Simulating intergalactic gas for DESI-like small scale Lymanα forest observations

Abstract

Measurements of the Lyα forest based on large numbers of quasar spectra from sky surveys such as SDSS/eBOSS accurately probe the distribution of matter on small scales and thus provide important constraints on several ingredients of the cosmological model. A main summary statistic derived from those measurements is the one-dimensional power spectrum, P1D, of the Lyα absorption. However, model predictions for P1D rely on expensive hydrodynamical simulations of the intergalactic medium, which was the limiting factor in previous analyses. Datasets from upcoming surveys such as DESI will push observational accuracy near the 1%-level and probe even smaller scales. This observational push mandates even more accurate simulations as well as more careful exploration of parameter space. In this work we evaluate the robustness and accuracy of simulations and the statistical framework used to constrain cosmological parameters. We present a comparison between the grid-based simulation code Nyx and SPH-based code Gadget in the context of P1D. In addition, we perform resolution and box-size convergence tests using Nyx code. We use a Gaussian process emulation scheme to reduce the number of simulations required for exploration of parameter space without sacrificing the model accuracy. We demonstrate the ability to produce unbiased parameter constraints in an end-to-end inference test using mock eBOSS- and DESI-like data, and we advocate for the usage of adaptive sampling schemes as opposed to using a fixed Latin hypercube design.

The effect of inhomogeneous reionisation on the Lyman-$α$ forest power spectrum at redshift $z>4$: implications for thermal parameter recovery

“…This implies that patchy reionisation effects should start to become important for Ly𝛼 forest power spectrum analyses at small scales, 𝑘 ∼ 0.1 km −1 s, if measurement uncertainties are at the ∼ 5 per cent level. At this level, however, several other effects may also become important in the future data sets, such as the accuracy of the power spectrum emulator (Bird et al 2019;Pedersen et al 2021), simulation initial conditions (Bird et al 2020), and hydrodynamical methods (Regan et al 2007;Walther et al 2021). Furthermore, the accuracy of the approximation for the patchy correction starts to approach (at most) the 5 per cent level at the smallest scales (see Figure 9).…”

Fast, high-fidelity Lyman $α$ forests with convolutional neural networks

“…The range of scales accurately mapped with this method is well beyond the resolution of next generation observations (Walther et al 2021). This method provides an alternative to running costly hydrodynamical simulations for mock catalog generation and the con- struction of Lyα power-spectra emulators.…”

“…In Nyx, the dark matter particles are evolved with a particle-mesh scheme, while the additional baryon hydrodynamics are modeled as an inviscid ideal fluid on a set of Eulerian grids. We neglect physics related to galaxy formation, but this is extremely common approach in Lyα cosmological simulations (Iršič et al 2017;Boera et al 2019;Walther et al 2019;Rogers & Peiris 2020;Walther et al 2021;Pedersen et al 2020). This is because regions transparent to Lyα photons at redhisfts z 2 are in low density regions (Lukić et al 2015;Mc-Quinn 2016), and poorly understood physical processes related to galaxy formation only play a minor role in those regions (Kollmeier et al 2006;Desjacques et al 2006).…”

“…More recently, recipes to model Lyα forest without running full hydrodynamical simulations have been provided by Peirani et al (2014) and Sorini et al (2016), as well as models for specific summary statistics in Seljak (2012), but they have not been really used in practice due to their complexity and/or applicability to only a limited range of scales. Thus, Lyα forest forward models rely on computationally expensive hydrodynamical simulations (Iršič et al 2017;Boera et al 2019;Walther et al 2019;Palanque-Delabrouille et al 2020;Rogers & Peiris 2020;Walther et al 2021), and solving inverse problems commonly requires running dozens of simulations, at a rough cost of 2 million CPU hours per simulation (see, for example, Walther et al 2021). Of course, if one wants to consider cosmological models beyond standard ΛCDM or to model complex astrophysical phenomena affecting the Lyα signal like high column density absorbers (Rogers et al 2018), UV background fluctuations (Oñorbe et al 2019), or HeII reionization (Upton Sanderbeck & Bird 2020) the number of needed simulations, and thus the computational costs, will increase significantly.…”

HyPhy: Deep Generative Conditional Posterior Mapping of Hydrodynamical Physics

“…However, the power of these hydrodynamical simulations comes with significant computational cost, and the next generation of cosmological surveys will require unprecedented precision across a wide range of scales (e.g. Walther et al (2021)). In this regime, computing quantities like covariance matrices (which require large numbers of simulations) becomes an increasingly daunting task, so there is a clear need for approximate methods that can ease some of the computational burden.…”