def test_init():
with pytest.raises(TypeError):
p = likelihood.FitParameter()
p = likelihood.FitParameter(name="test")
assert np.array_equal(p.bounds, np.array((-np.inf, np.inf)))
assert p.guess == 1.0
assert p.value == p.guess
assert p.fixed == False
def test_name_getter_setter():
# Default (must be set via __init__)
p = likelihood.FitParameter(name="test")
assert p.name == "test"
# Set directly
p.name = 5
assert p.name == "5"
# Set directly
p.name = "junk"
assert p.name == "junk"
def test_bounds_getter_setter():
# Default
p = likelihood.FitParameter(name="test")
assert np.array_equal(p.bounds, np.array((-np.inf, np.inf)))
# Set via __init__
bounds = [1, 2]
p = likelihood.FitParameter(name="test", bounds=bounds)
assert np.array_equal(p.bounds, np.array(bounds))
# Set directly
bounds = [1, 2]
p.bounds = bounds
assert np.array_equal(p.bounds, np.array(bounds))
# --- bad value checks ---
with pytest.raises(ValueError):
p.bounds = "test"
with pytest.raises(ValueError):
p.bounds = "te"
with pytest.raises(ValueError):
p.bounds = 1.0
with pytest.raises(ValueError):
p.bounds = [1.0]
with pytest.raises(ValueError):
p.bounds = [1.0, 1.0]
with pytest.raises(ValueError):
p.bounds = [1.0, -1.0]
with pytest.raises(ValueError):
p.bounds = ["a", "b"]
def test_guess_getter_setter():
# Default
p = likelihood.FitParameter(name="test")
assert np.array_equal(p.bounds, np.array((-np.inf, np.inf)))
assert p.guess == 1.0
# Set via __init__
p = likelihood.FitParameter(name="test", guess=12)
assert p.guess == 12
# Set directly
p = likelihood.FitParameter(name="test")
p.guess = 22
assert p.guess == 22
# --- bad value checks ---
p = likelihood.FitParameter(name="test")
with pytest.raises(ValueError):
p.guess = "test"
with pytest.raises(ValueError):
p.guess = [1.0]
with pytest.raises(ValueError):
p.guess = np.arange(10)
# --- bound check ---
with pytest.raises(ValueError):
p = likelihood.FitParameter(name="test")
p.bounds = [-10, 10]
p.guess = -20
with pytest.raises(ValueError):
p = likelihood.FitParameter(name="test")
p.bounds = [-10, 10]
p.guess = 20
def test_load_clear_fit_results(fitter_object):
# --- Make sure we can load fit result into parameter ---
p = likelihood.FitParameter(name="test")
assert p.value == p.guess
assert p.stdev is None
assert p.ninetyfive is None
assert not p.is_fit_result
p.load_fit_result(fitter_object, 0)
assert p.value == fitter_object.estimate[0]
assert p.stdev == fitter_object.stdev[0]
assert np.array_equal(p.ninetyfive[0], fitter_object.ninetyfive[0, 0])
assert np.array_equal(p.ninetyfive[1], fitter_object.ninetyfive[1, 0])
assert p.is_fit_result
# --- Make sure setting guess wipes out fit ---
p.guess = 22
assert p.guess == 22
assert p.value == p.guess
assert p.stdev is None
assert p.ninetyfive is None
assert not p.is_fit_result
# --- Make sure setting bounds wipes out fit ---
p = likelihood.FitParameter(name="test")
assert p.value == p.guess
assert p.stdev is None
assert p.ninetyfive is None
assert np.array_equal(p.bounds, np.array((-np.inf, np.inf)))
assert not p.is_fit_result
p.load_fit_result(fitter_object, 0)
assert p.value == fitter_object.estimate[0]
assert p.stdev == fitter_object.stdev[0]
assert np.array_equal(p.ninetyfive[0], fitter_object.ninetyfive[0, 0])
assert np.array_equal(p.ninetyfive[1], fitter_object.ninetyfive[1, 0])
assert p.is_fit_result
p.bounds = [-100, 100]
assert np.array_equal(p.bounds, np.array((-100, 100)))
assert p.value == p.guess
assert p.stdev is None
assert p.ninetyfive is None
assert not p.is_fit_result
# --- Make sure setting fixed wipes out fit ---
p = likelihood.FitParameter(name="test")
assert p.value == p.guess
assert p.stdev is None
assert p.ninetyfive is None
assert p.fixed == False
assert not p.is_fit_result
p.load_fit_result(fitter_object, 0)
assert p.value == fitter_object.estimate[0]
assert p.stdev == fitter_object.stdev[0]
assert np.array_equal(p.ninetyfive[0], fitter_object.ninetyfive[0, 0])
assert np.array_equal(p.ninetyfive[1], fitter_object.ninetyfive[1, 0])
assert p.is_fit_result
p.fixed = True
assert p.fixed == True
assert p.value == p.guess
assert p.stdev is None
assert p.ninetyfive is None
assert not p.is_fit_result
def test_interaction_bounds_guesses():
# --- default guess depends on bounds; try different bounds scenarios ---
# Two positive bounds
bounds = [10, 20]
p = likelihood.FitParameter(name="test", bounds=bounds)
assert np.array_equal(p.bounds, np.array(bounds))
assert np.isclose(p.guess, np.exp(np.mean(np.log(bounds))))
# Two negative bounds
bounds = [-20, -10]
p = likelihood.FitParameter(name="test", bounds=bounds)
assert np.array_equal(p.bounds, np.array(bounds))
assert np.isclose(p.guess, -np.exp(np.mean(np.log(np.abs(bounds)))))
# One negative, one positive
bounds = [-10, 10]
p = likelihood.FitParameter(name="test", bounds=bounds)
assert np.array_equal(p.bounds, np.array(bounds))
assert np.isclose(p.guess, np.sum(bounds) / 2)
# negative infinity, positive real
bounds = [-np.inf, 10]
internal_bounds = [-likelihood.fit_param._INFINITY_PROXY, bounds[1]]
expected = np.sum(internal_bounds) / 2
p = likelihood.FitParameter(name="test", bounds=bounds)
assert np.array_equal(p.bounds, np.array(bounds))
assert np.isclose(p.guess, expected)
# negative real, positive infinity
bounds = [-10, np.inf]
p = likelihood.FitParameter(name="test", bounds=bounds)
internal_bounds = [bounds[0], likelihood.fit_param._INFINITY_PROXY]
expected = np.sum(internal_bounds) / 2
assert np.array_equal(p.bounds, np.array(bounds))
assert np.isclose(p.guess, expected)
# negative infinity, positive infinity
bounds = [-np.inf, np.inf]
p = likelihood.FitParameter(name="test", bounds=bounds)
assert np.array_equal(p.bounds, np.array(bounds))
assert p.guess == 1.0
# --- Update bounds such that guess is outside the new bounds ---
bounds = [-10, 10]
p = likelihood.FitParameter(name="test", bounds=bounds, guess=-5)
assert np.array_equal(p.bounds, np.array(bounds))
assert p.guess == -5.0
new_bounds = [0, 10]
with pytest.warns(UserWarning):
p.bounds = new_bounds
assert np.array_equal(p.bounds, np.array(new_bounds))
assert p.guess == new_bounds[0]
bounds = [-10, 10]
p = likelihood.FitParameter(name="test", bounds=bounds, guess=5)
assert np.array_equal(p.bounds, np.array(bounds))
assert p.guess == 5.0
new_bounds = [-10, 0]
with pytest.warns(UserWarning):
p.bounds = new_bounds
assert np.array_equal(p.bounds, np.array(new_bounds))
assert p.guess == new_bounds[1]
# --- Upddate bounds such tthat guess remains in the new bounds ---
bounds = [-10, 10]
p = likelihood.FitParameter(name="test", bounds=bounds, guess=5)
assert np.array_equal(p.bounds, np.array(bounds))
assert p.guess == 5.0
new_bounds = [0, 10]
p.bounds = new_bounds
assert np.array_equal(p.bounds, np.array(new_bounds))
assert p.guess == 5.0