FastEMRIWaveforms

Overview

FastEMRIWaveforms (FEW) generates fully relativistic extreme mass-ratio inspiral (EMRI) waveforms quickly enough for LISA data analysis — seconds on a CPU, and faster still on a GPU. The framework is modular: a trajectory module evolves the inspiral, an amplitude module supplies the mode amplitudes, and a summation module assembles the waveform. Modules can be swapped to trade accuracy against speed.

Installation

The CPU build installs from PyPI:

pip install fastemriwaveforms

GPU acceleration requires a build matched to your CUDA toolkit; see the full documentation for the right package and the supported model names.

Usage

Generate a waveform from a high-level model. Parameter names and conventions follow the FEW documentation:

from few.waveform import GenerateEMRIWaveform

wave = GenerateEMRIWaveform("FastSchwarzschildEccentricFlux")

h = wave(
    M=1e6,      # primary mass (solar masses)
    mu=1e1,     # secondary mass (solar masses)
    a=0.0,      # primary spin
    p0=12.0,    # initial semi-latus rectum
    e0=0.3,     # initial eccentricity
    x0=1.0,     # cosine of the inclination
    dist=1.0,   # luminosity distance (Gpc)
    dt=10.0,    # time step (s)
    T=1.0,      # observation time (years)
)

h is the complex strain time series; take its real and imaginary parts for the two polarisations.

Examples

The repository ships tutorial notebooks covering trajectory generation, individual modules, and GPU usage. Start from the documentation.

Citing FastEMRIWaveforms

If you use FastEMRIWaveforms in your research, please acknowledge the Toolkit:

FastEMRIWaveforms also requests the following citations:

• Katz et al., FastEMRIWaveforms: New tools for millihertz gravitational-wave data analysis, arXiv:2104.04582
• Chua et al., Rapid generation of fully relativistic extreme-mass-ratio-inspiral waveform templates for LISA data analysis, arXiv:2008.06071

See how to cite for the BibTeX entry and guidance.