[17:31:22] Amir, https://github.com/aetilley/sigclust [17:31:24] now exists [17:31:28] Aaron ^ [17:31:37] hey aetilley [17:32:04] in order to ping, write "Amir1" (if you use IRC cloud or chat zilla, use tab) [17:32:27] and Aaron is halfak (which he is not around) [17:33:16] * Amir1 is trying to open github :| [17:33:31] of course. woops. [17:33:46] it's an empty repo. just fyi [17:34:51] it still something we can work on :) [17:35:04] yah [17:35:14] So I'm guessing numpy is going to be best for this? [17:37:26] depends [17:37:45] I think scipy has methods related to sigclust [17:37:51] that numpy doesn't have [17:38:05] oh I just meant numpy arrays [17:38:13] which are supposed to be fast [17:38:46] naturally I'd use any relevent scipy methods. [17:39:11] brb [17:42:21] yes, we must use numpy arrays, otherwise it would be pretty difficult to handle [17:51:58] got it [17:52:51] I think the cluster index is alread part of sklearn [17:53:32] and there are tools for generating the Monte Carlo datasets once we have the covariance matrix for the null distribution [17:53:50] the rest is just computing eigenvalues [17:54:20] and this one pesky detail of this multivariate MAD which isn't totally clear [17:55:14] anyway, at this point I don't have any specific questions, I'm just doing research on what num/scipy already has and planning how to impliment it. [17:55:19] implement* [17:56:50] I have been uite slient. Let me know if you need anything from me [17:57:07] will do [17:57:59] ToAruShiroiNeko: you're gediz right? [17:58:00] :) [17:59:11] that is correct [17:59:24] That'd be my human designation :p [18:00:43] rodger [18:19:04] ok, for what it's worth that SigClust summary is now in the repo's docs dir. [18:19:12] who needs google docs.