Inferring dark matter substructure with astrometric lensing

Siddharth Mishra-Sharma

Abstract

Astrometry—the precise measurement of positions and motions of celestial objects—has emerged as a promising avenue for characterizing the dark matter population in our Galaxy. By leveraging recent advances in simulation-based inference and neural network architectures, we introduce a novel method to search for global dark matter-induced gravitational lensing signatures in astrometric datasets. Our method based on neural likelihood-ratio estimation shows significantly enhanced sensitivity to a cold dark matter population and more favorable scaling with measurement noise compared to existing approaches based on two-point correlation statistics, establishing machine learning as a powerful tool for characterizing dark matter using astrometric data.

Code

Note: This code uses a custom version of PyGSP, which can be installed as follows:

git clone https://github.com/smsharma/pygsp.git -b sphere-graphs
cd pygsp
python setup.py install

• simulate.py produces full-sky astrometric maps for training. In the scripts folder, sbatch --array=0-999 simulate.sh parallelizes sample generation in a SLURM HPC environment.
• combine_samples.py combines the generated samples into single files in order to use them for training. scripts/combine_samples.sh submits this as a SLURM job.
• train.py trains the likelihood-ratio estimator. Experiments are managed using `MLflow'. scripts/submit_train.py can loop over a grid of analysis configurations and submit a SLURM script for each; see options in train.py.

This notebooks analyzes the results and produces the plots in the paper. There, trained neural networks are loaded using their saved MLflow IDs.

Citation

@article{Mishra-Sharma:2021xyz,
      author         = "Mishra-Sharma, Siddharth",
      title          = "{Inferring dark matter substructure with astrometric lensing beyond the power spectrum}",
      year           = "2021",
      eprint         = "2110.01620",
      archivePrefix  = "arXiv",
      primaryClass   = "astro-ph.CO",
      SLACcitation   = "%%CITATION = ARXIV:2110.01620;%%"
}

The repository contains

• Code that is part of sbi for inference,
• Code associated with 2003.02264 for forward modeling,
• Code associated with 1909.02005 for scripting and data processing, and
• Code associated with 2012.15000 for constructing the feature extractor network.