An efficient linear programming solver for optimal filter synthesis
Jihong Ren, Chen Greif, Mark R. Greenstreet · Numerical Linear Algebra with Applications · 2007
Abstract We consider the problem of l∞ optimal deconvolution arising in high data‐rate communication between integrated circuits. The optimal deconvolver can be found by solving a linear program for which we use Mehrotra's interior‐point approach. The critical step is solving the linear system for the normal equations in each iteration. We show that this linear system has a special block structure that can be exploited to obtain a fast solution technique whose overall computational cost depends mostly on the number of design variables, and only linearly on the number of constraints. Numerical experiments validate our findings and illustrate the merits of our approach. Copyright © 2007 John Wiley & Sons, Ltd.