Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Low redshift observational constraints on dark energy models using ANN - CosmicANNEstimator

Ashly Joseph, Albin Joseph, Christina Terese Joseph, John Paul Martin, Sunil Kumar PV, Sarthak Giri

TL;DR

Cosmological parameters in the $\Lambda$CDM framework are inferred at low redshift using CosmicANNEstimator, a dual artificial neural network approach that separately analyzes $H(z)$ and Pantheon SN data. The models are trained on $\sim10^5$ synthetic realizations with Gaussian noise and optimized with a heteroscedastic loss, then validated against Metropolis–Hastings MCMC results. The ANN predictions for $(H_0,\Omega_{m0},\Omega_{\Lambda0})$ show good agreement with traditional methods, with orders-of-magnitude faster inference once trained, enabling rapid analyses for large surveys. Limitations include diagonal uncertainty estimates lacking full covariance and opportunities for joint $SN$+$H$ analyses; future work may address covariance reconstruction and physics-informed architectures to improve interpretability and precision.

Abstract

We present CosmicANNEstimator (Cosmological Parameters Artificial Neural Network Estimator), a machine learning approach for constraining cosmological parameters within the Lambda Cold Dark Matter ($Λ$CDM) framework. Our methodology employs two specialized artificial neural networks (ANNs) designed to analyze Hubble parameter and Supernova data independently. The estimator is trained on synthetic data covering broad parameter ranges, with Gaussian random noise incorporated to simulate observational uncertainties. Our results demonstrate parameter estimates and associated uncertainties comparable to traditional Markov Chain Monte Carlo (MCMC) methods, establishing machine learning as an efficient alternative for cosmological parameter estimation. This work underscores the potential of neural network-based inference to complement traditional Bayesian methods and accelerate future cosmological analyses.

Low redshift observational constraints on dark energy models using ANN - CosmicANNEstimator

TL;DR

Cosmological parameters in the CDM framework are inferred at low redshift using CosmicANNEstimator, a dual artificial neural network approach that separately analyzes and Pantheon SN data. The models are trained on synthetic realizations with Gaussian noise and optimized with a heteroscedastic loss, then validated against Metropolis–Hastings MCMC results. The ANN predictions for show good agreement with traditional methods, with orders-of-magnitude faster inference once trained, enabling rapid analyses for large surveys. Limitations include diagonal uncertainty estimates lacking full covariance and opportunities for joint + analyses; future work may address covariance reconstruction and physics-informed architectures to improve interpretability and precision.

Abstract

We present CosmicANNEstimator (Cosmological Parameters Artificial Neural Network Estimator), a machine learning approach for constraining cosmological parameters within the Lambda Cold Dark Matter (CDM) framework. Our methodology employs two specialized artificial neural networks (ANNs) designed to analyze Hubble parameter and Supernova data independently. The estimator is trained on synthetic data covering broad parameter ranges, with Gaussian random noise incorporated to simulate observational uncertainties. Our results demonstrate parameter estimates and associated uncertainties comparable to traditional Markov Chain Monte Carlo (MCMC) methods, establishing machine learning as an efficient alternative for cosmological parameter estimation. This work underscores the potential of neural network-based inference to complement traditional Bayesian methods and accelerate future cosmological analyses.

Paper Structure

This paper contains 33 sections, 26 equations, 11 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Cosmological Dynamics
  3. Methodology
  4. Datasets
  5. H(z) Compilation:
  6. Pantheon Supernova Dataset:
  7. Data Simulations
  8. H(z) Compilation:
  9. Pantheon Supernova Dataset:
  10. Adding Noise to Data
  11. Data Preprocessing
  12. Data Preprocessing for the Hubble Compilation:
  13. Proposed Model
  14. Proposed Model for Hubble Compilation - CosmicANNEstimator-Hubble:
  15. Proposed Model for Pantheon Supernova Dataset - CosmicANNEstimator-SN:
  16. ...and 18 more sections

Figures (11)

  • Figure 1: Schematic representation of the CosmicANNEstimator framework integrating observational data processing, neural network training, and MCMC validation
  • Figure 2: The architecture of CosmicANNEstimator for Hubble Compilation. The network architecture consists of an input layer with 31 z-values, four hidden layers with 22, 16, 11, and 8 neurons respectively (with ReLU activation), and an output layer with 6 cosmological parameters.
  • Figure 3: Figure showing trace plots of Hubble Parameter Data after performing thinning process
  • Figure 4: Figure showing trace plots of Supernova Data after performing thinning process
  • Figure 5: Figure showing Learning Curve of CosmicANNEstimator-Hubble Model
  • ...and 6 more figures