A globally convergent flow for computing the best low rank approximation of a matrix

Mathematics – Optimization and Control

Scientific paper

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

Scientific paper

We work in the space of $m$-by-$n$ real matrices with the Frobenius inner product. Consider the following Problem: Given an m-by-n real matrix A and a positive integer k, find the m-by-n matrix with rank k that is closest to A. I discuss a rank-preserving differential equation (d.e.) which solves this problem. If X(t) is a solution of this d.e., then the distance between $X(t)$ and $A$ decreases as t increases; this distance function is a Lyapunov function for the d.e. If $A$ has distinct positive singular values (which is a generic condition) then this d.e. has only one stable equilibrium point. The other equilibrium points are finite in number and unstable. In other words, the basin of attraction of the stable equilibrium point on the manifold of matrices with rank $k$ consists of almost all matrices. This special equilibrium point is the solution of the given problem. Usually constrained optimization problems have many local minimums (most of which are undesirable). So the constrained optimization problem considered here is very special.

No associations

LandOfFree

Say what you really think

Search LandOfFree.com for scientists and scientific papers. Rate them and share your experience with other people.

Rating

A globally convergent flow for computing the best low rank approximation of a matrix does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with A globally convergent flow for computing the best low rank approximation of a matrix, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and A globally convergent flow for computing the best low rank approximation of a matrix will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFWR-SCP-O-84416

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.