rapidtide 3.0.10__py3-none-any.whl → 3.1__py3-none-any.whl

rapidtide/util.py

@@ -27,13 +27,17 @@ import sys
 import time
 from datetime import datetime
 from multiprocessing import shared_memory
+from typing import Any
 
 import matplotlib.pyplot as plt
 import numpy as np
 import pandas as pd
+from numpy.typing import NDArray
 
 import rapidtide._version as tide_versioneer
+import rapidtide.decorators
 import rapidtide.io as tide_io
+from rapidtide.decorators import getdecoratorvars
 
 try:
     import mkl
@@ -50,16 +54,33 @@ MemoryLGR = logging.getLogger("MEMORY")
 # ---------------------------------------- Global constants -------------------------------------------
 defaultbutterorder = 6
 MAXLINES = 10000000
-donotbeaggressive = True
 
-# ----------------------------------------- Conditional imports ---------------------------------------
-try:
-    from numba import jit
-except ImportError:
+
+def disablenumba() -> None:
+    """
+    Set a global variable to disable numba.
+
+    This function sets the global variable `donotusenumba` to `True`, which
+    effectively disables the use of numba in subsequent operations that check
+    this variable.
+
+    Notes
+    -----
+    This function modifies a global variable. The variable `donotusenumba` should
+    be checked by other functions in the codebase to determine whether to use
+    numba or not.
+
+    Examples
+    --------
+    >>> disablenumba()
+    >>> print(donotusenumba)
+    True
+    """
+    global donotusenumba
     donotusenumba = True
-else:
-    donotusenumba = False
 
+
+# ----------------------------------------- Conditional imports ---------------------------------------
 try:
     import pyfftw
 except ImportError:
@@ -68,7 +89,52 @@ else:
     pyfftwpresent = True
 
 
-def checkimports(optiondict):
+def checkimports(optiondict: dict[str, Any]) -> None:
+    """
+    Check availability of optional dependencies and optimization settings.
+
+    This function verifies the presence of optional packages and optimization
+    settings, printing status messages and updating the provided dictionary with
+    the results. It checks for pyfftw, aggressive optimization flags, and numba
+    usage settings.
+
+    Parameters
+    ----------
+    optiondict : dict[str, Any]
+        Dictionary to be updated with boolean values indicating the status of
+        optional dependencies and optimization settings. The dictionary will be
+        modified in-place with the following keys:
+
+        - "pfftwexists": bool, True if pyfftw is available, False otherwise
+        - "donotbeaggressive": bool, True if aggressive optimization is disabled,
+          False if enabled
+        - "donotusenumba": bool, True if numba usage is disabled, False if numba
+          will be used when available
+
+    Returns
+    -------
+    None
+        This function does not return a value but modifies the input dictionary
+        in-place.
+
+    Notes
+    -----
+    The function relies on global variables:
+    - `pyfftwpresent`: Indicates if pyfftw is available
+    - `donotbeaggressive`: Controls aggressive optimization flag
+    - `donotusenumba`: Controls numba usage flag
+
+    Examples
+    --------
+    >>> options = {}
+    >>> checkimports(options)
+    pfftw does not exist
+    aggressive optimization
+    using numba if present
+    >>> print(options)
+    {'pfftwexists': False, 'donotbeaggressive': False, 'donotusenumba': False}
+    """
+    donotusenumba, donotbeaggressive = getdecoratorvars()
     if pyfftwpresent:
         print("pfftw exists")
     else:
@@ -81,7 +147,6 @@ def checkimports(optiondict):
         print("aggressive optimization")
     optiondict["donotbeaggressive"] = donotbeaggressive
 
-    global donotusenumba
     if donotusenumba:
         print("will not use numba even if present")
     else:
@@ -89,31 +154,41 @@ def checkimports(optiondict):
     optiondict["donotusenumba"] = donotusenumba
 
 
-# ----------------------------------------- Conditional jit handling ----------------------------------
-def conditionaljit():
-    def resdec(f):
-        if donotusenumba:
-            return f
-        return jit(f, nopython=True)
-
-    return resdec
+def disablemkl(numprocs: int, debug: bool = False) -> None:
+    """
+    Disable MKL threading for parallel execution.
 
+    This function configures Intel MKL (Math Kernel Library) to use only a single
+    thread when the number of processes exceeds 1. This is useful for avoiding
+    oversubscription of CPU resources in parallel computing environments.
 
-def conditionaljit2():
-    def resdec(f):
-        if donotusenumba or donotbeaggressive:
-            return f
-        return jit(f, nopython=True)
+    Parameters
+    ----------
+    numprocs : int
+        Number of processes to check against. If greater than 1, MKL threading
+        will be disabled by setting the number of threads to 1.
+    debug : bool, optional
+        If True, prints debug information about the threading configuration
+        (default is False).
 
-    return resdec
+    Returns
+    -------
+    None
+        This function does not return any value.
 
+    Notes
+    -----
+    This function only has an effect if MKL is available (mklexists is True).
+    The function uses mkl.set_num_threads(1) to disable parallel threading in MKL.
 
-def disablenumba():
-    global donotusenumba
-    donotusenumba = True
+    Examples
+    --------
+    >>> disablemkl(numprocs=4, debug=True)
+    disablemkl: setting threads to 1
 
-
-def disablemkl(numprocs, debug=False):
+    >>> disablemkl(numprocs=1)
+    # No output, no threading changes
+    """
     if mklexists:
         if numprocs > 1:
             if debug:
@@ -121,7 +196,37 @@ def disablemkl(numprocs, debug=False):
             mkl.set_num_threads(1)
 
 
-def enablemkl(numthreads, debug=False):
+def enablemkl(numthreads: int, debug: bool = False) -> None:
+    """
+    Enable Intel MKL threading with specified number of threads.
+
+    This function configures the Intel MKL (Math Kernel Library) to use the
+    specified number of threads for parallel execution. It only has an effect
+    if MKL is available in the current environment.
+
+    Parameters
+    ----------
+    numthreads : int
+        Number of threads to use for MKL operations. Must be a positive integer.
+    debug : bool, optional
+        If True, print debug information about the thread setting operation.
+        Default is False.
+
+    Returns
+    -------
+    None
+        This function does not return any value.
+
+    Notes
+    -----
+    This function only has an effect if MKL is available (mklexists is True).
+    The function uses mkl.set_num_threads() internally to configure the threading.
+
+    Examples
+    --------
+    >>> enablemkl(4)
+    >>> enablemkl(8, debug=True)
+    """
     if mklexists:
         if debug:
             print(f"enablemkl: setting threads to {numthreads}")
@@ -129,7 +234,32 @@ def enablemkl(numthreads, debug=False):
 
 
 # --------------------------- Utility functions -------------------------------------------------
-def findavailablemem():
+def findavailablemem() -> tuple[int, int]:
+    """
+    Get available memory information from system resources.
+
+    This function retrieves memory information from either cgroup limits or system
+    free memory statistics. It returns a tuple containing the memory limit and
+    swap information, both in bytes.
+
+    Returns
+    -------
+    tuple[int, int]
+        A tuple containing two integers:
+        - First integer: Memory limit in bytes (from cgroup or total memory)
+        - Second integer: Swap available in bytes (from cgroup or swap memory)
+
+    Notes
+    -----
+    The function first checks for cgroup memory limits at "/sys/fs/cgroup/memory/memory.limit_in_bytes".
+    If found, it returns the limit for both values in the tuple. Otherwise, it uses the "free" command
+    to retrieve system memory information, specifically the free memory and swap memory values.
+
+    Examples
+    --------
+    >>> findavailablemem()
+    (8589934592, 2147483648)
+    """
     if os.path.isfile("/sys/fs/cgroup/memory/memory.limit_in_bytes"):
         with open("/sys/fs/cgroup/memory/memory.limit_in_bytes") as limit:
             mem = int(limit.read())
@@ -141,13 +271,45 @@ def findavailablemem():
         return free, swap
 
 
-def checkifincontainer():
-    # Determine if the program is running in a container.  If so, we may need to adjust the python memory
-    # limits because they are not set properly.  But check if we're running on CircleCI - it does not seem
-    # to like you twiddling with the container parameters.
-    #
-    # possible return values are: None, "Docker", "Singularity", and "CircleCI"
-    #
+def checkifincontainer() -> str | None:
+    """
+    Determine if the program is running in a container and identify the container type.
+
+    This function checks environment variables to detect whether the program is running
+    inside a container environment. It specifically looks for indicators of Docker,
+    Singularity, and CircleCI environments. The function returns the container type
+    as a string, or None if running outside any container.
+
+    Returns
+    -------
+    str or None
+        Container type if running in a container, otherwise None. Possible return values:
+        - "Docker": Running in a Docker container (indicated by RUNNING_IN_CONTAINER env var)
+        - "Singularity": Running in a Singularity container (indicated by SINGULARITY_CONTAINER env var)
+        - "CircleCI": Running in CircleCI environment (indicated by CIRCLECI env var)
+        - None: Not running in any container environment
+
+    Notes
+    -----
+    The function prioritizes detection in the following order:
+    1. Singularity containers (SINGULARITY_CONTAINER env var)
+    2. Docker containers (RUNNING_IN_CONTAINER env var)
+    3. CircleCI environment (CIRCLECI env var)
+
+    CircleCI detection takes precedence over other container types, as CircleCI
+    environments may not handle container parameter adjustments properly.
+
+    Examples
+    --------
+    >>> checkifincontainer()
+    'Docker'
+
+    >>> checkifincontainer()
+    'Singularity'
+
+    >>> checkifincontainer()
+    None
+    """
     if os.environ.get("SINGULARITY_CONTAINER") is not None:
         containertype = "Singularity"
     elif os.environ.get("RUNNING_IN_CONTAINER") is not None:
@@ -159,11 +321,75 @@ def checkifincontainer():
     return containertype
 
 
-def setmemlimit(memlimit):
+def setmemlimit(memlimit: int) -> None:
+    """
+    Set the memory limit for the current process.
+
+    This function sets the virtual memory limit (RLIMIT_AS) for the current process
+    using the resource module. The limit is specified in bytes and applies to both
+    soft and hard limits.
+
+    Parameters
+    ----------
+    memlimit : int
+        The memory limit in bytes. Setting this to -1 will remove the limit.
+        Values should be non-negative integers.
+
+    Returns
+    -------
+    None
+        This function does not return any value.
+
+    Notes
+    -----
+    - This function uses `resource.setrlimit()` with `resource.RLIMIT_AS`
+    - The memory limit is enforced by the operating system
+    - Setting memlimit to -1 removes any existing memory limit
+    - This function may raise `ValueError` or `OSError` if the limit cannot be set
+    - The limit applies to the current process and its children
+
+    Examples
+    --------
+    >>> setmemlimit(1024 * 1024 * 100)  # Set limit to 100 MB
+    >>> setmemlimit(-1)                 # Remove memory limit
+    """
     resource.setrlimit(resource.RLIMIT_AS, (memlimit, memlimit))
 
 
-def formatmemamt(meminbytes):
+def formatmemamt(meminbytes: int) -> str:
+    """
+    Format memory amount in bytes to human readable format.
+
+    Convert a memory size in bytes to a human readable string with appropriate units
+    (B, kB, MB, GB, TB).
+
+    Parameters
+    ----------
+    meminbytes : int
+        Memory amount in bytes to be formatted.
+
+    Returns
+    -------
+    str
+        Formatted memory amount with appropriate unit. The result is rounded to 3
+        decimal places and includes the unit suffix.
+
+    Notes
+    -----
+    The function uses binary units (1024-based) rather than decimal units (1000-based).
+    Units are: B (bytes), kB (kilobytes), MB (megabytes), GB (gigabytes), TB (terabytes).
+
+    Examples
+    --------
+    >>> formatmemamt(1024)
+    '1.000kB'
+
+    >>> formatmemamt(1048576)
+    '1.000MB'
+
+    >>> formatmemamt(1073741824)
+    '1.000GB'
+    """
     units = ["B", "kB", "MB", "GB", "TB"]
     index = 0
     unitnumber = np.uint64(1)
@@ -177,7 +403,44 @@ def formatmemamt(meminbytes):
     return f"{round(meminbytes/unitnumber, 3):.3f}{units[-1]}"
 
 
-def format_bytes(size):
+def format_bytes(size: float) -> tuple[float, str]:
+    """
+    Convert a size in bytes to a human-readable format with appropriate units.
+
+    Convert a size in bytes to a more readable format by scaling it to the
+    appropriate unit (bytes, kilobytes, megabytes, gigabytes, terabytes).
+
+    Parameters
+    ----------
+    size : float
+        The size in bytes to be converted. Should be a non-negative number.
+
+    Returns
+    -------
+    tuple[float, str]
+        A tuple containing the scaled size (float) and the corresponding unit (str).
+        The unit will be one of: 'bytes', 'kilobytes', 'megabytes', 'gigabytes', 'terabytes'.
+
+    Notes
+    -----
+    This function uses base-2 (binary) units where 1 kilobyte = 1024 bytes.
+    The conversion continues until the size is less than 1024, at which point
+    the appropriate unit is returned.
+
+    Examples
+    --------
+    >>> format_bytes(512)
+    (512.0, 'bytes')
+
+    >>> format_bytes(2048)
+    (2.0, 'kilobytes')
+
+    >>> format_bytes(1048576)
+    (1.0, 'megabytes')
+
+    >>> format_bytes(1073741824)
+    (1.0, 'gigabytes')
+    """
     # 2**10 = 1024
     power = 2**10
     n = 0
@@ -188,15 +451,43 @@ def format_bytes(size):
     return size, power_labels[n] + "bytes"
 
 
-def logmem(msg=None):
-    """Log memory usage with a logging object.
+def logmem(msg: str | None = None) -> None:
+    """
+    Log memory usage with a logging object.
+
+    This function logs detailed memory usage statistics for the current process
+    and its children, including resident set size (RSS), shared and unshared memory,
+    page faults, and swap usage. On Windows, memory statistics are not available
+    and a placeholder message is logged instead.
 
     Parameters
     ----------
-    msg : str or None, optional
-        A message to include in the first column.
-        If None, the column headers are logged.
-        Default is None.
+    msg : str, optional
+        A message to include in the first column of the logged output.
+        If None, column headers are logged instead. Default is None.
+
+    Returns
+    -------
+    None
+        This function does not return any value; it logs information to a global
+        logger named `MemoryLGR`.
+
+    Notes
+    -----
+    - On Unix-like systems (Linux, macOS), this function uses `resource.getrusage`
+      to retrieve memory usage details.
+    - On Windows, memory statistics are not supported and a placeholder message
+      is logged.
+    - The function maintains internal state (`lastmaxrss_parent`, `lastmaxrss_child`)
+      to compute differences in memory usage between calls.
+
+    Examples
+    --------
+    >>> logmem("Before loop")
+    # Logs memory usage with "Before loop" as the first column
+
+    >>> logmem()
+    # Logs column headers for memory usage statistics
     """
     global lastmaxrss_parent, lastmaxrss_child
     if platform.system() != "Windows":
@@ -252,16 +543,37 @@ def logmem(msg=None):
     MemoryLGR.info("\t".join(outvals))
 
 
-def findexecutable(command):
+def findexecutable(command: str) -> str | None:
     """
+    Locate an executable file in the system PATH.
+
+    This function searches for an executable file with the given name in the
+    system's PATH environment variable. It uses the most appropriate method
+    based on the Python version.
 
     Parameters
     ----------
-    command
+    command : str
+        The name of the executable command to search for.
 
     Returns
     -------
-
+    str or None
+        The full path to the executable if found, None otherwise.
+
+    Notes
+    -----
+    For Python 3.3 and later, this function uses `shutil.which()` which is the
+    recommended approach. For earlier Python versions, it manually searches
+    through the PATH environment variable and checks execute permissions.
+
+    Examples
+    --------
+    >>> findexecutable('python')
+    '/usr/bin/python'
+
+    >>> findexecutable('nonexistent_command')
+    None
     """
     import shutil
 
@@ -275,16 +587,39 @@ def findexecutable(command):
         return None
 
 
-def isexecutable(command):
+def isexecutable(command: str) -> bool:
     """
+    Check if a command is executable in the system's PATH.
+
+    This function determines whether a given command can be executed by checking
+    if it exists in the system's PATH and has execute permissions. For Python 3.3+
+    the function uses shutil.which() for cross-platform compatibility, while for
+    older versions it manually checks execute permissions in each PATH directory.
 
     Parameters
     ----------
-    command
+    command : str
+        The name of the command to check for executability.
 
     Returns
     -------
-
+    bool
+        True if the command is executable, False otherwise.
+
+    Notes
+    -----
+    This function provides cross-platform compatibility by using different
+    approaches depending on the Python version. For Python 3.3 and later,
+    shutil.which() is used which handles platform-specific path searching.
+    For older Python versions, the function manually checks execute permissions
+    in each directory listed in the PATH environment variable.
+
+    Examples
+    --------
+    >>> isexecutable('python')
+    True
+    >>> isexecutable('nonexistent_command')
+    False
     """
     import shutil
 
@@ -301,7 +636,42 @@ def isexecutable(command):
         )
 
 
-def makeadir(pathname):
+def makeadir(pathname: str) -> bool:
+    """
+    Create a directory if it doesn't already exist.
+
+    This function attempts to create a directory at the specified path. If the
+    directory already exists, it returns True without raising an error. If the
+    directory cannot be created due to a permissions error or other OS-related
+    issues, it returns False.
+
+    Parameters
+    ----------
+    pathname : str
+        The path of the directory to create. Can be a relative or absolute path.
+
+    Returns
+    -------
+    bool
+        True if the directory exists or was successfully created, False otherwise.
+
+    Notes
+    -----
+    This function uses `os.makedirs()` which creates all intermediate-level
+    directories needed to contain the leaf directory. If the directory already
+    exists, no error is raised.
+
+    Examples
+    --------
+    >>> makeadir('test_directory')
+    True
+
+    >>> makeadir('path/to/new/directory')
+    True
+
+    >>> makeadir('/root/protected_directory')
+    False  # Will fail due to insufficient permissions
+    """
     try:
         os.makedirs(pathname)
     except OSError:
@@ -315,7 +685,33 @@ def makeadir(pathname):
     return True
 
 
-def findreferencedir():
+def findreferencedir() -> str:
+    """
+    Find and return the path to the rapidtide reference data directory.
+
+    This function locates the site-packages directory and constructs the path to
+    the rapidtide reference data folder. It searches through all site-packages
+    directories to find the one ending with "site-packages" and then builds
+    the reference directory path relative to that location.
+
+    Returns
+    -------
+    str
+        Absolute path to the rapidtide reference data directory, typically
+        structured as: {site-packages-dir}/rapidtide/data/reference/
+
+    Notes
+    -----
+    This function is designed to work within the rapidtide package environment
+    and assumes that rapidtide is installed in a standard Python site-packages
+    location. The function will return None if no site-packages directory is found.
+
+    Examples
+    --------
+    >>> ref_dir = findreferencedir()
+    >>> print(ref_dir)
+    '/usr/local/lib/python3.8/site-packages/rapidtide/data/reference'
+    """
     # Get the list of directories
     site_packages_dirs = site.getsitepackages()
 
@@ -335,33 +731,89 @@ def findreferencedir():
     return referencedir
 
 
-def savecommandline(theargs, thename):
+def savecommandline(theargs: list[str], thename: str) -> None:
     """
+    Save command line arguments to a text file.
+
+    This function takes a list of command line arguments and saves them
+    as a single line in a text file with a specified name.
 
     Parameters
     ----------
-    theargs
-    thename
+    theargs : list[str]
+        List of command line arguments to be saved
+    thename : str
+        Base name for the output file (without extension)
 
     Returns
     -------
-
+    None
+        This function does not return any value
+
+    Notes
+    -----
+    The function creates a file named ``{thename}_commandline.txt`` containing
+    the command line arguments joined by spaces on a single line.
+
+    Examples
+    --------
+    >>> savecommandline(['python', 'script.py', '--verbose'], 'myrun')
+    # Creates file 'myrun_commandline.txt' with content: "python script.py --verbose"
     """
     tide_io.writevec([" ".join(theargs)], thename + "_commandline.txt")
 
 
-def startendcheck(timepoints, startpoint, endpoint):
+def startendcheck(timepoints: int, startpoint: int, endpoint: int) -> tuple[int, int]:
     """
+    Validate and adjust start and end points for time series processing.
+
+    This function checks if the provided start and end points are within valid
+    bounds for a time series with the specified number of time points. It handles
+    edge cases by adjusting values to reasonable defaults and raises errors for
+    invalid configurations.
 
     Parameters
     ----------
-    timepoints
-    startpoint
-    endpoint
+    timepoints : int
+        Total number of time points in the series. Must be positive.
+    startpoint : int
+        Starting index for the time series segment. If negative, set to 0.
+        If greater than timepoints-1, the program exits with an error.
+    endpoint : int
+        Ending index for the time series segment. If -1, set to a large default value.
+        If greater than timepoints-1, set to timepoints-1.
 
     Returns
     -------
-
+    tuple[int, int]
+        A tuple containing (realstart, realend) where both values are valid
+        indices for the time series. realstart <= realend and both are within
+        the valid range [0, timepoints-1].
+
+    Notes
+    -----
+    - If startpoint is negative, it's automatically set to 0
+    - If endpoint is -1, it's set to 100000000 (large default value)
+    - If endpoint exceeds timepoints-1, it's set to timepoints-1
+    - The function exits with sys.exit() if startpoint >= endpoint or if
+      startpoint exceeds the maximum valid index
+
+    Examples
+    --------
+    >>> startendcheck(10, 2, 5)
+    startpoint set to 2
+    endpoint set to 5
+    (2, 5)
+
+    >>> startendcheck(5, -1, 3)
+    startpoint set to minimum, (0)
+    endpoint set to 3
+    (0, 3)
+
+    >>> startendcheck(5, 2, -1)
+    startpoint set to 2
+    endpoint set to maximum, (4)
+    (2, 4)
     """
     if startpoint > timepoints - 1:
         print("startpoint is too large (maximum is ", timepoints - 1, ")")
@@ -387,33 +839,61 @@ def startendcheck(timepoints, startpoint, endpoint):
 
 
 def valtoindex(
-    thearray,
-    thevalue,
-    evenspacing=True,
-    discrete=True,
-    discretization="round",
-    debug=False,
-):
+    thearray: NDArray,
+    thevalue: float,
+    evenspacing: bool = True,
+    discrete: bool = True,
+    discretization: str = "round",
+    debug: bool = False,
+) -> int:
     """
+    Find the index of the closest value in an ordered array to a given value.
+
+    This function computes the index of the element in `thearray` that is closest
+    to `thevalue`. It supports both evenly spaced and unevenly spaced arrays,
+    with options for discrete or continuous index output and different rounding
+    methods.
 
     Parameters
     ----------
-    thearray: array-like
-        An ordered list of values (does not need to be equally spaced)
-    thevalue: float
-        The value to search for in the array
-    evenspacing: boolean, optional
+    thearray : array-like
+        An ordered list of values (does not need to be equally spaced).
+    thevalue : float
+        The value to search for in the array.
+    evenspacing : bool, optional
         If True (default), assume data is evenly spaced for faster calculation.
-    discrete: boolean, optional
-        If True make the index an integer (round by default).
-    discretization: string, optional
-        Select rounding method - floor, ceiling, or round(default)
+    discrete : bool, optional
+        If True (default), the returned index is an integer.
+    discretization : str, optional
+        Select rounding method when `discrete=True`. Options are:
+        - "round" (default): round to nearest integer
+        - "floor": round down to nearest integer
+        - "ceiling": round up to nearest integer
+    debug : bool, optional
+        If True, print debug information during execution.
 
     Returns
     -------
-    closestidx: int
-        The index of the sample in thearray that is closest to val
-
+    int or float
+        The index of the closest value in `thearray` to `thevalue`. If `discrete=False`,
+        the index may be a float.
+
+    Notes
+    -----
+    When `evenspacing=True`, the function assumes uniform spacing between elements
+    and calculates the index using a linear interpolation formula. This is faster
+    than the default method but only accurate for evenly spaced data.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> arr = np.array([0, 1, 2, 3, 4])
+    >>> valtoindex(arr, 2.3)
+    2
+    >>> valtoindex(arr, 2.7, discretization="ceil")
+    3
+    >>> valtoindex(arr, 2.5, evenspacing=False)
+    2
     """
     if evenspacing:
         limval = np.max([thearray[0], np.min([thearray[-1], thevalue])])
@@ -444,19 +924,43 @@ def valtoindex(
         return int((np.abs(thearray - thevalue)).argmin())
 
 
-def progressbar(thisval, end_val, label="Percent", barsize=60):
+def progressbar(thisval: int, end_val: int, label: str = "Percent", barsize: int = 60) -> None:
     """
+    Display a progress bar in the terminal.
+
+    This function creates a visual progress indicator that updates in place
+    on the terminal. It shows a bar filled according to the progress percentage
+    and displays the percentage value.
 
     Parameters
     ----------
-    thisval
-    end_val
-    label
-    barsize
+    thisval : int
+        Current progress value. Should be less than or equal to ``end_val``.
+    end_val : int
+        Total value representing 100% progress.
+    label : str, optional
+        Label to display before the progress bar (default is "Percent").
+    barsize : int, optional
+        Size of the progress bar in characters (default is 60).
 
     Returns
     -------
-
+    None
+        This function does not return any value. It prints directly to stdout.
+
+    Notes
+    -----
+    The progress bar updates in place using carriage return (`\\r`) to overwrite
+    the previous output. The function uses ``sys.stdout.flush()`` to ensure
+    immediate display updates.
+
+    Examples
+    --------
+    >>> progressbar(25, 100, "Loading", 30)
+    Loading: [##############################                          ] 25.00%
+
+    >>> progressbar(50, 50)
+    Percent: [##################################################] 100.00%
     """
     percent = float(thisval) / end_val
     hashes = "#" * int(round(percent * barsize))
@@ -465,18 +969,40 @@ def progressbar(thisval, end_val, label="Percent", barsize=60):
     sys.stdout.flush()
 
 
-def makelaglist(lagstart, lagend, lagstep):
+def makelaglist(lagstart: float, lagend: float, lagstep: float) -> NDArray:
     """
+    Create a list of lag values from start to end with specified step size.
+
+    This function generates an array of evenly spaced lag values starting from
+    `lagstart` up to (and including) `lagend` with increments of `lagstep`.
 
     Parameters
     ----------
-    lagstart
-    lagend
-    lagstep
+    lagstart : float
+        The starting value of the lag sequence.
+    lagend : float
+        The ending value of the lag sequence (inclusive).
+    lagstep : float
+        The step size between consecutive lag values.
 
     Returns
     -------
-
+    NDArray
+        Array of lag values from `lagstart` to `lagend` with step size `lagstep`.
+
+    Notes
+    -----
+    The function adjusts the `lagend` value to ensure that the last value in the
+    sequence is exactly `lagend` if it's a valid step from `lagstart`. The actual
+    number of steps is calculated as ``(lagend - lagstart) // lagstep + 1``.
+
+    Examples
+    --------
+    >>> makelaglist(0.0, 1.0, 0.2)
+    array([0. , 0.2, 0.4, 0.6, 0.8, 1. ])
+
+    >>> makelaglist(1.0, 5.0, 1.5)
+    array([1. , 2.5, 4. ])
     """
     numsteps = int((lagend - lagstart) // lagstep + 1)
     lagend = lagstart + lagstep * (numsteps - 1)
@@ -497,12 +1023,42 @@ def makelaglist(lagstart, lagend, lagstep):
 
 
 # ------------------------------------------ Version function ----------------------------------
-def version():
+def version() -> tuple[str, str, str, bool | str]:
     """
+    Retrieve version information for the package, including version string,
+    Git SHA, commit date, and dirty status.
+
+    This function attempts to retrieve version information from environment
+    variables when running inside a container. If not in a container, it falls
+    back to using `tide_versioneer.get_versions()` to obtain version details
+    from the Git repository.
 
     Returns
     -------
-
+    tuple of (str, str, str, bool or str)
+        A tuple containing:
+        - version (str): The version string, potentially modified for container builds.
+        - sha (str): The Git commit SHA, or "UNKNOWN" if not available.
+        - date (str): The Git commit date, or "UNKNOWN" if not available.
+        - isdirty (bool or str): Indicates whether the working directory is dirty
+          (i.e., has uncommitted changes). Returns `True`, `False`, or `"UNKNOWN"`
+          if the information is not available.
+
+    Notes
+    -----
+    - In containerized environments, version information is expected to be
+      provided via environment variables: `GITVERSION`, `GITDIRECTVERSION`,
+      `GITSHA`, and `GITDATE`.
+    - If the environment variable `RUNNING_IN_CONTAINER` is not set, the function
+      attempts to use `tide_versioneer` to extract version information from the
+      Git repository.
+    - If `tide_versioneer` is not available or fails, the function returns
+      `("UNKNOWN", "UNKNOWN", "UNKNOWN", "UNKNOWN")`.
+
+    Examples
+    --------
+    >>> version()
+    ('1.2.3', 'a1b2c3d', '2023-04-05', False)
     """
     try:
         dummy = os.environ["RUNNING_IN_CONTAINER"]
@@ -579,25 +1135,80 @@ def version():
 
 
 # --------------------------- timing functions -------------------------------------------------
-def timefmt(thenumber):
+def timefmt(thenumber: float) -> str:
     """
+    Format a floating-point number as a string with fixed width and 2 decimal places.
 
     Parameters
     ----------
-    thenumber
+    thenumber : float
+        The numeric value to be formatted as a string.
 
     Returns
     -------
-    outputlines:
-        The formatted lines to save to the formatted runtimings file
-    totaldiff:
-        The total time from start to finish, in seconds
-
+    str
+        A string representation of the input number formatted to 2 decimal places
+        with a minimum width of 10 characters, right-aligned.
+
+    Notes
+    -----
+    The formatting uses "{:10.2f}".format() which ensures:
+    - Fixed width of 10 characters
+    - 2 decimal places
+    - Right alignment (default for numeric formats)
+
+    Examples
+    --------
+    >>> timefmt(123.456)
+    '   123.46'
+    >>> timefmt(1.234)
+    '     1.23'
+    >>> timefmt(0.0)
+    '     0.00'
     """
     return "{:10.2f}".format(thenumber)
 
 
-def proctiminglogfile(logfilename, timewidth=10):
+def proctiminglogfile(logfilename: str, timewidth: int = 10) -> tuple[list[str], float]:
+    """
+    Process a timing log file and return formatted timing information.
+
+    This function reads a timing log file, calculates cumulative and incremental
+    time differences from the start time, and formats the output into a list of
+    strings. If numerical data and units are present in the log, they are used
+    to compute and display processing speeds.
+
+    Parameters
+    ----------
+    logfilename : str
+        Path to the timing log file. The file should be a CSV with columns:
+        'time', 'description', 'number', 'units'.
+    timewidth : int, optional
+        Width for right-justifying time values in the output (default is 10).
+
+    Returns
+    -------
+    tuple[list[str], float]
+        A tuple containing:
+        - List of formatted timing lines as strings.
+        - Total elapsed time in seconds as a float.
+
+    Notes
+    -----
+    The log file is expected to follow the format:
+    `YYYYMMDDTHHMMSS.ffffff` for timestamps.
+    The function assumes the first row is the starting point for all time calculations.
+
+    Examples
+    --------
+    >>> lines, total_time = proctiminglogfile('timing.log', timewidth=12)
+    >>> for line in lines:
+    ...     print(line)
+    Total (s)    Diff. (s)    Description
+    0.00         0.00         Start process
+    5.20         5.20         Load data
+    12.40        7.20         Process data (1000 items @ 138.89 items/s)
+    """
     timingdata = pd.read_csv(
         logfilename,
         sep=None,
@@ -635,18 +1246,51 @@ def proctiminglogfile(logfilename, timewidth=10):
     return outputlines, totaldiff
 
 
-def proctiminginfo(thetimings, outputfile="", extraheader=None):
+def proctiminginfo(
+    thetimings: list[tuple[str, float, float | None, str | None]],
+    outputfile: str = "",
+    extraheader: str | None = None,
+) -> None:
     """
+    Process and display timing information for program execution.
+
+    This function takes a list of timing events and displays them in a formatted table
+    showing clock time, program time, duration, and event descriptions. Optional
+    output to file and additional header information can also be specified.
 
     Parameters
     ----------
-    thetimings
-    outputfile
-    extraheader
+    thetimings : list of tuple of (str, float, float | None, str | None)
+        List of timing events where each event is a tuple containing:
+        - Event description (str)
+        - Timestamp (float)
+        - Events per second (float or None)
+        - Unit of measurement (str or None)
+    outputfile : str, optional
+        Path to output file for writing timing information (default is "")
+    extraheader : str, optional
+        Additional header text to be printed before timing information (default is None)
 
     Returns
     -------
-
+    None
+        This function does not return any value but prints timing information to stdout
+        and optionally writes to a file.
+
+    Notes
+    -----
+    The function formats timestamps using YYYYMMDDTHHMMSS format and calculates
+    durations between consecutive events. If event rate information is provided,
+    it will be displayed in the format "(rate unit/second)".
+
+    Examples
+    --------
+    >>> timings = [
+    ...     ("Start", 1640995200.0, None, None),
+    ...     ("Process A", 1640995205.5, 100.0, "events"),
+    ...     ("End", 1640995210.0, None, None)
+    ... ]
+    >>> proctiminginfo(timings, "timing_output.txt", "Execution Timing Report")
     """
     theinfolist = []
     start = thetimings[0]
@@ -680,7 +1324,52 @@ def proctiminginfo(thetimings, outputfile="", extraheader=None):
 
 
 # timecourse functions
-def maketcfrom3col(inputdata, timeaxis, outputvector, debug=False):
+def maketcfrom3col(
+    inputdata: NDArray, timeaxis: NDArray, outputvector: NDArray, debug: bool = False
+) -> NDArray:
+    """
+    Create temporal output vector from 3-column input data.
+
+    This function processes input data containing start times, durations, and values,
+    and maps these to an output vector based on a time axis. Each input row defines
+    a time interval [start_time, start_time + duration] that is mapped to the output
+    vector by setting the corresponding elements to the specified value.
+
+    Parameters
+    ----------
+    inputdata : array-like
+        3-column input data where:
+        - First column: start times
+        - Second column: durations
+        - Third column: values to assign
+    timeaxis : array-like
+        Time axis defining the temporal resolution of the output vector
+    outputvector : array-like
+        Output vector to be populated with values from inputdata
+    debug : bool, optional
+        If True, displays a plot of the output vector (default is False)
+
+    Returns
+    -------
+    ndarray
+        The populated output vector with values assigned according to input intervals
+
+    Notes
+    -----
+    - Intervals are clipped to the bounds of the time axis
+    - Only intervals that overlap with the time axis (0 to max(timeaxis)) are processed
+    - The function modifies the outputvector in-place
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> timeaxis = np.linspace(0, 10, 11)
+    >>> inputdata = np.array([[1, 2, 5], [3, 1, 10]])
+    >>> outputvector = np.zeros(11)
+    >>> result = maketcfrom3col(inputdata, timeaxis, outputvector)
+    >>> print(result)
+    [0. 5. 5. 10. 10.  0.  0.  0.  0.  0.  0.]
+    """
     theshape = np.shape(inputdata)
     for idx in range(0, theshape[1]):
         starttime = inputdata[0, idx]
@@ -699,7 +1388,53 @@ def maketcfrom3col(inputdata, timeaxis, outputvector, debug=False):
     return outputvector
 
 
-def maketcfrom2col(inputdata, timeaxis, outputvector, debug=False):
+def maketcfrom2col(
+    inputdata: NDArray, timeaxis: NDArray, outputvector: NDArray, debug: bool = False
+) -> NDArray:
+    """
+    Create a temporal output vector from 2-column input data.
+
+    This function processes input data consisting of two columns where the first column
+    represents time ranges and the second column represents the value to be assigned to
+    corresponding time intervals in the output vector. The function iterates through
+    the input data and assigns values to contiguous ranges in the output vector.
+
+    Parameters
+    ----------
+    inputdata : NDArray
+        2D array with shape (2, n) where first row contains start/end time indices
+        and second row contains corresponding values to assign.
+    timeaxis : NDArray
+        1D array representing time values for plotting (used only in debug mode).
+    outputvector : NDArray
+        1D array to be populated with values from inputdata. This array is modified
+        in-place and returned.
+    debug : bool, optional
+        If True, enables debug output including range assignments and plots the
+        resulting output vector. Default is False.
+
+    Returns
+    -------
+    NDArray
+        The modified outputvector with values assigned from inputdata.
+
+    Notes
+    -----
+    The function processes inputdata by iterating through columns and assigning
+    values to ranges in outputvector. Each column represents a time interval
+    [start, end) where start is taken from inputdata[0, i-1] and end from
+    inputdata[0, i]. The value assigned is from inputdata[1, i-1].
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> inputdata = np.array([[0, 5, 10, 15], [1, 2, 3, 4]])
+    >>> timeaxis = np.arange(20)
+    >>> outputvector = np.zeros(20)
+    >>> result = maketcfrom2col(inputdata, timeaxis, outputvector, debug=False)
+    >>> print(result[:15])
+    [1. 1. 1. 1. 1. 2. 2. 2. 2. 2. 3. 3. 3. 3. 3.]
+    """
     theshape = np.shape(inputdata)
     rangestart = int(inputdata[0, 0])
     for i in range(1, theshape[1]):
@@ -721,7 +1456,62 @@ def maketcfrom2col(inputdata, timeaxis, outputvector, debug=False):
 
 
 # --------------------------- simulation functions ----------------------------------------------
-def makeslicetimes(numslices, sliceordertype, tr=1.0, multibandfac=1, debug=False):
+def makeslicetimes(
+    numslices: int,
+    sliceordertype: str,
+    tr: float = 1.0,
+    multibandfac: int = 1,
+    debug: bool = False,
+) -> NDArray | None:
+    """
+    Generate slice timing list for MRI data acquisition based on slice ordering type.
+
+    Parameters
+    ----------
+    numslices : int
+        Number of slices in the volume.
+    sliceordertype : str
+        Type of slice ordering. Valid options are:
+        - 'ascending': slices acquired in ascending order
+        - 'descending': slices acquired in descending order
+        - 'ascending_interleaved': interleaved ascending order
+        - 'descending_interleaved': interleaved descending order
+        - 'ascending_sparkplug': sparkplug ascending order
+        - 'descending_sparkplug': sparkplug descending order
+        - 'ascending_interleaved_siemens': Siemens-style interleaved ascending
+        - 'descending_interleaved_siemens': Siemens-style interleaved descending
+        - 'ascending_interleaved_philips': Philips-style interleaved ascending
+        - 'descending_interleaved_philips': Philips-style interleaved descending
+    tr : float, optional
+        Repetition time in seconds (default is 1.0).
+    multibandfac : int, optional
+        Multiband factor (default is 1).
+    debug : bool, optional
+        If True, print debug information (default is False).
+
+    Returns
+    -------
+    NDArray | None
+        Array of slice times in seconds. Returns None if an error occurs.
+
+    Notes
+    -----
+    The function computes slice acquisition times based on the specified slice order
+    and multiband factor. It supports various slice ordering strategies commonly used
+    in MRI pulse sequences.
+
+    Examples
+    --------
+    >>> makeslicetimes(32, 'ascending_interleaved', tr=2.0, multibandfac=2)
+    array([0.   , 0.25 , 0.5  , 0.75 , 1.   , 1.25 , 1.5  , 1.75 , 2.   ,
+           2.25 , 2.5  , 2.75 , 3.   , 3.25 , 3.5  , 3.75 , 4.   , 4.25 ,
+           4.5  , 4.75 , 5.   , 5.25 , 5.5  , 5.75 , 6.   , 6.25 , 6.5  ,
+           6.75 , 7.   , 7.25 , 7.5  , 7.75 ])
+
+    >>> makeslicetimes(16, 'descending_sparkplug', multibandfac=2)
+    array([0.   , 0.5  , 1.   , 1.5  , 2.   , 2.5  , 3.   , 3.5  , 4.   ,
+           4.5  , 5.   , 5.5  , 6.   , 6.5  , 7.   , 7.5  ])
+    """
     outlist = np.zeros((numslices), dtype=np.float)
     if (numslices % multibandfac) != 0:
         print("ERROR: numslices is not evenly divisible by multband factor")
@@ -795,7 +1585,64 @@ def makeslicetimes(numslices, sliceordertype, tr=1.0, multibandfac=1, debug=Fals
 
 
 # --------------------------- testing functions -------------------------------------------------
-def comparemap(map1, map2, mask=None, debug=False):
+def comparemap(
+    map1: NDArray, map2: NDArray, mask: NDArray | None = None, debug: bool = False
+) -> tuple[float, float, float, float, float, float, float, float]:
+    """
+    Compare two arrays (maps) and compute various difference statistics.
+
+    This function computes multiple metrics comparing two input arrays, `map1` and `map2`.
+    It supports optional masking to focus comparisons on specific regions of the arrays.
+    The function handles both 1D and multi-dimensional arrays, with support for different
+    mask dimensions (either matching the map dimensions or one less).
+
+    Parameters
+    ----------
+    map1 : NDArray
+        First input array to compare. Can be 1D or multi-dimensional.
+    map2 : NDArray
+        Second input array to compare. Must have the same shape as `map1`.
+    mask : NDArray, optional
+        A boolean or numeric mask to select valid voxels for comparison.
+        If provided, its shape must either match `map1` or be one dimension smaller.
+        If `None`, all voxels are compared.
+    debug : bool, optional
+        If True, print debug information during execution. Default is False.
+
+    Returns
+    -------
+    tuple of float
+        A tuple containing the following statistics in order:
+        - `mindiff`: Minimum absolute difference between `map1` and `map2`.
+        - `maxdiff`: Maximum absolute difference between `map1` and `map2`.
+        - `meandiff`: Mean absolute difference between `map1` and `map2`.
+        - `mse`: Mean squared error between `map1` and `map2`.
+        - `minreldiff`: Minimum relative difference (relative to `map1`).
+        - `maxreldiff`: Maximum relative difference (relative to `map1`).
+        - `meanreldiff`: Mean relative difference (relative to `map1`).
+        - `relmse`: Mean squared relative error between `map1` and `map2`.
+
+    Notes
+    -----
+    - If `map1` contains zero values, relative differences are set to 0 to avoid division by zero.
+    - When `mask` is provided and has one fewer dimension than `map1`, it is reshaped to match
+      the first dimension of `map1` before comparison.
+    - The function exits with an error if shapes are incompatible or if masks are not valid.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> map1 = np.array([1.0, 2.0, 3.0])
+    >>> map2 = np.array([1.1, 2.2, 2.9])
+    >>> result = comparemap(map1, map2)
+    >>> print(result)
+    (-0.1, 0.1, 0.0, 0.006666666666666667, -0.1, 0.1, 0.0, 0.006666666666666667)
+
+    >>> mask = np.array([1, 1, 0])
+    >>> result = comparemap(map1, map2, mask=mask)
+    >>> print(result)
+    (-0.1, 0.1, 0.0, 0.006666666666666667, -0.1, 0.1, 0.0, 0.006666666666666667)
+    """
     ndims = len(map1.shape)
     if debug:
         print("map has", ndims, "axes")
@@ -861,7 +1708,54 @@ def comparemap(map1, map2, mask=None, debug=False):
     return mindiff, maxdiff, meandiff, mse, minreldiff, maxreldiff, meanreldiff, relmse
 
 
-def comparerapidtideruns(root1, root2, debug=False):
+def comparerapidtideruns(root1: str, root2: str, debug: bool = False) -> dict[str, Any]:
+    """
+    Compare results from two rapidtide runs by evaluating corresponding maps and timecourses.
+
+    This function compares NIfTI maps and text-based timecourses from two different rapid tide
+    processing runs. It evaluates differences between the corresponding files using various
+    statistical measures such as mean difference, max difference, mean squared error, and
+    relative versions of these metrics.
+
+    Parameters
+    ----------
+    root1 : str
+        The base filename (without extension) for the first rapid tide run.
+    root2 : str
+        The base filename (without extension) for the second rapid tide run.
+    debug : bool, optional
+        If True, print detailed debug information during execution. Default is False.
+
+    Returns
+    -------
+    dict[str, Any]
+        A dictionary containing comparison results for each map and timecourse.
+        Keys are map or timecourse names, and values are dictionaries with the following keys:
+        - 'mindiff': minimum difference
+        - 'maxdiff': maximum difference
+        - 'meandiff': mean difference
+        - 'mse': mean squared error
+        - 'relmindiff': relative minimum difference
+        - 'relmaxdiff': relative maximum difference
+        - 'relmeandiff': relative mean difference
+        - 'relmse': relative mean squared error
+
+    Notes
+    -----
+    - The function assumes that both runs have corresponding mask files named
+      ``<root>_desc-corrfit_mask.nii.gz``.
+    - For each map, the function checks if the corresponding NIfTI files exist and match
+      in spatial dimensions.
+    - For each timecourse, the function reads from JSON files and compares the time series
+      only if both files are present and have matching lengths.
+    - If spatial or temporal dimensions do not match, the function exits with an error.
+
+    Examples
+    --------
+    >>> results = comparerapidtideruns("run1", "run2", debug=True)
+    >>> print(results["maxtime"])
+    {'mindiff': -0.01, 'maxdiff': 0.02, 'meandiff': 0.005, 'mse': 0.0001, ...}
+    """
     results = {}
     maskname1 = f"{root1}_desc-corrfit_mask.nii.gz"
     (
@@ -989,7 +1883,50 @@ def comparerapidtideruns(root1, root2, debug=False):
     return results
 
 
-def comparehappyruns(root1, root2, debug=False):
+def comparehappyruns(root1: str, root2: str, debug: bool = False) -> dict[str, Any]:
+    """
+    Compare results from two happy runs by comparing output maps and timecourses.
+
+    This function compares neuroimaging maps (e.g., app_info, vessels_mask) and
+    cardiac timecourses (e.g., cardfromfmri_25.0Hz.txt) between two datasets
+    identified by their root names. It performs spatial alignment checks and
+    computes various statistical differences between corresponding files.
+
+    Parameters
+    ----------
+    root1 : str
+        Root name of the first dataset (e.g., 'subject01_run1').
+    root2 : str
+        Root name of the second dataset (e.g., 'subject01_run2').
+    debug : bool, optional
+        If True, print debug information during execution. Default is False.
+
+    Returns
+    -------
+    dict[str, Any]
+        A dictionary containing comparison results for each processed map and
+        timecourse. Each entry includes:
+        - mindiff: minimum absolute difference
+        - maxdiff: maximum absolute difference
+        - meandiff: mean absolute difference
+        - mse: mean squared error
+        - relmindiff: relative minimum difference
+        - relmaxdiff: relative maximum difference
+        - relmeandiff: relative mean difference
+        - relmse: relative mean squared error
+
+    Notes
+    -----
+    - The function expects specific file naming conventions for both maps and
+      timecourses.
+    - Spatial dimensions of masks and data must match for comparison to proceed.
+    - If any file is missing or mismatched, the function will exit with an error.
+
+    Examples
+    --------
+    >>> results = comparehappyruns('subject01_run1', 'subject01_run2', debug=True)
+    >>> print(results['app_info']['meandiff'])
+    """
     results = {}
     if debug:
         print("comparehappyruns rootnames:", root1, root2)
@@ -1084,7 +2021,53 @@ def comparehappyruns(root1, root2, debug=False):
 
 
 # shared memory routines
-def numpy2shared(inarray, theouttype, name=None):
+def numpy2shared(
+    inarray: NDArray, theouttype: type, name: str | None = None
+) -> tuple[NDArray, shared_memory.SharedMemory]:
+    """
+    Convert a numpy array to a shared memory array.
+
+    This function creates a shared memory block and copies the data from the input
+    numpy array to the shared memory array. The returned array and shared memory
+    object can be used for inter-process communication or memory sharing.
+
+    Parameters
+    ----------
+    inarray : numpy.ndarray
+        Input numpy array to be converted to shared memory.
+    theouttype : type
+        Data type of the output shared memory array.
+    name : str, optional
+        Name of the shared memory block. If None, an anonymous shared memory
+        block is created.
+
+    Returns
+    -------
+    tuple[numpy.ndarray, multiprocessing.shared_memory.SharedMemory]
+        A tuple containing:
+        - The shared memory array with the same shape as input array
+        - The shared memory object that manages the memory block
+
+    Notes
+    -----
+    The returned shared memory object must be explicitly closed and unlink
+    when no longer needed to free system resources. The shared memory block
+    will be automatically unlinked upon creation but can be accessed by other
+    processes using the same name.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> from multiprocessing import shared_memory
+    >>> arr = np.array([1, 2, 3, 4, 5])
+    >>> shared_arr, shm = numpy2shared(arr, np.int32)
+    >>> print(shared_arr)
+    [1 2 3 4 5]
+    >>> # Clean up when done
+    >>> shared_arr.close()
+    >>> shm.close()
+    >>> shm.unlink()
+    """
     # Create a shared memory block to store the array data
     outnbytes = np.dtype(theouttype).itemsize * inarray.size
     shm = shared_memory.SharedMemory(name=None, create=True, size=outnbytes)
@@ -1094,7 +2077,54 @@ def numpy2shared(inarray, theouttype, name=None):
     return inarray_shared, shm  # Return both the array and the shared memory object
 
 
-def allocshared(theshape, thetype, name=None):
+def allocshared(
+    theshape: tuple[int, ...], thetype: type, name: str | None = None
+) -> tuple[NDArray, shared_memory.SharedMemory]:
+    """
+    Allocate shared memory for a numpy array.
+
+    This function creates a shared memory block and returns both the numpy array
+    backed by this shared memory and the shared memory object itself. The array
+    can be accessed from different processes, making it useful for inter-process
+    communication.
+
+    Parameters
+    ----------
+    theshape : tuple of int
+        The shape of the array to be created. Must be a tuple of integers.
+    thetype : type
+        The data type of the array elements. Can be any numpy-compatible dtype.
+    name : str, optional
+        Name of existing shared memory object. If None, a new shared memory
+        block is created. Default is None.
+
+    Returns
+    -------
+    tuple[NDArray, shared_memory.SharedMemory]
+        A tuple containing:
+        - The numpy array backed by shared memory
+        - The shared_memory.SharedMemory object
+
+    Notes
+    -----
+    The returned shared memory object should be explicitly closed and unlink
+    when no longer needed to free system resources. The array can be accessed
+    from multiple processes, but care should be taken to avoid race conditions.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> from multiprocessing import shared_memory
+    >>> arr, shm = allocshared((3, 4), np.float64)
+    >>> arr[0, 0] = 42.0
+    >>> print(arr)
+    [[42.  0.  0.  0.]
+     [ 0.  0.  0.  0.]
+     [ 0.  0.  0.  0.]]
+    >>> # Don't forget to clean up
+    >>> shm.close()
+    >>> shm.unlink()
+    """
     # Calculate size based on shape
     thesize = np.prod(theshape)
     # Determine the data type size
@@ -1106,14 +2136,83 @@ def allocshared(theshape, thetype, name=None):
     return outarray, shm  # Return both the array and the shared memory object
 
 
-def allocarray(theshape, thetype, shared=False, name=None):
+def allocarray(
+    theshape: tuple[int, ...], thetype: type, shared: bool = False, name: str | None = None
+) -> tuple[NDArray, shared_memory.SharedMemory | None]:
+    """
+    Allocate and return a numpy array with specified shape and type.
+
+    Parameters
+    ----------
+    theshape : tuple[int, ...]
+        Shape of the array to be allocated.
+    thetype : type
+        Data type of the array elements.
+    shared : bool, optional
+        If True, allocate the array in shared memory. Default is False.
+    name : str | None, optional
+        Name for the shared memory segment. Required if shared=True. Default is None.
+
+    Returns
+    -------
+    tuple[NDArray, shared_memory.SharedMemory | None]
+        A tuple containing:
+        - The allocated numpy array filled with zeros
+        - The shared memory object if shared=True, otherwise None
+
+    Notes
+    -----
+    When ``shared=True``, the function delegates to ``allocshared`` to create
+    a shared memory array. Otherwise, it creates a regular numpy array using
+    ``np.zeros`` with the specified shape and dtype.
+
+    Examples
+    --------
+    >>> import numpy as np
+    >>> arr, shm = allocarray((3, 4), np.float64)
+    >>> print(arr.shape)
+    (3, 4)
+    >>> print(arr.dtype)
+    float64
+
+    >>> # For shared memory allocation
+    >>> arr, shm = allocarray((2, 3), np.int32, shared=True, name="my_array")
+    >>> print(shm is not None)
+    True
+    """
     if shared:
         return allocshared(theshape, thetype, name)
     else:
         return np.zeros(theshape, dtype=thetype), None
 
 
-def cleanup_shm(shm):
+def cleanup_shm(shm: shared_memory.SharedMemory | None) -> None:
+    """
+    Clean up shared memory object.
+
+    Parameters
+    ----------
+    shm : shared_memory.SharedMemory or None
+        Shared memory object to clean up. If None, no action is taken.
+
+    Returns
+    -------
+    None
+        This function does not return any value.
+
+    Notes
+    -----
+    This function is designed to properly release shared memory resources.
+    It should be called to clean up shared memory objects to prevent resource leaks.
+    If the shared memory object is None, the function performs no operation.
+
+    Examples
+    --------
+    >>> from multiprocessing import shared_memory
+    >>> shm = shared_memory.SharedMemory(create=True, size=1024)
+    >>> cleanup_shm(shm)
+    >>> # Shared memory is now cleaned up
+    """
     # Cleanup
     pass
     # if shm is not None: