arXiv subjects

James M. Sullivan

Publications and source records attributed to James M. Sullivan.

At least 19 recordsLinked to original sources

arXiv2026

Toward Precision Kinetic Sunyaev-Zel'dovich Cosmology---1. The Matter - Momentum Bispectrum at One-Loop Order

The kinetic Sunyaev-Zeldovich (kSZ) effect is a sensitive probe of the free-electron momentum field, tapping into spatial correlations relevant for both large-scale cosmology and smaller-scale astrophysics. Precision kSZ measurements will provide a robust channel for constraining inflationary physics and equivalence principle violation, while simultaneously characterizing the small-scale gas distribution and associated baryonic feedback. However, robust kSZ measurement for cosmological applications is challenging, as the primary source of large-scale information is convolved with unknown small-scale dynamics in a degenerate combination. In this paper we address this challenge using the Effective Field Theory (EFT) framework that provides a principled way of treating the effects of small-scale dynamics on large scale-structure. Extracting the kSZ signal can be reformulated as a measurement of the one-loop galaxy-galaxy-electron momentum bispectrum. We compute here a pure dark matter version of this bispectrum in EFT and compare it to simulations. We derive the EFT counterterms that capture small-scale backreaction and find that two of them acquire a transverse direction. One transverse counterterm gets generated entirely by the vorticity of the velocity field. Our computation of the bispectrum dipole agrees with simulation data to 5% up to k_max = 0.23 h/Mpc, which is significantly larger than the reach of tree level theory, which breaks down beyond k_max = 0.07 h/Mpc (at redshift z = 0.5). For the bispectrum computed with the projected momentum field, relevant to measuring the kSZ field, we find a similar k_max, and make the first detection of the transverse counterterms from simulations. Our results suggest that one-loop bispectrum computations can play an important role in advancing kSZ cosmology.

astro-ph.CO↗

arXiv2026

Reanalyzing DESI DR1: 4. Percent-Level Cosmological Constraints from Combined Probes and Robust Evidence for the Normal Neutrino Mass Hierarchy

We present cosmological parameter measurements from the full combination of DESI DR1 galaxy clustering data, described with large-scale structure effective field theory. By incorporating photometric galaxies and CMB lensing cross-correlations, and extending the bispectrum likelihood to smaller scales with a consistent one-loop computation, we achieve substantial gains in constraining power. Combined with the latest DESI baryon acoustic oscillation (BAO) data and cosmic microwave background (CMB) priors on the spectral tilt and baryon density, we find, in $Λ$CDM, $H_0=69.08\pm 0.37~\mathrm{km}\,\mathrm{s}^{-1}\mathrm{Mpc}^{-1}$, $Ω_m=0.2974\pm 0.0050$, and $σ_8 = 0.838\pm 0.017$ ($S_8 = σ_8\sqrt{Ω_m/0.3} =0.834\pm 0.018$). Adding the Pantheon+ supernovae (SNe), we find a $2.2σ$ preference for the $w_0w_a$ dynamical dark energy model from low-redshift data alone, rising to $2.7σ$ when exchanging the SNe for \textit{Planck} CMB data. Combining the full-shape, BAO, CMB, and SNe likelihoods improves the dark energy figure-of-merit by $15\%$ and bounds the neutrino mass sum to $M_ν<0.049$ eV ($Λ$CDM) and $M_ν<0.077$ eV ($w_0w_a$CDM) at 95\% CL. This is the strongest $w_0w_a$CDM bound to date, $37\%$ tighter than from the background expansion data alone. The preference for the normal neutrino mass ordering thus holds regardless of the background model: the inverted hierarchy is disfavored at ${\approx}\,3.5σ$ in $Λ$CDM and ${\approx}\,2.4σ$ in $w_0w_a$CDM, with the latter constraint free of the geometric tension between CMB and BAO that is known to sharpen the $Λ$CDM bound.

astro-ph.CO↗

arXiv2026

Sample Variance Cancellation for Future Spectroscopic Surveys

High-redshift spectroscopic galaxy surveys will be the scientific engines of the next generation of large-scale structure cosmology. The clustering signal of high redshift, star-forming Lyman-$α$ emitters (LAEs) will be of key importance for obtaining high-redshift constraints on the growth of structure and redshift-space distortions. The complex radiative transfer (RT) of Lyman-$α$ photons alters the symmetry group respected by the overdensity field constructed from these galaxies, and so the observed large-scale clustering of LAEs may have an angular dependence that differs significantly from that of linear theory, possibly biasing inference of cosmological parameters. While such an effect has been seen in simulations, its amplitude in nature and its detailed form remains unclear. In the restricted context of a linear, Gaussian model, we outline a procedure for pinning down the type and amplitude of such changes in angular dependence on large scales due to unknown RT or a more general unmodeled angular effect in the hypothetical scenario in which an observer is presented with LAE data containing such an effect. We show that if a second tracer without the modified angular dependence is available for cross correlation with the LAEs at the same redshifts (e.g., Lyman-break galaxies), then, with a high redshift survey of modest size, it is possible to rapidly identify: 1) the presence of a nontrivial angular functional form of radiative transfer (by a conditional field-level realization), 2) the functional form itself (with an optimal filter that we derive), and 3) the value of its amplitude with an uncertainty (via an adaptation of the standard quadratic estimator). Such sample-variance-cancellation strategies therefore provide a statistical solution to unknown astrophysical or systematic angular clustering dependence, including from RT.

astro-ph.CO↗

arXiv2026

Cosmological Concordance in an Especially Opaque Universe: A Tentative Cosmological Detection of Physical Neutrino Mass in $Λ$CDM

The measurement of the sum of neutrino masses is among the primary promises of precision cosmology, achievable by combining complementary early- and late-Universe probes. However, these datasets currently exhibit mild-to-strong disagreements within $Λ$CDM and its simplest extensions, giving rise to multiple tensions, including the Hubble tension, the preference for "negative" neutrino mass, and indications of evolving dark energy. It has recently been shown that these tensions can be alleviated by adopting a higher value of the optical depth to reionization parameter, $τ$, when large-scale cosmic microwave background (CMB) polarization data are ignored. We extend this proposal and show that an especially high prior on $τ= 0.11 \pm 0.006$ simultaneously addresses all three of these tensions, significantly reducing the need for new physics beyond $Λ$CDM. We determine the "concordance" value of $τ$ by requiring physical neutrino mass and consistency of the Hubble constant, $H_0$, inferred from the CMB and that preferred by the Dark Energy Spectroscopic Instrument (DESI) baryon acoustic oscillation (BAO) and full-shape measurements. Within this high-$τ$ Universe, we obtain the first $2σ$ detection of a positive neutrino mass, $Σm_ν = 0.10^{+0.04}_{-0.05}$~eV at 68\% C.L., while restoring cosmological concordance between datasets within $Λ$CDM. In particular, low-redshift distance predictions are consistent with DESI BAO observations and the inferred dark-energy equation-of-state parameters are consistent with a cosmological constant, both with and without supernovae data. The concordance power of our $τ$ prior further motivates new measurements of $τ$, e.g., through large angular scale CMB polarization observations with the \textit{LiteBIRD}, CLASS, or proposed PICO experiments. (Abridged)

astro-ph.CO↗

arXiv2026

Lyman-Alpha Forest and its Cross-Correlation with High-Redshift Galaxies in Effective Field Theory at the Field Level

We present a field-level perturbative forward model for the Lyman-alpha (Lya) forest flux decrement. We validate it on two simulation suites: large-volume AbacusSummit N-body simulations with the Lya forest painted onto the dark matter field, and the Sherwood hydrodynamic simulations. Across the redshift range of the simulations (z=2.0-3.2), the 3D and 1D power spectra of the model match the simulated Lya fields at the 1% (5%) level up to k <= 0.3 (1.0) h/Mpc, with similar performance for the cross-correlation with massive dark matter halos. The counts-in-cells statistic shows excellent agreement down to cell radii of 2 Mpc/h. Leveraging cosmic variance cancellation, the model enables precision measurements of Lya bias parameters and robustly detects the full set of quadratic line-of-sight bias operators, consistent with the notion of naturalness in effective field theory (EFT). We quantify the stochasticity of the Lya forest (the analog to the one-halo term), and find it to be white (scale- and orientation-independent) on large scales, matching EFT predictions. We further find that phenomenological flux power spectrum models, based on modulations of the linear-theory power spectrum, fail at the field level even on quasi-linear scales. For the currently observing Dark Energy Spectroscopic Instrument (DESI), we generate large-scale clustering mocks of the Lya forest to validate cosmological parameter inference pipelines. Looking ahead to its successor, DESI-II, we produce large-volume mocks of representative samples of Lyman-break galaxies (LBGs) and Lya emitters (LAEs), calibrated on Astrid hydrodynamic simulations and matched to observations at z=3, enabling joint analyses of Lya forest and high-redshift galaxy data.

astro-ph.CO↗

arXiv2026

ArkenstoneBH. A model for high-specific energy black hole feedback in cosmological simulations

AGN feedback is a key piece of galaxy evolution but is difficult to model due to its high specific energies, multiphase nature, and limited simulation resolutions. Arkenstone is a subgrid framework for representing multiphase flows in coarse resolution simulations that has been used to model stellar feedback driven galactic winds. It ensures the correct treatment of high specific energy feedback that would otherwise be challenging to model accurately in Lagrangian simulations. We introduce the new Arkenstone BH model, which extends the Arkenstone framework to model black hole feedback. We focus on describing the first piece of this framework, which follows the hot, high specific energy phase of these outflows. The second piece, which treats their multiphase structure with a scheme for modeling unresolved cold clouds, will be implemented and described in a later paper. We present Arkenstone BH in simulations of an isolated galaxy to demonstrate the framework and its ability to capture high specific energy feedback that interacts only weakly with cold, dense gas. We show how these energetic outflows suppress star formation in our isolated galaxy by counteracting the inflow of gas from the circumgalactic medium into the interstellar medium. This work is part of the "Learning the Universe" collaboration, which aims to understand the Universe's underlying physics and initial conditions.

astro-ph.GA↗

arXiv2026

Fast radio burst dispersion is an unbiased tracer of matter on large scales

The dispersion of fast radio bursts (FRBs) measures the column density of free electrons, tracing the diffuse ionized gas that contains more than $90\%$ of all baryons. On linear scales the FRB dispersion field is an approximately unbiased tracer of the matter distribution, an idea long assumed in the FRB large-scale structure literature and recently formalized by Zhou and Zhang [arXiv:2510.11022]. This follows from baryon-mass conservation, which forces the total baryon field to have unit linear bias, with dispersion inheriting this bias up to small corrections from the stellar and neutral-gas components. We show these corrections can be bounded at the percent level using existing galaxy and 21 cm surveys, and confirm with the FLAMINGO hydrodynamical simulations that the electron bias varies at the percent level across a wide range of feedback prescriptions. The dispersion-galaxy cross-power spectrum at linear scales directly constrains $B_8 \equiv σ_8(Ω_b/0.05)^{1/2}$, a baryonic analog of $S_8$, independently of feedback physics. Because most of the per-object variance in dispersion is cosmological signal rather than noise, $\sim\!10^5$ localized FRBs can match the statistical power of $\sim\!10^8$ weak-lensing galaxy shape measurements. FRB dispersion thus joins weak lensing and redshift-space distortions as a new unbiased tracer of matter on large scales.

astro-ph.CO↗

arXiv2025

Kinetic Mixing and the Phantom Illusion: Axion-Dilaton Quintessence in Light of DESI DR2

Recent results from DESI BAO analyses suggest that dark energy may not be a cosmological constant and is in fact dynamical. Furthermore, the data suggest that the equation of state may have been in the phantom regime in the distant past, recently undergoing a phantom crossing. In this work, we investigate whether this preference can be realized within a kinetically mixed axion-dilaton (KMIX) quintessence model, a string-motivated system in which an axion-like field couples exponentially to a dilaton-like (moduli) field. Crucially, KMIX can appear phantom in a standard Chevallier-Polarski-Linder (CPL) based analysis. To confront the model with data, we develop a fast pipeline based on normalizing flows that (i) learns a theory-informed prior on $(w_0,w_a)$ from KMIX realizations and (ii) provides an inverse mapping from CPL parameters back to the physical KMIX parameters. By importance-sampling pre-computed CPL chains using this framework, we effectively transform generic phenomenological constraints into direct, computationally efficient constraints on the underlying KMIX theory, avoiding the prohibitive cost of full parameter space exploration. Applied to Planck+DESI DR2 BAO measurements, our framework finds support for KMIX at $2.5σ$ compared to the base CPL fit at $3.1σ$, demonstrating that KMIX may account for the DESI preference without invoking true phantom behavior. When additionally including Type Ia supernovae data, we find that the preference remains above $3σ$ for Union3 and DES Y5, but drops to $2.1σ$ with Pantheon+. The latter, combined with the DESI full-shape power spectrum and bispectrum data, further reduces the preference to $1.7σ$. Ultimately, should the DESI deviation persist with future data, KMIX may offer a theoretically well-motivated explanation for the phantom-like signatures inferred from phenomenological fits.

astro-ph.CO↗

arXiv2025

Equilateral non-Gaussian Bias at the Field Level

Primordial non-Gaussianity (PNG) is a common prediction of a wide class of inflationary models. Equilateral-type PNG, generically predicted by single-field inflationary models with higher-derivative interactions, imprints subtle but measurable signatures on the large-scale distribution of matter. An important parameter of these imprints is the PNG-induced bias coefficient $b_ψ$, which quantifies how the abundance and clustering of dark matter halos and galaxies respond to mode coupling in the initial conditions. Measuring $b_ψ$ is important for constraining equilateral PNG, yet it is notoriously challenging due to its degeneracy with Gaussian scale-dependent bias contributions. In this work, we present the first precision measurements of equilateral $b_ψ$ for dark matter halos using effective field theory at the field level. We show that this approach disentangles PNG effects from those of the Gaussian bias by virtue of noise variance cancellation. We compare our results with the phenomenological predictions based on the Peak-Background Split model, finding some agreement at the qualitative level on the redshift and mass dependence, but poor agreement at the quantitative level. We present a fitting formula for $b_ψ$ as a function of the linear bias, which can be used to set priors in PNG searches with ongoing and future galaxy surveys.

astro-ph.CO↗

arXiv2025

The Compressed 3D Lyman-Alpha Forest Bispectrum

Cosmological studies of the Lyman-Alpha (Lya) forest typically constrain parameters using two-point statistics. However, higher-order statistics, such as the three-point function (or its Fourier counterpart, the bispectrum) offer additional information and help break the degeneracy between the mean flux and power spectrum amplitude, albeit at a significant computational cost. To address this, we extend an existing highly informative compression of the bispectrum, the skew spectra, to the Lya forest. We derive the tree-level bispectrum of Lya forest fluctuations in the framework of effective field theory (EFT) directly in redshift space and validate our methodology on synthetic Lya forest data. We measure the anisotropic cross-spectra between the transmitted flux fraction and all quadratic operators arising in the bispectrum, yielding a set of 26 skew spectra. Using idealized 3D Gaussian smoothing (R=10 Mpc/h), we find good agreement (1-2 sigma level based on the statistical errors of the mocks) with the theoretical tree-level bispectrum prediction for monopole and quadrupole up to k <= 0.17 h/Mpc. To enable the cosmological analysis of Lya forest data from the currently observing Dark Energy Spectroscopic Instrument (DESI), where we cannot do 3D smoothing, we use a line-of-sight smoothing and introduce a new statistic, the shifted skew spectra. These probe non-squeezed bispectrum triangles and avoid locally applying quadratic operators to the field by displacing one copy of the field in the radial direction. Using a fixed displacement of 40 Mpc/h (and line-of-sight smoothing of 10 Mpc/h) yields a similar agreement with the theory prediction. For the special case of correlating the squared (and displaced) field with the original one, we analytically forward model the window function making this approach readily applicable to DESI data.

astro-ph.CO↗

arXiv2025

Modeling the Cosmological Lyman-$α$ Forest at the Field Level

The distribution of absorption lines in the spectra of distant quasars, called the Lyman-$α$ (Ly-$α$) forest, is a unique probe of cosmology and the intergalactic medium at high redshifts and small scales. The statistical power of ongoing redshift surveys demands precise theoretical tools to model the Ly-$α$ forest. We address this challenge by developing an analytic, perturbative forward model to predict the Ly-$α$ forest at the field level for a given set of cosmological initial conditions. Our model shows a remarkable performance when compared with the Sherwood hydrodynamic simulations: it reproduces the flux distribution, the Ly-$α$ - dark matter halo cross-correlations, and the count-in-cell statistics at the percent level down to scales of a few Mpc. Our work provides crucial tools that bridge analytic modeling on large scales with simulations on small-scales, enabling field-level inference from Ly-$α$ forest data and simulation-based priors for cosmological analyses. This is especially timely for realizing the full scientific potential of the Ly-$α$ forest measurements by the Dark Energy Spectroscopic Instrument.

astro-ph.CO↗

arXiv2025

High-redshift Millennium and Astrid galaxies in effective field theory at the field level

Effective Field Theory (EFT) modeling is expected to be a useful tool in the era of future higher-redshift galaxy surveys such as DESI-II and Spec-S5 due to its robust description of various large-scale structure tracers. However, large values of EFT bias parameters of higher-redshift galaxies could jeopardize the convergence of the perturbative expansion. In this paper we measure the bias parameters and other EFT coefficients from samples of two types of star-forming galaxies in the state-of-the-art MilleniumTNG and Astrid hydrodynamical simulations. Our measurements are based on the field-level EFT forward model that allows for precision EFT parameter measurements by virtue of cosmic variance cancellation. Specifically, we consider approximately representative samples of Lyman-break galaxies (LBGs) and Lyman-alpha emitters (LAEs) that are consistent with the observed (angular) clustering and number density of these galaxies at $z=3$. Reproducing the linear biases and number densities observed from existing LAE and LBG data, we find quadratic bias parameters that are roughly consistent with those predicted from the halo model coupled with a simple halo occupation distribution model. We also find non-perturbative velocity contributions (Fingers of God) of a similar size for LBGs to the familiar case of Luminous Red Galaxies. However, these contributions are quite small for LAEs despite their large satellite fraction values of up to $\sim 30\%$. Our results indicate that the effective momentum reach $k_{\rm{Max}}$ at $z=3$ for LAEs (LBGs) will be in the range $0.3-0.6 ~h\rm{Mpc}^{-1}$ ($0.2-0.8~h\rm{Mpc}^{-1}$), suggesting that EFT will perform well for high redshift galaxy clustering. This work provides the first step toward obtaining realistic simulation-based priors on EFT parameters for LAEs and LBGs.

astro-ph.CO↗

arXiv2025

Local Primordial non-Gaussian Bias from Time Evolution

Primordial non-Gaussianity (PNG) is a signature of fundamental physics in the early universe that is probed by cosmological observations. It is well known that the local type of PNG generates a strong signal in the two-point function of large-scale structure tracers, such as galaxies. This signal, often termed ``scale-dependent bias'' is a generic feature of modulation of gravitational structure formation by a large-scale mode. It is less well-appreciated that the coefficient controlling this signal, $b_ϕ$, is closely connected to the time evolution of the tracer number density. This correspondence between time evolution and local PNG can be simply explained for a universal tracer whose mass function only depends on peak height, and more generally for non-universal tracers in the separate universe picture, which we validate in simulations. We also describe how to recover the bias of tracers subject to a survey selection function, and perform a simple demonstration on simulated galaxies. Since the local PNG amplitude in $n-$point statistics ($f_{\rm NL}$) is largely degenerate with the coefficient $b_ϕ$, this proof of concept study demonstrates that galaxy survey data can allow for more optimal and robust extraction of local PNG information from upcoming surveys.

astro-ph.CO↗

arXiv2025

Local Primordial Non-Gaussian Bias at the Field Level

Local primordial non-Gaussianity (LPNG) couples long-wavelength cosmological fluctuations to the short-wavelength behavior of galaxies. This coupling is encoded in bias parameters including $b_ϕ$ and $b_{δϕ}$ at linear and quadratic order in the large-scale biasing framework. We perform the first field-level measurement of $b_ϕ$ and $b_{δϕ}$ using Lagrangian bias and non-linear displacements from N-body simulations. We compare our field level measurements with universality predictions and separate universe results, finding qualitative consistency, but disagreement in detail. We also quantify the information on $f_{\mathrm{NL}}^{(\mathrm{loc})}$ available in the field given various assumptions on knowledge of $b_ϕ$ at fixed initial conditions. We find that it is not possible to precisely constrain $f_{\mathrm{NL}}^{(\mathrm{loc})}$ when marginalizing over $b_ϕ f_{\mathrm{NL}}^{(\mathrm{loc})}$ even at the field level, observing a 2-3X degradation in constraints between a linear and quadratic biasing model on perturbative field-level mocks, suggesting that a $b_ϕ$ prior is necessary to meaningfully constrain $f_{\mathrm{NL}}^{(\mathrm{loc})}$ at the field level even in this idealized scenario. For simulated dark matter halos, the pure $f_{\mathrm{NL}}^{(\mathrm{loc})}$ constraints from both linear and quadratic field-level models appear biased when marginalizing over bias parameters including $b_ϕ$ and $b_{δϕ}$ due largely to the $f_{\mathrm{NL}}^{(\mathrm{loc})} - b_ϕ$ degeneracy. Our results are an important consistency test of the large-scale bias framework for LPNG and highlight the importance of physically motivated priors on LPNG bias parameters for future surveys.

astro-ph.CO↗

arXiv2024

Emulating Recombination with Neural Networks using Universal Differential Equations

With an aim towards modeling cosmologies beyond the $Λ$CDM paradigm, we demonstrate the automatic construction of recombination history emulators while enforcing a prior of causal dynamics. These methods are particularly useful in the current era of precision cosmology, where extremely constraining datasets provide insights into a cosmological model dominated by unknown contents. Cosmic Microwave Background (CMB) data in particular provide a clean glimpse into the interaction of dark matter, baryons, and radiation in the early Universe, but interpretation of this data requires knowledge of the Universe's ionization history. The exploration of new physics with new CMB data will require fast and flexible calculation of this ionization history. We develop a differentiable machine learning model for recombination physics using a neural network ordinary differential equation architecture (Universal Differential Equations, UDEs), building towards automatic dimensionality reduction and the avoidance of manual tuning based on cosmological model.

astro-ph.CO↗

arXiv2024

An Analytic Hybrid Halo + Perturbation Theory Model for Small-scale Correlators: Baryons, Halos, and Galaxies

We update Halo Zeldovich Perturbation Theory (HZPT), an analytic model for the two-point statistics of dark matter, to describe halo and galaxy clustering, and galaxy-matter cross-correlation on nonlinear scales. The model correcting Zeldovich has an analytic Fourier transform, and therefore is valid in both configuration space and Fourier space. The model is accurate at the $2\%$-level or less for $P_{mm}$ (k < 1 h/Mpc), $P_{hm}$ (k < 1 h/Mpc), $P_{hh}$ (k < 2 h/Mpc), $P_{gm}$ (k < 1 h/Mpc), $P_{gg}$ (k < 1 h/Mpc), $ξ_{mm}$ (r > 1 Mpc/h), $ξ_{hm}$ (r > 2 Mpc/h), $ξ_{hh}$ (r > 2 Mpc/h), $ξ_{gm}$ (r > 1 Mpc/h), $ξ_{gg}$ (r > 2 Mpc/h), for LRG-like mock galaxies. We show that the HZPT model for matter correlators can account for the effects of a wide range of baryonic feedback models and provide extended dark matter models which are of $1\% ~(3\%)$ accuracy for k < 10 (8) h/Mpc. We explicitly model the non-perturbative features of halo exclusion for the halo-halo and galaxy-galaxy correlators, as well as the presence of satellites for galaxy-matter and galaxy-galaxy correlation functions. We perform density estimation using N-body simulations and a wide range of HOD galaxy mocks to obtain correlations of model parameters with the cosmological parameters $Ω_{m}$ and $σ_{8}$. HZPT can provide a fast, interpretable, and analytic model for combined-probe analyses of redshift surveys using scales well into the non-linear regime.

astro-ph.CO↗

arXiv2024

Improving initialization and evolution accuracy of cosmological neutrino simulations

Neutrino mass constraints are a primary focus of current and future large-scale structure (LSS) surveys. Non-linear LSS models rely heavily on cosmological simulations -- the impact of massive neutrinos should therefore be included in these simulations in a realistic, computationally tractable, and controlled manner. A recent proposal to reduce the related computational cost employs a symmetric neutrino momentum sampling strategy in the initial conditions. We implement a modified version of this strategy into the Hardware/Hybrid Accelerated Cosmology Code (HACC) and perform convergence tests on its internal parameters. We illustrate that this method can impart $\mathcal{O}(1\%)$ numerical artifacts on the total matter field on small scales, similar to previous findings, and present a method to remove these artifacts using Fourier-space filtering of the neutrino density field. Moreover, we show that the converged neutrino power spectrum does not follow linear theory predictions on relatively large scales at early times at the $15\%$ level, prompting a more careful study of systematics in particle-based neutrino simulations. We also present an improved method for backscaling linear transfer functions for initial conditions in massive neutrino cosmologies that is based on achieving the same relative neutrino growth as computed with Boltzmann solvers. Our self-consistent backscaling method yields sub-percent accuracy in the total matter growth function. Comparisons for the non-linear power spectrum with the Mira-Titan emulator at a neutrino mass of $m_ν=0.15~\mathrm{eV}$ are in very good agreement with the expected level of errors in the emulator and in the direct N-body simulation.

astro-ph.CO↗

arXiv2023

Deterministic Langevin Unconstrained Optimization with Normalizing Flows

We introduce a global, gradient-free surrogate optimization strategy for expensive black-box functions inspired by the Fokker-Planck and Langevin equations. These can be written as an optimization problem where the objective is the target function to maximize minus the logarithm of the current density of evaluated samples. This objective balances exploitation of the target objective with exploration of low-density regions. The method, Deterministic Langevin Optimization (DLO), relies on a Normalizing Flow density estimate to perform active learning and select proposal points for evaluation. This strategy differs qualitatively from the widely-used acquisition functions employed by Bayesian Optimization methods, and can accommodate a range of surrogate choices. We demonstrate superior or competitive progress toward objective optima on standard synthetic test functions, as well as on non-convex and multi-modal posteriors of moderate dimension. On real-world objectives, such as scientific and neural network hyperparameter optimization, DLO is competitive with state-of-the-art baselines.

cs.LG↗