sage.data.waveform.approximants.phenom

Filename : helper.py Description : Short description of the file

Created on 2026-01-23 03:24:33

__author__ = Narenraju Nagarajan __copyright__ = Copyright 2026, ProjectName __license__ = MIT Licence __version__ = 0.0.1 __maintainer__ = Narenraju Nagarajan __affiliation__ = N/A __email__ = N/A __status__ = [‘inProgress’, ‘Archived’, ‘inUsage’, ‘Debugging’]

GitHub Repository: NULL

Documentation: NULL

Classes

PhenomConstants

Device-resident pre-allocated constants for IMRPhenom waveform generation.

Module Contents

class PhenomConstants(device='cuda', batch_size=None, dtype=None, **kwargs)[source]

Device-resident pre-allocated constants for IMRPhenom waveform generation.

Stores all scalar constants, fractions, QNM interpolation tables, and PhenomD coefficient tables as torch.Tensor objects on the target device. This avoids creating tensors inside the hot-path iteration loop, which would break torch.compile graph capture.

QNM (quasi-normal mode) ringdown frequency and damping time tables are pre-interpolated onto a fine 500 000-point grid via torch_scipylike_cubic_interp() so that ringdown frequency lookups can be done with a simple linear-slope computation at runtime.

Parameters:

device (str or torch.device) – Target device for all tensors (default "cuda").
batch_size (int or None) – Batch size used to pre-allocate ONES and ZEROS tensors.
dtype (torch.dtype or None) – Floating-point precision for all tensors.
**kwargs – Ignored; accepted for forward-compatibility.