You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
…les()`
python#108324 switched to a faster
implementation, but a caveat is that it changes the specific outputs.
This preserves backwards compatibility when the user has asked for
specific outputs by supplying a seed value. With an option for them
to ask for the new algorithm implementation if desired.
When return random values anyway (no seed), the new faster algorithm is
used by default.
Currently,
samples()
callsrandom.gauss()
. It would be several times faster to use the inverse CDF method instead.Linked PRs
NormalDist.samples()
when a seed is given #108658The text was updated successfully, but these errors were encountered: