| --- Log opened Sat Jul 06 00:00:19 2013 | ||
| -!- sanyam [uid10602@gateway/web/irccloud.com/x-dagcmudckvzdcxma] has quit [Remote host closed the connection] | 00:43 | |
| -!- hushell [~hushell@c-24-21-141-32.hsd1.or.comcast.net] has joined #shogun | 00:49 | |
| -!- nube [~rho@49.244.14.60] has joined #shogun | 00:53 | |
| -!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 276 seconds] | 01:10 | |
| -!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 01:21 | |
| -!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 01:31 | |
| -!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Client Quit] | 01:31 | |
| -!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 01:31 | |
| -!- mode/#shogun [+o iglesiasg] by ChanServ | 01:31 | |
| @iglesiasg | have you guys seen this? http://jmlr.org/papers/volume14/curtin13a/curtin13a.pdf | 01:43 |
|---|---|---|
| @iglesiasg | I bet they didn't use our cover tree for the benchmark, but yes theirs ?_? | 01:43 |
| -!- FSCV [~FSCV@187.210.54.166] has joined #shogun | 02:03 | |
| -!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 02:40 | |
| -!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 02:46 | |
| -!- nube [~rho@49.244.14.60] has quit [Quit: Leaving.] | 02:48 | |
| -!- nube [~rho@49.244.14.60] has joined #shogun | 02:49 | |
| -!- van51 [~van51@athedsl-408350.home.otenet.gr] has quit [Quit: Leaving.] | 03:19 | |
| -!- FSCV [~FSCV@187.210.54.166] has quit [Quit: Leaving] | 03:48 | |
| -!- nube [~rho@49.244.14.60] has quit [Quit: Leaving.] | 03:54 | |
| -!- lambday [67157d36@gateway/web/freenode/ip.103.21.125.54] has joined #shogun | 04:14 | |
| shogun-buildbot | build #449 of nightly_default is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/449 | 04:15 |
| lambday | pickle27: hi | 05:18 |
| pickle27 | lambday: hey | 05:18 |
| lambday | pickle27: how's it going? | 05:19 |
| pickle27 | pretty good you? | 05:19 |
| lambday | pickle27: pretty good too :D | 05:19 |
| lambday | pickle27: hey could you please have a look at this - https://travis-ci.org/shogun-toolbox/shogun/jobs/8787066 | 05:19 |
| pickle27 | lambday: yeah I just fixed that earlier this eve but it hasn't been merged yet | 05:20 |
| lambday | pickle27: okay, thanks man :) | 05:20 |
| pickle27 | yeah heres the latest | 05:21 |
| pickle27 | https://travis-ci.org/shogun-toolbox/shogun/builds/8785919 | 05:21 |
| lambday | pickle27: feels so good seeing travis green :D | 05:22 |
| pickle27 | yeah haha | 05:22 |
| pickle27 | lambday: I hate trying to debug travis when things run on my machines but Im getting better at it | 05:22 |
| lambday | tell me about it !! :( | 05:22 |
| pickle27 | that unit test was a weird fix | 05:22 |
| lambday | but that worked :D | 05:23 |
| pickle27 | yeah I think it was a bad memcpy on my part from eigen3 back to an SGMatrix | 05:24 |
| pickle27 | I don't know why my machines are so resilient to these bugs lol and Travis picks them up | 05:24 |
| lambday | travis is like an annoyed girlfriend.. gets mad at pretty much every single mistake :D | 05:25 |
| pickle27 | haha | 05:25 |
| lambday | btw.. you were using eigen3 for matrix, and then copied the results back to SGMatrix? | 05:26 |
| pickle27 | yeah thats what I had, turns out it was easier and cleaner to just leave it as eigen3 | 05:27 |
| pickle27 | I usually create SGMatrix and then use eigen::map but had this one the other way around | 05:27 |
| pickle27 | worked on my machine but I don't think it memcpy'd fully on travis which was the issue | 05:28 |
| lambday | I see.. yes I too use map, seems easier :) | 05:28 |
| pickle27 | yeah talking about it now Im really not sure why I had it that way in the first place | 05:29 |
| lambday | hehe... been there just 2 days back! :D | 05:30 |
| lambday | pickle27: will be back after breakfast.. for you its dinner time I guess | 05:31 |
| pickle27 | haha its like almost bedtime lol | 05:31 |
| pickle27 | lambday: catch ya later | 05:31 |
| lambday | pickle27: hehe... ciao man! | 05:31 |
| -!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 06:11 | |
| -!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 06:36 | |
| -!- sanyam [uid10602@gateway/web/irccloud.com/x-qnjvagnlypgbldrp] has joined #shogun | 06:52 | |
| -!- gsomix [~gsomix@109.188.126.242] has joined #shogun | 07:30 | |
| gsomix | sonney2k, good morning. finally I fixed my reader. preparing PR. | 07:32 |
| gsomix | I'm sorry for what I wasn't yesterday on the channel | 07:33 |
| lambday | sonney2k: moin moin | 07:48 |
| lambday | sonney2k: SGSparseMatrix is not under parameter framework, shall I add that? | 07:49 |
| * gsomix away | 08:20 | |
| -!- nube [~rho@49.244.47.119] has joined #shogun | 09:00 | |
| -!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 09:03 | |
| shogun-notifier- | shogun: lambday :develop * 350616d / / (4 files): https://github.com/shogun-toolbox/shogun/commit/350616dbfdd96135f5f330f49ff810dff145c32c | 09:03 |
| shogun-notifier- | shogun: rational approximation unit-test added | 09:03 |
| shogun-notifier- | shogun: Heiko Strathmann :develop * eddf5ca / / (4 files): https://github.com/shogun-toolbox/shogun/commit/eddf5ca6d6412d8dcff95300c1e191a48423e32a | 09:03 |
| shogun-notifier- | shogun: Merge pull request #1212 from lambday/feature/log_determinant | 09:03 |
| shogun-notifier- | shogun: | 09:03 |
| shogun-notifier- | shogun: rational approximation unit-test added | 09:03 |
| shogun-buildbot | build #1198 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1198 blamelist: lambday <heavensdevil6909@gmail.com> | 09:17 |
| shogun-buildbot | build #1199 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1199 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 09:29 |
| shogun-buildbot | build #1321 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/1321 blamelist: lambday <heavensdevil6909@gmail.com> | 09:45 |
| -!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun | 09:58 | |
| -!- foulwall [~user@2001:da8:215:6100:acb5:df7a:527c:2a23] has joined #shogun | 10:17 | |
| shogun-buildbot | build #1322 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/1322 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 10:20 |
| -!- foulwall` [~user@2001:da8:215:6100:acb5:df7a:527c:2a23] has joined #shogun | 10:33 | |
| -!- gsomix [~gsomix@109.188.126.242] has quit [Remote host closed the connection] | 10:43 | |
| -!- foulwall` [~user@2001:da8:215:6100:acb5:df7a:527c:2a23] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] | 10:44 | |
| -!- foulwall [~user@2001:da8:215:6100:acb5:df7a:527c:2a23] has quit [Remote host closed the connection] | 10:44 | |
| -!- foulwall [~user@2001:da8:215:6100:acb5:df7a:527c:2a23] has joined #shogun | 10:44 | |
| -!- foulwall [~user@2001:da8:215:6100:acb5:df7a:527c:2a23] has quit [Ping timeout: 252 seconds] | 11:33 | |
| -!- lisitsyn [~lisitsyn@109-226-115-134.clients.tlt.100megabit.ru] has left #shogun [] | 11:37 | |
| -!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 12:03 | |
| @sonney2k | lambday, that would be excellent! | 12:15 |
| @sonney2k | lambday, and btw regarding implementing DenseFeatures for the complex type - so far everything I implemented in shogun got used sooner or later. In particular if it is data type related. | 12:16 |
| @sonney2k | So it is not a time waste! | 12:17 |
| van51 | sonney2k: hey | 12:22 |
| van51 | sonney2k: should I make an option to normalize the vector if the user wants so and make a PR? | 12:23 |
| van51 | sonney2k: and if so I'll normalize with just vector.size() which is an upper bound | 12:25 |
| lambday | sonney2k: hey.. sorry I was away.. | 12:29 |
| lambday | sonney2k: alright I'll add support for complex64 in DenseFeatures... (and there were a few others) | 12:29 |
| -!- lambday [67157d36@gateway/web/freenode/ip.103.21.125.54] has quit [] | 13:04 | |
| -!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Remote host closed the connection] | 13:06 | |
| -!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 13:07 | |
| -!- van51 [~van51@athedsl-408350.home.otenet.gr] has left #shogun ["PING 1373112359"] | 14:06 | |
| -!- foulwall [~user@2001:da8:215:6100:1422:e7f8:7364:2ed6] has joined #shogun | 14:26 | |
| -!- foulwall [~user@2001:da8:215:6100:1422:e7f8:7364:2ed6] has quit [Ping timeout: 245 seconds] | 14:43 | |
| -!- foulwall [~user@2001:da8:215:6100:895c:6f5:d9c:ece9] has joined #shogun | 14:56 | |
| -!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 15:13 | |
| -!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun | 15:25 | |
| -!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 15:27 | |
| -!- foulwall [~user@2001:da8:215:6100:895c:6f5:d9c:ece9] has quit [Ping timeout: 264 seconds] | 16:05 | |
| -!- lambday [67157f36@gateway/web/freenode/ip.103.21.127.54] has joined #shogun | 16:16 | |
| -!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 16:56 | |
| shogun-notifier- | shogun: Kevin Hughes :develop * d40ed8e / / (5 files): https://github.com/shogun-toolbox/shogun/commit/d40ed8e4bc8749fc82d5e4ae334ca606b8d647ef | 16:56 |
| shogun-notifier- | shogun: added FFDiag AJD algorithm and a unit test. | 16:56 |
| shogun-notifier- | shogun: Sergey Lisitsyn :develop * d38be81 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/d38be81e4a61bd4dd73d657041523ae0bdc4b268 | 16:56 |
| shogun-notifier- | shogun: Merge pull request #1211 from pickle27/develop | 16:56 |
| shogun-notifier- | shogun: | 16:56 |
| shogun-notifier- | shogun: added FFDiag AJD algorithm and a unit test. | 16:56 |
| -!- FSCV [~FSCV@189.139.250.209] has joined #shogun | 17:07 | |
| shogun-buildbot | build #1200 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1200 blamelist: Kevin Hughes <kevin@kevin-M3920.(none)> | 17:12 |
| -!- lisitsyn [~lisitsyn@109-226-115-134.clients.tlt.100megabit.ru] has joined #shogun | 17:14 | |
| shogun-buildbot | build #1323 of deb3 - modular_interfaces is complete: Failure [failed compile python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1323 blamelist: Kevin Hughes <kevin@kevin-M3920.(none)> | 17:15 |
| -!- lambday [67157f36@gateway/web/freenode/ip.103.21.127.54] has quit [] | 17:24 | |
| shogun-buildbot | build #1201 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1201 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 17:25 |
| -!- nube [~rho@49.244.47.119] has quit [Quit: Leaving.] | 17:31 | |
| shogun-buildbot | build #1324 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/1324 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 17:45 |
| -!- foulwall [~user@li379-21.members.linode.com] has joined #shogun | 18:18 | |
| foulwall | ping sonney2k | 18:18 |
| -!- van51 [~van51@athedsl-408350.home.otenet.gr] has quit [Read error: No route to host] | 18:29 | |
| -!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun | 18:31 | |
| -!- foulwall [~user@li379-21.members.linode.com] has quit [Ping timeout: 268 seconds] | 18:41 | |
| -!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 18:45 | |
| -!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Client Quit] | 18:49 | |
| van51 | sonney2k: I | 19:40 |
| van51 | sonney2k: I've started working on a streaming version of the HashedDocFeatures | 19:40 |
| van51 | sonney2k: do these streaming have features *have* to use the CInputParser? | 19:41 |
| van51 | sonney2k: because I don't see the current version of the Parser to be able to be used easily in text collections like the webspam | 19:42 |
| van51 | I may be missing something though | 19:43 |
| van51 | when you see this (and what I asked before) please comment and I'll check the logs later | 19:43 |
| -!- van51 [~van51@athedsl-408350.home.otenet.gr] has left #shogun ["PART #gsoc-gr :PING 1373132672"] | 19:44 | |
| -!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 19:56 | |
| @sonney2k | van51, yes streamingfeatures is what you would have to support. if it doesn't work with webspam just yet then we just use a different dataset for now | 20:06 |
| @sonney2k | van51, so just do the online streamingdoc*features trying to not have too much code duplication | 20:55 |
| @sonney2k | hurray | 21:05 |
| @sonney2k | the cygwin buildbot find g++ again | 21:05 |
| @sonney2k | no more compiler RIP | 21:05 |
| lisitsyn | sonney2k: what's the reason? | 21:06 |
| @sonney2k | oh I am upgrading cygwin | 21:06 |
| @sonney2k | new one has swig 2.0.9 | 21:06 |
| @sonney2k | and lua etc etc | 21:06 |
| @sonney2k | votjakovr, any idea why regression_gaussian_process_modular.py fails on http://shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1324/steps/test%20python_modular/logs/stdio ? | 21:10 |
| @sonney2k | votjakovr, could you add some debug code to let us figure out what is wrong? | 21:10 |
| votjakovr | sonney2k: hi, have the same problem with numpy 1.7.1 | 21:11 |
| @sonney2k | sonney2k, what do you mean? it is failing or what? | 21:14 |
| votjakovr | sonney2k: i mean, that i think that this problem depends on numpy | 21:15 |
| lisitsyn | votjakovr: really? why do you think so? | 21:15 |
| @sonney2k | votjakovr, Checking for Python numpy support ... 1.6.2 | 21:15 |
| @sonney2k | votjakovr, seems an old version is on the buildbot | 21:15 |
| @sonney2k | votjakovr, IIRC you changed some things in GPs recently no? | 21:16 |
| @sonney2k | couldn't that be the reason? | 21:16 |
| @sonney2k | I mean examples also needed to be changed so maybe it is just missing a test data regenerate | 21:16 |
| votjakovr | sonney2k: hmm... so why is diff empty? | 21:17 |
| @sonney2k | votjakovr, no idea | 21:20 |
| @sonney2k | votjakovr, sounds like bug in test framework | 21:20 |
| @sonney2k | votjakovr, last working version was | 21:20 |
| @sonney2k | c868efbc9dc180433fce5b8bf04f77d56fc59c2a | 21:20 |
| @sonney2k | votjakovr, git diff c868efbc9dc180433fce5b8bf04f77d56fc59c2a shogun/regression/GaussianProcessRegression.* | 21:22 |
| @sonney2k | votjakovr, maybe this helps ... | 21:22 |
| votjakovr | sonney2k: but i didn't change implementation of algorithms | 21:28 |
| votjakovr | sonney2k: so i think, that data should stay the same | 21:29 |
| @sonney2k | votjakovr, who changed it then? | 21:29 |
| votjakovr | sonney2k: sorry, what is changed? | 21:30 |
| @sonney2k | votjakovr, well please checkout c868efbc9dc180433fce5b8bf04f77d56fc59c2a and see if it still passes the test | 21:31 |
| @sonney2k | votjakovr, if it does figure out what has changed | 21:31 |
| @sonney2k | it is bad to have a broken test ... | 21:31 |
| @sonney2k | so it is important to get it back to work | 21:31 |
| -!- gsomix [~gsomix@109.188.125.161] has joined #shogun | 21:32 | |
| gsomix | good evening | 21:32 |
| @sonney2k | gsomix, good evening! good to see you alive | 21:33 |
| gsomix | sonney2k, yeah. There was a storm in Moscow. However, I was saved by the castle of Catherine the Great. :) | 21:35 |
| gsomix | sonney2k, btw, https://github.com/shogun-toolbox/shogun/pull/1130 | 21:36 |
| votjakovr | sonney2k: i've rolled back to c868efbc9dc180433fce5b8bf04f77d56fc59c2a, then i've run tests, and this test is failed and diff is empty (i've reinstalled shogun) :( | 21:40 |
| @sonney2k | gsomix, did you just update? | 21:41 |
| gsomix | sonney2k, yep | 21:41 |
| @sonney2k | gsomix, looks pretty good now. lets see what travis says and I will merge | 21:45 |
| @sonney2k | votjakovr, weird | 21:46 |
| @sonney2k | votjakovr, only explanation would be some update of data or library then | 21:46 |
| votjakovr | sonney2k: i've figured, that on line 29 of tester.py: return numpy.all(a == b) is false | 21:50 |
| @sonney2k | votjakovr, tester.py has some option to -d IIRC to stop when sth differs | 21:51 |
| @sonney2k | you can then use | 21:51 |
| @sonney2k | p a | 21:51 |
| @sonney2k | p b | 21:51 |
| @sonney2k | to see the two objects | 21:51 |
| votjakovr | hmm, yep, they are different | 22:01 |
| @sonney2k | ok then now you 'only' have to figure out whats wrong | 22:02 |
| votjakovr | sonney2k: maximal difference is approx. 1e-12 | 22:03 |
| @sonney2k | votjakovr, if it is just a numerical issue then use some numpy.round trick! | 22:06 |
| votjakovr | sonney2k: Btw regenerating of the data helps :) | 22:11 |
| @sonney2k | votjakovr, | 22:20 |
| @sonney2k | heh | 22:20 |
| @sonney2k | if the 1e-12 is justified then we can use this as a fix | 22:20 |
| shogun-buildbot | build #996 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/996 blamelist: lambday <heavensdevil6909@gmail.com>, Kevin Hughes <kevin@kevin-M3920.(none)>, foulwall <liuzhengyang@outlook.com>, Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein | 22:31 |
| shogun-buildbot | <thoralf.klein@zib.de>, Viktor Gal <vigsterkr@gmail.com>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>, Kevin <kevinhughes27@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, van51 <vangelis_51@hotmail.com>, Roman Votyakov <votjakovr@gmail.com> | 22:31 |
| lisitsyn | hahha | 22:31 |
| gsomix | wow, how many | 22:31 |
| votjakovr | sonney2k: ok:) i'll try to find the reason... | 22:31 |
| @sonney2k | haha | 22:32 |
| @sonney2k | yeah I managed to get the buildbot back to fail | 22:32 |
| @sonney2k | looks like the update did help a lot :`-( | 22:32 |
| gsomix | sonney2k, green. https://travis-ci.org/shogun-toolbox/shogun/builds/8803120 | 22:32 |
| votjakovr | wow! travis is green :) | 22:33 |
| @sonney2k | shogun-buildbot, force build 'cyg1 - libshogun' | 22:35 |
| shogun-buildbot | build #997 forced | 22:35 |
| shogun-buildbot | I'll give a shout when the build finishes | 22:35 |
| @sonney2k | gsomix, yeah travis doesn't do everything and no cygwin in particular | 22:35 |
| -!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 22:40 | |
| shogun-notifier- | shogun: Soeren Sonnenburg :develop * c392690 / examples/undocumented/python_modular/serialization_complex_example.py: https://github.com/shogun-toolbox/shogun/commit/c39269021fbe8983b9a372320a9a49a8302c6421 | 22:40 |
| shogun-notifier- | shogun: don't fail bsd bot when json is not available | 22:40 |
| shogun-notifier- | shogun: Evgeniy Andreev :develop * 6e4b7db / / (13 files): https://github.com/shogun-toolbox/shogun/commit/6e4b7db37174e8c6b60989966fb086e455212d86 | 22:41 |
| shogun-notifier- | shogun: added CircularBuffer and LineReader | 22:41 |
| shogun-notifier- | shogun: Soeren Sonnenburg :develop * db2aecf / / (13 files): https://github.com/shogun-toolbox/shogun/commit/db2aecff490d83ff12c5b6d595613b58373c9172 | 22:41 |
| shogun-notifier- | shogun: Merge pull request #1130 from gsomix/feature/getline | 22:41 |
| shogun-notifier- | shogun: | 22:41 |
| shogun-notifier- | shogun: LineReader class for reading lines from file stream | 22:41 |
| @sonney2k | gsomix, finally! | 22:41 |
| gsomix | sonney2k, thank God. :) | 22:42 |
| lisitsyn | there is no such thing! | 22:42 |
| @sonney2k | gsomix, lisitsyn thank $DEITY then | 22:44 |
| @sonney2k | in my case DEITY=vapnik | 22:44 |
| -!- lambday [67157c36@gateway/web/freenode/ip.103.21.124.54] has joined #shogun | 22:44 | |
| lisitsyn | well that's better than some unexistent thing ;) | 22:44 |
| @sonney2k | lambday, what is your $DEITY? DEITY=HeikoS or vapnik? | 22:45 |
| @sonney2k | lambday, and good evening! | 22:45 |
| gsomix | lol | 22:45 |
| @sonney2k | lambday, you are free to choose | 22:46 |
| @sonney2k | lambday, I mean choose me ;) | 22:46 |
| lambday | sonney2k: :-o | 22:46 |
| lambday | :-s | 22:46 |
| @sonney2k | lisitsyn, I manually rm -rf'd the build directory on cyg1 | 22:47 |
| @sonney2k | lisitsyn, and now it builds | 22:47 |
| * sonney2k sighs | 22:47 | |
| lisitsyn | sonney2k: haha | 22:47 |
| @sonney2k | there might be some weird race condition | 22:47 |
| lambday | export DEITY="sonney2k" :-o | 22:48 |
| lambday | echo $DEITY | 22:49 |
| lambday | sonney2k: | 22:49 |
| lambday | sonney2k: lisitsyn I have a problem :- | 22:49 |
| lisitsyn | lambday: houston on the line | 22:49 |
| gsomix | in my case DEITY=random cool physicist or mathematician. And it seems I know not many cool computer scientists, hm. | 22:50 |
| * gsomix z-z-z-z | 22:50 | |
| @sonney2k | lisitsyn, https://plus.google.com/photos/111049168280159033135/albums/5897522656212965009/5897522659811359522?pid=5897522659811359522&oid=111049168280159033135 | 22:50 |
| lambday | lisitsyn: huh? | 22:50 |
| @sonney2k | lisitsyn, baikonur rather no? | 22:50 |
| * lambday max(confused) | 22:50 | |
| shogun-buildbot | build #997 of cyg1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/997 | 22:50 |
| @sonney2k | lambday, or that https://plus.google.com/photos/111049168280159033135/albums/5897522656212965009/5897522659811359522?pid=5897522659811359522&oid=111049168280159033135 | 22:50 |
| lisitsyn | lambday: what is the problem? | 22:51 |
| lisitsyn | sonney2k: baikonur is kazakhstan actually you know ;) | 22:51 |
| @sonney2k | lisitsyn, yes since borat | 22:52 |
| lambday | sonney2k: whoa!! what was that? looks like our Indian festival "Diwali" | 22:52 |
| lambday | :-o | 22:52 |
| @sonney2k | lambday, a notebook! | 22:52 |
| shogun-buildbot | build #1202 of bsd1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1202 | 22:52 |
| @sonney2k | lambday, I thought that was your problem | 22:52 |
| @sonney2k | all hail shogun-buildbot ! | 22:52 |
| gsomix | nite guys. | 22:52 |
| lambday | sonney2k: HAHAHAHA =) | 22:52 |
| @sonney2k | he is good | 22:52 |
| @sonney2k | gsomix, nite | 22:52 |
| lambday | luckily, I use insti's computer for shogun stuff... my laptop is safe so far :D | 22:53 |
| @sonney2k | lisitsyn, this one Borat: Cultural Learnings of America for Make Benefit Glorious Nation of Kazakhstan (2006) | 22:53 |
| @sonney2k | lambday, heh | 22:53 |
| lambday | sonney2k: lisitsyn: the problem is - | 22:53 |
| lambday | borat? | 22:53 |
| lisitsyn | sonney2k: is that the full name of the movie? | 22:53 |
| lisitsyn | haha | 22:53 |
| lambday | isn't that the movie maker or something | 22:53 |
| lambday | he's hilarious | 22:53 |
| lambday | well, the problem is :( | 22:54 |
| lisitsyn | that's a role of sacha baron-cohen lambday | 22:54 |
| @sonney2k | lisitsyn, yes | 22:54 |
| lisitsyn | just like admiral general aladeen :D | 22:54 |
| lisitsyn | I don't know what is more hilarious | 22:54 |
| @sonney2k | lambday, just tell us | 22:55 |
| lambday | lisitsyn: I only know "that's" "a" "role" "of" "lambday" of what you just said :D | 22:55 |
| lambday | sonney2k: I need to work with sparse matrices | 22:55 |
| @sonney2k | we are drunk and entropy increasing at this time of day anyways | 22:55 |
| @sonney2k | lambday, do it! | 22:55 |
| @sonney2k | so what is the problem | 22:55 |
| @sonney2k | I do that to | 22:55 |
| @sonney2k | o | 22:55 |
| @sonney2k | too | 22:55 |
| @sonney2k | too tuuut | 22:55 |
| lambday | sonney2k: the problem is... :D | 22:55 |
| lambday | sonney2k: I need to convert it to eigen3 sparse each time I use that | 22:56 |
| * sonney2k grabs a beer and prepares for the worst | 22:56 | |
| lisitsyn | we had a problem with performance and we went for multithreading | 22:56 |
| @sonney2k | lambday, uhh ok big one | 22:56 |
| lambday | sonney2k: you better.. I'm gonna drop a bomb | 22:56 |
| lisitsyn | nwo ew h vea a wto bplemros | 22:56 |
| @sonney2k | lisitsyn, any experience with that so far | 22:57 |
| lisitsyn | sonney2k: sparse? | 22:57 |
| lisitsyn | yes | 22:57 |
| -!- gsomix [~gsomix@109.188.125.161] has quit [Ping timeout: 268 seconds] | 22:57 | |
| @sonney2k | sparse from eigen3 I mean | 22:57 |
| lisitsyn | yes | 22:57 |
| @sonney2k | what format do they internally use | 22:57 |
| lambday | so, the number of times I have to convert that is HUGE.. and each time new memory will be allocated and will be real bad performance wise | 22:57 |
| lambday | and here is the bomb | 22:57 |
| lisitsyn | CCR sonney2k | 22:57 |
| @sonney2k | lambday, yeah not possible that is clear | 22:57 |
| lambday | I want to use eigen | 22:58 |
| lisitsyn | ehm | 22:58 |
| lambday | 3s' sparsematrix as a member | 22:58 |
| lisitsyn | CC/CR | 22:58 |
| lisitsyn | any | 22:58 |
| lambday | and want serialization/param registration :( | 22:58 |
| lisitsyn | uh | 22:58 |
| lambday | with that | 22:58 |
| * lisitsyn kills himself | 22:58 | |
| lambday | lisitsyn: no don't die :( | 22:58 |
| lambday | me cries | 22:58 |
| lambday | fuck! | 22:58 |
| * lambday cries | 22:59 | |
| lisitsyn | lambday: hah | 22:59 |
| lisitsyn | okay | 22:59 |
| @sonney2k | lets all wine like real men! | 22:59 |
| lisitsyn | we should go for SURGERY | 22:59 |
| * sonney2k grabs another beer | 22:59 | |
| lambday | shit man! I am missing all the drinks.. :( | 23:00 |
| lambday | I don't have any atm :'( | 23:00 |
| lisitsyn | lambday: I am into the eigen3 code now | 23:00 |
| lambday | lisitsyn: which code? | 23:00 |
| @sonney2k | lambday, how big is that matrix? | 23:01 |
| lisitsyn | lambday: sparse | 23:01 |
| @sonney2k | lambday, there are a couple of options | 23:01 |
| lambday | sonney2k: say, a mil x a mil | 23:01 |
| @sonney2k | a) not use eigen3 for sparse | 23:01 |
| lambday | sonney2k: what else we're gonna use? | 23:01 |
| lambday | I need to run iterative solvers with that :( | 23:01 |
| @sonney2k | b) use eigen3 for sparse but convert it *once* from SGSparseMatrix | 23:02 |
| lambday | I like the second.. | 23:02 |
| lambday | but then I need to keep it as a param | 23:02 |
| lisitsyn | lambday: all you need is love^W | 23:02 |
| lisitsyn | is | 23:02 |
| @sonney2k | c) not use SGSparseMatrix except for serialization | 23:02 |
| @sonney2k | lambday, no I mean how big is the matrix in terms of non-zero elements | 23:02 |
| lisitsyn | CompressedStorage::Map(Index* indices, Scalar* values, size_t size) | 23:02 |
| @sonney2k | lambday, mil x mil is nothing for a really sparse matrix :) | 23:02 |
| lambday | sonney2k: wait let me check | 23:03 |
| @sonney2k | lisitsyn, shogun uses LIL :) https://en.wikipedia.org/wiki/Sparse_matrix#List_of_lists_.28LIL.29 | 23:03 |
| lambday | 200,000 nz elements | 23:03 |
| lisitsyn | sonney2k: I know | 23:04 |
| @sonney2k | lambday, so small | 23:04 |
| lisitsyn | sonney2k: I believe its conversion is fast | 23:04 |
| lisitsyn | nobody would ever notice that | 23:04 |
| lambday | sonney2k: but that's the demo that we're gonna prepare | 23:04 |
| @sonney2k | lambday, in this case conversion is very cheap | 23:04 |
| lisitsyn | I construct sparse matrices in tapkee dynamically | 23:04 |
| @sonney2k | lambday, do you know how big matrices might become | 23:04 |
| lisitsyn | it is like less than 0.01s or so | 23:05 |
| lambday | Heiko knows | 23:05 |
| @sonney2k | lisitsyn, yeah even less I think | 23:05 |
| lisitsyn | lambday: serious things happen in solvers I guess rgiht? | 23:05 |
| @sonney2k | lambday, conversion is only a problem when you have hard memory constraints | 23:05 |
| @sonney2k | lambday, but this here is just 1MB | 23:05 |
| lambday | sonney2k: wait, that can't be right! the matrix I have for ozone is about 32 mb | 23:06 |
| lisitsyn | lambday: what I suggest now | 23:06 |
| lisitsyn | use construct from triplets thingy | 23:06 |
| lisitsyn | if it is slow (measure first!) | 23:06 |
| @sonney2k | lisitsyn, does eigen3 support using CRS / CCS based matrices | 23:06 |
| lisitsyn | sonney2k: if you have arrays of indices and values | 23:06 |
| lisitsyn | yes | 23:06 |
| lambday | sonney2k: lisitsyn and the main problem is - I need to do this convert this like 1,000,000x20 times | 23:06 |
| @sonney2k | lambday, lisitsyn, if so we should have a converter from LIL -> CRS / CCS | 23:06 |
| lisitsyn | lambday: why? | 23:06 |
| lambday | lisitsyn: sonney2k I already added an util to convert to eigen3 sparse in eigen3.h | 23:07 |
| @sonney2k | lambday, yeah why? | 23:07 |
| lambday | sonney2k: I can convert it once and store inside the class as a member.. but that requires eigen3 sparse to be make it as a member | 23:07 |
| lambday | no serialization no cloning nothing is possible | 23:08 |
| lambday | but | 23:08 |
| lisitsyn | still? | 23:08 |
| lisitsyn | 20M times? | 23:08 |
| lambday | we'll be needing that I guess for parallel computation thing - same framework shoudl work | 23:08 |
| lambday | sonney2k: lisitsyn we actually need to compute many many estimates of log-det | 23:08 |
| lambday | 20 for each | 23:08 |
| lambday | 1 mil is the number of estimates I am guessing | 23:09 |
| lisitsyn | what 1M comes from? | 23:09 |
| lisitsyn | ah | 23:09 |
| lambday | oh no | 23:09 |
| lambday | wait! | 23:09 |
| lambday | I was wrong! | 23:09 |
| lambday | it should be 1,000,000x20x100 | 23:09 |
| lisitsyn | ehm | 23:09 |
| @sonney2k | superpowers! | 23:10 |
| lisitsyn | you need to solve | 23:10 |
| lisitsyn | 1M x 1M sparse matric | 23:10 |
| lisitsyn | matrix | 23:10 |
| lisitsyn | 1M times | 23:10 |
| lisitsyn | *20*100 | 23:10 |
| @sonney2k | lisitsyn, don't forget it is /really/ sparse | 23:10 |
| lisitsyn | sonney2k: no I understand but it is like a lot of time anyway | 23:10 |
| @sonney2k | lambday, I still don't understand how exactly this is working | 23:11 |
| lambday | I don't need to solve, but I need to convert it to eigen3-sparse from SGSparse this number of time | 23:11 |
| lambday | sonney2k: well | 23:11 |
| lisitsyn | lambday: okay why to convert it so many times? | 23:11 |
| lambday | initial plan was to keep SGSparseMatrix as a member of our opertor class | 23:11 |
| lisitsyn | ahh so you apply operator | 23:11 |
| lambday | yes | 23:11 |
| lambday | exactly | 23:11 |
| lisitsyn | and when you apply operator | 23:11 |
| lisitsyn | you convert | 23:11 |
| @sonney2k | lambday, ahh ok | 23:11 |
| lambday | each time I need to apply, I need to convert | 23:11 |
| @sonney2k | lambday, well a few more ideas | 23:12 |
| lisitsyn | yes I have too | 23:12 |
| lambday | and if I don't get to keep eigen3's sparse inside the class, we're stuck | 23:12 |
| lisitsyn | sonney2k: you first | 23:12 |
| lisitsyn | :D | 23:12 |
| lisitsyn | lambday: there is a way to put it inside | 23:12 |
| lambday | sonney2k: please :( | 23:12 |
| @sonney2k | a) you could call finalize() or sth to make the conversion back to SGSparseMatrix | 23:12 |
| shogun-buildbot | build #1325 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/1325 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:12 |
| @sonney2k | b) we could add a different storage layout to SGSparseMatrix | 23:12 |
| @sonney2k | (as in CCS, CRS) | 23:13 |
| @sonney2k | if eigen3 can immediately use this | 23:13 |
| lambday | that would be great! | 23:13 |
| lambday | like map for dense | 23:13 |
| lambday | ours is an adjacency list, right | 23:13 |
| lisitsyn | my other solution is | 23:14 |
| @sonney2k | lisitsyn, your ideas? | 23:14 |
| lisitsyn | just add a base class in operator | 23:14 |
| lisitsyn | like MatrixHandler | 23:14 |
| lambday | then? | 23:14 |
| lisitsyn | and declare EigenMatrixHandler in .cpp | 23:14 |
| lisitsyn | and use it inside | 23:14 |
| @sonney2k | lisitsyn, so it would be making eigen3 work with LIL? | 23:15 |
| lambday | lisitsyn: I didn't get.. I need to store it as a member.. AND also probably need serialization | 23:15 |
| lisitsyn | sonney2k: I mean it would convert once | 23:15 |
| lisitsyn | lambday: ahh serialization sorry | 23:15 |
| lambday | sonney2k: lisitsyn can't we add Eigen::SparseMatrix as a ctype or something? | 23:15 |
| lambday | :( | 23:15 |
| lisitsyn | ctype? | 23:15 |
| @sonney2k | taht would be option c) then - make eigen3 support LIL | 23:16 |
| lambday | LIL=? :( | 23:16 |
| lisitsyn | list of lists | 23:16 |
| @sonney2k | lambday, our sparse matrix format | 23:16 |
| lambday | ohkay | 23:16 |
| lambday | but when we need to multiply vectors with it, I think LIL is not a good option | 23:16 |
| @sonney2k | lambday, why not? | 23:17 |
| lambday | I mean, if I can directly use eigen3's sparse, I don't need SGSparseMatrix at all for our purpose | 23:17 |
| lisitsyn | lambday: what do you use eigen3 here ? | 23:18 |
| lisitsyn | for* | 23:18 |
| @sonney2k | lambday, yeah problem is with serializtion only | 23:18 |
| lambday | mat-vec multiplications | 23:18 |
| lambday | which is really fast | 23:18 |
| lisitsyn | just multiply? | 23:18 |
| lambday | for iterative solvers, yes | 23:18 |
| lisitsyn | there is no special tricks here | 23:19 |
| lambday | and that's what we'll be using sparse for, so yes | 23:19 |
| lisitsyn | lambday: are you computing largest eigenvalues? | 23:19 |
| @sonney2k | lambday, ok in this case we have d) do multply with shogun's code | 23:19 |
| lambday | lisitsyn: yes that, and also writing a cg solver | 23:19 |
| lisitsyn | lambday: are you going to estimate smallest too? | 23:19 |
| lambday | sonney2k:you mean, add a new method with SGSparse? | 23:20 |
| lambday | lisitsyn: yes, extremal | 23:20 |
| lambday | but that's not where I'll be using it only.. | 23:21 |
| @sonney2k | lambday, it has it already IIRC | 23:21 |
| * sonney2k checks | 23:21 | |
| lisitsyn | lambday: then you'd need to solve systems too? | 23:21 |
| lambday | lisitsyn: yes that's the main part | 23:21 |
| lambday | each estimate generates like a lot of systems which need to be solved using iterative solvers | 23:21 |
| lambday | for sparse | 23:21 |
| lisitsyn | are you writing your own solver for that? | 23:21 |
| @sonney2k | lambday, yes you could do sparse_matrix[index].dense_dot() | 23:21 |
| lambday | yep, have to | 23:21 |
| lambday | (well, taking krylstat's) | 23:21 |
| shogun-buildbot | build #1203 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1203 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org> | 23:21 |
| lisitsyn | lambday: why have to btw? | 23:22 |
| lambday | sonney2k: that computes mat-vec product? | 23:22 |
| lambday | lisitsyn: we need a solver that solves a family of shifted systems simultaneously | 23:22 |
| lambday | COCG_M | 23:22 |
| lambday | for complex | 23:22 |
| lisitsyn | I see | 23:23 |
| lambday | I don't think eigen3 provides that | 23:23 |
| lisitsyn | no | 23:23 |
| lisitsyn | haven't seen such thing there | 23:23 |
| lambday | but mat-vec product is the key here and we gotta do it many many times | 23:23 |
| @sonney2k | lambday, well just for one row. | 23:23 |
| lambday | I want it to be fast :( | 23:23 |
| * lambday scratches his head | 23:23 | |
| @sonney2k | lambday, but you could easily add a function to SGSparseMatrix using that then | 23:24 |
| lisitsyn | most of things that make code fast are not that obvious and you'd have to measure anyway | 23:24 |
| @sonney2k | it is easy and will be fast | 23:24 |
| @sonney2k | the dense_dot has jus non-zero elements in vector operations | 23:24 |
| lambday | sonney2k: we'll be needing sparse-matrix-dense vec products | 23:25 |
| lambday | and also sparse-matrix-sparse-vector products | 23:25 |
| lisitsyn | bah boeing 777 crashed at SFO | 23:25 |
| lambday | what! | 23:25 |
| @sonney2k | lambday, we have that | 23:26 |
| @sonney2k | lambday, sparse_dot() | 23:26 |
| lisitsyn | *just* two dead | 23:26 |
| lambday | sonney2k: alright I'll check then | 23:26 |
| lambday | so, not use eigen3 at all?? | 23:26 |
| @sonney2k | lisitsyn, that is where we will arrive for GSoC summit | 23:27 |
| lambday | when is it? | 23:27 |
| lisitsyn | sonney2k: I am on the waiting list :P | 23:27 |
| @sonney2k | lisitsyn, lucky you | 23:27 |
| @sonney2k | lambday, it should take you no more than a few minutes to add these 2 funcitons to SGSparseMatrix | 23:27 |
| lambday | sonney2k: lisitsyn oh but the problem is.. we don't have serialization with SGSparseMatrix either | 23:28 |
| @sonney2k | lambday, just use dense_dot and sparse_dot from SGVector for that | 23:28 |
| lambday | is there any reason it wasn't added? | 23:28 |
| lisitsyn | lambday: oh well it is easier though | 23:28 |
| lambday | sonney2k: alright then | 23:28 |
| @sonney2k | lambday, then you can measure how fast things are for eigen3 / sgsparse* for your application | 23:28 |
| lisitsyn | lambday: the only reason is not enough hours in day of sonney2k and lazyness of me | 23:28 |
| lambday | :D | 23:29 |
| lambday | sonney2k: okay | 23:29 |
| @sonney2k | no idea if eigen3 is doing it faster currenlty | 23:29 |
| lambday | sonney2k: I think they do :-/ | 23:29 |
| lisitsyn | sonney2k: what exactly? | 23:29 |
| lambday | mat-vec products, right? | 23:29 |
| @sonney2k | lambday, problem with sparse matrices is that you don't have contiguous memory access | 23:29 |
| lambday | yes | 23:29 |
| @sonney2k | so I don't expect taht | 23:29 |
| lisitsyn | I think their storage is more cache-efficient | 23:30 |
| lisitsyn | it is contiguous | 23:30 |
| @sonney2k | lisitsyn, ours is contiguous | 23:30 |
| @sonney2k | they have to look into different index tables all the time | 23:30 |
| lisitsyn | sonney2k: but our columns are not continuous right? | 23:30 |
| @sonney2k | but we have index / value contiguos in memory | 23:31 |
| lisitsyn | of a column | 23:31 |
| lisitsyn | not of the whole matrix, right? | 23:31 |
| -!- lambday_ [67157f4c@gateway/web/freenode/ip.103.21.127.76] has joined #shogun | 23:31 | |
| lambday_ | sonney2k: lisitsyn sorry I got dc | 23:31 |
| lisitsyn | eigen's storage is just two arrays | 23:31 |
| @sonney2k | lisitsyn, well malloc will usually allocate it contigously | 23:31 |
| lisitsyn | we have an arrays of contiguous arrays | 23:31 |
| @sonney2k | yes true | 23:32 |
| lisitsyn | sonney2k: still a room for cache misses | 23:32 |
| @sonney2k | lisitsyn, for a nnz=100k matrix theirs might be in cvache completely | 23:32 |
| lambday_ | yep | 23:32 |
| @sonney2k | lisitsyn, I don't expect CCS / CSR to be better to the contrary | 23:32 |
| lisitsyn | I have no idea what is better | 23:33 |
| @sonney2k | lisitsyn, I mean if we allocated one block of memory | 23:33 |
| @sonney2k | and then had the lists in there contigously | 23:33 |
| @sonney2k | then we would certainly win | 23:33 |
| @sonney2k | and I guess malloc does just that normally | 23:33 |
| -!- lambday [67157c36@gateway/web/freenode/ip.103.21.124.54] has quit [Ping timeout: 250 seconds] | 23:34 | |
| lisitsyn | I don't know if compilers optimize that | 23:34 |
| lisitsyn | but you still kind of jump here | 23:34 |
| lambday_ | is there any way we can test that? | 23:34 |
| lisitsyn | from array of pointers to the array of values | 23:34 |
| lisitsyn | lambday_: hmm you can indeed | 23:35 |
| lisitsyn | lambday_: multiply some matrix as SGSparsematrix and as Eigen3 sparsematrix | 23:35 |
| lisitsyn | with some vector or so | 23:35 |
| lisitsyn | and check cache misses with valgrind ;) | 23:35 |
| @sonney2k | lisitsyn, look at this | 23:36 |
| lisitsyn | and walltime and other things like that | 23:36 |
| lambday_ | I was trying to find some page that I saw probably, regarding mat-vec products | 23:36 |
| @sonney2k | int main(int argc, char** argv) | 23:36 |
| @sonney2k | { | 23:36 |
| @sonney2k | void* a=malloc(256); | 23:36 |
| @sonney2k | void* b=malloc(256); | 23:36 |
| @sonney2k | printf("%p %p\n", a,b); | 23:36 |
| lambday_ | not sure where did I see that | 23:36 |
| @sonney2k | } | 23:36 |
| @sonney2k | 0x1dcb010 0x1dcb120 | 23:36 |
| lisitsyn | sonney2k: yes I understand that | 23:36 |
| @sonney2k | contigous! | 23:36 |
| @sonney2k | but there might be other tricks | 23:37 |
| @sonney2k | I think cache misses might rather come from accessing the million dim vector | 23:37 |
| lisitsyn | I am just afraid it reloads the cache | 23:37 |
| @sonney2k | it is randomly jumping in there | 23:38 |
| lisitsyn | when working with list of lists and the list | 23:38 |
| @sonney2k | lambday_, ok we are guessing too much now. just try it out | 23:38 |
| lambday_ | but the entries are SparseVectorEntries, can be created randomly, not col by col.. and then added to particular row or col.. so when we try to access it sequentially, it can be slow | 23:38 |
| @sonney2k | lisitsyn, should we have some directory in shogun/benchmarks? | 23:39 |
| @sonney2k | or so where we do such experiments? | 23:39 |
| lisitsyn | sonney2k: not in code right? | 23:39 |
| lisitsyn | in examples I guess | 23:39 |
| @sonney2k | lisitsyn, no | 23:39 |
| @sonney2k | benchmarks | 23:39 |
| lisitsyn | they are typically not in code I think | 23:39 |
| @sonney2k | as in shogun/tests shogun/examples shogun/benchmarks | 23:39 |
| lisitsyn | ahh | 23:39 |
| lisitsyn | yes | 23:39 |
| lisitsyn | didn't get you | 23:39 |
| @sonney2k | lambday_, so please add it there | 23:40 |
| @sonney2k | lambday_, van51 can add his dotfeatures benchmarks there too | 23:40 |
| lambday_ | sonney2k: umm... what will go there in the benchmarks? | 23:41 |
| lambday_ | I mean, I shouldn't add the dot in lib/SGSparseMatrix.h ? | 23:41 |
| @sonney2k | lambday_, the mil x mil SGSparsematrix with nnz=100k times dense vector run say 100 times | 23:41 |
| lambday_ | or just the experiment code goes there | 23:41 |
| @sonney2k | lambday_, yes | 23:41 |
| @sonney2k | to both | 23:41 |
| lambday_ | okay | 23:42 |
| @sonney2k | lambday_, so feel free to do the same test with eigen3 there and compare | 23:42 |
| @sonney2k | I am curious what the result will be | 23:42 |
| lambday_ | yes... | 23:43 |
| @sonney2k | lambday_, so just do it for the most needed operation for now (and only implement that) | 23:43 |
| @sonney2k | so when we figure out it doesn't work we can do it differently again | 23:43 |
| lambday_ | sonney2k: alright then | 23:43 |
| lambday_ | in the krylstat code (that erlend wrote) they use eigen3 everywhere.. ours shouldn't be slower than that | 23:44 |
| lambday_ | I'm sorry again. our mil x mil sparse matrix would have at least a mil nz entries (main diagonal) | 23:46 |
| lambday_ | anyway, I'm trying | 23:47 |
| @sonney2k | lambday_, try to create a 'realistic' sparse matrix ohh and note that sparse vector indices must be sorted | 23:50 |
| lambday_ | sonney2k: yes | 23:51 |
| shogun-buildbot | build #1326 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/1326 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org> | 23:59 |
| --- Log closed Sun Jul 07 00:00:20 2013 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!