def test_UnbinnedNLL_mask():
c = UnbinnedNLL([1, np.nan, 2], lambda x, a: x + a)
assert c.mask is None
assert np.isnan(c(0)) == True
c.mask = np.arange(3) != 1
assert_equal(c.mask, (True, False, True))
assert np.isnan(c(0)) == False
def test_UnbinnedNLL(unbinned, verbose):
mle, x = unbinned
def pdf(x, mu, sigma):
return norm(mu, sigma).pdf(x)
cost = UnbinnedNLL(x, pdf, verbose=verbose)
m = Minuit(cost, mu=0, sigma=1, limit_sigma=(0, None))
m.migrad()
assert_allclose(m.args, mle[1:], atol=1e-3)
assert m.errors["mu"] == pytest.approx(1000**-0.5, rel=0.05)
def test_UnbinnedNLL(unbinned, verbose):
mle, x = unbinned
def pdf(x, mu, sigma):
return norm(mu, sigma).pdf(x)
cost = UnbinnedNLL(x, pdf, verbose=verbose)
assert cost.ndata == np.inf
m = Minuit(cost, mu=0, sigma=1)
m.limits["sigma"] = (0, None)
m.migrad()
assert_allclose(m.values, mle[1:], atol=1e-3)
assert m.errors["mu"] == pytest.approx(1000**-0.5, rel=0.05)
assert_equal(m.fmin.reduced_chi2, np.nan)
def test_UnbinnedNLL_properties():
def pdf(x, a, b):
return 0
c = UnbinnedNLL([1, 2], pdf)
assert c.pdf is pdf
with pytest.raises(AttributeError):
c.pdf = None
assert_equal(c.data, [1, 2])
c.data = [2, 3]
assert_equal(c.data, [2, 3])
with pytest.raises(ValueError):
c.data = [1, 2, 3]
assert c.verbose == 0
c.verbose = 1
assert c.verbose == 1
def test_UnbinnedNLL_mask():
c = UnbinnedNLL([1, np.nan, 2], lambda x, a: x + a)
assert np.isnan(c(0)) == True
c.mask = np.arange(3) != 1
assert np.isnan(c(0)) == False