[mlpack-svn] [MLPACK] #378: sparse matrix support for SDP-LR

[MLPACK Trac]

#378: sparse matrix support for SDP-LR
----------------------------------------+-----------------------------------
  Reporter:  stephentu                  |        Owner:     
      Type:  enhancement                |       Status:  new
  Priority:  minor                      |    Milestone:     
 Component:  mlpack                     |   Resolution:     
  Keywords:  lrsdp, sdp, auglagrangian  |     Blocking:     
Blocked By:                             |  
----------------------------------------+-----------------------------------
Changes (by rcurtin):

  * keywords:  => lrsdp, sdp, auglagrangian


Comment:

 Hi Stephen,

 Judging by your recent work on large-scale SDPs, you look like the right
 person for the job.  Can you explain more of what your plan is?  Do you
 mean that you plan to adapt LRSDP to work with arma::sp_mat as
 coordinates, or to adapt LRSDP so that the constraint matrices are sparse?
 (By your recent paper, it seems like the latter is the goal.)

 My original intention with this code was to be able to perform LMNN and
 MVU with LRSDP as the SDP solver, but I was never able to get it to work
 and my research went different directions anyway.  If you're interested in
 picking up where I left off --- or otherwise building upon the SDP solver
 framework we have in place --- I'd be more than happy to help out where I
 can.

 Also, I know you know this given your pull request from earlier today, but
 we are in the process of transitioning to Github (albeit slowly, because
 collecting Github user info to do the name changes correctly is very
 tedious), so at some point in the next few weeks this ticket will be
 transferred.  Alternately I guess we could close this one now, open an
 issue now on Github, and work like that.  Doesn't matter to me. :)

 Thanks,

 Ryan

-- 
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/378#comment:2>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed at Georgia Tech.