On the structure of the solutions to the matrix equation \(G^\ast JG = J\) (Q2100082)

The authors study the matrix equation \(G^*JG = J\), where \(J\) is a given square matrix, focusing on the real and complex cases. As this equation describes the isometry group for the bilinear or sesquilinear form associated with matrix \(J\), they use as their main tool the linear matrix equation \(X^*J + JX = 0\) in the tangent space. They find reasonable to consider first the auxiliary equation \(X^*J - JX = 0\) together with \(X^*J + JX = 0\), obtaining then the two solution sets from projection maps. Finally, they discuss the numerical implementation of the computation and show how to overcome stability problems.