Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural Networks
We present a Bayesian graph neural network (BGNN) that can estimate the weak lensing convergence ( κ ) from photometric measurements of galaxies along a given line of sight (LOS). The method is of particular interest in strong gravitational time-delay cosmography (TDC), where characterizing the “ext...
| Published in: | The Astrophysical Journal |
|---|---|
| Main Authors: | , , , , , , , , |
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2023-01-01
|
| Subjects: | |
| Online Access: | https://doi.org/10.3847/1538-4357/acdc25 |
| _version_ | 1850309694579539968 |
|---|---|
| author | Ji Won Park Simon Birrer Madison Ueland Miles Cranmer Adriano Agnello Sebastian Wagner-Carena Philip J. Marshall Aaron Roodman the LSST Dark Energy Science Collaboration |
| author_facet | Ji Won Park Simon Birrer Madison Ueland Miles Cranmer Adriano Agnello Sebastian Wagner-Carena Philip J. Marshall Aaron Roodman the LSST Dark Energy Science Collaboration |
| author_sort | Ji Won Park |
| collection | DOAJ |
| container_title | The Astrophysical Journal |
| description | We present a Bayesian graph neural network (BGNN) that can estimate the weak lensing convergence ( κ ) from photometric measurements of galaxies along a given line of sight (LOS). The method is of particular interest in strong gravitational time-delay cosmography (TDC), where characterizing the “external convergence” ( κ _ext ) from the lens environment and LOS is necessary for precise Hubble constant ( H _0 ) inference. Starting from a large-scale simulation with a κ resolution of ∼1′, we introduce fluctuations on galaxy–galaxy lensing scales of ∼1″ and extract random sight lines to train our BGNN. We then evaluate the model on test sets with varying degrees of overlap with the training distribution. For each test set of 1000 sight lines, the BGNN infers the individual κ posteriors, which we combine in a hierarchical Bayesian model to yield constraints on the hyperparameters governing the population. For a test field well sampled by the training set, the BGNN recovers the population mean of κ precisely and without bias (within the 2 σ credible interval), resulting in a contribution to the H _0 error budget well under 1%. In the tails of the training set with sparse samples, the BGNN, which can ingest all available information about each sight line, extracts a stronger κ signal compared to a simplified version of the traditional method based on matching galaxy number counts, which is limited by sample variance. Our hierarchical inference pipeline using BGNNs promises to improve the κ _ext characterization for precision TDC. The code is available as a public Python package, Node to Joy https://github.com/jiwoncpark/node-to-joy . |
| format | Article |
| id | doaj-art-e632574e7a1e4e4aa9bbccb5d0a63de6 |
| institution | Directory of Open Access Journals |
| issn | 1538-4357 |
| language | English |
| publishDate | 2023-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| spelling | doaj-art-e632574e7a1e4e4aa9bbccb5d0a63de62025-08-19T23:27:36ZengIOP PublishingThe Astrophysical Journal1538-43572023-01-01953217810.3847/1538-4357/acdc25Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural NetworksJi Won Park0https://orcid.org/0000-0002-0692-1092Simon Birrer1https://orcid.org/0000-0003-3195-5507Madison Ueland2Miles Cranmer3https://orcid.org/0000-0002-6458-3423Adriano Agnello4https://orcid.org/0000-0001-9775-0331Sebastian Wagner-Carena5https://orcid.org/0000-0001-5039-1685Philip J. Marshall6https://orcid.org/0000-0002-0113-5770Aaron Roodman7https://orcid.org/0000-0001-5326-3486the LSST Dark Energy Science CollaborationKavli Institute for Particle Astrophysics and Cosmology, Department of Physics, Stanford University , Stanford, CA 94025, USA ; jp281@slac.stanford.edu; SLAC National Accelerator Laboratory , Menlo Park, CA 94025, USAKavli Institute for Particle Astrophysics and Cosmology, Department of Physics, Stanford University , Stanford, CA 94025, USA ; jp281@slac.stanford.edu; SLAC National Accelerator Laboratory , Menlo Park, CA 94025, USAKavli Institute for Particle Astrophysics and Cosmology, Department of Physics, Stanford University , Stanford, CA 94025, USA ; jp281@slac.stanford.eduPrinceton University , Princeton, NJ 08544, USADARK, Niels Bohr Institute, University of Copenhagen , Copenhagen, DenmarkKavli Institute for Particle Astrophysics and Cosmology, Department of Physics, Stanford University , Stanford, CA 94025, USA ; jp281@slac.stanford.edu; SLAC National Accelerator Laboratory , Menlo Park, CA 94025, USAKavli Institute for Particle Astrophysics and Cosmology, Department of Physics, Stanford University , Stanford, CA 94025, USA ; jp281@slac.stanford.edu; SLAC National Accelerator Laboratory , Menlo Park, CA 94025, USAKavli Institute for Particle Astrophysics and Cosmology, Department of Physics, Stanford University , Stanford, CA 94025, USA ; jp281@slac.stanford.edu; SLAC National Accelerator Laboratory , Menlo Park, CA 94025, USAWe present a Bayesian graph neural network (BGNN) that can estimate the weak lensing convergence ( κ ) from photometric measurements of galaxies along a given line of sight (LOS). The method is of particular interest in strong gravitational time-delay cosmography (TDC), where characterizing the “external convergence” ( κ _ext ) from the lens environment and LOS is necessary for precise Hubble constant ( H _0 ) inference. Starting from a large-scale simulation with a κ resolution of ∼1′, we introduce fluctuations on galaxy–galaxy lensing scales of ∼1″ and extract random sight lines to train our BGNN. We then evaluate the model on test sets with varying degrees of overlap with the training distribution. For each test set of 1000 sight lines, the BGNN infers the individual κ posteriors, which we combine in a hierarchical Bayesian model to yield constraints on the hyperparameters governing the population. For a test field well sampled by the training set, the BGNN recovers the population mean of κ precisely and without bias (within the 2 σ credible interval), resulting in a contribution to the H _0 error budget well under 1%. In the tails of the training set with sparse samples, the BGNN, which can ingest all available information about each sight line, extracts a stronger κ signal compared to a simplified version of the traditional method based on matching galaxy number counts, which is limited by sample variance. Our hierarchical inference pipeline using BGNNs promises to improve the κ _ext characterization for precision TDC. The code is available as a public Python package, Node to Joy https://github.com/jiwoncpark/node-to-joy .https://doi.org/10.3847/1538-4357/acdc25CosmologyHubble constantHierarchical modelsBayesian statisticsNeural networksWeak gravitational lensing |
| spellingShingle | Ji Won Park Simon Birrer Madison Ueland Miles Cranmer Adriano Agnello Sebastian Wagner-Carena Philip J. Marshall Aaron Roodman the LSST Dark Energy Science Collaboration Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural Networks Cosmology Hubble constant Hierarchical models Bayesian statistics Neural networks Weak gravitational lensing |
| title | Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural Networks |
| title_full | Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural Networks |
| title_fullStr | Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural Networks |
| title_full_unstemmed | Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural Networks |
| title_short | Hierarchical Inference of the Lensing Convergence from Photometric Catalogs with Bayesian Graph Neural Networks |
| title_sort | hierarchical inference of the lensing convergence from photometric catalogs with bayesian graph neural networks |
| topic | Cosmology Hubble constant Hierarchical models Bayesian statistics Neural networks Weak gravitational lensing |
| url | https://doi.org/10.3847/1538-4357/acdc25 |
| work_keys_str_mv | AT jiwonpark hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT simonbirrer hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT madisonueland hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT milescranmer hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT adrianoagnello hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT sebastianwagnercarena hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT philipjmarshall hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT aaronroodman hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks AT thelsstdarkenergysciencecollaboration hierarchicalinferenceofthelensingconvergencefromphotometriccatalogswithbayesiangraphneuralnetworks |