[mlpack-git] [mlpack] Low-rank constraint matrices in LRSDP (#578)

[Stephen Tu]

Cool. I don't think it will be that hard to get it into LRSDP. I'd punt on supporting the IP solver for now (just throw a runtime exception if a user sets these generalize constraints and calls the IP solver). I think the easiest way to do it is provide a `LinearConstraint` callback which supports two operations: 1) evaluation of `Tr(A_i * UU^T)`, which takes as input `U` and returns a double, and 2) evaluation of `(Tr(A_i * UU^T) - b_i) A_i U`, which takes as input a `U` and `b_i` and returns (or say does `+=` to an input) an n x r matrix. This is needed for computing gradients . 

---
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/issues/578#issuecomment-197462026
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20160316/012272ae/attachment.html>