Cosmology

Abstract:

ABSTRACT We introduce the ‘Engineering Dwarfs at Galaxy Formation’s Edge’ (EDGE) project to study the cosmological formation and evolution of the smallest galaxies in the Universe. In this first paper, we explore the effects of resolution and sub-grid physics on a single low-mass halo ($M_{\rm halo}=10^{9}{\, \rm M}_\odot$), simulated to redshift z = 0 at a mass and spatial resolution of $\sim 20{\, \rm M}_\odot$ and ∼3 pc. We consider different star formation prescriptions, supernova feedback strengths, and on-the-fly radiative transfer (RT). We show that RT changes the mode of galactic self-regulation at this halo mass, suppressing star formation by causing the interstellar and circumgalactic gas to remain predominantly warm (∼104 K) even before cosmic reionization. By contrast, without RT, star formation regulation occurs only through starbursts and their associated vigorous galactic outflows. In spite of this difference, the entire simulation suite (with the exception of models without any feedback) matches observed dwarf galaxy sizes, velocity dispersions, V-band magnitudes, and dynamical mass-to-light-ratios. This is because such structural scaling relations are predominantly set by the host dark matter halo, with the remaining model-to-model variation being smaller than the observational scatter. We find that only the stellar mass–metallicity relation differentiates the galaxy formation models. Explosive feedback ejects more metals from the dwarf, leading to a lower metallicity at a fixed stellar mass. We conclude that the stellar mass–metallicity relation of the very smallest galaxies provides a unique constraint on galaxy formation physics.

Abstract:

Computer simulations are invaluable tools for scientific discovery. However, accurate simulations are often slow to execute, which limits their applicability to extensive parameter exploration, large-scale data analysis, and uncertainty quantification. A promising route to accelerate simulations by building fast emulators with machine learning requires large training datasets, which can be prohibitively expensive to obtain with slow simulations. Here we present a method based on neural architecture search to build accurate emulators even with a limited number of training data. The method successfully accelerates simulations by up to 2 billion times in 10 scientific cases including astrophysics, climate science, biogeochemistry, high energy density physics, fusion energy, and seismology, using the same super-architecture, algorithm, and hyperparameters. Our approach also inherently provides emulator uncertainty estimation, adding further confidence in their use. We anticipate this work will accelerate research involving expensive simulations, allow more extensive parameters exploration, and enable new, previously unfeasible computational discovery.

Abstract:

Tighter constraints on measurements of primordial non-Gaussianity (PNG) will allow the differentiation of inflationary scenarios. The cosmic microwave background bispectrum – the standard method of measuring the local non-Gaussianity – is limited by cosmic variance. Therefore, it is sensible to investigate measurements of non-Gaussianity using the large-scale structure. This can be done by investigating the effects of non-Gaussianity on the power spectrum on large scales. In this study, we forecast the constraints on the local PNG parameter fNL that can be obtained with future radio surveys. We utilize the multitracer method that reduces the effect of cosmic variance and takes advantage of the multiple radio galaxy populations that are differently biased tracers of the same underlying dark matter distribution. Improvements on previous work include the use of observational bias and halo mass estimates, updated simulations, and realistic photometric redshift expectations, thus producing more realistic forecasts. Combinations of Square Kilometre Array simulations and radio observations were used as well as different redshift ranges and redshift bin sizes. It was found that in the most realistic case the 1σ error on fNL falls within the range 4.07–6.58, rivalling the tightest constraints currently available.

Abstract:

<jats:title>ABSTRACT</jats:title> <jats:p>Accurate methods for reverberation mapping using photometry are highly sought after since they are inherently less resource intensive than spectroscopic techniques. However, the effectiveness of photometric reverberation mapping for estimating black hole masses is sparsely investigated at redshifts higher than z ≈ 0.04. Furthermore, photometric methods frequently assume a damped random walk (DRW) model, which may not be universally applicable. We perform photometric reverberation mapping using the javelin photometric DRW model for the QSO SDSS-J144645.44+625304.0 at z = 0.351 and estimate the Hβ lag of $65^{+6}_{-1}$ d and black hole mass of $10^{8.22^{+0.13}_{-0.15}}\, \mathrm{M_{\odot }}$. An analysis of the reliability of photometric reverberation mapping, conducted using many thousands of simulated CARMA process light curves, shows that we can recover the input lag to within 6 per cent on average given our target’s observed signal-to-noise of &amp;gt;20 and average cadence of 14 d (even when DRW is not applicable). Furthermore, we use our suite of simulated light curves to deconvolve aliases and artefacts from our QSO’s posterior probability distribution, increasing the signal-to-noise on the lag by a factor of ∼2.2. We exceed the signal-to-noise of the Sloan Digital Sky Survey Reverberation Mapping Project (SDSS-RM) campaign with a quarter of the observing time per object, resulting in a ∼200 per cent increase in signal-to-noise efficiency over SDSS-RM.</jats:p>