A canonical form under \(\phi\)-equivalence (Q753895): Difference between revisions

The matrices \(A,B\in M_ n=\{set\) of \(n\times n\) complex matrices\(\}\) are said to be \(\phi\)-equivalent if there is a nonsingular \(R\in M_ n\) such that \(A=RB\phi (R)\), where \(\phi\) : \(M_ k\to M_ k\) satisfies: 1) \(\phi (\phi (X))=X\), 2) \(\phi (XY)=\phi (Y)\phi (X)\), 3) either \(\phi (X\oplus Y)=\phi (X)\oplus \phi (Y)\) or \(\phi (X\oplus Y)=\phi (Y)\oplus \phi (X)\) for all \(X,Y\in M_ k\), \(k=1,...,n\), with \(\oplus\) denoting the direct-sum. One shows that an almost explicit canonical form of a nonsingular A under \(\phi\)-equivalence can be obtained from a canonical solution of the equation \(X\phi (X^{-1})\in Ker J(A(\phi (A^{-1}))\), where J(\(\cdot)\) denotes the Jordan canonical form. As special cases one considers three types of congruence and the cosimilarity of matrices.