Removed qiskit as a dependency and added seeds to randomness-related functions

pyproject.toml

@@ -24,7 +24,6 @@ numpy = "2.1.2"
 scipy = "1.14.1"
 scs = "3.2.7"
 picos = "2.4.17"
-qiskit = "1.2.4"
 
 
 [tool.poetry.group.dev.dependencies]

toqito/rand/random_circulant_gram_matrix.py

@@ -3,7 +3,7 @@
 import numpy as np
 
 
-def random_circulant_gram_matrix(dim: int) -> np.ndarray:
+def random_circulant_gram_matrix(dim: int, seed: int | None = None) -> np.ndarray:
     r"""Generate a random circulant Gram matrix of specified dimension.
 
     A circulant matrix is a square matrix where the elements of each row are identical to the elements of the
@@ -38,6 +38,16 @@ def random_circulant_gram_matrix(dim: int) -> np.ndarray:
            [0.04257471, 0.21058986, 0.42351891, 0.21058986],
            [0.21058986, 0.04257471, 0.21058986, 0.42351891]])
 
+    It is also possible to pass a seed to this function for reproducibility.
+
+    >>> from toqito.rand import random_circulant_gram_matrix
+    >>> circulant_matrix = random_circulant_gram_matrix(4, seed=42)
+    >>> circulant_matrix
+    array([[ 0.69220011, -0.02116047,  0.12407687, -0.02116047],
+           [-0.02116047,  0.69220011, -0.02116047,  0.12407687],
+           [ 0.12407687, -0.02116047,  0.69220011, -0.02116047],
+           [-0.02116047,  0.12407687, -0.02116047,  0.69220011]])
+
 
     References
     ==========
@@ -46,13 +56,16 @@ def random_circulant_gram_matrix(dim: int) -> np.ndarray:
 
     :param dim: int
         The dimension of the circulant matrix to generate.
+    :param seed: int | None
+        A seed used to instantiate numpy's random number generator.
 
     :return: numpy.ndarray
         A `dim` x `dim` real, symmetric, circulant matrix.
 
     """
+    gen = np.random.default_rng(seed=seed)
     # Step 1: Generate a random diagonal matrix with non-negative entries
-    diag_mat = np.diag(np.random.rand(dim))
+    diag_mat = np.diag(gen.random(dim))
 
     # Step 2: Construct the normalized DFT matrix
     dft_mat = np.fft.fft(np.eye(dim)) / np.sqrt(dim)

toqito/rand/random_density_matrix.py

@@ -10,6 +10,7 @@ def random_density_matrix(
     is_real: bool = False,
     k_param: list[int] | int = None,
     distance_metric: str = "haar",
+    seed: int | None = None,
 ) -> np.ndarray:
     r"""Generate a random density matrix.
 
@@ -74,6 +75,21 @@ def random_density_matrix(
     >>> is_density(bures_mat)
     np.True_
 
+    It is also possible to pass a seed to this function for reproducibility.
+
+    >>> from toqito.rand import random_density_matrix
+    >>> seeded = random_density_matrix(2, seed=42)
+    >>> seeded
+    array([[0.82448019+0.j        , 0.14841568-0.33318114j],
+           [0.14841568+0.33318114j, 0.17551981+0.j        ]])
+
+    We can once again verify that this is in fact a valid density matrix using the
+    :code:`is_density` function from :code:`toqito` as follows
+
+    >>> from toqito.matrix_props import is_density
+    >>> is_density(seeded)
+    np.True_
+
 
     :param dim: The number of rows (and columns) of the density matrix.
     :param is_real: Boolean denoting whether the returned matrix will have all
@@ -83,20 +99,22 @@ def random_density_matrix(
                             density matrix. This metric is either the Haar
                             measure or the Bures measure. Default value is to
                             use the Haar measure.
+    :param seed: A seed used to instantiate numpy's random number generator.
     :return: A :code:`dim`-by-:code:`dim` random density matrix.
 
     """
+    gen = np.random.default_rng(seed=seed)
     if k_param is None:
         k_param = dim
 
     # Haar / Hilbert-Schmidt measure.
-    gin = np.random.rand(dim, k_param)
+    gin = gen.random((dim, k_param))
 
     if not is_real:
-        gin = gin + 1j * np.random.randn(dim, k_param)
+        gin = gin + 1j * gen.standard_normal((dim, k_param))
 
     if distance_metric == "bures":
-        gin = random_unitary(dim, is_real) + np.identity(dim) @ gin
+        gin = random_unitary(dim, is_real, seed=seed) + np.identity(dim) @ gin
 
     rho = gin @ np.array(gin).conj().T
 

toqito/rand/random_ginibre.py

@@ -3,7 +3,7 @@
 import numpy as np
 
 
-def random_ginibre(dim_n: int, dim_m: int) -> np.ndarray:
+def random_ginibre(dim_n: int, dim_m: int, seed: int | None = None) -> np.ndarray:
     r"""Generate a Ginibre random matrix :cite:`WikiCircLaw`.
 
     Generates a random :code:`dim_n`-by-:code:`dim_m` Ginibre matrix.
@@ -23,6 +23,12 @@ def random_ginibre(dim_n: int, dim_m: int) -> np.ndarray:
     array([[0.39166472-1.54657971j, 0.36538245+0.23324642j],
            [0.50103695-0.25857737j, 0.8357054 +0.31404353j]])
 
+    It is also possible to pass a seed to this function for reproducibility.
+
+    >>> from toqito.rand import random_ginibre
+    >>> random_ginibre(2, 2, seed=42)
+    array([[ 0.21546751-1.37959021j, -0.73537981-0.92077996j],
+           [ 0.53064913+0.09039682j,  0.66507969-0.22361728j]])
 
 
     References
@@ -33,7 +39,9 @@ def random_ginibre(dim_n: int, dim_m: int) -> np.ndarray:
 
     :param dim_n: The number of rows of the Ginibre random matrix.
     :param dim_m: The number of columns of the Ginibre random matrix.
+    :param seed: A seed used to instantiate numpy's random number generator.
     :return: A :code:`dim_n`-by-:code:`dim_m` Ginibre random density matrix.
 
     """
-    return (np.random.randn(dim_n, dim_m) + 1j * np.random.randn(dim_n, dim_m)) / np.sqrt(2)
+    gen = np.random.default_rng(seed=seed)
+    return (gen.standard_normal((dim_n, dim_m)) + 1j * gen.standard_normal((dim_n, dim_m))) / np.sqrt(2)