qec 0.0.11__py3-none-any.whl → 0.2.1__py3-none-any.whl

qec/utils/binary_pauli_utils.py

@@ -0,0 +1,403 @@
+import numpy as np
+import scipy
+import scipy.sparse
+from qec.utils.sparse_binary_utils import convert_to_binary_scipy_sparse
+
+
+def GF4_to_binary(GF4_matrix: np.typing.ArrayLike) -> scipy.sparse.csr_matrix:
+    """
+    Convert a matrix over GF4 (elements {0,1,2,3}) to a binary sparse matrix in CSR format.
+
+    Each entry (row i, column j) is mapped as follows:
+      - 0 => no 1's (row has [0, 0])
+      - 1 => one 1 in column 2*j ([1, 0])
+      - 2 => two 1's in columns 2*j and 2*j + 1 ([1, 1])
+      - 3 => one 1 in column 2*j + 1 ([0, 1])
+
+    Parameters
+    ----------
+    GF4_matrix : ArrayLike
+        Input matrix of shape (M, N) containing only elements from {0, 1, 2, 3}.
+        Can be a dense array-like or any SciPy sparse matrix format.
+
+    Returns
+    -------
+    scipy.sparse.csr_matrix
+        Binary sparse matrix in CSR format, of shape (M, 2*N).
+
+    Raises
+    ------
+    ValueError
+        If the input matrix has elements outside {0, 1, 2, 3}.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> from scipy.sparse import scipy.sparse.csr_matrix
+    >>> mat = np.array([[0, 1],
+    ...                 [2, 3]])
+    >>> GF4_to_binary(mat).toarray()
+    array([[0, 1, 1, 0],
+           [1, 1, 0, 1]], dtype=uint8)
+    """
+    if scipy.sparse.issparse(GF4_matrix):
+        mat_coo = GF4_matrix.tocoo(copy=False)
+
+        if not np.all(np.isin(mat_coo.data, [1, 2, 3])):
+            raise ValueError(
+                "Input matrix must contain only elements from GF4: {0, 1, 2, 3}"
+            )
+
+        row_ids = []
+        col_ids = []
+        rows, cols = mat_coo.shape
+
+        for r, c, val in zip(mat_coo.row, mat_coo.col, mat_coo.data):
+            if val == 1:
+                row_ids.append(r)
+                col_ids.append(c)
+            elif val == 2:
+                row_ids.extend([r, r])
+                col_ids.extend([c, cols + c])
+            elif val == 3:
+                row_ids.append(r)
+                col_ids.append(cols + c)
+
+        data = np.ones(len(row_ids), dtype=np.uint8)
+        return scipy.sparse.csr_matrix(
+            (data, (row_ids, col_ids)), shape=(rows, 2 * cols)
+        )
+
+    GF4_matrix = np.asanyarray(GF4_matrix, dtype=int)
+    if not np.all(np.isin(GF4_matrix, [0, 1, 2, 3])):
+        raise ValueError(
+            "Input matrix must contain only elements from GF4: {0, 1, 2, 3}"
+        )
+
+    row_ids = []
+    col_ids = []
+    rows, cols = GF4_matrix.shape
+
+    for i in range(rows):
+        for j in range(cols):
+            val = GF4_matrix[i, j]
+            if val == 1:
+                row_ids.append(i)
+                col_ids.append(j)
+            elif val == 2:
+                row_ids.extend([i, i])
+                col_ids.extend([j, j + cols])
+            elif val == 3:
+                row_ids.append(i)
+                col_ids.append(j + cols)
+
+    data = np.ones(len(row_ids), dtype=np.uint8)
+    return scipy.sparse.csr_matrix((data, (row_ids, col_ids)), shape=(rows, 2 * cols))
+
+
+def pauli_str_to_binary_pcm(
+    pauli_strings: np.typing.ArrayLike,
+) -> scipy.sparse.csr_matrix:
+    """
+    Convert an (M x 1) array of Pauli strings, where each string has length N, corresponding to the number of physical qubits, into a binary parity-check matrix (PCM) with dimensions (M x 2*N).
+
+    The mapping for each qubit j in the string is:
+      - 'I' => (0|0)
+      - 'X' => (1|0)
+      - 'Z' => (0|1)
+      - 'Y' => (1|1)
+    where the first element (a),  in (a|b) is at column j and the second element (b) is at column j + N.
+
+    Parameters
+    ----------
+    pauli_strings : ArrayLike
+        Array of shape (M, 1), where each element is a string of Pauli operators
+        ('I', 'X', 'Y', 'Z'). Can be dense or any SciPy sparse matrix format with
+        an object/string dtype.
+
+    Returns
+    -------
+    scipy.sparse.csr_matrix
+        Binary parity-check matrix of shape (M, 2*N) in CSR format, where M is the number of stabilisers and
+        N is the number of physical qubits.
+    Raises
+    ------
+    ValueError
+        If any character in the Pauli strings is not one of {'I', 'X', 'Y', 'Z'}.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> paulis = np.array([["XIZ"], ["YYI"]], dtype=object)
+    >>> pcm = pauli_str_to_binary_pcm(paulis)
+    >>> pcm.toarray()
+    array([[1, 0, 0, 0, 0, 1],
+           [1, 1, 0, 1, 1, 0]], dtype=uint8)
+    """
+
+    if scipy.sparse.issparse(pauli_strings):
+        if pauli_strings.dtype == object:
+            mat_coo = pauli_strings.tocoo(copy=False)
+            dense = np.full(pauli_strings.shape, "I", dtype=str)
+            for r, c, val in zip(mat_coo.row, mat_coo.col, mat_coo.data):
+                dense[r, c] = val
+            pauli_strings = dense
+        else:
+            pauli_strings = pauli_strings.toarray()
+
+    pauli_strings = np.asanyarray(pauli_strings, dtype=str)
+
+    if pauli_strings.size == 0:
+        return scipy.sparse.csr_matrix((0, 0))
+
+    row_ids = []
+    col_ids = []
+
+    m_stabilisers = pauli_strings.shape[0]
+    n_qubits = len(pauli_strings[0, 0])
+
+    for i, string in enumerate(pauli_strings):
+        if len(string[0]) != n_qubits:
+            raise ValueError("The Pauli strings do not have equal length.")
+        for j, char in enumerate(string[0]):
+            if char == "I":
+                continue
+            elif char == "X":
+                row_ids.append(i)
+                col_ids.append(j)
+            elif char == "Z":
+                row_ids.append(i)
+                col_ids.append(j + n_qubits)
+            elif char == "Y":
+                row_ids.extend([i, i])
+                col_ids.extend([j, j + n_qubits])
+            else:
+                raise ValueError(f"Invalid Pauli character '{char}' encountered.")
+
+    data = np.ones(len(row_ids), dtype=np.uint8)
+
+    return scipy.sparse.csr_matrix(
+        (data, (row_ids, col_ids)), shape=(m_stabilisers, 2 * n_qubits), dtype=np.uint8
+    )
+
+
+def binary_pcm_to_pauli_str(binary_pcm: np.typing.ArrayLike) -> np.ndarray:
+    """
+    Convert a binary (M x 2*N) PCM corresponding to M stabilisers acting on N physical qubits,
+    back into an array (M x 1) of Pauli strings that have length N.
+
+    For each qubit j, columns (j | j + N) of the PCM encode:
+      - (0|0) => 'I'
+      - (1|0) => 'X'
+      - (0|1) => 'Z'
+      - (1|1) => 'Y'
+
+    Parameters
+    ----------
+    binary_pcm : ArrayLike
+        Binary matrix of shape (M, 2*N), in dense or any SciPy sparse matrix format.
+
+    Returns
+    -------
+    np.ndarray
+        Array of shape (M, 1), where each element is a string of Pauli operators with length N.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> from scipy.sparse import scipy.sparse.csr_matrix
+    >>> pcm = np.array([[1, 0, 0, 0, 0, 1],
+    ...                 [1, 1, 0, 1, 1, 0]], dtype=np.uint8)
+    >>> pauli_str_to_return = binary_pcm_to_pauli_str(pcm)
+    >>> pauli_str_to_return
+    array([['XIZ'],
+           ['YYI']], dtype='<U3')
+    """
+    if scipy.sparse.issparse(binary_pcm):
+        binary_pcm = binary_pcm.toarray()
+
+    binary_pcm = np.asanyarray(binary_pcm, dtype=int)
+    n_rows, n_cols = binary_pcm.shape
+    n_qubits = n_cols // 2
+    pauli_strings = [""] * n_rows
+
+    for i in range(n_rows):
+        row = binary_pcm[i]
+        x_bits = row[:n_qubits]
+        z_bits = row[n_qubits:]
+        for x_bit, z_bit in zip(x_bits, z_bits):
+            if x_bit == 0 and z_bit == 0:
+                pauli_strings[i] += "I"
+            elif x_bit == 1 and z_bit == 0:
+                pauli_strings[i] += "X"
+            elif x_bit == 0 and z_bit == 1:
+                pauli_strings[i] += "Z"
+            else:
+                pauli_strings[i] += "Y"
+
+    return np.array(pauli_strings, dtype=str).reshape(-1, 1)
+
+
+def symplectic_product(
+    a: np.typing.ArrayLike, b: np.typing.ArrayLike
+) -> scipy.sparse.csr_matrix:
+    """
+    Compute the symplectic product of two binary matrices in CSR format.
+
+    The input matrices (A,B) are first converted to binary sparse format (modulo 2)
+    and then partitioned into `x` and `z` components, where x and z have the same shape:
+
+        A = (a_x|a_z)
+        B = (b_x|b_z)
+
+    Then the symplectic product is computed as: (a_x * b_z^T + a_z * b_x^T) mod 2.
+
+    Parameters
+    ----------
+    a : array_like
+        A 2D array-like object with shape (M, 2N), which will be converted to
+        a binary sparse matrix (mod 2).
+    b : array_like
+        A 2D array-like object with shape (M, 2N), which will be converted to
+        a binary sparse matrix (mod 2). Must have the same shape as `a`.
+
+    Returns
+    -------
+    scipy.sparse.csr_matrix
+        The symplectic product of the two input matrices, stored in CSR format.
+
+    Raises
+    ------
+    AssertionError
+        If the shapes of `a` and `b` do not match.
+    AssertionError
+        If the number of columns of `a` (and `b`) is not even.
+
+    Notes
+    -----
+    This function is particularly useful for calculating commutation between Pauli operators,
+    where a result of 0 indicates commuting operators, and 1 indicates anti-commuting operators.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> from qec.utils.sparse_binary_utils import convert_to_binary_scipy_sparse
+    >>> a_data = np.array([[1, 0, 0, 1],
+    ...                    [0, 1, 1, 0],
+    ...                    [1, 1, 0, 0]], dtype=int)
+    >>> b_data = np.array([[0, 1, 1, 0],
+    ...                    [1, 0, 0, 1],
+    ...                    [0, 1, 1, 0]], dtype=int)
+    >>> # Compute symplectic product
+    >>> sp = symplectic_product(a_data, b_data)
+    >>> sp.toarray()
+    array([[0, 0, 0],
+           [0, 0, 0],
+           [1, 1, 1]], dtype=int8)
+    """
+
+    a = convert_to_binary_scipy_sparse(a)
+    b = convert_to_binary_scipy_sparse(b)
+
+    assert (
+        a.shape[1] == b.shape[1]
+    ), "Input matrices must have the same number of columns."
+    assert a.shape[1] % 2 == 0, "Input matrices must have an even number of columns."
+
+    n = a.shape[1] // 2
+
+    ax = a[:, :n]
+    az = a[:, n:]
+    bx = b[:, :n]
+    bz = b[:, n:]
+
+    sp = ax @ bz.T + az @ bx.T
+    sp.data %= 2
+
+    return sp
+
+
+def check_binary_pauli_matrices_commute(
+    mat1: scipy.sparse.spmatrix, mat2: scipy.sparse.spmatrix
+) -> bool:
+    """
+    Check if two binary Pauli matrices commute.
+    """
+    symplectic_product_result = symplectic_product(mat1, mat2)
+    symplectic_product_result.eliminate_zeros()
+    return not np.any(symplectic_product_result.data)
+
+
+def binary_pauli_hamming_weight(
+    mat: scipy.sparse.spmatrix,
+) -> np.ndarray:
+    """
+    Compute the row-wise Hamming weight of a binary Pauli matrix.
+
+    A binary Pauli matrix has 2*n columns, where the first n columns encode
+    the X part and the second n columns encode the Z part. The Hamming weight
+    for each row is the number of qubits that are acted upon by a non-identity
+    Pauli operator (X, Y, or Z). In other words, for each row, we count the
+    number of columns where either the X part or the Z part has a 1.
+
+    Parameters
+    ----------
+    mat : scipy.sparse.spmatrix
+        A binary Pauli matrix with an even number of columns (2*n). Each entry
+        must be 0 or 1, indicating whether the row has an X or Z component
+        for the corresponding qubit.
+
+    Returns
+    -------
+    np.ndarray
+        A 1D NumPy array of length `mat.shape[0]`, where the i-th entry is
+        the Hamming weight of the i-th row in `mat`.
+
+    Raises
+    ------
+    AssertionError
+        If the matrix does not have an even number of columns.
+
+    Notes
+    -----
+    Internally, this function:
+      1. Splits the matrix into the X and Z parts.
+      2. Computes an elementwise OR of the X and Z parts.
+      3. Counts the non-zero entries per row (i.e., columns where the row has a 1).
+
+    Because the bitwise OR operator `|` is not directly supported for CSR
+    matrices, we achieve the OR operation by adding the two sparse matrices
+    and capping the sum at 1. Any entries with a value >= 1 in the sum
+    are set to 1, which corresponds to OR semantics for binary data.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> from scipy.sparse import csr_matrix
+    >>> # Create a 2-row matrix, each row having 6 columns (for n=3 qubits).
+    >>> # Row 0: columns [0,2] are set -> X on qubits 0 and 2.
+    >>> # Row 1: columns [3,4,5] are set -> Z on qubit 1, Y on qubit 2.
+    >>> mat_data = np.array([[1,0,1,0,0,0],
+    ...                      [0,0,0,1,1,1]], dtype=np.uint8)
+    >>> mat_sparse = csr_matrix(mat_data)
+    >>> binary_pauli_hamming_weight(mat_sparse)
+    array([2, 2], dtype=int32)
+    """
+    assert mat.shape[1] % 2 == 0, "Input matrix must have an even number of columns."
+
+    # Determine the number of qubits from the total columns.
+    n = mat.shape[1] // 2
+
+    # Partition the matrix into X and Z parts.
+    x_part = mat[:, :n]
+    z_part = mat[:, n:]
+
+    # We want a bitwise OR. Since CSR matrices do not support a direct OR,
+    # we add and then cap at 1: (x_part + z_part >= 1) -> 1
+    xz_or = x_part.copy()
+    xz_or += z_part
+    # Clip values greater than 1 to 1.
+    xz_or.data[xz_or.data > 1] = 1
+
+    # The row-wise Hamming weight is the number of non-zero columns in each row.
+    return xz_or.getnnz(axis=1)

qec/utils/codetables_de_utils.py

@@ -0,0 +1,274 @@
+import logging
+
+# Suppress debug and info messages from urllib3 and requests libraries
+logging.getLogger("urllib3").setLevel(logging.WARNING)
+logging.getLogger("requests").setLevel(logging.WARNING)
+
+from scipy.sparse import csr_matrix
+
+import requests
+from bs4 import BeautifulSoup
+import json
+
+
+def get_codetables_de_matrix(q, n, k, output_json_path=None, write_to_file=False):
+    """
+    Retrieve quantum code data from Markus Grassl's codetables.de website.
+
+    This function queries the URL:
+    ``https://codetables.de/QECC/QECC.php?q={q}&n={n}&k={k}``,
+    attempting to fetch data for a quantum code with the specified parameters
+    over GF(q). The HTML response is parsed to extract:
+
+    - The lower bound (``d_lower``) and upper bound (``d_upper``) on the code distance.
+    - The stabilizer matrix (as lines within a ``<pre>`` block).
+
+    The stabilizer matrix is then converted into a list of rows, each containing
+    the column indices of any '1' entries (the ``pcm``). The result is returned
+    as a dictionary, and optionally written to a JSON file.
+
+    Parameters
+    ----------
+    q : int
+        The field size (e.g. 2, 4, etc.).
+    n : int
+        The length of the code (number of physical qubits).
+    k : int
+        The dimension of the code (number of logical qubits).
+    output_json_path : str or None, optional
+        File path to which the resulting dictionary will be written if
+        ``write_to_file`` is set to True. If None and ``write_to_file`` is True,
+        raises a ValueError.
+    write_to_file : bool, optional
+        Whether to write the resulting dictionary to a JSON file.
+
+    Returns
+    -------
+    dict
+        A dictionary with the fields:
+        ``{"n", "k", "d_upper", "d_lower", "url", "pcm"}``.
+
+        - ``pcm`` is a list of lists, where each inner list contains the column
+          indices of '1's for that row of the stabilizer matrix.
+        - ``url`` is the codetables.de URL used for the query.
+        - ``d_upper`` and ``d_lower`` are the distance bounds, if found.
+
+    Raises
+    ------
+    ValueError
+        If the server response is not 200 OK, or if no valid stabilizer matrix
+        lines could be found in the HTML (i.e., no code data for those parameters).
+        Also raised if ``write_to_file`` is True and ``output_json_path`` is None.
+
+    Notes
+    -----
+    - Data is sourced from `codetables.de <https://codetables.de>`__,
+      maintained by Markus Grassl.
+    - The function does not return an actual matrix but rather a convenient
+      representation of it (the ``pcm``). Use ``pcm_to_csr_matrix`` or another
+      helper to convert it into a numerical/sparse form.
+    """
+    url = f"https://codetables.de/QECC/QECC.php?q={q}&n={n}&k={k}"
+    resp = requests.get(url)
+    if resp.status_code != 200:
+        raise ValueError(
+            f"Failed to retrieve data (status code: {resp.status_code}). URL was: {url}"
+        )
+
+    soup = BeautifulSoup(resp.text, "html.parser")
+
+    # 1) Extract lower and upper distance bounds from <table> elements
+    lower_bound = None
+    upper_bound = None
+    tables = soup.find_all("table")
+    for table in tables:
+        rows = table.find_all("tr")
+        for row in rows:
+            cells = row.find_all("td")
+            if len(cells) == 2:
+                heading = cells[0].get_text(strip=True).lower()
+                value = cells[1].get_text(strip=True)
+                if "lower bound" in heading:
+                    lower_bound = value
+                elif "upper bound" in heading:
+                    upper_bound = value
+
+    # 2) Extract the stabilizer matrix lines from <pre> tags
+    matrix_lines = []
+    for tag in soup.find_all("pre"):
+        text = tag.get_text()
+        if "stabilizer matrix" in text.lower():
+            lines = text.splitlines()
+            capture = False
+            for line in lines:
+                if "stabilizer matrix" in line.lower():
+                    capture = True
+                    continue
+                if capture:
+                    # Stop at 'last modified:' or if the line is empty
+                    if "last modified:" in line.lower():
+                        break
+                    if line.strip() != "":
+                        matrix_lines.append(line.strip())
+
+    if not matrix_lines:
+        raise ValueError(f"No valid stabilizer matrix found at {url}")
+
+    # 3) Convert lines -> list of column-index lists
+    pcm_list = []
+    for line in matrix_lines:
+        line = line.strip().strip("[]").replace("|", " ")
+        elements = line.split()
+        row_cols = [i for i, val in enumerate(elements) if val == "1"]
+        pcm_list.append(row_cols)
+
+    if not pcm_list:
+        raise ValueError(f"No valid rows containing '1' found at {url}")
+
+    # 4) Build final dictionary
+    result_dict = {
+        "n": n,
+        "k": k,
+        "d_upper": upper_bound,
+        "d_lower": lower_bound,
+        "url": url,
+        "pcm": pcm_list,
+    }
+
+    # 5) Optionally write to JSON file
+    if write_to_file:
+        if output_json_path is None:
+            raise ValueError("output_json_path must be provided if write_to_file=True.")
+        with open(output_json_path, "w") as out_file:
+            json.dump(result_dict, out_file, indent=2)
+
+    return result_dict
+
+
+def pcm_to_csr_matrix(pcm, num_cols=None):
+    """
+    Convert a "pcm" to a SciPy CSR matrix.
+
+    Each inner list of ``pcm`` is interpreted as the column indices in which
+    row `i` has a value of 1. The resulting CSR matrix will thus have as many
+    rows as ``len(pcm)``. The number of columns can either be:
+
+    - Inferred automatically (``num_cols=None``) by taking 1 + max(column index).
+    - Specified by the user. If a column index is >= num_cols, a ValueError is raised.
+
+    Parameters
+    ----------
+    pcm : list of lists of int
+        Each element ``pcm[i]`` is a list of column indices where row i has '1'.
+    num_cols : int or None, optional
+        The desired number of columns (width of the matrix).
+        If None, the width is auto-detected from the maximum column index.
+
+    Returns
+    -------
+    csr_matrix
+        A sparse matrix of shape ``(len(pcm), num_cols)``.
+
+    Raises
+    ------
+    ValueError
+        If any column index exceeds the specified ``num_cols``.
+        Also raised if no rows or invalid columns exist.
+
+    See Also
+    --------
+    get_codetables_de_matrix : Returns a dictionary with ``pcm`` field from codetables.de.
+
+    Notes
+    -----
+    Data is typically retrieved from `codetables.de <https://codetables.de>`__
+    and fed into this function to produce a numerical/sparse representation.
+    """
+    if not pcm:
+        # No rows at all => shape (0, num_cols) or (0, 0) if num_cols is None
+        if num_cols is None:
+            return csr_matrix((0, 0), dtype=int)
+        else:
+            return csr_matrix((0, num_cols), dtype=int)
+
+    row_indices = []
+    col_indices = []
+    data = []
+
+    max_col_found = -1
+
+    # Collect row/col for each '1' entry
+    for row_idx, col_list in enumerate(pcm):
+        for c in col_list:
+            row_indices.append(row_idx)
+            col_indices.append(c)
+            data.append(1)
+            if c > max_col_found:
+                max_col_found = c
+
+    num_rows = len(pcm)
+
+    # Auto-detect columns if not specified
+    if num_cols is None:
+        num_cols = max_col_found + 1
+    else:
+        # If the user specified num_cols, ensure the data fits
+        if max_col_found >= num_cols:
+            raise ValueError(
+                f"Column index {max_col_found} is out of range for a matrix of width {num_cols}."
+            )
+
+    return csr_matrix(
+        (data, (row_indices, col_indices)), shape=(num_rows, num_cols), dtype=int
+    )
+
+
+def load_codetables_de_matrix_from_json(json_data):
+    """
+    Construct a CSR matrix from a codetables.de JSON/dict output.
+
+    This function takes either a dictionary (as returned by
+    ``get_codetables_de_matrix``) or a JSON string that decodes to the same
+    structure, and converts the ``pcm`` field into a SciPy CSR matrix.
+
+    Parameters
+    ----------
+    json_data : dict or str
+        Must contain at least the following keys:
+        ``{"n", "k", "d_upper", "d_lower", "url", "pcm"}``.
+        - ``pcm`` is a list of lists of column indices.
+
+    Returns
+    -------
+    csr_matrix
+        The stabilizer matrix in CSR format.
+    dict
+        The original dictionary that was passed in (or parsed from JSON).
+
+    Raises
+    ------
+    ValueError
+        If ``json_data`` is not a dict, if it cannot be parsed into one,
+        or if required keys are missing.
+
+    Notes
+    -----
+    - Data is assumed to come from Markus Grassl's `codetables.de <https://codetables.de>`__.
+    - This utility is helpful when the data is stored or transmitted in JSON form
+      but needs to be loaded back into a matrix representation for further processing.
+    """
+    if isinstance(json_data, str):
+        json_data = json.loads(json_data)
+
+    if not isinstance(json_data, dict):
+        raise ValueError(
+            "json_data must be a dict or a JSON string that decodes to a dict."
+        )
+
+    required_keys = {"n", "k", "d_upper", "d_lower", "url", "pcm"}
+    if not required_keys.issubset(json_data.keys()):
+        raise ValueError(f"JSON data missing required keys: {required_keys}")
+
+    pcm = json_data["pcm"]
+    sparse_matrix = pcm_to_csr_matrix(pcm)
+    return sparse_matrix, json_data