mlgw-bns 0.11.0a0

Accelerating gravitational wave template generation with machine learning.

Machine Learning for Gravitational Waves from Binary Neutron Star mergers

This package's purpose is to speed up the generation of template gravitational waveforms for binary neutron star mergers by training a machine learning model on a dataset of waveforms generated with some physically-motivated surrogate.

It is able to reconstruct them with mismatches lower than 1/10000, with as little as 1000 training waveforms; the accuracy then steadily improves as more training waveforms are used.

Currently, the only model used for training is TEOBResumS, but it is planned to introduce the possibility to use others.

The documentation can be found here.

Installation

To install the package, use

pip install mlgw-bns

For more details see the documentation.

Changelog

Changes across versions are documented since version 0.10.1 in the CHANGELOG.

Inner workings

The main steps taken by mlgw_bns to train on a dataset are as follows:

• generate the dataset, consisting of EOB waveforms
• decompose the Fourier transforms of the waveforms into phase and amplitude
• downsample the dataset to a few thousand points
• compute the residuals of the EOB waveforms from PN ones
• apply a PCA to reduce the dimensionality to a few tens of real numbers
• train a neural network on the relation between the waveform parameters and the PCA components

After this, the model can reconstruct a waveform within its parameter space, resampled at arbitrary points in frequency space.

In several of the training steps data-driven optimizations are performed:

• the points at which the waveforms are downsampled are not uniformly chosen: instead, a greedy downsampling algorithm determines them
• the hyperparameters for the neural network are optimized, according to both the time taken for the training and the estimated reconstruction error, also varying the number of training waveforms available.