def test_rightmult(backend, dtype, chi, d):
lam = tn.randn((chi, chi), dtype=dtype, backend=backend, seed=10)
gam = tn.randn((chi, d, chi), dtype=dtype, backend=backend, seed=10)
result = ct.rightmult(gam, lam)
compare = tn.linalg.operations.ncon([gam, lam], [[-1, -2, 1],
[1, -3]])
np.testing.assert_allclose(result.array, compare.array)
def test_projdiag(backend, dtype, chi): A = tn.randn((chi, chi), dtype=dtype, backend=backend, seed=10) B = tn.randn((chi, chi), dtype=dtype, backend=backend, seed=10) result = ct.projdiag(A, B) compare_val = tn.linalg.operations.ncon([A, B], [[1, 2], [1, 2]]) compare = compare_val * tn.eye(chi, dtype=dtype, backend=backend) rtol = (A.size + B.size) * np.finfo(dtype).eps np.testing.assert_allclose(result.array, compare.array, rtol=rtol)
def test_gauge_transform(backend, dtype, chi, d):
lamL = tn.randn((chi, chi), dtype=dtype, backend=backend, seed=10)
lamR = tn.randn((chi, chi), dtype=dtype, backend=backend, seed=10)
gam = tn.randn((chi, d, chi), dtype=dtype, backend=backend, seed=10)
result = ct.gauge_transform(lamL, gam, lamR)
compare = tn.linalg.operations.ncon([lamL, gam, lamR],
[[-1, 1], [1, -2, 2], [2, -3]])
rtol = (lamL.size + lamR.size + gam.size) * np.finfo(dtype).eps
np.testing.assert_allclose(result.array, compare.array, rtol=rtol)
def test_XopR(backend, dtype, chi, d):
A = tn.randn((chi, d, chi), dtype=dtype, backend=backend, seed=10)
X = tn.randn((chi, chi), dtype=dtype, backend=backend, seed=10)
result = ct.XopR(A, X)
compare = tn.linalg.operations.ncon([A, A.conj(), X],
[[-2, 3, 1],
[-1, 3, 2],
[2, 1]])
rtol = (2*A.size + X.size) * np.finfo(dtype).eps
np.testing.assert_allclose(result.array, compare.array, rtol=rtol)
def test_rho_loc(backend, dtype, chi, d):
A = tn.randn((chi, d, chi), dtype=dtype, backend=backend, seed=10)
B = tn.randn((chi, d, chi), dtype=dtype, backend=backend, seed=10)
result = ct.rholoc(A, B)
As = A.conj()
Bs = B.conj()
to_contract = [A, B, As, Bs]
idxs = [(1, -3, 2),
(2, -4, 3),
(1, -1, 4),
(4, -2, 3)]
compare = tn.linalg.operations.ncon(to_contract, idxs).reshape((d**2, d**2))
rtol = 2 * (A.size + B.size) * np.finfo(dtype).eps
np.testing.assert_allclose(result.array, compare.array, rtol=rtol)
def random_hermitian_system(backend, dtype, chi, d): """ Return A_L, C, A_R representing a normalized quantum state and a Hermitian H. """ A_1 = tn.randn((chi, d, chi), dtype=dtype, backend=backend, seed=10) A_L, _ = tn.linalg.linalg.qr(A_1, pivot_axis=2, non_negative_diagonal=True) C, A_R = tn.linalg.linalg.rq(A_L, pivot_axis=1, non_negative_diagonal=True) C /= tn.linalg.linalg.norm(C) H = tn.randn((d, d, d, d), dtype=dtype, backend=backend, seed=10) H = H.reshape((d*d, d*d)) H = 0.5*(H + H.H) H = H.reshape((d, d, d, d)) return (A_L, C, A_R, H)
def test_minimize_Hc(backend, dtype, chi, d):
"""
The result of minimize_Hc solves the eigenproblem for Hc.
"""
A_L, C, A_R, H = random_hermitian_system(backend, dtype, chi, d)
LH = tn.randn((chi, chi), dtype=dtype, backend=backend)
RH = tn.randn((chi, chi), dtype=dtype, backend=backend)
LH = 0.5 * (LH + LH.H)
RH = 0.5 * (RH + RH.H)
params = tn_vumps.params.krylov_params(n_krylov=chi * chi)
ev, newC = tn_vumps.vumps.minimize_Hc([A_L, C, A_R], [H, LH, RH], 1E-5,
params)
C_prime = ct.apply_Hc(newC, A_L, A_R, [H, LH, RH])
EC = ev * newC
np.testing.assert_allclose(C_prime.array, EC.array, rtol=1E-4, atol=1E-4)
def test_minimize_HAc(backend, dtype, chi, d):
"""
The result of minimize_HAc solves the eigenproblem for HAc.
"""
A_L, C, A_R, H = random_hermitian_system(backend, dtype, chi, d)
A_C = ct.leftmult(C, A_R)
LH = tn.randn((chi, chi), dtype=dtype, backend=backend)
LH = 0.5 * (LH + LH.H)
RH = tn.randn((chi, chi), dtype=dtype, backend=backend)
RH = 0.5 * (RH + RH.H)
params = tn_vumps.params.krylov_params(n_krylov=chi * chi, max_restarts=20)
ev, newAC = tn_vumps.vumps.minimize_HAc([A_L, C, A_R], A_C, [H, LH, RH],
1E-5, params)
AC_prime = ct.apply_HAc(newAC, A_L, A_R, H, LH, RH)
EAC = ev * newAC
np.testing.assert_allclose(AC_prime.array, EAC.array, rtol=1E-4, atol=1E-4)
def vumps_initialization(
d: int,
chi: int,
dtype: Optional[DtypeType] = None,
backend: Optional[Text] = None
) -> Tuple[ThreeTensors, tn.Tensor, TwoTensors]:
"""
Generate a random uMPS in mixed canonical forms, along with the left
dominant eV L of A_L and right dominant eV R of A_R.
Args:
d: Physical dimension.
chi: Bond dimension.
dtype: Data dtype of tensors.
backend: The backend.
Returns:
mps = (A_L, C, A_R): A_L and A_R have shape (chi, d, chi), and are
respectively left and right isometric. C is the (chi, chi)
centre of orthogonality.
A_C: A_L @ C. One of the equations vumps minimizes is
A_L @ C = C @ A_R = A_C.
fpoints = [rL, lR]: C^dag @ C and C @ C^dag respectively. Will converge
to the left and right fixed points of A_R and A_L. Both are chi x chi.
"""
A_1 = tn.randn((chi, d, chi), dtype=dtype, backend=backend)
A_L, _ = tn.linalg.linalg.qr(A_1, pivot_axis=2, non_negative_diagonal=True)
C, A_R = tn.linalg.linalg.rq(A_L, pivot_axis=1, non_negative_diagonal=True)
C /= tn.linalg.linalg.norm(C)
A_C = ct.rightmult(A_L, C)
L0, R0 = vumps_approximate_tm_eigs(C)
fpoints = (L0, R0)
mps = [A_L, C, A_R]
benchmark.block_until_ready(R0)
return (mps, A_C, fpoints)
def test_polar(backend, dtype, chi): A = tn.randn((chi, chi), backend=backend, dtype=dtype, seed=10) U = polar.polarU(A) UUdag = U @ U.H UdagU = U.H @ U eye = tn.eye(chi, dtype, backend=backend) atol = A.size * np.finfo(dtype).eps np.testing.assert_allclose(eye.array, UUdag.array, atol=atol) np.testing.assert_allclose(UUdag.array, eye.array, atol=atol)
def test_polar_mps(backend, dtype, chi, d): A = tn.randn((chi, d, chi), backend=backend, dtype=dtype, seed=10) UR = polar.polarU(A, pivot_axis=1) UL = polar.polarU(A, pivot_axis=2) eye = tn.eye(chi, dtype, backend=backend) testR = ct.XopR(UR, eye) testL = ct.XopL(UL, eye) atol = A.size * np.finfo(dtype).eps np.testing.assert_allclose(eye.array, testR.array, atol=atol) np.testing.assert_allclose(eye.array, testL.array, atol=atol)
def test_gauge_match_minimizes(backend, dtype, chi, d):
"""
Gauge matching decreases the values of ||A_C - A_L C ||
and ||A_C - C A_R||.
"""
A_L, C, A_R, _ = random_hermitian_system(backend, dtype, chi, d)
A_C = tn.randn((chi, d, chi), dtype=dtype, backend=backend)
epsL = tn.linalg.linalg.norm(A_C - ct.rightmult(A_L, C))
epsR = tn.linalg.linalg.norm(A_C - ct.leftmult(C, A_R))
A_L2, A_R2 = vumps.gauge_match(A_C, C, True)
epsL2 = tn.linalg.linalg.norm(A_C - ct.rightmult(A_L2, C))
epsR2 = tn.linalg.linalg.norm(A_C - ct.leftmult(C, A_R2))
assert epsL2 < epsL
assert epsR2 < epsR
def test_randn(backend):
"""
Tests tensornetwork.randn against the backend code.
"""
shape = (5, 10, 3, 2)
seed = int(time.time())
np.random.seed(seed=seed)
backend_obj = backends.backend_factory.get_backend(backend)
for dtype in dtypes[backend]["rand"]:
tnI = tensornetwork.randn(shape,
dtype=dtype,
seed=seed,
backend=backend)
npI = backend_obj.randn(shape, dtype=dtype, seed=seed)
np.testing.assert_allclose(tnI.array, npI)
