On Thursday, March 10 at 3 PM, Dang Phan will present a talk titled

Some Approaches To Solving Quadratically Constrained
Quadratic Programs

We will meet in Little 305 this week; note the CHANGE IN LOCATION!

Abstract: This talk considers the problem of minimizing a
quadratic function subject to infinitely many strictly convex
quadratic constraints. Two approaches have been investigated for
two types of quadratic problems: non-convex and convex. In the
first approach, an improvement of the difference of convex
optimization algorithm (DCA) has been studied to the case with
general quadratic objective function for locally solving. The
second approach is dedicated to the case with convex objective
function. The proposed algorithm is based on the new idea of using
balls to approximate ellipsoids. It is an iterative method which
solves, at each iteration, a ball constrained subproblem. These
algorithms are reliable, efficient and extremely easy to
implement. They use only matrix-vector products and do not need to
solve systems of linear equations, hence are very suitable for
large scale problems.

Chris Tweddle
VP Gator-SIAM