[mlpack-git] [mlpack] Add a primal-dual interior point SDP solver (#387)
notifications at github.com
Fri Feb 13 04:17:12 EST 2015
alright, so I believe that there are some starting points for which, despite being positive, result in some poor numerical issues with the algorithm. I worked around this issue in e4330544cb04ac6d2bb38e1372bffd8c6cef6bcb by simply using a more numerically stable starting point (the condition # of some of the random starting points was around 400 FYI).
to actually solve this problem in general, i'm not sure of how to do this at the moment. there are some sophisticated pre-conditioning techniques to try and make the problem more well behaved. these techniques are far beyond the realm of my knowledge. i think we can leave it for now and wait for the kinds of problems people actually try to solve SDPs with to decide how to proceed.
Reply to this email directly or view it on GitHub:
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mlpack-git