Abstract

In the standard cosmological model, cold dark matter gives rise to small-scale structure problems which Warm Dark Matter (WDM) in the keV range may address. Previous studies have narrowed the mass range for WDM , but further testing is increasingly challenging due to resolution limits of current spectrographs. Given the lack of success in laboratory dark matter searches, an intriguing alternative has recently re-emerged—a mixed dark matter scenario that combines cold and warm components, known as Cold+Warm Dark Matter (CWDM). In this framework, structure formation is influenced by the free-streaming of the warm component at the small scales probed by the Lyman-α forest, requiring hydrodynamical simulations to account for non-linear structure evolution and gas dynamics.

This presentation focuses on updated constraints on CWDM derived from high resolution and high-redshift (z=4.2−5.0) spectra from UVES and HIRES spectrographs. We use the 1D flux power spectrum to compare simulated Lyman-α forest to observational data in a Bayesian inference framework. The grid of simulations spans a high-dimensional parameter space, which we efficiently sample by implementing a neural network emulator at the likelihood level.

Beyond 1D clustering statistics, these allowed mixed dark matter models can accommodate small-scale deviations in the high-k regime of the matter power spectrum, potentially alleviating the S8 tension.