def check_fpp_kuiper(F, N, M, thresh=1.):
ps = []
while len(ps) < M:
p = F(np.random.random(N))
if p < thresh:
ps.append(p / thresh)
assert kuiper.kuiper(ps)[1] > 0.01
def check_fpp_kuiper(F,N,M,thresh=1.):
ps = []
while len(ps)<M:
p = F(np.random.random(N))
if p<thresh:
ps.append(p/thresh)
assert kuiper.kuiper(ps)[1]>0.01
def get_marg_info(sampler):
'''makes chains into marginalize postierior and does KL and kuiper test against uniform'''
chains = sampler.flatchain[:,:3]
out_post = []
out_div = []
# sfh
for i in range(3):
x, y = np.histogram(chains[:,i], bins=int(np.sqrt(len(chains))))
post = np.vstack((y[:-1], x/np.sum(x,dtype=float))).T
out_post.append(post)
unif = np.ones_like(post[:,1])/post[:,0].ptp()
out_div.append([kl_diverg(post[:,1], unif), kuiper(post[:,1], unif)[0]])
#return sfh_post, age_post, met_post, sfh_div, age_div, met_div
return out_post[0], out_post[1], out_post[2], out_div[0], out_div[1], out_div[1]
def test_detect_nonuniform():
D, f = kuiper.kuiper(np.random.random(500) * 0.5)
assert f < 0.01
def F(x):
return kuiper.kuiper(x)[1]
def test_uniform():
for N in [10, 100, 1000, 10000]:
yield check_uniform, lambda x: kuiper.kuiper(x)[1], N
def test_weighted():
a = (np.random.random(100) * 3.4 + 0.8) % 1
i = (0.8, 4.2, 1)
b, t = kuiper.fold_intervals([i])
cdf = kuiper.cdf_from_intervals(b, t)
assert kuiper.kuiper(a, cdf)[1] > 0.01
def test_weighted():
a = (np.random.random(100) * 3.4 + 0.8)%1
i = (0.8,4.2,1)
b, t = kuiper.fold_intervals([i])
cdf = kuiper.cdf_from_intervals(b,t)
assert kuiper.kuiper(a,cdf)[1]>0.01
def test_detect_nonuniform():
D, f = kuiper.kuiper(np.random.random(500)*0.5)
assert f<0.01
def F(x):
return kuiper.kuiper(x)[1]
def test_uniform():
for N in [10,100,1000,10000]:
yield check_uniform, lambda x: kuiper.kuiper(x)[1], N
