Computing a "cheap" upper bound on the norm of the solution to a linear system

293 Views Asked by At

Consider the linear system $A x = b$, where $A$ is an invertible, $n \times n$, real matrix. I would like to compute a "cheap" upper bound on the (p-)norm of the solution; i.e. $\|x\|_p$. One can, of course, solve the system and get the exact value in $O(n^3)$ time. Can we get an upper bound on $\|x\|_p$ any quicker than this?