07:30 < Suryo> Hello zoq! I have exams and loads of homework this week so I haven't been able to submit the PSO pull request yet because I'm still going through the code to make sure everything is okay. I've pushed the changes but still reviewing at my end.

07:31 < Suryo> I also have a question for zoq and rcurtin - when I spoke about parallelization, you suggested two things: (i) taking a look at OpenMP (ii) utilizing the native OpenBLAS parallelization capabilities.

07:32 < Suryo> As far as I'm aware, a lot of parallelization that is done in C++ in enterprise level software uses posix threads for parallelization. This is what I wanted to discuss earlier but I was just going through either method.

07:33 < Suryo> But OpenMP is a lot easier to deal with.

07:34 < Suryo> So if we have to include some parallelization in ensmallen (not in any method in armadillo), would it be okay to use OpenMP or would it be preferable to use posix threads? I've used OpenMP in the past and it's quite easy to use for embarassingly parallel loops.

07:34 < Suryo> Kindly let me know. Thank you!

13:26 < rcurtin> Suryo: I'd definitely prefer OpenMP because it makes the code maintenance a lot easier

15:23 < jenkins-mlpack2> Project docker mlpack monthly build build #8: ABORTED in 5 days 6 hr: http://ci.mlpack.org/job/docker%20mlpack%20monthly%20build/8/

15:23 < rcurtin> three straight days of green daily matrix build :)

15:23 < rcurtin> the changes for that haven't yet propagated to the weekly or monthly builds, so we'll see what happens with those

15:27 < zoq> Will take a look into the GradientTransposedConvolutionLayerTest

15:27 < zoq> or perhaps ShikharJ already did that

15:30 < rcurtin> yeah, I remember there is something weird going on there but I haven't looked into it much

15:52 < ShikharJ> zoq: I tried to investigate but didn't get too far. I have my exams in the coming week, so it'll be difficult to find the time.

16:04 < Mehdi_> Thanks in advance!

16:06 < Mehdi_> Hello there! I was wondering if the c++ code for deep reinforcement learning with continuous action domain is available anywhere.

16:10 < Mehdi_> Hello there! I was wondering if the c++ code for deep reinforcement learning with continuous action domain is available anywhere. Thank you in advance!

16:17 < rcurtin> Mehdi_: a little patience might help you out, someone will answer when they have a chance

16:17 < rcurtin> I know that there is reinforcement learning code in src/mlpack/methods/, but I don't know if it is exactly what you are looking for

16:17 < rcurtin> so I did not answer; maybe someone else knows better

16:20 < Mehdi_> Hi, thanks a lot for your reply. I have already checked that and it doesn't include what I am looking for. I hope someone could help me with that.

16:28 < rcurtin> if the code isn't there, it's not likely to exist, but I am not sure because I don't work with that code

16:28 < rcurtin> maybe it is in some PR somewhere, I am not sure

16:28 < rcurtin> I'd suggest searching or waiting for another response

16:50 < Mehdi_> Thank you for your response. I'll do what you mentioned.

18:27 < rcurtin> zoq: do you want me to release ensmallen 1.14.0 with DE after merging #83 (the Cholesky decomposition fix for CMAES)?

18:28 < zoq> rcurtin: I'd like to fix some DE related style issues first.

18:28 < rcurtin> sure, sounds good---just let me know when you're ready, then I'll merge #83 and release

18:28 < rcurtin> there is no hurry of course :)

19:56 < zoq> rcurtin: done

19:58 < zoq> rcurtin: Do you update CONTRIBUTING.md in the merge process?

19:58 < zoq> rcurtin: I was talking about the Developers and Contributors list.

19:59 < rcurtin> sure, I can do that

21:01 < zoq> rcurtin: Not sure I get the meaning of 'Difficult Crimp', maybe there is no deeper meaning :)

21:10 < rcurtin> I was doing some bouldering, and there was some hold that is a crimp (I think this is a reasonable term) that was really hard to hold on to

21:10 < rcurtin> so when I came back to do the release the challenge was on my mind :)

21:11 < zoq> ahh, I see makes sense :)