On the condition number of linear least squares problems in a weighted Frobenius norm (Q1923873): Difference between revisions

Let \(A\) be an \(m\times n\), \(m\geq n\), full rank real matrix and \(b\) a real \(m\)-vector. An explicit formula for the condition number of the linear least squares problem (LLSP) defined by \(\min|Ax-b|_2\), \(x\in\mathbb{R}^n\) is given. The weighted Frobenius norm on the data is taken in the form \(|[\alpha A,\beta b]|_F\) where \(\alpha\) and \(\beta\) are two positive real numbers. A straightforward generalization of the normwise backward error in the solution of LLSP to this norm is provided. This allows to carry out a first-order estimate of the forward error for the LLSP with weighted Frobenius norm. Numerical results obtained by QR factorization, the method of normal equations and the method of semi-normal equations are presented.