The first line of the nan functions (such as nansum, nanmin, nanmax) is y = array(a) That leads to matrix in, array out. Is there some way to make it matrix in, matrix out? Here, for example, is nansum: def nansum(a, axis=None): """Sum the array over the given axis, treating NaNs as 0. """ y = array(a) if not issubclass(y.dtype.type, _nx.integer): y[isnan(a)] = 0 return y.sum(axis)