In general, a computed solution to a problem solves many perturbed problems. the optimal backward perturbation bound for the problem with respect to the computed solution is a measure of the nearness between the perturbed problems and the original problem. An explicit expression of the optimal backward perturbation bound may provide a convenient way of testing the stability of practical algorithms. This paper surveys some new results on optimal backward perturbation bounds for linear systems, underdetermined systems, and linear least squares problems.
展开▼