def test_hf_state_plane_wave_basis_lowest_single_determinant_state(self):
grid_length = 7
dimension = 1
spinless = True
n_particles = 4
length_scale = 2.0
grid = Grid(dimension, grid_length, length_scale)
hamiltonian = jellium_model(grid, spinless)
hamiltonian_sparse = get_sparse_operator(hamiltonian)
hf_state = hartree_fock_state_jellium(grid,
n_particles,
spinless,
plane_wave=True)
HF_energy = expectation(hamiltonian_sparse, hf_state)
for occupied_orbitals in permutations([1] * n_particles + [0] *
(grid_length - n_particles)):
state_index = numpy.sum(2**numpy.array(occupied_orbitals))
HF_competitor = numpy.zeros(2**grid_length)
HF_competitor[state_index] = 1.0
self.assertLessEqual(HF_energy,
expectation(hamiltonian_sparse, HF_competitor))
def test_hf_state_energy_same_in_plane_wave_and_dual_basis(self):
grid_length = 4
dimension = 1
wigner_seitz_radius = 10.0
spinless = False
n_orbitals = grid_length**dimension
if not spinless:
n_orbitals *= 2
n_particles = n_orbitals // 2
length_scale = wigner_seitz_length_scale(wigner_seitz_radius,
n_particles, dimension)
grid = Grid(dimension, grid_length, length_scale)
hamiltonian = jellium_model(grid, spinless)
hamiltonian_dual_basis = jellium_model(grid, spinless, plane_wave=False)
# Get the Hamiltonians as sparse operators.
hamiltonian_sparse = get_sparse_operator(hamiltonian)
hamiltonian_dual_sparse = get_sparse_operator(hamiltonian_dual_basis)
hf_state = hartree_fock_state_jellium(grid,
n_particles,
spinless,
plane_wave=True)
hf_state_dual = hartree_fock_state_jellium(grid,
n_particles,
spinless,
plane_wave=False)
E_HF_plane_wave = expectation(hamiltonian_sparse, hf_state)
E_HF_dual = expectation(hamiltonian_dual_sparse, hf_state_dual)
self.assertAlmostEqual(E_HF_dual, E_HF_plane_wave)
def test_hf_state_energy_close_to_ground_energy_at_high_density(self):
grid_length = 8
dimension = 1
spinless = True
n_particles = grid_length**dimension // 2
# High density -> small length_scale.
length_scale = 0.25
grid = Grid(dimension, grid_length, length_scale)
hamiltonian = jellium_model(grid, spinless)
hamiltonian_sparse = get_sparse_operator(hamiltonian)
hf_state = hartree_fock_state_jellium(grid,
n_particles,
spinless,
plane_wave=True)
restricted_hamiltonian = jw_number_restrict_operator(
hamiltonian_sparse, n_particles)
E_g = get_ground_state(restricted_hamiltonian)[0]
E_HF_plane_wave = expectation(hamiltonian_sparse, hf_state)
self.assertAlmostEqual(E_g, E_HF_plane_wave, places=5)
def test_apply_constraints(self):
# Get norm of original operator.
original_norm = 0.
for term, coefficient in self.fermion_hamiltonian.terms.items():
if term != ():
original_norm += abs(coefficient)
# Get modified operator.
modified_operator = apply_constraints(self.fermion_hamiltonian,
self.n_fermions)
modified_operator.compress()
# Get norm of modified operator.
modified_norm = 0.
for term, coefficient in modified_operator.terms.items():
if term != ():
modified_norm += abs(coefficient)
self.assertTrue(modified_norm < original_norm)
# Map both to sparse matrix under Jordan-Wigner.
sparse_original = get_sparse_operator(self.fermion_hamiltonian)
sparse_modified = get_sparse_operator(modified_operator)
# Check spectra.
sparse_original = jw_number_restrict_operator(sparse_original,
self.n_fermions)
sparse_modified = jw_number_restrict_operator(sparse_modified,
self.n_fermions)
original_spectrum = sparse_eigenspectrum(sparse_original)
modified_spectrum = sparse_eigenspectrum(sparse_modified)
spectral_deviation = numpy.amax(
numpy.absolute(original_spectrum - modified_spectrum))
self.assertAlmostEqual(spectral_deviation, 0.)
# Check expectation value.
energy, wavefunction = get_ground_state(sparse_original)
modified_energy = expectation(sparse_modified, wavefunction)
self.assertAlmostEqual(modified_energy, energy)
def test_constraint_matrix(self):
# Randomly project operator with constraints.
numpy.random.seed(8)
constraints = constraint_matrix(self.n_orbitals, self.n_fermions)
n_constraints, n_terms = constraints.shape
self.assertEqual(1 + self.n_orbitals**2 + self.n_orbitals**4, n_terms)
random_weights = numpy.random.randn(n_constraints)
vectorized_operator = operator_to_vector(self.fermion_hamiltonian)
modification_vector = constraints.transpose() * random_weights
new_operator_vector = vectorized_operator + modification_vector
modified_operator = vector_to_operator(new_operator_vector,
self.n_orbitals)
# Map both to sparse matrix under Jordan-Wigner.
sparse_original = get_sparse_operator(self.fermion_hamiltonian)
sparse_modified = get_sparse_operator(modified_operator)
# Check expectation value.
energy, wavefunction = get_ground_state(sparse_original)
modified_energy = expectation(sparse_modified, wavefunction)
self.assertAlmostEqual(modified_energy, energy)
