Abstract:In this paper, we present an analysis of Supernova Ia (SNIa) distance moduli $\mu(z)$ and dark energy using an Artificial Neural Network (ANN) reconstruction based on LSST simulated three-year SNIa data. The ANNs employed in this study utilize genetic algorithms for hyperparameter tuning and Monte Carlo Dropout for predictions. Our ANN reconstruction architecture is capable of modeling both the distance moduli and their associated statistical errors given redshift values. We compare the performance of the ANN-based reconstruction with two theoretical dark energy models: $\Lambda$CDM and Chevallier-Linder-Polarski (CPL). Bayesian analysis is conducted for these theoretical models using the LSST simulations and compared with observations from Pantheon and Pantheon+ SNIa real data. We demonstrate that our model-independent ANN reconstruction is consistent with both theoretical models. Performance metrics and statistical tests reveal that the ANN produces distance modulus estimates that align well with the LSST dataset and exhibit only minor discrepancies with $\Lambda$CDM and CPL.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to conduct a reconstruction analysis of the distance modulus ($\mu(z)$) of Supernova Ia (SNIa) and dark energy by using artificial neural networks (ANN). Specifically, the goals of the paper include: 1. **Distance modulus reconstruction using ANN**: - The author uses an ANN architecture with hyper - parameter tuning based on genetic algorithms and prediction using Monte Carlo Dropout to model the distance modulus and its statistical error for three - year SNIa data from Rubin Observatory simulations. 2. **Comparison with theoretical models**: - Compare the ANN - reconstructed results with two theoretical dark - energy models (ΛCDM and Chevallier - Linder - Polarski, CPL). Through Bayesian analysis, evaluate the performance of these theoretical models on LSST simulation data and compare them with the actual SNIa observation data of Pantheon and Pantheon +. 3. **Model - independence verification**: - Verify whether the ANN reconstruction method can provide results consistent with existing theoretical models independently of specific theoretical models. Through performance metrics and statistical tests, evaluate the consistency between the distance modulus estimates generated by ANN and the LSST dataset. 4. **Explore the properties of dark energy**: - Through the non - parametric ANN reconstruction method, directly extract the properties of dark energy from the data, reducing the dependence on specific theoretical models, and thus analyze the characteristics of dark energy more freely. ### Main research content - **Data sources**: - Use three - year SNIa data simulated by LSST, containing 5,785 data points, with a redshift range from 0.01 to 1.4. - Comparison data include the actual SNIa observation data of Pantheon and Pantheon +. - **Methods**: - **ANN architecture**: Use genetic algorithms for hyper - parameter tuning and Monte Carlo Dropout for prediction. - **Training and validation**: Divide the dataset into 80% for training and 20% for validation, with the loss function being Mean Squared Error (MSE), and train for 2,000 epochs. - **Bayesian analysis**: Use nested sampling and SimpleMC code for Bayesian parameter estimation, and compare the ΛCDM and CPL models under different datasets. - **Results**: - The ANN reconstruction results are highly consistent with the LSST dataset, and also show small differences from the theoretical predictions of the ΛCDM and CPL models. - Bayesian analysis shows that LSST data provides better constraints on the ΛCDM model, especially on the $\Omega_m$ parameter. For the CPL model, LSST data provides better constraints on the $w_0$ parameter. - Model comparison shows that although the CPL model has a slightly better fitting effect in some cases, the ΛCDM model has a certain advantage in Bayesian evidence, especially on LSST data. ### Conclusion The paper demonstrates the effectiveness of the ANN reconstruction method in handling SNIa data and dark - energy research, verifies its consistency with existing theoretical models, and provides a basis for future model - selection tests using larger - scale datasets (such as LSST ten - year data).

Dark energy reconstruction analysis with artificial neural networks: Application on simulated Supernova Ia data from Rubin Observatory

AI-assisted Reconstruction of Cosmic Velocity Field from Redshift-Space Spatial Distribution of Halos

Role of Future SNIa Data from Rubin LSST in Reinvestigating Cosmological Models

Estimating Cosmological Parameters and Reconstructing Hubble Constant with Artificial Neural Networks: A Test with Reconstructed $H(z)$

PkANN - I. Non-linear Matter Power Spectrum Interpolation Through Artificial Neural Networks

Likelihood-free Cosmological Constraints with Artificial Neural Networks: An Application on Hubble Parameters and SNe Ia

Distinguishing Coupled Dark Energy Models with Neural Networks

Inferring IGM parameters from the redshifted 21-cm Power Spectrum using Artificial Neural Networks

ParamANN: A Neural Network to Estimate Cosmological Parameters for $Λ$CDM Universe Using Hubble Measurements

Dynamical dark energy in light of cosmic distance measurements I: a demonstration using simulated datasets

Model-independent calibration of Gamma-Ray Bursts with neural networks

Emulating Radiative Transfer with Artificial Neural Networks

ParamANN: a neural network to estimate cosmological parameters for ΛCDM universe using hubble measurements

Machine Learning Constraints on Dark Matter Annihilation during the Epoch of Reionization: A CNN Analysis of the 21-cm Signal

Uncorrelated Measurements of the Cosmic Expansion History and Dark Energy from Supernovae

Not Hydro: Using Neural Networks to estimate galaxy properties on a Dark-Matter-Only simulation

SIDE-real: Supernova Ia Dust Extinction with truncated marginal neural ratio estimation applied to real data

The cosmological analysis of X-ray cluster surveys: VI. Inference based on analytically simulated observable diagrams