Transformation to versal deformations of matrices (Q5954126)

The author considers smooth deformations \(A(p)\) of a square matrix \(A_{0}=A(0)\), where \(p=(p_{1},p_{2},\dots,p_{n})\) denotes a vector of \(n\) real or complex parameters. A deformation \(B(q)\) of the initial matrix \(A_{0}\) is called versal if, in a neighbourhood of the origin \(0\), an arbitrary deformation \(A(p)\) can be written as \(A(p)=C(p)B(q(p))C^{-1}(p)\). Thus, up to a smooth change of coordinates (i.e., a deformation \(C(p)\) of the identity matrix) and a smooth transformation of parameters, a versal deformation is already the ``most general'' deformation of \(A_{0}\). Both the change of coordinates and the change of parameters can be written as a Taylor series; it is shown how to calculate the coefficients of these expansions by recurrence. Several examples are given and studied in detail.