Closed nkoskelo closed 1 month ago
I don't see the need. The 1-norm is the only other mainstream norm that's missing/would be practically relevant, but its integrand is non-smooth at the zero crossings , so we cannot accurately approximate it. I don't foresee anyone urgently needing a 4-norm.
Currently, the operator,
grudge.op.norm = grudge.reductions.norm
is specified as returning "the vector norm of a function represented by its vector of degrees of freedom." However, only the 2-norm and the infinity norms are supported. Should we generalize the function to support all integral "p-norms"?