Simulation Dataset for BBHnet

NOTE: OLD README, UPDATE IN PROGRESS

We generate simulation dataset to train BBHnet, our deep learning framework for detection of compact binary coalescene (CBC) gravitational-wave (GW) signals .

Example

To generate a noise dataset, simply run generateRealNoise.py:

python generateRealNoise.py -t0 1186729980 -t1 1186734086 -t0-psd 1186729980 -t1-psd 1186734086
    -fs 1024 -fmin 20 -o test_noise.h5

To also add CBC signals, enable the flag -S and add the prior distribution file in Bilby format with -p

python generateRealNoise.py -t0 1186729980 -t1 1186734086 -t0-psd 1186729980 -t1-psd 1186734086
    -fs 1024 -fmin 20 -S -p config/priors/nonspin_BBH.prior -o test_signal.h5

A full list of generateRealNoise.py arguments can be found below:

usage: generateRealNoise.py [-h] -t0 FRAME_START -t1 FRAME_STOP -t0-psd FRAME_START_PSD -t1-psd FRAME_STOP_PSD -o OUTFILE [-S]
                            [-fs SAMPLE_RATE] [-fmin HIGH_PASS] [-T SAMPLE_DURATION] [-dt TIME_STEP] [-p PRIOR_FILE]
                            [--correlation-shift CORRELATION_SHIFT] [--min-trigger MIN_TRIGGER] [--max-trigger MAX_TRIGGER]
                            [-s SEED]

optional arguments:
  -h, --help            show this help message and exit
  -t0 FRAME_START, --frame-start FRAME_START
                        starting GPS time of strain
  -t1 FRAME_STOP, --frame-stop FRAME_STOP
                        stopping GPS time of strain
  -t0-psd FRAME_START_PSD, --frame-start-psd FRAME_START_PSD
                        starting GPS time of strain for PSD estimation
  -t1-psd FRAME_STOP_PSD, --frame-stop-psd FRAME_STOP_PSD
                        stopping GPS time of strain for PSD estimation
  -o OUTFILE, --outfile OUTFILE
                        path to write output file in HDF5 format
  -S, --signal          Enable to add GW signal on top of background noise
  -fs SAMPLE_RATE, --sample-rate SAMPLE_RATE
                        sampling rate of strain
  -fmin HIGH_PASS, --high-pass HIGH_PASS
                        frequency of highpass filter
  -T SAMPLE_DURATION, --sample-duration SAMPLE_DURATION
                        duration in seconds of each sample
  -dt TIME_STEP, --time-step TIME_STEP
                        time step size in seconds between consecutive samples
  -p PRIOR_FILE, --prior-file PRIOR_FILE
                        path to prior config file. Required for signal simulation
  --correlation-shift CORRELATION_SHIFT
                        if given, also compute the correlation with given shift value
  --min-trigger MIN_TRIGGER
                        mininum trigger time w.r.t to sample. must be within [0, sample_duration]
  --max-trigger MAX_TRIGGER
                        maximum trigger time w.r.t to sample. must be within [0, sample_duration]
  -s SEED, --seed SEED  random seed for reproducibility

Simulation code and tutorial for BBHnet training data

Related tags

Overview

Simulation Dataset for BBHnet

NOTE: OLD README, UPDATE IN PROGRESS

Example

Owner

A Fast and Stable GAN for Small and High Resolution Imagesets - pytorch

Voice Gender Recognition

Out-of-Distribution Generalization of Chest X-ray Using Risk Extrapolation

Self-training with Weak Supervision (NAACL 2021)

Multi Task Vision and Language

Towards Understanding Quality Challenges of the Federated Learning: A First Look from the Lens of Robustness

Pytorch port of Google Research's LEAF Audio paper

🐤 Nix-TTS: An Incredibly Lightweight End-to-End Text-to-Speech Model via Non End-to-End Distillation

[ICML 2021] “ Self-Damaging Contrastive Learning”, Ziyu Jiang, Tianlong Chen, Bobak Mortazavi, Zhangyang Wang

An implementation of shampoo

High dimensional black-box optimizer using Latent Action Monte Carlo Tree Search algorithm

Official implementation of NeurIPS 2021 paper "One Loss for All: Deep Hashing with a Single Cosine Similarity based Learning Objective"

RETRO-pytorch - Implementation of RETRO, Deepmind's Retrieval based Attention net, in Pytorch

Symbolic Parallel Adaptive Importance Sampling for Probabilistic Program Analysis in JAX

DIR-GNN - Discovering Invariant Rationales for Graph Neural Networks

Code for a seq2seq architecture with Bahdanau attention designed to map stereotactic EEG data from human brains to spectrograms, using the PyTorch Lightning.

Script that receives an Image (original) and a set of images to be used as "pixels" in reconstruction of the Original image using the set of images as "pixels"

Implementation of the 😇 Attention layer from the paper, Scaling Local Self-Attention For Parameter Efficient Visual Backbones

Website for D2C paper

Official PyTorch implementation of "Physics-aware Difference Graph Networks for Sparsely-Observed Dynamics".