Cosmic Inference: Constraining Parameters With Observations and Highly Limited Number of Simulations

Takhtaganov, Timur; Lukić, Zarija; Mueller, Juergen; Morozov, Dmitriy

doi:10.48550/arxiv.1905.07410

Cited by 4 publications

(8 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Rogers et al (2019) proposed using Bayesian optimization to improve emulator accuracy by a sequential choice of new simulation points designed to globally optimize the emulator function. Similar approaches to iterative selection of training data in a cosmological parameter space have been presented by Takhtaganov et al (2019); Pellejero-Ibañez et al (2020). Computer scientists and engineers, including Huang et al (2006); Forrester et al (2007); Lam et al (2015); Poloczek et al (2016);McLeod et al (2017), have extensively studied combining multi-fidelity methods with Bayesian optimization.…”

Section: Introductionmentioning

confidence: 93%

Multi-Fidelity Emulation for the Matter Power Spectrum using Gaussian Processes

Ho,

Bird,

Shelton

2021

Preprint

View full text Add to dashboard Cite

We present methods for emulating the matter power spectrum which effectively combine information from cosmological N -body simulations at different resolutions. An emulator allows estimation of simulation output by interpolating across the parameter space of a handful of simulations. We present the first implementation of multi-fidelity emulation in cosmology, where many low-resolution simulations are combined with a few high-resolution simulations to achieve an increased emulation accuracy. The power spectrum's dependence on cosmology is learned from the low-resolution simulations, which are in turn calibrated using high-resolution simulations. We show that our multifidelity emulator can achieve percent-level accuracy on average with only 3 high-fidelity simulations and outperforms a single-fidelity emulator that uses 11 simulations. With a fixed number of high-fidelity training simulations, we show that our multi-fidelity emulator is 100 times better than a single-fidelity emulator at k 2 hMpc −1 , and 20 times better at 3 k < 6.4 hMpc −1 . Multi-fidelity emulation is fast to train, using only a simple modification to standard Gaussian processes. Our proposed emulator shows a new way to predict non-linear scales by fusing simulations from different fidelities.

show abstract

Section: Introductionmentioning

confidence: 93%

Multi-Fidelity Emulation for the Matter Power Spectrum using Gaussian Processes

Ho,

Bird,

Shelton

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…While this approach will produce non-negligible artifacts for the smallest scales and is thus not usable for an interpretation of Lyα forest data from high-resolution datasets, the effects of this approach on large scales k 0.04 skm −1 as measured e.g. in the DESI survey are small [49].…”

Section: Adopted Parameters and Grid Designmentioning

confidence: 99%

“…A Bayesian optimization approach as described in e.g. [48,49] which additionally also takes into account the emulation error returned by the emulation scheme should allow such an iterative approach. While it is beyond the scope of this work to study this in more detail, we do plan to use such a technique for future analyses.…”

Section: Fitting a Fiducial Model In Different Configurationsmentioning

confidence: 99%

See 1 more Smart Citation

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

Walther,

Armengaud,

Ravoux

et al. 2020

Preprint

View full text Add to dashboard Cite

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, P 1D , of the Lyα absorption. However, model predictions for P 1D 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 P 1D . 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.

show abstract

“…In addition to cosmological parameter constraints and the characterization of systematics, the Sejong Suite can in fact be helpful for novel methods that aim at painting baryonic physics into DM-only simulations, that intend to add small-scale physics to lowresolution larger-box simulations, or that attempt to accurately reproduce observables from a limited number of realizations in parameter space -as suggested by several recent studies. In this regard, there has been interesting progress toward efficient emulators (Bird et al 2019;Giblin et al 2019;Rogers et al 2019;der Velden et al 2019;Zhai et al 2019), training sets, neural networks, and machine-learning techniques (Nadler et al 2018;Rodríguez et al 2018;Mustafa et al 2019;Ramanah et al 2019;Shirasaki et al 2019;Takhtaganov et al 2019;Zamudio-Fernandez et al 2019;Wibking et al 2020), or novel approaches such as the Cosmological Evidence Modeling (Lange et al 2019) or the Machineassisted Semi-Simulation Model (Jo & Kim 2019). The common aspect between these methods is the reliance on some (even partial) information derived from highresolution hydrodynamical realizations, and in this context, the Sejong Suite could be useful for calibrations purposes.…”

Section: Summary and Conclusion: Novelties Applications And Outlookmentioning

confidence: 99%

The Sejong Suite: Cosmological Hydrodynamical Simulations with Massive Neutrinos, Dark Radiation, and Warm Dark Matter

Rossi

2020

Preprint

View full text Add to dashboard Cite

We present the Sejong Suite, an extensive collection of state-of-the-art high-resolution cosmological hydrodynamical simulations spanning a variety of cosmological and astrophysical parameters, primarily developed for modeling the Lyman-α (Lyα) forest. The suite is organized into three main categories (Grid Suite, Supporting Suite, and Systematics Suite), addressing different science targets. Adopting a particle-based implementation, we follow the evolution of gas, dark matter (cold and warm), massive neutrinos, and dark radiation, and consider several combinations of box sizes and number of particles. With additional enhancing techniques, we are able to reach an equivalent resolution up to 3 × 3328 3 = 110 billion particles in a (100h −1 Mpc) 3 box size, ideal for current and future surveys (e.g., eBOSS, DESI). Noticeably, for the first time, we simulate extended mixed scenarios describing the combined effects of warm dark matter, neutrinos, and dark radiation, modeled consistently by taking into account the neutrino mass splitting. In addition to providing multicomponent snapshots from z = 5.0 to z = 2.0 in intervals of ∆z = 0.2 for all of the models considered, we produced over 288 million Lyα skewers in the same z-range and extended parameter space. The skewers are well suited for Lyα forest science studies, for mapping the high-z cosmic web and the matter-to-flux relation and bias, and for quantifying the critical role of baryons at small scales. We also present a first analysis of the suite focused on the matter and flux statistics, and show that we are able to accurately reproduce the 1D flux power spectrum down to scales k = 0.06 [km/s] −1 as mapped by recent high-resolution quasar data, as well as the thermal history of the intergalactic medium. The simulations and products described here will be progressively made available.

show abstract

Cosmic Inference: Constraining Parameters With Observations and Highly Limited Number of Simulations

Cited by 4 publications

References 35 publications

Multi-Fidelity Emulation for the Matter Power Spectrum using Gaussian Processes

Multi-Fidelity Emulation for the Matter Power Spectrum using Gaussian Processes

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

The Sejong Suite: Cosmological Hydrodynamical Simulations with Massive Neutrinos, Dark Radiation, and Warm Dark Matter

Contact Info

Product

Resources

About