In the note at the very end, it says “In calculating condition number, errors and residuals, use one-norms, and exact arithmetic”.
If it asks me to find the condition number of a matrix, does it mean to find $k_1(A)$, that is, to use the column norm to find the condition number?
Sorry, I’m a little bit confused with these definitions, could anyone help me out? Thanks a lot!
Yes, norms should be used consistently in condition numbers.
When one-norm (column-norm) is use for A, so it should be for A^{-1}.
1 Like