flopscope.accounting.norm_cost

shape:tuple

Shape of the array passed to the analytical cost formula.

ord:object, optional

Argument forwarded to the analytical linalg.norm cost formula. Defaults to None.

:int

Weighted public cost estimate, floored to match runtime accounting.

This helper multiplies the analytical FLOP count by the active weight from flopscope._weights and then applies int(...) so public estimates match budget deductions.

Cost depends on the ord parameter and input dimensionality.

• Elementwise norms (Frobenius, L1, Linf, etc.): 2 * numel (FMA=2, weight=1 baked in).

• SVD-based norms (2-norm, nuclear norm): values-only SVD cost 2*a*b^2 + 2*b^3 where a=max(m,n), b=min(m,n).

• General-p vector norms (ord not in {None, 0, 1, 2, inf, -inf}): 18 * numel + 16 (abs + pow per elem + sum-reduce + root pow).