def objective(phi00):
phis = [phi00.item(k) for k in range(2 * d)]
if any([x < 0 or x > 1 for x in phis + [sum(phis)]]):
return 0 # used if finish != None in brute
phim1 = 1 - sum(phis)
pPhi = {k - d: phis[k] for k in range(2 * d)}
pPhi[d] = phim1
pPhis = [pPhi]
pZsp = pZs + pPhis
e = entropy.wme(g, pYs, pZsp, X)
return -e
support = [f(*(X + Y + Z)) for Y in iproduct(*pYs) for Z in iproduct(*pZs)]
j1 = merge_MPC.dmerge(support, d)
def j(*x):
return j1(f(*x))
do = merge_MPC.do(f, pYs, pZs, X)
k1 = merge_MPC.dyn_merge(do, d)
def k(*x):
return k1(f(*x))
## compute H(Y|O) and corresponding H(Y|O')'s
xfs.append(iii)
e1 = entropy.wme(f, pYs, pZs, X)
yfs.append(e1)
e2 = entropy.wme(g, pYs, pZs2, X)
ygs.append(e2)
e3 = entropy.wme(h, pYs, pZs, X)
yhs.append(e3)
e5 = entropy.wme(j, pYs, pZs, X)
yjs.append(e5)
e6 = entropy.wme(k, pYs, pZs, X)
yks.append(e6)
e7 = entropy.wme(g, pYs, pZs3, X)
yls.append(e7)
# compute optimal randomisation with independent noise
optimal = True