def test_get_expectation_values_adaptive_factories(factory):
num_steps = 8
fac = AdaExpFactory(steps=num_steps, scale_factor=2.0, asymptote=None)
executor = apply_seed_to_func(f_exp_up, seed=1)
# Expectation values haven't been computed at any scale factors yet
assert isinstance(fac.get_expectation_values(), np.ndarray)
assert len(fac.get_expectation_values()) == 0
# Compute expectation values at all the scale factors
fac.run_classical(executor)
assert isinstance(fac.get_scale_factors(), np.ndarray)
# Given this seeded executor, the scale factors should be as follows
correct_scale_factors = np.array(
[
1.0,
2.0,
4.0,
4.20469548,
4.20310693,
4.2054822,
4.2031916,
4.2052843,
]
)
correct_expectation_values = np.array(
[executor(scale) for scale in correct_scale_factors]
)
assert len(fac.get_expectation_values()) == num_steps
assert np.allclose(
fac.get_expectation_values(), correct_expectation_values, atol=1e-3
)
def test_ada_exp_factory_no_asympt_more_steps(test_f: Callable[[float],
float], ):
"""Test of the adaptive exponential extrapolator."""
seeded_f = apply_seed_to_func(test_f, SEED)
fac = AdaExpFactory(steps=8, scale_factor=2.0, asymptote=None)
fac.run_classical(seeded_f)
assert np.isclose(fac.reduce(), seeded_f(0, err=0), atol=CLOSE_TOL)
def test_adaptive_factory_max_iteration_warnings():
"""Test that the correct warning is raised beyond the iteration limit."""
fac = AdaExpFactory(steps=10)
with warns(
ConvergenceWarning,
match=r"Factory iteration loop stopped before convergence.",
):
fac.run_classical(lambda scale_factor: 1.0, max_iterations=3)
def test_ada_exp_factory_with_asympt(
test_f: Callable[[float], float], avoid_log: bool
):
"""Test of the adaptive exponential extrapolator."""
seeded_f = apply_seed_to_func(test_f, SEED)
fac = AdaExpFactory(
steps=3, scale_factor=2.0, asymptote=A, avoid_log=avoid_log
)
# Note: run_classical calls next which calls reduce, so calling
# fac.run_classical with an AdaExpFactory sets the optimal parameters as
# well. Hence we check that the opt_params are empty before
# AdaExpFactory.run_classical is called.
assert not fac._opt_params
fac.run_classical(seeded_f)
assert np.isclose(fac.reduce(), seeded_f(0, err=0), atol=CLOSE_TOL)
# There are three parameters to fit for the (adaptive) exponential ansatz
assert len(fac._opt_params) == 3