symmray.utils_test

Utility functions specifically for testing.

Attributes

rand_herm

rand_posdef

Functions

hash_kwargs_to_int(→ int)

Hash a set of keyword arguments to a deterministic integer, which can be

rand_valid_tensordot(symmetry[, ndim_a, ndim_b, ncon, ...])

Generate two random symmetric arrays and valid contraction axes.

rand_matrix(symmetry, d[, seed, subsizes, dtype, ...])

Create a general, hermitian, or positive definite matrix (with conjugate

Module Contents

symmray.utils_test.hash_kwargs_to_int(**kwargs) → int[source]

Hash a set of keyword arguments to a deterministic integer, which can be used for seeding random number generators in tests, for example, allowing a varying but reproducible seed.

symmray.utils_test.rand_valid_tensordot(symmetry, ndim_a=None, ndim_b=None, ncon=None, charge_a=None, charge_b=None, min_ndim=1, max_ndim=5, min_d=1, max_d=3, dimension_multiplier=1, subsizes=None, seed=None, **kwargs)[source]

Generate two random symmetric arrays and valid contraction axes. For testing purposes.

Parameters:

  • symmetry (str or Symmetry) – Symmetry to use.

  • ndim_a (int or None) – Number of dimensions for the first array. If None, a random number is chosen between min_ndim and max_ndim (inclusive).

  • ndim_b (int or None) – Number of dimensions for the second array. If None, a random number is chosen between min_ndim and max_ndim (inclusive).

  • ncon (int or None) – Number of axes to contract. If None, a random number is chosen between 0 and min(ndim_a, ndim_b) (inclusive).

  • charge_a (charge_like or None) – Charge for the first array. If None, a random charge is chosen.

  • charge_b (charge_like or None) – Charge for the second array. If None, a random charge is chosen.

  • min_ndim (int) – Minimum number of dimensions for each array.

  • max_ndim (int) – Maximum number of dimensions for each array.

  • min_d (int) – Minimum dimension for each index. This is multiplied by dimension_multiplier if that is not None.

  • max_d (int) – Maximum dimension for each index. This is multiplied by dimension_multiplier if that is not None.

  • dimension_multiplier (int, optional) – Each index will have dimension d * dimension_multiplier where d is a random integer in [min_d, max_d]. Useful for generating flat arrays with exactly equally distributed charge sizes.

  • subsizes (str or None) – Passed to rand_index. If “equal”, all sectors will have equal size, required for flat arrays.

  • seed (int or np.random.Generator or None) – Random seed or generator. If None, a new generator is created.

  • **kwargs – Additional arguments passed to get_rand.

Returns:

  • a (AbelianArray) – First random array.

  • b (AbelianArray) – Second random array.

  • axes (tuple[tuple[int, …], tuple[int, …]]) – Axes to contract.

symmray.utils_test.rand_matrix(symmetry, d, seed=None, subsizes='equal', dtype='float64', fermionic=False, flat=False, matrix_type='general', d_per_charge=False)[source]

Create a general, hermitian, or positive definite matrix (with conjugate indices) with the given symmetry and shape (d, d). For testing purposes.

Parameters:

  • symmetry (str or Symmetry) – Symmetry to use.

  • d (int) – Dimension of each index. If d_per_charge is True, this is the dimension per charge sector, else it is the total dimension.

  • seed (int or np.random.Generator or None) – Random seed or generator. If None, a new generator is created.

  • subsizes (str or None) – Passed to rand_index. If “equal”, all sectors will have equal size, required for flat arrays.

  • dtype (str or dtype) – Data type of the array.

  • fermionic (bool) – Whether the array should be fermionic.

  • flat (bool) – Whether to return a flat array (with a single index) instead of a block array. If True, subsizes must be “equal”.

  • matrix_type ({"general", "hermitian", "posdef"}) – Type of matrix to generate. “general” generates a general matrix, “hermitian” generates a hermitian matrix, and “posdef” generates a positive definite matrix.

  • d_per_charge (bool) – Whether d specifies the dimension per charge sector (True) or the total dimension (False). If True, the total dimension is determined by the symmetry and d, only valid for ZN.

symmray.utils_test.rand_herm
symmray.utils_test.rand_posdef