078ce2b23:08
shogun-notifier-shogun: Add an initial cookbook page for quadratic time mmd23:08
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * fcee3d0 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/fcee3d07c93707b5a8a82c61b716bdcf3f864aac23:08
shogun-notifier-shogun: add linear time mmd api example draft23:08
shogun-notifier-shogun: lambday :feature/bigtest * cb75b45 / / (127 files): https://github.com/shogun-toolbox/shogun/commit/cb75b452dbe8e84c3dcd6c244d178d811a81048a23:08
shogun-notifier-shogun: fix build issues23:08
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]23:10
--- Log closed Wed Mar 30 00:00:52 2016
« Monday, 2016-03-28 Index Wednesday, 2016-03-30 »

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!