scientific article; zbMATH DE number 17464
[For the entire collection see Zbl 0728.00018.] The authors consider the problem \(x^ TAx=\min\) subject to the constraints \(N^ Tx=t\), \(x^ Tx=1\), where \(A\) is a symmetric \((n+m)\times (n+m)\) matrix, \(n>0\), \(N\) is a \((n+m)\times n\) matrix with full rank, and \(t\) is a vector of dimension \(m\) with \(\| (N^ T)^ +t\| < 1\). They show that the original minimization can be reduced to a problem of the form \(\lambda = \min\) subject to \(Cz=\lambda z+b\), \(z^ Tz=s^ 2\). This problem, after an eigenvalue decomposition, is reduced to a problem of finding the solution of a one-dimensional secular equation. From another point of view the last problem can be reduced to a quadratic eigenvalue problem, which is transformed into an equivalent linear one.
- A constrained eigenvalue problem
- New zero-finders for trust-region computations
- Eigenvalue analysis of constrained minimization problem for homogeneous polynomial
- scientific article; zbMATH DE number 1159562 (Why is no real title available?)
- Quadratic minimisation problems in statistics
- Eigenvalue programming beyond matrices
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3974784)