counts = numpy.zeros(size, dtype=numpy.int32)
flags = numpy.zeros(size, dtype=numpy.uint8)
for a, b in enumerate(pos_to_use):
counts[b] = draw[a]
flags[b] = 1
sam.append((counts, flags))
# Construct the flag index array, put relevant index into each sample...
fia = FlagIndexArray(size)
fia.addSingles()
for i in xrange(len(sam)):
ind = fia.flagIndex(sam[i][1])
sam[i] = (sam[i][0], sam[i][1], ind)
# Construct the SMP object...
smp = SMP(fia)
smp.setSampleCount(samCount)
for s in sam:
smp.add(s[2], s[0])
# Get and print out the mean and its distance from the actual multinomial...
mean = smp.mean()
print 'Mean =', mean
print 'error =', numpy.fabs(mn - mean).sum() / mn.shape[0]
print 'cError =\n', numpy.fabs(mn - mean)
sam.append((counts,flags)) # Construct the flag index array, put relevant index into each sample... fia = FlagIndexArray(size) fia.addSingles() for i in xrange(len(sam)): ind = fia.flagIndex(sam[i][1]) sam[i] = (sam[i][0],sam[i][1],ind) # Construct the SMP object... smp = SMP(fia) smp.setSampleCount(samCount) for s in sam: smp.add(s[2],s[0]) # Get and print out the mean and its distance from the actual multinomial... mean = smp.mean() print 'Mean =', mean print 'error =', numpy.fabs(mn-mean).sum()/mn.shape[0] print 'cError =\n', numpy.fabs(mn-mean)
