flopscope.stats.expon
flopscope.stats.expon
Exponential continuous random variable.
This object mirrors scipy.stats.expon.
Notes
loc shifts the origin and scale is the reciprocal of the rate.
Per-method FLOP costs (weight 1.0): pdf deducts 22 * numel(input)
(composite: z=(x-loc)/scale(2) + exp(-z)(17) + /scale(1) + where(2),
FMA=2), cdf deducts 22 * numel(input) (composite: z(2) + exp(-z)(17)
+ 1-exp(1) + where(2), FMA=2), ppf deducts 27 * numel(input)
(composite: log1p(-q)(19) + 3 where/cmp/and(8), FMA=2).