A branch-and-bound algorithm for solving BMIs taking account of activities of relaxed LMIs
Hiroyoshi Fujioka · 2003
An optimization problem related to bilinear matrix inequalities (BMIs) is considered. A branch-and-bound type global algorithm is obtained based on a linear matrix inequality (LMI)-relaxation and an upper bound of its worst case performance is derived. Taking account of activities of the relaxed problem, an improved algorithm is also proposed.