The geometry of linear infeasibility
作者:
Highlights:
•
摘要
The best approximation of an unsolvable system of linear equations is shown to lie in a set that is bounded by finite many hyperplanes but need not be convex. This candidate set, defined to be the polyhedral interior of the linear system, is the same for the best approximations with respect to all p-norms, 1⩽p⩽∞. Polyhedral considerations allow the treatment of several issues including the removal of linear equations to render the remaining system feasible.
论文关键词:Best approximation,Generalized inverse,Linear programming
论文评审过程:Available online 14 May 2002.
论文官网地址:https://doi.org/10.1016/S0096-3003(01)00042-X