Data analysis

grid_catalogue(x, y, z, w=None, xlim=None, ylim=None, zlim=None, nx=None, ny=None, nz=None)

Bin a catalogue of 3D positions onto a regular grid.

Parameters:

Name	Type	Description	Default
x,	y, z (array_like	3D positions of objects.	required
w	array_like	Weight assigned to each object. Can be None.	None
xlim,	ylim, zlim (tuple	Tuples of the minimum and maximum coordinate values of the grid in each dimension, e.g. (xmin, xmax). If not specified, the ranges of the data will be used.	required
nx,	ny, nz (int	Number of grid cells in each direction.	required

Source code in fastbox/analysis.py

def grid_catalogue(
    x, y, z, w=None, xlim=None, ylim=None, zlim=None, nx=None, ny=None, nz=None
):
    """
    Bin a catalogue of 3D positions onto a regular grid.

    Parameters:
        x, y, z (array_like):
            3D positions of objects.
        w (array_like):
            Weight assigned to each object. Can be `None`.
        xlim, ylim, zlim (tuple):
            Tuples of the minimum and maximum coordinate values of the grid in
            each dimension, e.g. `(xmin, xmax)`. If not specified, the ranges
            of the data will be used.
        nx, ny, nz (int):
            Number of grid cells in each direction.
    """
    assert (
        (nx is not None) and (ny is not None) and (nz is not None)
    ), "nx, ny, and nz must be specified."

    # Get ranges
    if xlim is None:
        xlim = (np.min(x), np.max(x))
    if ylim is None:
        ylim = (np.min(y), np.max(y))
    if zlim is None:
        zlim = (np.min(z), np.max(z))
    xmin, xmax = xlim
    ymin, ymax = ylim
    zmin, zmax = zlim

    # Define grid
    xgrid = np.linspace(xmin, xmax, nx)  # bin centres
    ygrid = np.linspace(ymin, ymax, ny)
    zgrid = np.linspace(zmin, zmax, nz)

    # Grid catalogue
    grid, _ = np.histogramdd(
        np.vstack([x, y, z]).T,
        bins=(nx, ny, nz),
        range=[(xmin, xmax), (ymin, ymax), (zmin, zmax)],
        weights=w,
    )
    return grid, (xgrid, ygrid, zgrid)

interpolate_onto_grid(field, coords_orig, coords_new)

Interpolate a 3D field onto a grid with a different pixel/channel spacing.

NOTE: The input set of coordinates should be in ascending order.

Parameters:

Name	Type	Description	Default
field	array_like	3D array of data.	required
coords_orig	tuple	Tuple of coordinates of the original grid, in the order (x, y, z), corresponding to field axes (0, 1, 2).	required
coords_new	tuple	Tuple of coordinates of the new grid, in the order (x, y, z), corresponding to field axes (0, 1, 2).	required

Returns:

Type	Description
field_new (array_like)	3D array of data interpolated onto new grid shape.

Source code in fastbox/analysis.py

def interpolate_onto_grid(field, coords_orig, coords_new):
    """
    Interpolate a 3D field onto a grid with a different pixel/channel spacing.

    NOTE: The input set of coordinates should be in ascending order.

    Parameters:
        field (array_like):
            3D array of data.
        coords_orig (tuple):
            Tuple of coordinates of the original grid, in the order (x, y, z),
            corresponding to field axes (0, 1, 2).
        coords_new (tuple):
            Tuple of coordinates of the new grid, in the order (x, y, z),
            corresponding to field axes (0, 1, 2).

    Returns:
        field_new (array_like):
            3D array of data interpolated onto new grid shape.
    """
    # Extract original and new coordinates
    x, y, z = coords_orig
    x_new, y_new, z_new = coords_new

    # Replace NaN values in data
    field_nonan = replace_nan_with_channel_mean(field)

    # Construct interpolator for data
    # (Each array in coords_orig has to be ascending order)
    interp = scipy.interpolate.RegularGridInterpolator(
        (x, y, z), field_nonan, method="linear", bounds_error=False, fill_value=np.nan
    )

    # Build new grid with chosen resolution
    x3d, y3d, z3d = np.meshgrid(x_new, y_new, z_new)
    pts = np.array([x3d.flatten(), y3d.flatten(), z3d.flatten()])

    # Interpolate onto new fine grid
    field_new = interp(pts.T).reshape(x3d.shape)
    return field_new

replace_nan_with_channel_mean(field)

Replace NaN values in a datacube with the mean of the non-NaN values in each channel (frequency slice).

Parameters:

Name	Type	Description	Default
field	array_like	3D datacube. The -1 axis will be assumed to be the frequency direction.	required

Returns:

Type	Description
field_replaced (array_like)	3D datacube with NaN values replaced.

Source code in fastbox/analysis.py

def replace_nan_with_channel_mean(field):
    """
    Replace NaN values in a datacube with the mean of the non-NaN values in
    each channel (frequency slice).

    Parameters:
        field (array_like):
            3D datacube. The -1 axis will be assumed to be the frequency
            direction.

    Returns:
        field_replaced (array_like):
            3D datacube with NaN values replaced.
    """
    # Work on a copy
    field_repl = field.copy().reshape((-1, field.shape[-1]))

    # Loop over frequency channels
    for j in range(field_repl.shape[-1]):
        idxs = np.isnan(field_repl[:, j])  # idxs of NaN values
        avg = np.mean(field_repl[~idxs, j])

        field_repl[idxs, j] = avg
    return field_repl.reshape(field.shape)