synthesizer.emissions.scaling

Shared helpers for scaling emission arrays.

Provides normalised mask handling and a unified scaling dispatcher that hands the common 2D scaling cases over to the specialised C++ kernels, while retaining a NumPy fallback for the more general broadcast shapes. Both Sed.scale and LineCollection.scale delegate here for the single-array path.

Example usage:

from synthesizer.emissions.scaling import (
    normalise_scale_masks,
    scale_array,
)

scaled = scale_array(array, scaling, mask=mask, nthreads=4)

Functions

synthesizer.emissions.scaling.normalise_line_scaling(scaling, get_nu, lum_units, cont_units)[source]

Resolve a line scaling into luminosity and continuum factors.

Line luminosity and continuum carry different units (erg/s and erg/s/Hz respectively), so a single unit-bearing scaling may need to be converted into two different raw arrays. The frequency coordinate is only constructed when we actually need one of those unit conversions.

Parameters:

  • scaling (float, np.ndarray, or unyt quantity) – Scaling factor supplied by the caller.

  • get_nu (Callable[[], unyt_array]) – Callable returning the frequency coordinate for the lines.

  • lum_units (unyt.Unit) – Units of the luminosity array.

  • cont_units (unyt.Unit) – Units of the continuum array.

Returns:

(scaling_lum, scaling_cont) as raw values.

Return type:

tuple

synthesizer.emissions.scaling.normalise_scale_masks(mask, lam_mask, shape)[source]

Normalise row and wavelength masks for scaling helpers.

Parameters:

  • mask (np.ndarray or None) – Optional mask supplied by the caller.

  • lam_mask (np.ndarray or None) – Optional explicit wavelength mask.

  • shape (tuple) – Target array shape.

Returns:

(mask, lam_mask) in a form the scaling helper understands.

Return type:

tuple

synthesizer.emissions.scaling.normalise_scaling_for_units(scaling, units)[source]

Convert a scaling factor into raw values compatible with units.

Parameters:

  • scaling (float, np.ndarray, or unyt quantity) – Scaling factor supplied by the caller.

  • units (unyt.Unit) – Units expected by the target array.

Returns:

Raw scaling values with any units stripped.

Return type:

float or np.ndarray

synthesizer.emissions.scaling.scale_array(array, scaling, mask=None, lam_mask=None, nthreads=1, out=None)[source]

Scale an emission array with optional row and wavelength masks.

Parameters:

  • array (np.ndarray) – The array to scale.

  • scaling (float or np.ndarray) – The scaling to apply.

  • mask (np.ndarray or None) – Optional row or element mask.

  • lam_mask (np.ndarray or None) – Optional wavelength mask.

  • nthreads (int) – The number of OpenMP threads available to compatible kernels.

  • out (np.ndarray or None) – Optional output buffer. When provided and compatible with the fast 2D path the result is written directly into this array, avoiding an allocation.

Returns:

The scaled array (may be out if the fast path was used).

Return type:

np.ndarray

synthesizer.emissions.scaling.scale_inplace(array, scaling, mask=None, lam_mask=None, nthreads=1)[source]

Scale an array into its existing buffer.

Parameters:

  • array (np.ndarray) – Raw array to mutate.

  • scaling (float or np.ndarray) – Raw scaling values.

  • mask (np.ndarray or None) – Optional row or element mask.

  • lam_mask (np.ndarray or None) – Optional wavelength mask.

  • nthreads (int) – The number of OpenMP threads available to compatible kernels.

Returns:

The mutated array buffer.

Return type:

np.ndarray

synthesizer.emissions.scaling.scale_line_arrays(luminosity, continuum, scaling_lum, scaling_cont, mask=None, lam_mask=None, nthreads=1, out_lum=None, out_cont=None)[source]

Scale a luminosity/continuum pair, using the fused kernel when useful.

This keeps the Python-side dispatch rules for line scaling in one place. When both scaling arrays are 1D row factors we hand straight over to the fused C++ kernel; otherwise we delegate to scale_array for each array.

Parameters:

  • luminosity (np.ndarray) – Raw luminosity array.

  • continuum (np.ndarray) – Raw continuum array.

  • scaling_lum (float or np.ndarray) – Raw luminosity scaling values.

  • scaling_cont (float or np.ndarray) – Raw continuum scaling values.

  • mask (np.ndarray or None) – Optional row or element mask.

  • lam_mask (np.ndarray or None) – Optional wavelength mask.

  • nthreads (int) – The number of OpenMP threads available to compatible kernels.

  • out_lum (np.ndarray or None) – Optional output buffer for luminosity.

  • out_cont (np.ndarray or None) – Optional output buffer for continuum.

Returns:

Tuple of scaled (luminosity, continuum) arrays.

Return type:

tuple

synthesizer.emissions.scaling.scale_to_quantity(array, scaling, units, mask=None, lam_mask=None, nthreads=1)[source]

Scale an array and wrap the result in units without copying again.

Parameters:

  • array (np.ndarray) – Raw array to scale.

  • scaling (float or np.ndarray) – Raw scaling values.

  • units (unyt.Unit) – Units to attach to the scaled result.

  • mask (np.ndarray or None) – Optional row or element mask.

  • lam_mask (np.ndarray or None) – Optional wavelength mask.

  • nthreads (int) – The number of OpenMP threads available to compatible kernels.

Returns:

Scaled result wrapped in units.

Return type:

unyt_array