Inverse problems for nonsymmetric matrices with a submatrix constraint (Q2383644)

The paper deals with the following three problems arising from finite model updating in structural dynamics: Problem I(a): Given matrices \(X\in R^{n\times p}\) with full column rank, \(B\in R^{p\times p}\) and \(A_0\in R^{r\times r}\), find \(A\in R^{n\times n}\) such that \(X^TAX=B\), \(A([1,r])=A_0\) where \(A([1,r])\) is the \(r\times r\) leading principal submatrix of \(A\). Problem I(b): Given \(X\in R^{n\times p}\), \(B\in R^{p\times p}\) and \(A_0\in R^{r\times r}\), find \(A\in R^{n\times n}\) such that the norm \(\| X^TAX-B\| \) is minimal and \(A([1,r])=A_0\). Problem II: Given \(\tilde{A}\in R^{n\times n}\) with \(\tilde{A}([1,r])=A_0\), find \(\hat{A}\in S_E\) such that \(\| \tilde{A}-\hat{A}\| =\inf _{A\in S_E}\| \tilde{A}-A\| \), where \(S_E\) is the solution set to Problem I(a). The authors apply the general singular value decomposition and the canonical correlation decomposition of a matrix pair to derive the solvability conditions for Problem I(a) and the general solution of Problem I. They present the solution to Problem II and provide a numerical algorithm for solving the problems.