stats broadcasting in rvs (Trac #1544) #2069
Description
Original ticket http://projects.scipy.org/scipy/ticket/1544 on 2011-10-30 by @josef-pkt, assigned to unknown.
I thought that rvs would create a broadcasted random array for different loc
import numpy as np
from scipy import stats
bins = stats.norm.ppf(np.linspace(0,1,11))
loc = np.zeros(1000)
scale = 1
print bins
print
np.random.seed(7654321)
print np.histogram(stats.norm.rvs(size=loc.shape), bins)[0]
np.random.seed(7654321)
print np.histogram(loc + scale * stats.norm.rvs(size=loc.shape), bins)[0]
np.random.seed(7654321)
print np.histogram(stats.norm.rvs(loc=loc, scale=scale), bins)[0]
np.random.seed(7654321)
print np.histogram(stats.norm(loc=loc, scale=scale).rvs(), bins)[0]
np.random.seed(7654321)
however it just draws one rvs and broadcasts that, so we get all identical rvs
[ -inf -1.28155157 -0.84162123 -0.52440051 -0.2533471 0. 0.2533471
0.52440051 0.84162123 1.28155157 inf]
[ 99 104 92 100 103 95 96 95 108 108]
[ 99 104 92 100 103 95 96 95 108 108]
[ 0 0 1000 0 0 0 0 0 0 0]
[ 0 0 1000 0 0 0 0 0 0 0]
>>> rvs = stats.norm(loc=loc, scale=scale).rvs()
>>> rvs[:10]
array([-0.70256379, -0.70256379, -0.70256379, -0.70256379, -0.70256379,
-0.70256379, -0.70256379, -0.70256379, -0.70256379, -0.70256379])