Available at OO
Authors
Bruno F. Lourenço (ISM) and Gábor Pataki (UNC-Chapel Hill)
Abstract
In semidefinite programming the dual may fail to attain its optimal value and there could be a duality gap, i.e., the primal and dual optimal values may differ. In a striking paper, Ramana proposed a polynomial size extended dual that does not have these deficiencies and yields a number of fundamental results in complexity theory. The known expositions of Ramana’s dual tend to be somewhat technical; here we give a concise and self-contained derivation, relying mostly on elementary linear algebra.