Detection of Deviations from Lambda-CDM in Dark Matter Power Spectrum Using Cosmology Informed Neural Networks

This study investigates the application of Cosmology-Informed Neural Networks (CINNs) to detect deviations from the Lambda Cold Dark Matter (ACDM) model in dark matter power spectra. By incorporating Partial Differential Equations (PDEs) into the training process, CINNs are designed to differentiate between ACDM and alternative models of dark matter and dark energy. The network was trained to classify power spectra in the range of 0.01 hMpc<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−1</sup> to 2.5 hMpc<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−1</sup>, achieving a classification accuracy of approximately 97%. Additionally, the CINN demonstrates sensitivity to small deviations from the ACDM model, making it a promising tool for constraining cosmological parameters and enhancing the understanding of dark matter and dark energy. Future developments may involve expanding the model with larger datasets to further improve accuracy and performance for cosmological research.