--- Log opened Thu May 16 00:00:04 2013 | ||
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 00:09 | |
-!- goldbug [~quassel@global-2-1.nat.csx.cam.ac.uk] has quit [Read error: Connection reset by peer] | 00:17 | |
van51 | anyone here? | 00:24 |
---|---|---|
van51 | I am getting this weird error : | 01:14 |
van51 | error: mkl_classify_christmas_star.m: /usr/lib/x86_64-linux-gnu/octave/site/oct/api-v48+/x86_64-pc-linux-gnu/sg.oct: failed to load: /usr/lib/x86_64-linux-gnu/octave/site/oct/api-v48+/x86_64-pc-linux-gnu/sg.oct: undefined symbol: _ZN6shogun9CSGObject6equalsEPS0_e | 01:14 |
van51 | if someone has an idea let me know :) | 01:15 |
van51 | i'll be awake in a few hours again | 01:15 |
-!- iglesiasg [d58f3252@gateway/web/freenode/ip.213.143.50.82] has quit [Ping timeout: 250 seconds] | 01:38 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 03:27 | |
-!- FSCV [~FSCV@187.210.54.166] has joined #shogun | 03:30 | |
-!- nube [~rho@49.244.41.130] has quit [Quit: Leaving.] | 03:43 | |
shogun-buildbot | build #396 of nightly_default is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/396 | 03:45 |
-!- FSCV [~FSCV@187.210.54.166] has quit [Quit: Leaving] | 03:59 | |
-!- kakashi__ [~kakashi_@nltk/kakashi] has joined #shogun | 05:11 | |
-!- kakashi_ [~kakashi_@nltk/kakashi] has quit [Ping timeout: 245 seconds] | 05:13 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 05:22 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Remote host closed the connection] | 05:26 | |
@sonney2k | van51, CR == code review | 05:47 |
@sonney2k | van51, I guess git clean -dfx is your friend (will delete everything not in git!) and then configure/compile again | 05:48 |
-!- foulwall_ [~foulwall@2001:da8:215:503:f039:facc:3fd2:afa7] has joined #shogun | 06:10 | |
@sonney2k | moin foulwall_ | 06:13 |
foulwall_ | hello sonney2k. | 06:13 |
foulwall_ | almost done the visualization. | 06:13 |
-!- nube [~rho@116.90.239.3] has quit [Quit: Leaving.] | 06:18 | |
@sonney2k | foulwall_, for kernels you mean? | 06:18 |
foulwall_ | sonney2k: yes. | 06:18 |
-!- nube [~rho@116.90.239.3] has joined #shogun | 06:18 | |
foulwall_ | sonney2k: I use dvalcarce's heatmap code. | 06:19 |
@sonney2k | foulwall_, makes sense | 06:21 |
@sonney2k | foulwall_, however, for classifiers we need heatmap + contour lines | 06:22 |
foulwall_ | Do we need contour in kernel matrix? | 06:23 |
@sonney2k | foulwall_, well it wouldn't hurt but it might be good enough for now | 06:24 |
@sonney2k | foulwall_, basically what I have in mind is the heat map plus on top http://vis.lbl.gov/NERSC/Software/pvwave/docs/pvwavehtmlhelp/waveref/files/c2c-2-7.gif | 06:25 |
foulwall_ | wow, thats cool | 06:25 |
@sonney2k | but definitely a colorbar | 06:26 |
@sonney2k | showing the range | 06:26 |
foulwall_ | ok. | 06:27 |
@sonney2k | foulwall_, look at e.g. svm_classification.py example | 06:27 |
@sonney2k | or that http://shogun-toolbox.org/bigpicture/krr.png | 06:27 |
foulwall_ | I'll try that, but I'll focus on the heatmap first. | 06:32 |
-!- sumit [75e5a6f2@gateway/web/freenode/ip.117.229.166.242] has joined #shogun | 06:33 | |
foulwall_ | sonney2k: how about generate the image using gnuplot or plotting toolbox on the server? | 06:35 |
foulwall_ | sonney2k: ignore that, I'll try to draw that in js. | 06:36 |
@sonney2k | foulwall_, yeah that is what I did before... | 07:22 |
@sonney2k | but d3 would be nicer... | 07:22 |
foulwall_ | sonney2k: :) | 07:23 |
van51 | hi | 07:43 |
van51 | sonney2k: ok, i am re-compiling now | 07:43 |
van51 | but I also pulled the code from upstream yesterday and installed that and had the same error | 07:44 |
van51 | I did that in another directory | 07:44 |
-!- hoijui [~hoijui@dslb-088-074-126-103.pools.arcor-ip.net] has joined #shogun | 07:47 | |
van51 | now I am getting an error | 07:59 |
van51 | error: 'sg' undefined | 07:59 |
@sonney2k | van51, well make install | 08:07 |
van51 | sonney2k: yea I did | 08:07 |
@sonney2k | van51, and yes I have it all running here so all failures are self inflicted :D | 08:07 |
van51 | hehe | 08:07 |
@sonney2k | foulwall_, btw lisitsyn has some very nice d3 visualizations for dimension reduction methods | 08:09 |
foulwall_ | sonney2k: ok I'll ask him when he's online. | 08:10 |
@sonney2k | foulwall_, have a look at http://tapkee.lisitsyn.me/ | 08:10 |
@sonney2k | and then graphical examples | 08:10 |
@sonney2k | and click on any of them. they are all interactive | 08:10 |
@sonney2k | so mouse over will show you sth etc | 08:11 |
foulwall_ | sonney2k: cool | 08:11 |
@sonney2k | this is about as interactive is we should do it for everything | 08:11 |
@sonney2k | e.g. allow people to mouse-hover over points of the curve | 08:11 |
@sonney2k | and show the coordinates | 08:11 |
@sonney2k | dragging a point -> retrain etc | 08:11 |
foulwall_ | sonney2k: that's easy to add. | 08:12 |
foulwall_ | sonney2k: just write some callback functions and connect them to the mouse event...wait me finish this kernel demo, I'll add them later. | 08:15 |
@sonney2k | foulwall_, yeah | 08:17 |
van51 | g2g run some errands argh | 08:24 |
van51 | cu later :) | 08:24 |
-!- van51 [~van51@athedsl-320452.home.otenet.gr] has left #shogun ["QUIT :Leaving."] | 08:25 | |
-!- gsomix [~Miranda@185.6.239.216] has joined #shogun | 08:53 | |
-!- sumit [75e5a6f2@gateway/web/freenode/ip.117.229.166.242] has quit [Ping timeout: 250 seconds] | 08:53 | |
gsomix | hello | 08:53 |
sonne|work | gsomix: hullo! | 08:53 |
gsomix | Today I got interesting quest in my uni. | 08:54 |
gsomix | Need to collect five signatures to your proceeding for conference can be published. | 08:55 |
gsomix | sonne|work: how are you? | 08:56 |
@lisitsyn | back | 09:08 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 09:21 | |
sonne|work | gsomix: working :) | 09:28 |
sonne|work | and it is finally summer in berlin | 09:28 |
sonne|work | 25 C and more | 09:28 |
sonne|work | if it stays like this I am going to say hi to lake mueggelsee | 09:29 |
sonne|work | it is 15 C 'hot' already :D | 09:30 |
@lisitsyn | sonne|work: 31C here | 09:30 |
@lisitsyn | :D | 09:30 |
sonne|work | you should turn of your heaters! | 09:38 |
-!- sijin [~smuxi@144.214.222.109] has joined #shogun | 09:39 | |
foulwall_ | sonne|work: Have a look at this? http://foulwall.com/kernelmatrix.png | 09:54 |
foulwall_ | sonne|work: a output from the demo, and I don't sure if it's correct. | 09:55 |
-!- gsomix [~Miranda@185.6.239.216] has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org] | 09:55 | |
sonne|work | foulwall_: which kernel? | 09:56 |
foulwall_ | gussian kernel | 09:56 |
sonne|work | foulwall_: which input data? | 09:56 |
-!- mdragan [8d55007a@gateway/web/freenode/ip.141.85.0.122] has joined #shogun | 09:56 | |
sonne|work | foulwall_: it could be - looks a bit blocky though | 09:57 |
foulwall_ | toy data from http://shogun-toolbox.org/demos/gp/entrance | 09:57 |
foulwall_ | leave for class, will be back soon. | 10:00 |
sonne|work | foulwall_: could very well be | 10:00 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 250 seconds] | 10:05 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 10:06 | |
-!- foulwall_ [~foulwall@2001:da8:215:503:f039:facc:3fd2:afa7] has quit [Remote host closed the connection] | 10:12 | |
-!- iglesiasg [c1934d1a@gateway/web/freenode/ip.193.147.77.26] has joined #shogun | 10:44 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 10:44 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 10:49 | |
-!- iglesiasg [c1934d1a@gateway/web/freenode/ip.193.147.77.26] has quit [Quit: Page closed] | 10:53 | |
-!- iglesiasg [~iglesiasg@193.147.77.26] has joined #shogun | 10:53 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 10:53 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 250 seconds] | 11:10 | |
-!- nube [~rho@116.90.239.3] has quit [Quit: Leaving.] | 11:16 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 11:18 | |
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 246 seconds] | 11:25 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 11:31 | |
-!- lambday [e8b614c@gateway/web/cgi-irc/kiwiirc.com/x-bdtrtlfghxznuoya] has joined #shogun | 11:39 | |
-!- nube [~rho@116.90.239.3] has quit [Quit: Leaving.] | 11:50 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 11:50 | |
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 240 seconds] | 12:09 | |
-!- lambday [e8b614c@gateway/web/cgi-irc/kiwiirc.com/x-bdtrtlfghxznuoya] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 12:14 | |
-!- lambday [e8b614c@gateway/web/cgi-irc/kiwiirc.com/x-zpiudvqdzbkwuhpw] has joined #shogun | 12:22 | |
-!- foulwall [~foulwall@2001:da8:215:6110:90cc:bc83:e796:b175] has joined #shogun | 12:24 | |
-!- lambday [e8b614c@gateway/web/cgi-irc/kiwiirc.com/x-zpiudvqdzbkwuhpw] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 12:36 | |
-!- mdragan [8d55007a@gateway/web/freenode/ip.141.85.0.122] has quit [Ping timeout: 250 seconds] | 12:37 | |
@lisitsyn | ha votjakovr played a bad boy | 12:42 |
@lisitsyn | good | 12:42 |
@iglesiasg | lisitsyn, why so? | 12:47 |
@lisitsyn | iglesiasg: https://github.com/shogun-toolbox/shogun/pull/1100/files | 12:49 |
@lisitsyn | killer! | 12:49 |
@iglesiasg | hehe | 12:56 |
@iglesiasg | indeed | 12:57 |
sonne|work | votjakovr: excellent! | 13:30 |
votjakovr | :) | 13:45 |
sonne|work | you are one of us now - we want to see more :) | 13:46 |
-!- iglesiasg [~iglesiasg@193.147.77.26] has quit [Quit: Leaving] | 13:48 | |
votjakovr | Ок | 13:50 |
@wiking | :> | 14:03 |
@wiking | it's nice that somebody picked up the string and continued :P | 14:04 |
-!- hoijui [~hoijui@dslb-088-074-126-103.pools.arcor-ip.net] has quit [Quit: Leaving] | 14:23 | |
-!- iglesiasg [~iglesiasg@193.147.77.26] has joined #shogun | 14:23 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 14:23 | |
-!- foulwall [~foulwall@2001:da8:215:6110:90cc:bc83:e796:b175] has quit [Remote host closed the connection] | 14:42 | |
-!- lambday [3ba2174d@gateway/web/cgi-irc/kiwiirc.com/ip.59.162.23.77] has joined #shogun | 15:25 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 15:26 | |
-!- FSCV [~FSCV@187.210.54.166] has joined #shogun | 15:28 | |
-!- mdragan [8d55007a@gateway/web/freenode/ip.141.85.0.122] has joined #shogun | 15:33 | |
-!- nube [~rho@49.244.79.189] has joined #shogun | 15:36 | |
-!- gsomix [~Miranda@37.61.180.163] has joined #shogun | 15:43 | |
gsomix | only one signature... (~T___T)~ | 15:44 |
-!- foulwall [~foulwall@2001:da8:215:6100:b91c:8d63:d54f:d5f2] has joined #shogun | 15:47 | |
-!- kakashi__ [~kakashi_@nltk/kakashi] has quit [Read error: Connection reset by peer] | 15:49 | |
foulwall | sonney2k: hi sonney2k | 15:55 |
foulwall | sonney2k: I've sent a pull request on the kernel matrix visualization. | 15:55 |
foulwall | sonney2k: sorry there are some bugs on my server, I cant' make the demo run on server instantly, so merge it and have a look pls. | 15:58 |
foulwall | iglesiasg: lisitsyn: sonney2k: wiking: I've found that when I run some demo on my server(gentoo x86), when I create some object like GausianKernel, the django server always crash and give me a message of "illegal instruction" | 16:03 |
foulwall | that's strange. If there's no solution, I'll rebuild the server with ubuntu tomorrow... | 16:04 |
-!- foulwall [~foulwall@2001:da8:215:6100:b91c:8d63:d54f:d5f2] has quit [Remote host closed the connection] | 16:05 | |
@iglesiasg | foulwall: mm I see. Have you tried tracking where this error fires? | 16:15 |
-!- van51 [~van51@195.251.235.71] has joined #shogun | 16:17 | |
van51 | sonney2k: I am now making a PR about the CombinedFeatures class | 16:18 |
van51 | I ran the example and got the same w | 16:18 |
van51 | have a look at the changed files and if you think more examples are needed let me know | 16:19 |
-!- foulwall [~foulwall@2001:da8:215:c252:33:d3d0:c07a:558c] has joined #shogun | 16:19 | |
van51 | if everything is ok, I 'm thinking of starting work on the ProductKernel class that you mentioned | 16:19 |
van51 | and after that to make the unit tests for those three (changed classes) | 16:19 |
van51 | I can't stay for now, but I'll be checking my email for comments on the PR on github | 16:20 |
van51 | looking forward to your input :) | 16:20 |
van51 | bye 4 now | 16:21 |
-!- van51 [~van51@195.251.235.71] has left #shogun [] | 16:21 | |
foulwall | iglesiasg: havent tracked yet, looks like a memory failure. | 16:23 |
foulwall | iglesiasg: I'll track the error tomorrow. late in the night now. night | 16:24 |
-!- foulwall [~foulwall@2001:da8:215:c252:33:d3d0:c07a:558c] has left #shogun [] | 16:25 | |
@iglesiasg | fouldwall: good night | 16:40 |
sonne|work | foulwall, we need some minimal example.. it sounds a lot like you compiled on one machine and copied things over | 16:56 |
sonne|work | but lets see... | 16:57 |
sonne|work | van51 ohh cool. I will check | 16:57 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has left #shogun [] | 17:03 | |
-!- gsomix [~Miranda@37.61.180.163] has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org] | 17:21 | |
-!- iglesiasg [~iglesiasg@193.147.77.26] has quit [Quit: Leaving] | 17:28 | |
-!- mdragan [8d55007a@gateway/web/freenode/ip.141.85.0.122] has quit [Ping timeout: 250 seconds] | 17:33 | |
-!- iglesiasg [~fjig@share-01.csc.kth.se] has joined #shogun | 18:08 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 18:08 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 250 seconds] | 18:10 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 18:57 | |
-!- pickle27 [~kevin@rcv3-lab-pc.ee.queensu.ca] has joined #shogun | 19:04 | |
pickle27 | hey everyone | 19:04 |
-!- dvalcarce [5b757556@gateway/web/freenode/ip.91.117.117.86] has joined #shogun | 19:08 | |
dvalcarce | hi all | 19:09 |
dvalcarce | sonney2k, what do you think about these countour lines? https://github.com/shogun-toolbox/shogun-web/pull/27#issuecomment-18014790 | 19:09 |
@iglesiasg | daad/ | 19:39 |
-!- iglesiasg [~fjig@share-01.csc.kth.se] has quit [Quit: Lost terminal] | 19:39 | |
-!- iglesiasg [~fjig@share-01.csc.kth.se] has joined #shogun | 19:42 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 19:42 | |
@iglesiasg | hey dvalcarce | 19:42 |
@iglesiasg | I think it is looking very nice | 19:42 |
dvalcarce | igleasiasg, thanks! :) | 19:43 |
pickle27 | iglesiasg: how would you write code to subtract a vector from every row of a matrix using Eigen3? | 19:50 |
pickle27 | iglesiasg: I've done it with loops but I'm wondering if there is a better way | 19:50 |
@iglesiasg | pickle27: hey, yeah there is something to do that | 19:51 |
@iglesiasg | pickle27: take a look to rwise and cwise | 19:51 |
@iglesiasg | pickle27: let me check, I may have a minimal example doing sth similar | 19:51 |
pickle27 | iglesiasg: okay thanks | 19:52 |
@iglesiasg | pickle27: no I don't have one | 19:53 |
@iglesiasg | but check if you can do something like | 19:53 |
@iglesiasg | matrix.rowwise() - vector | 19:54 |
@iglesiasg | pickle27: let me know if it works | 19:54 |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: Leaving] | 19:55 | |
pickle27 | iglesiasg: that seems to have worked thanks | 20:00 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 250 seconds] | 20:02 | |
@iglesiasg | pickle27: cool | 20:03 |
pickle27 | iglesiasg: another quick one that you might be able to help with | 20:10 |
pickle27 | iglesiasg: I need to replace any 0's with 1's in my matrix | 20:10 |
pickle27 | iglesiasg: right now I'm looping but I'd like to try and use eigen3 | 20:10 |
@iglesiasg | pickle27: what other values can that matrix have? | 20:15 |
pickle27 | iglesiasg: its a matrix of standard deviation so positive values probably between 0 and 5 | 20:15 |
@iglesiasg | I see | 20:16 |
@iglesiasg | there must be some kind of find method or even some functionality via operators that return a Matrix::Index | 20:17 |
@iglesiasg | I don't recall to have done it before though | 20:17 |
pickle27 | iglesiasg: okay I'll look into it | 20:19 |
-!- lisitsyn [~blackburn@188-122-243-31.clients.tlt.100megabit.ru] has joined #shogun | 20:33 | |
-!- mode/#shogun [+o lisitsyn] by ChanServ | 20:33 | |
pickle27 | iglesiasg: have you had a chance to check out my eigen3 multi class lda at all? | 20:44 |
-!- gsomix [~Miranda@83.234.54.253] has joined #shogun | 20:46 | |
gsomix | good evening | 20:46 |
pickle27 | lisitsyn: I started doing a bit of work on my GSoC proposed project yesterday and it occurred to me that we're probably going to want ADJ and ICA as part of Tapkee is that right? | 20:46 |
-!- HeikoS [~heiko@nat-183-152.internal.eduroam.ucl.ac.uk] has joined #shogun | 20:46 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 20:46 | |
@lisitsyn | pickle27: oh good question | 20:47 |
@lisitsyn | pickle27: I am afraid of that actually | 20:47 |
lambday | HeikoS: hi :) | 20:47 |
@HeikoS | lambday: hi! | 20:47 |
@HeikoS | coming for you! :) | 20:47 |
@HeikoS | how is it going? | 20:47 |
lambday | I am thinking of getting the elliptic curve thing done asap | 20:48 |
@lisitsyn | HeikoS: what about meeee | 20:48 |
@HeikoS | lambday: yeah thats a good one to start since no thinking involved | 20:48 |
@HeikoS | lisitsyn: oh yes, and of course for you :D | 20:48 |
lambday | lol :P | 20:48 |
lambday | HeikoS:yes.. | 20:48 |
lambday | should we use arprec too in the project? | 20:49 |
@HeikoS | doesnt krylstat do everything for you? | 20:49 |
@HeikoS | or was it based on arprec | 20:50 |
@HeikoS | forgot | 20:50 |
lambday | HeikoS: no I mean, they use std:complex and double datatypes for few methods | 20:50 |
lambday | and in other places, they redefined those methods using arprec | 20:50 |
lambday | its for arbitrary precision stuff I guess | 20:50 |
@HeikoS | I see | 20:50 |
@HeikoS | so they just overload the std types and use that? | 20:50 |
lambday | actually, the code is quite complicated | 20:51 |
lambday | they have written everything separately | 20:51 |
lambday | and redefined stuffs in many other places | 20:51 |
@HeikoS | Erlend said that this would be almost copy/paste didnt he? | 20:51 |
lambday | yes in few places they used overloading | 20:51 |
-!- iglesiasg [~fjig@share-01.csc.kth.se] has quit [Quit: leaving] | 20:51 | |
lambday | ummm... yes | 20:52 |
lambday | but those methods, they haven't used in later parts | 20:52 |
lambday | :-/ | 20:52 |
@HeikoS | I see | 20:52 |
@HeikoS | whats your suggestion then? | 20:52 |
lambday | we can start with float64_t | 20:52 |
pickle27 | lisitsyn: afraid lol, how so? | 20:52 |
lambday | as of now | 20:52 |
@HeikoS | lambday why would we need anything more precicse than float64_t | 20:53 |
lambday | later we can use arprec if needed | 20:53 |
lambday | I guess | 20:53 |
@HeikoS | is this really the reason why they use arpec? | 20:53 |
@lisitsyn | pickle27: I do not want to replace shogun by tapkee inside of shogun | 20:53 |
@lisitsyn | :D | 20:53 |
@HeikoS | maybe something else? | 20:53 |
@lisitsyn | pickle27: we have to ask heiko and soeren what do they think | 20:53 |
lambday | HeikoS: they use arprec for mp_real and mp_complex datatypes | 20:53 |
lambday | nothing else (at least I haven't seen anything else) | 20:53 |
lambday | we can use float64_t for real | 20:54 |
@HeikoS | which is arbritary precision stuff right? | 20:54 |
lambday | and std::complex | 20:54 |
lambday | yes | 20:54 |
@HeikoS | so the complex thing is the one that worries me | 20:54 |
lambday | me too :D | 20:54 |
@HeikoS | maybe there is something going on with arpec in there | 20:54 |
pickle27 | lisitsyn: yeah that makes sense, okay. Just seems like Tapkee has more of that style of preprocessor than Shogun does | 20:54 |
@HeikoS | for sure not for the float precision | 20:54 |
@lisitsyn | HeikoS: lambday: guys sorry to interrupt you - just a quick question | 20:54 |
@HeikoS | shoot | 20:54 |
@lisitsyn | are you going to compute some eigenvalues | 20:54 |
lambday | hmm? | 20:54 |
@lisitsyn | like partial eigendecomposition | 20:55 |
@HeikoS | lisitsyn: yes that might be necessary | 20:55 |
@HeikoS | for example largest and smallest | 20:55 |
lambday | yes | 20:55 |
lambday | lancozos | 20:55 |
@lisitsyn | HeikoS: alright, lets discuss possible approaches later then | 20:55 |
@lisitsyn | HeikoS: we need to unify that stuff for tapkee and shogun | 20:55 |
@HeikoS | lancozos is the only way I guess, but yes agreed on making things unified | 20:56 |
@lisitsyn | lancozos?? | 20:56 |
@lisitsyn | lanczos you mean? | 20:56 |
lambday | yes my mistake :( | 20:56 |
@HeikoS | copy/paste >:D | 20:56 |
lambday | :D | 20:56 |
@lisitsyn | well pure lanczos makes a little sense as there is arpack | 20:56 |
@lisitsyn | which is implicitly restarted lanczos | 20:56 |
@lisitsyn | HeikoS: http://www.ecs.umass.edu/~polizzi/feast/ | 20:57 |
@HeikoS | lambday: you see, we should ask lisitsyn when it comes to this :) | 20:57 |
@lisitsyn | HeikoS: this is quite interesting I think | 20:57 |
lambday | HeikoS: :) | 20:57 |
@HeikoS | lisitsyn: yes I want that | 20:57 |
@lisitsyn | HeikoS: I imagine two cases | 20:57 |
@lisitsyn | a) | 20:57 |
@HeikoS | looks maintained and stuff | 20:57 |
@lisitsyn | if we select some guy for dr he is going to do that | 20:58 |
@HeikoS | what? | 20:58 |
@lisitsyn | b) if we select no one for DR lambday may integrate that | 20:58 |
@HeikoS | dr? | 20:58 |
@lisitsyn | HeikoS: dim red | 20:58 |
@HeikoS | I see | 20:58 |
@HeikoS | yes good idea | 20:58 |
@HeikoS | we need it anyway | 20:58 |
lambday | hmm... okay I'll check this out | 20:59 |
@HeikoS | but I want to talk to the other mentor first | 20:59 |
@lisitsyn | HeikoS: alright sure | 20:59 |
@HeikoS | he know a lot how to compute eigenvalues | 20:59 |
@HeikoS | lambday: so for the std::complex replacement, maybe send an email to erlend and ask whether its fine to go with the not-overloaded standard types | 20:59 |
@HeikoS | and if yes, go for it | 20:59 |
@HeikoS | this should not involve any planning | 20:59 |
@HeikoS | just a few functions of statistics | 20:59 |
lambday | it should go in statistics? | 21:00 |
@HeikoS | lambday: maybe, what do you have in mind? | 21:00 |
lambday | HeikoS: umm... I was thinking of a separate class but that won't be necessary I think | 21:01 |
lambday | statistics is good | 21:01 |
@sonney2k | HeikoS, lisitsyn good evening gentlemen | 21:01 |
@HeikoS | sonney2k: hi! | 21:01 |
@HeikoS | lambday, so how many methods are we talking about? | 21:01 |
@HeikoS | two or three? | 21:01 |
lambday | three | 21:01 |
@lisitsyn | sonney2k: abend | 21:01 |
lambday | and few overloaded stuffs | 21:02 |
@HeikoS | lambday I dont think this stuff will ever be used somewhere else, so Statistics should be fine | 21:02 |
@HeikoS | we can always move it later | 21:02 |
lambday | HeikoS: yup | 21:02 |
@sonney2k | I just finished mowing the lawn so lisitsyn HeikoS lets discuss? | 21:02 |
@HeikoS | ok good | 21:02 |
lambday | oh and one more thing | 21:02 |
lambday | (actually many more things) | 21:02 |
@lisitsyn | sonney2k: just in public yes? :D | 21:02 |
lambday | but | 21:02 |
lambday | they used normal vectors for the sampling thing | 21:03 |
@HeikoS | lambday go ahead | 21:03 |
lambday | which they generate within the rational sampler .. | 21:03 |
lambday | but | 21:03 |
@HeikoS | sonney2k: give me a few minutes to catch up with lambday and to work through some PRs | 21:03 |
lambday | they don't use that later too | 21:03 |
lambday | rather rewritten those stuffs for probing vectors | 21:03 |
lambday | so, I was thinking that vectors should be provided externally to the class | 21:03 |
lambday | so that initially we can work with normal vectors later we can replace with probing vec stuff | 21:04 |
@HeikoS | lambday this issues can easily be solbed with a class for the Hutchison estimator for the trace which is then overloaded | 21:04 |
@HeikoS | so yes, start with std Gaussian samples | 21:04 |
@HeikoS | but add the possibility to use other samplers instead | 21:04 |
@HeikoS | like probing | 21:04 |
lambday | HeikoS: yes... | 21:04 |
lambday | hmm | 21:04 |
@HeikoS | but also random [-1,1] vectors | 21:05 |
lambday | yes... | 21:05 |
@HeikoS | which have a lower variance than the Gaussian | 21:05 |
@HeikoS | start with the simplest | 21:05 |
lambday | okay | 21:05 |
@sonney2k | lambday, can you discussion with HeikoS wait a bit (say <30mins?) - we have a meeting with him just now? | 21:05 |
lambday | sonney2k: oh no problem :( sorry :( | 21:05 |
@HeikoS | but dont use a flag for this, rather polymorphic thing | 21:05 |
lambday | HeikoS: alright :) | 21:05 |
@sonney2k | HeikoS, please ... | 21:06 |
@HeikoS | sonney2k ok lets go | 21:06 |
@sonney2k | thx | 21:06 |
@HeikoS | lambday, we can continue in private chat | 21:06 |
lambday | HeikoS: okie | 21:06 |
@sonney2k | pickle27, hey there ! on you blas vs eigen3 thing | 21:12 |
@sonney2k | pickle27, did you compile without optimizations perhaps? | 21:12 |
pickle27 | pickle27: yeah? | 21:12 |
pickle27 | pickle27: I can rebuild shogun and run it again | 21:13 |
@sonney2k | pickle27, running w/o optimizations for eigen3 has a massive impact - but close to none for blas | 21:13 |
@lisitsyn | haha! | 21:13 |
@lisitsyn | without optimizations!?!?! | 21:14 |
@lisitsyn | eigen?! | 21:14 |
pickle27 | sonney2k: right, also the debugging flag was probably on too | 21:14 |
@lisitsyn | ohoh | 21:14 |
@lisitsyn | pickle27: you basically get like 17 calls in stack | 21:14 |
pickle27 | sonney2k: lisitsyn does shogun build without eigen debug by default or do I need to set that | 21:14 |
@lisitsyn | just to compute norm | 21:14 |
@lisitsyn | pickle27: should build without debu | 21:15 |
@lisitsyn | but -O3 is most important | 21:15 |
@lisitsyn | pickle27: it makes no sense to compare blas with -O0 | 21:15 |
@sonney2k | pickle27, if you did ./configure --disable-optimizations then -> slow | 21:15 |
pickle27 | I'll rebuild and double check | 21:15 |
@lisitsyn | eigen is even more slow | 21:15 |
@lisitsyn | pickle27: it heavily uses metaprogramming | 21:15 |
@lisitsyn | so with O0 nothing goes away | 21:16 |
@sonney2k | pickle27, debugging flag doesn't hurt | 21:16 |
@lisitsyn | with -O3 it goes down to a few SSE calls | 21:16 |
@lisitsyn | so matters crazy a lot | 21:16 |
pickle27 | lisitsyn: do I need to set anything for -03? | 21:17 |
@lisitsyn | pickle27: no just do not put --disable-optimizations here | 21:17 |
pickle27 | k | 21:17 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 264 seconds] | 21:27 | |
-!- van51 [~van51@athedsl-320452.home.otenet.gr] has joined #shogun | 21:33 | |
pickle27 | lisitsyn: okay yeah now I get that eigen3 at 0.17 seconds | 21:33 |
@lisitsyn | pickle27: what is blas again? | 21:33 |
pickle27 | blas was 0.25 before but its not easy to re-test it | 21:34 |
@lisitsyn | alright | 21:34 |
@lisitsyn | pickle27: here we go then | 21:34 |
pickle27 | lisitsyn: sonney2k so I'll remove the blas version of MCLDA and then where do we sit with merging the eigen3 version? | 21:35 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 21:36 | |
vgorbati | lisitsyn: hello, here? | 21:37 |
@lisitsyn | vgorbati: hey, yes | 21:37 |
pickle27 | lisitsyn: I've got it ifdef'd right now to default to eigen3 if both libs are present otherwise if only blas is there it will use blas. But it seemed the consensus was to keep just the eigen3 version | 21:37 |
@sonney2k | pickle27, yes we intend to switch to eigen3 completely | 21:37 |
@sonney2k | and drop all lapack/blas dependencies in the long term | 21:38 |
vgorbati | lisitsyn: here' s the result of applying manifold sculpting to swissroll with small gaps https://www.dropbox.com/s/tdlpwup7fwsl8qi/manifoldSculptingEmbedding.png :) | 21:38 |
@sonney2k | pickle27, eigen3 is much more readable and heavility developed (part of kde) | 21:38 |
pickle27 | sonney2k: yeah I much prefer it :) | 21:38 |
@lisitsyn | vgorbati: ha looks crazy | 21:38 |
vgorbati | lisitsyn: but some dependency is captured | 21:39 |
@lisitsyn | vgorbati: what about 2d embedding? | 21:39 |
vgorbati | lisitsyn: what data should I try? | 21:39 |
-!- dvalcarce [5b757556@gateway/web/freenode/ip.91.117.117.86] has quit [Quit: Page closed] | 21:39 | |
@lisitsyn | vgorbati: swissroll 3d->2d | 21:39 |
-!- lambday [3ba2174d@gateway/web/cgi-irc/kiwiirc.com/ip.59.162.23.77] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 21:40 | |
vgorbati | lisitsyn: this is swissroll 3d-2d, the result is just plotted in 3D | 21:40 |
van51 | hello all | 21:40 |
@lisitsyn | vgorbati: ehmmm where is the embedding? | 21:41 |
vgorbati | lisitsyn: all points on the left lie on the same plane | 21:41 |
@lisitsyn | ohh | 21:41 |
vgorbati | lisitsyn: left is the embedding | 21:41 |
@lisitsyn | why did you plot it 3d then :D | 21:41 |
vgorbati | lisitsyn: so that I can rotate:) | 21:41 |
-!- lambday [3ba2174c@gateway/web/cgi-irc/kiwiirc.com/ip.59.162.23.76] has joined #shogun | 21:41 | |
van51 | sonney2k: by your note 'no newline here', you mean it needs a newline after the SG_INFO() call? | 21:42 |
@sonney2k | errm there was some unneeded newline | 21:43 |
@sonney2k | van51, let me check | 21:43 |
vgorbati | lisitsyn: ok, there's now also a small unit-test for manifold sculpting (smoke test), I have and idea of adding the test, that checks whether average neighbor distance is preserved - should I add it also, or skip for now? | 21:43 |
@lisitsyn | vgorbati: no just skip for now | 21:44 |
vgorbati | lisitsyn: so, can I issue a new PR? | 21:44 |
@lisitsyn | vgorbati: yes sure | 21:44 |
vgorbati | lisitsyn: were there any tapkee updates recently?:) | 21:44 |
@lisitsyn | vgorbati: yes | 21:44 |
@lisitsyn | so you'd have to rebase | 21:44 |
vgorbati | :( | 21:45 |
@lisitsyn | sorry :D | 21:45 |
-!- mdragan [8d55007a@gateway/web/freenode/ip.141.85.0.122] has joined #shogun | 21:45 | |
vgorbati | lisitsyn: Are you talking about this one https://github.com/shogun-toolbox/shogun/commit/01205a7527236ea5f5e95cccf167650c038b80a7, or there were another? | 21:46 |
@lisitsyn | vgorbati: no, that's the last one we have | 21:46 |
vgorbati | lisitsyn: ah, good, I've already rebased that one:) | 21:46 |
@sonney2k | van51, line 177 is a newline | 21:47 |
@sonney2k | jsut not needed... | 21:47 |
vgorbati | lisitsyn: I have just understood, that I could look at the latest commit in shogun/tapkee:) | 21:47 |
@lisitsyn | of course | 21:47 |
van51 | sonney2k: ah hehe, ok! | 21:48 |
van51 | I believe I've made the changes | 21:49 |
van51 | let me re-check that everything works ok and I'll push | 21:49 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 21:50 | |
shogun-notifier- | shogun: abinashpanda :develop * 2f35129 / examples/undocumented/python_modular/regression_gaussian_process_modelselection.py: https://github.com/shogun-toolbox/shogun/commit/2f351293ea210f2709577498913a9334facef086 | 21:50 |
shogun-notifier- | shogun: example ML2 parameter selection for GP Regression | 21:50 |
shogun-notifier- | shogun: abinashpanda :develop * 237b7dc / examples/undocumented/python_modular/graphical/regression_gaussian_process_modelselection.py: https://github.com/shogun-toolbox/shogun/commit/237b7dc12e43d156305bb259a89c20dc744eec88 | 21:50 |
shogun-notifier- | shogun: example ML2 parameter selection for GP Regression | 21:50 |
shogun-notifier- | shogun: abinashpanda :develop * e4e5a97 / examples/undocumented/python_modular/regression_gaussian_process_modelselection.py: https://github.com/shogun-toolbox/shogun/commit/e4e5a970485ff9b23678a5300e9f07968107828d | 21:50 |
shogun-notifier- | shogun: file shifted to grahical | 21:50 |
shogun-notifier- | shogun: Heiko Strathmann :develop * aea5afa / examples/undocumented/python_modular/graphical/regression_gaussian_process_modelselection.py: https://github.com/shogun-toolbox/shogun/commit/aea5afa6c5cc3d2485c7622cbd949f5ce44fe6c3 | 21:50 |
shogun-notifier- | shogun: Merge pull request #1100 from abinashpanda/develop | 21:50 |
shogun-notifier- | shogun: | 21:50 |
shogun-notifier- | shogun: example ML2 parameter selection for GP Regression | 21:50 |
pickle27 | question about coding practice for shogun - in QDA code all loop counters (i,j,k) were defined at the beginning and then re-used throughout | 21:55 |
pickle27 | is this the preferred way? is this actually any faster? | 21:55 |
@lisitsyn | pickle27: not faster | 21:55 |
pickle27 | lisitsyn: so can I define my loop counter inside my for loop then? | 21:56 |
@lisitsyn | pickle27: yes that's the same | 21:56 |
pickle27 | lisitsyn: and style wise that is okay with you guys? | 21:56 |
@lisitsyn | pickle27: yes totally | 21:57 |
pickle27 | lisitsyn: okay, I personally think it easier to read that way | 21:57 |
@sonney2k | pickle27, my rule is to define variables at the lates possible moment | 21:58 |
@sonney2k | otherwise it is too hard to figure out when the variable is used | 21:59 |
pickle27 | okay right on, just doing some clean up stuff in mclda | 21:59 |
@lisitsyn | sonney2k: it is also an optimization rule | 21:59 |
shogun-buildbot | build #983 of deb2 - static_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/983 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, abinashpanda <abinash.panda.ece10@itbhu.ac.in> | 21:59 |
@sonney2k | lisitsyn, don't we have that in our coding guidelines? | 21:59 |
@lisitsyn | no idea | 21:59 |
@lisitsyn | :D | 21:59 |
@sonney2k | no | 22:00 |
@sonney2k | I guess this needs an update | 22:00 |
shogun-buildbot | build #993 of bsd1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/993 blamelist: abinashpanda <abinash.panda.ece10@itbhu.ac.in> | 22:04 |
@sonney2k | Ohh I am lost in this new g+ | 22:11 |
@sonney2k | so many columns and random switches around | 22:11 |
@lisitsyn | sonney2k: haha | 22:11 |
gsomix | sonne|work: oh, new g+... | 22:12 |
gsomix | sonney2k | 22:12 |
gsomix | btw, g+ is slow on my machine | 22:13 |
@sonney2k | on mine two | 22:13 |
@sonney2k | but travis is even worse | 22:13 |
@HeikoS | lisitsyn: do you take care of the KMM exsample? | 22:14 |
@lisitsyn | travis eats 100% of cpu | 22:14 |
@sonney2k | It makes my machines fans go airplane | 22:14 |
@lisitsyn | HeikoS: I am not applying for gsoc this year | 22:14 |
@lisitsyn | :D | 22:14 |
@HeikoS | ha! | 22:15 |
@sonney2k | lisitsyn, HeikoS thanks for the meeting if you come earlier to berlin we should have a beer :) | 22:15 |
@sonney2k | lisitsyn, you lazy mentor! | 22:15 |
@sonney2k | and org admin. | 22:15 |
@HeikoS | always up for beer :) | 22:15 |
* sonney2k is in kick ass ass kick mood today | 22:15 | |
-!- travis-ci [~travis-ci@ec2-54-234-86-18.compute-1.amazonaws.com] has joined #shogun | 22:15 | |
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/7231160 | 22:15 |
-!- travis-ci [~travis-ci@ec2-54-234-86-18.compute-1.amazonaws.com] has left #shogun [] | 22:15 | |
@lisitsyn | sonney2k: should I solve tasks instead of students? | 22:15 |
@lisitsyn | :D | 22:15 |
van51 | sonney2k: I updated it, have a look when you can | 22:15 |
@sonney2k | lisitsyn, don't let your students suffer alone - suffer with them - so yes! | 22:16 |
@sonney2k | van51, does the octave one still work? | 22:16 |
shogun-buildbot | build #984 of deb2 - static_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/984 | 22:16 |
van51 | sonney2k: yeap | 22:16 |
pickle27 | sonney2k: I pushed a final version of MCLDA with Eigen3 and a unit test, let me know what you think it might be ready to merge | 22:16 |
@HeikoS | lisitsyn: so there is a PR on this right | 22:16 |
@sonney2k | HeikoS, lisitsyn btw I was analyzing memory consumption on one of our servers at work today | 22:16 |
@HeikoS | and? | 22:17 |
@sonney2k | and guess what I could come up with a 100% reliable predictor for non-weekdays / weekdays | 22:17 |
@lisitsyn | HeikoS: ahhhhhh | 22:17 |
@lisitsyn | I didn't know | 22:17 |
@sonney2k | just based on memory consumption | 22:17 |
@lisitsyn | :D | 22:17 |
@HeikoS | sonney2k, haha thats cool | 22:17 |
@HeikoS | how? | 22:17 |
@HeikoS | whats with travis? | 22:18 |
@HeikoS | cannot see certain pages | 22:18 |
@sonney2k | well you know traffic is a lot less on non-weekdays/in holiday season | 22:18 |
@HeikoS | brwoser crashes | 22:18 |
@sonney2k | van51, alright then lets bite the apple | 22:18 |
@HeikoS | I like my weekends :) | 22:19 |
@sonney2k | van51, the code is really much easier w/ arrays... | 22:21 |
shogun-notifier- | shogun: van51 :develop * 5a36f0e / src/shogun/ (6 files): https://github.com/shogun-toolbox/shogun/commit/5a36f0e77e5455d78b9bc0b90f1e18360d758bf6 | 22:21 |
shogun-notifier- | shogun: Conversion from CList to CDynamicObjectArray in CCombinedFeatures | 22:21 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 34c42ef / src/shogun/ (6 files): https://github.com/shogun-toolbox/shogun/commit/34c42ef2f4080b3fd3f86a24d1fe50955129832c | 22:21 |
shogun-notifier- | shogun: Merge pull request #1101 from van51/feature/comb_feat | 22:21 |
shogun-notifier- | shogun: | 22:21 |
shogun-notifier- | shogun: Conversion from CList to CDynamicObjectArray in CCombinedFeatures | 22:21 |
van51 | yea, pretty straightforward | 22:21 |
-!- mdragan [8d55007a@gateway/web/freenode/ip.141.85.0.122] has quit [Ping timeout: 250 seconds] | 22:22 | |
@sonney2k | van51, I would be happy if you did this product kernel finally | 22:22 |
-!- nube [~rho@49.244.79.189] has quit [Ping timeout: 276 seconds] | 22:22 | |
@sonney2k | and then remove the obsolete crap stuff like get_first/last* | 22:22 |
van51 | sonney2k: I'll get started with ProductKernel now then | 22:23 |
van51 | but I also have a report due for tomorrow and it might take me a little longer | 22:24 |
@sonney2k | van51, no problem | 22:24 |
shogun-buildbot | build #1122 of deb3 - modular_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1122 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, abinashpanda <abinash.panda.ece10@itbhu.ac.in> | 22:25 |
@sonney2k | lisitsyn, haha maybe we shoudl switch to using http://www.scons.org/ | 22:28 |
@sonney2k | instead of cmake | 22:28 |
@lisitsyn | heard about that | 22:28 |
@lisitsyn | sonney2k: is it better/ | 22:28 |
@sonney2k | no idea | 22:29 |
@sonney2k | gsomix, btw any news on the getline reader? | 22:30 |
vgorbati | when I try to push to my fork repository, I get 'To prevent you from losing history, non-fast-forward updates were rejected'. How could it be, regarding that nobody else works with this repo? | 22:30 |
gsomix | sonney2k: yep, I'll push it tomorrow at evening. | 22:31 |
shogun-buildbot | build #985 of deb2 - static_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/985 blamelist: Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com> | 22:31 |
@sonney2k | van51, look the shogun-buildbot is your fan now | 22:32 |
van51 | sonney2k: not sure I enjoy that | 22:33 |
@sonney2k | yeah well some examples no longer run | 22:34 |
@sonney2k | http://shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/985/steps/test%20libshogun/logs/stdio | 22:34 |
vgorbati | lisitsyn: anyway, I mad a force push. Could you please close the old PR, so that I send a new one and see if build passes? | 22:34 |
van51 | some examples probably need rewriting | 22:34 |
van51 | yea | 22:34 |
shogun-buildbot | build #1123 of deb3 - modular_interfaces is complete: Failure [failed compile lua_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1123 blamelist: abinashpanda <abinash.panda.ece10@itbhu.ac.in> | 22:34 |
van51 | should I deal with that first? | 22:34 |
@lisitsyn | vgorbati: just close it :) | 22:34 |
shogun-buildbot | build #994 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/994 blamelist: Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com> | 22:34 |
@lisitsyn | you can do that too | 22:34 |
@sonney2k | van51, I looked at this particular example but failed fixing it | 22:34 |
@sonney2k | van51, I will need HeikoS' help here | 22:34 |
@HeikoS | ? | 22:35 |
@sonney2k | maybe I for now comment the failing parts in the example | 22:35 |
@sonney2k | HeikoS, I thought you don't have any time? | 22:35 |
-!- hoijui [~hoijui@dslb-088-074-126-103.pools.arcor-ip.net] has joined #shogun | 22:35 | |
@HeikoS | sonney2k: exactly, I just saw my name, whats up? | 22:36 |
@sonney2k | HeikoS, still the same issue with the example | 22:36 |
@HeikoS | sonney2k: I see, yes will do next week | 22:36 |
@HeikoS | comment out for now | 22:36 |
-!- nube [~rho@49.244.92.7] has joined #shogun | 22:36 | |
@HeikoS | I will tell you when I did it | 22:36 |
@HeikoS | maybe add an issue and assign it to me | 22:36 |
@sonney2k | van51, look problem resolved :D | 22:36 |
van51 | haha | 22:37 |
@sonney2k | HeikoS, done | 22:39 |
@HeikoS | thanks! | 22:39 |
@sonney2k | there will be 2-3 similar examples | 22:39 |
* sonney2k sth is weird today | 22:40 | |
vgorbati | lisitsyn: ok, I have just send the updated PR | 22:40 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Quit: Page closed] | 22:40 | |
@sonney2k | HeikoS, I can give you more issues if it makes you happy :D | 22:40 |
@HeikoS | sonney2k go ahead | 22:40 |
@HeikoS | :) | 22:40 |
@HeikoS | but not more than 3 :D | 22:40 |
@sonney2k | HeikoS, I am going to select the toughest I can find :D | 22:45 |
@HeikoS | then only 2.5, sonney2k | 22:46 |
-!- HeikoS [~heiko@nat-183-152.internal.eduroam.ucl.ac.uk] has left #shogun [] | 23:01 | |
-!- travis-ci [~travis-ci@ec2-54-242-254-73.compute-1.amazonaws.com] has joined #shogun | 23:01 | |
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/7232011 | 23:01 |
-!- travis-ci [~travis-ci@ec2-54-242-254-73.compute-1.amazonaws.com] has left #shogun [] | 23:01 | |
shogun-buildbot | build #1124 of deb3 - modular_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1124 blamelist: Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com> | 23:08 |
shogun-notifier- | shogun: vladislav.horbatiuk@gmail.com :develop * 87ebf66 / src/shogun/lib/tapkee/ (9 files): https://github.com/shogun-toolbox/shogun/commit/87ebf66ae41a7b0522f8ca8b159f772f0e048269 | 23:10 |
shogun-notifier- | shogun: Added an implementation of manifold sculpting DR method (see http://axon.cs.byu.edu/papers/gashler2011smc.pdf); | 23:10 |
shogun-notifier- | shogun: Added a method in tapkee library for converting FeatureCallback into a DenseMatrix (it was used previously a | 23:10 |
shogun-notifier- | shogun: few times); refactored a little; added all the stuff needed for the new method to be 'visible' from outside the tapkee. | 23:10 |
shogun-notifier- | shogun: vladislav.horbatiuk@gmail.com :develop * b99f50d / src/shogun/lib/tapkee/ (8 files): https://github.com/shogun-toolbox/shogun/commit/b99f50de8fa426c323eef1339b27b68b6be33cae | 23:10 |
shogun-notifier- | shogun: Fixed issues and refactored according to the comments in PR (https://github.com/shogun-toolbox/shogun/pull/1086). | 23:10 |
shogun-notifier- | shogun: vladislav.horbatiuk@gmail.com :develop * f6db86c / / (6 files): https://github.com/shogun-toolbox/shogun/commit/f6db86c7824b1a302e9ab1262b2077c25cab4865 | 23:10 |
shogun-notifier- | shogun: Added a wrapper for a manifold sculpting DR method implemented in tapkee; added a modular wrapper for the wrapper:); fixed some bugs in the algorithm itself. | 23:10 |
shogun-notifier- | shogun: Sergey Lisitsyn :develop * 1cfce0c / / (17 files): https://github.com/shogun-toolbox/shogun/commit/1cfce0cd08ba6383cd093fb2a64edb1dc54c2585 | 23:10 |
shogun-notifier- | shogun: Merge pull request #1105 from vladislav-horbatiuk/vgorbati-develop | 23:10 |
shogun-notifier- | shogun: | 23:10 |
shogun-notifier- | shogun: Added full implementation of manifold sculpting DR method and a smoke test for it. | 23:10 |
@sonney2k | van51, ohh I figured another crasher/hang for you: run kernel_combined_custom_poly_modular.py - you will get an infinite loop | 23:13 |
van51 | sonney2k: I'll compile the module and check it | 23:15 |
shogun-buildbot | build #1187 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1187 blamelist: vladislav.horbatiuk@gmail.com <vladislav.horbatiuk@gmail.com> | 23:15 |
@sonney2k | van51, here is a backtrace http://dpaste.com/1169591/ | 23:17 |
@lisitsyn | oops | 23:17 |
@lisitsyn | fixing | 23:17 |
@sonney2k | lisitsyn, ? | 23:18 |
@lisitsyn | sonney2k: ahh that's just a crasher inside a PR | 23:19 |
shogun-buildbot | build #831 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/831 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 23:19 |
shogun-buildbot | build #986 of deb2 - static_interfaces is complete: Failure [failed test libshogun] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/986 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 23:22 |
shogun-buildbot | build #1188 of deb1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1188 | 23:24 |
shogun-buildbot | build #995 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/995 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 23:25 |
van51 | sonney2k: this is from the latest submittion, eh? | 23:29 |
-!- lambday [3ba2174c@gateway/web/cgi-irc/kiwiirc.com/ip.59.162.23.76] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 23:30 | |
shogun-buildbot | build #996 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/996 blamelist: vladislav.horbatiuk@gmail.com <vladislav.horbatiuk@gmail.com> | 23:32 |
@sonney2k | van51, I din't check the other one | 23:33 |
@sonney2k | van51, we had this failure with the test before but wait maybe we see sth w/ debug output on | 23:34 |
van51 | sonney2k: yea I was going to do that when the compilation finished | 23:35 |
@sonney2k | http://dpaste.com/1169691/ | 23:36 |
van51 | cool | 23:36 |
@sonney2k | segfault now hurray :) | 23:36 |
@sonney2k | let me valgrind this | 23:36 |
@sonney2k | [INFO] using a kernel cache of size 0 MB (33856 bytes) for CombinedKernel Kernel | 23:37 |
@sonney2k | ==4924== Invalid read of size 8 | 23:37 |
@sonney2k | ==4924== at 0xBB225B0: shogun::DynArray<shogun::CSGObject*>::get_element(int) const (DynArray.h:142) | 23:37 |
@sonney2k | ==4924== by 0xBB2277A: shogun::DynArray<shogun::CSGObject*>::insert_element(shogun::CSGObject*, int) (DynArray.h:223) | 23:37 |
@sonney2k | ==4924== by 0xBB07368: shogun::CDynamicObjectArray::insert_element(shogun::CSGObject*, int) (DynamicObjectArray.h:267) | 23:37 |
@sonney2k | ==4924== by 0xBB10FBC: shogun::CCombinedKernel::insert_kernel(shogun::CKernel*, int) (CombinedKernel.h:151) | 23:37 |
@sonney2k | ==4924== by 0xBB10FF0: shogun::CCombinedKernel::append_kernel(shogun::CKernel*) (CombinedKernel.h:162) | 23:37 |
@sonney2k | so indeed | 23:37 |
@sonney2k | out of memory read | 23:37 |
van51 | how do you valgrind this? | 23:37 |
@sonney2k | van51, valgrind python bla.py | 23:38 |
shogun-buildbot | build #987 of deb2 - static_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/987 | 23:38 |
van51 | ah I didn't know it worked with python as weel | 23:38 |
@sonney2k | van51, sure anything | 23:38 |
-!- hoijui [~hoijui@dslb-088-074-126-103.pools.arcor-ip.net] has quit [Quit: Leaving] | 23:40 | |
shogun-buildbot | build #832 of cyg1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/832 | 23:40 |
-!- iglesiasg [d58f3231@gateway/web/freenode/ip.213.143.50.49] has joined #shogun | 23:45 | |
@sonney2k | van51, ohh it could be a dynarray issue - you call insert_element and that might not work for appending an element to the end | 23:45 |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 23:46 | |
@sonney2k | van51, maybe just create a minimal test/example for jsut dynarray to validate that | 23:46 |
@sonney2k | then it is also easy to fix | 23:46 |
van51 | sonney2k: I think I had tested it | 23:46 |
van51 | but let me do one now | 23:46 |
@sonney2k | van51, you don't really need insert_element with dynarray | 23:47 |
@sonney2k | just always call set_element | 23:47 |
@sonney2k | it will grow everything automatically | 23:47 |
@sonney2k | and we certainly need no insert* for the kernel stuff | 23:47 |
@sonney2k | all we want is to extend the list of kernels - oh well you never know | 23:48 |
van51 | sonney2k: set_element wouldn't replace the object at that index though? | 23:48 |
@sonney2k | it would | 23:48 |
@sonney2k | van51, it is just array[i] = obj | 23:48 |
@sonney2k | but with growing array | 23:48 |
@iglesiasg | sonney2k, van51 : hey guys | 23:50 |
@iglesiasg | let me join you in the conversation | 23:50 |
@sonney2k | iglesiasg, evening! | 23:50 |
van51 | hi | 23:50 |
@iglesiasg | I think a method like insert_element is a good idea to have | 23:50 |
@iglesiasg | I think it makes sense to have a method that allows you put elements without the need of giving the index where to put them explicitly | 23:50 |
@lisitsyn | push_back please | 23:51 |
@iglesiasg | yeah, exactly | 23:51 |
@sonney2k | iglesiasg, what should insert_element do when you give it an index too above anything | 23:51 |
@sonney2k | lisitsyn, we have that | 23:51 |
@lisitsyn | sonney2k: what is insert_element then | 23:51 |
@lisitsyn | ahhhh | 23:51 |
@lisitsyn | index based insert? | 23:51 |
@sonney2k | yes | 23:51 |
@iglesiasg | ah I see | 23:51 |
@lisitsyn | then please call it insert | 23:52 |
@lisitsyn | not insert_element or anything | 23:52 |
@lisitsyn | sonney2k: how's your vim setup, is it working nice? | 23:52 |
@sonney2k | lisitsyn, it is all long there | 23:53 |
@lisitsyn | long like? | 23:53 |
@sonney2k | lisitsyn, re vim not really using it | 23:53 |
@sonney2k | but it works yes | 23:53 |
@sonney2k | better than before | 23:53 |
@lisitsyn | sonney2k: much better than before yes | 23:53 |
@lisitsyn | sonney2k: due to no stl it is really fast in shogun | 23:54 |
@lisitsyn | a bit slower in tapkee with eigen+stl | 23:54 |
@sonney2k | well you are hardcore using templates etc | 23:54 |
@lisitsyn | sonney2k: yeah some mpl | 23:54 |
@sonney2k | I think dynarray dates back to 1999 or even before | 23:55 |
@lisitsyn | sonney2k: oldschool flava | 23:56 |
@lisitsyn | sonney2k: okay I think we need to speed up compilation now | 23:57 |
@lisitsyn | sonney2k: btw no templates compare to swig | 23:58 |
@sonney2k | lisitsyn, swig is only slow due to our %template stuff | 23:58 |
@lisitsyn | I think even boost is faster to handle | 23:58 |
@lisitsyn | than swig | 23:58 |
@iglesiasg | speed up compilation sounds great hehe | 23:59 |
@lisitsyn | sonney2k: we should get rid of it then | 23:59 |
@sonney2k | mission impossible if we don't want to hack swig | 23:59 |
@sonney2k | lisitsyn, how? I don't see it | 23:59 |
@lisitsyn | sonney2k: me neither but I didn't think hard yet! | 23:59 |
--- Log closed Fri May 17 00:00:05 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!