On nonsymmetric saddle point matrices that allow conjugate gradient iterations (Q2480884)

Conditions for a nonsymmetric saddle point matrix \(\mathcal A\) to be diagonalizable with a real and positive spectrum are studied. These conditions are based on necessary and sufficient conditions for positive definiteness of a certain bilinear form, with respect to which \(\mathcal A\) is symmetric. It is shown that in case the latter conditions are satisfied, there exists a well defined conjugate gradient method for solving a linear system with matrix \(\mathcal A\). An efficient implementation of this method is proposed and practical issues such as error bounds are discussed. Numerical experiments confirming the efficiency of the proposed method are presented.