Simulation-Based Priors without Simulations: an Analytic Perspective on EFT Parameters of Galaxies
Mikhail M. Ivanov · 2025
Read Simulation-Based Priors without Simulations: an Analytic Perspective on EFT Parameters of Galaxies by Mikhail M. Ivanov (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-07270.txt.
Introduction
Mikhail Ivanov shows, in twenty twenty-five, how to work out simulation-based priors for galaxy clustering without running simulations. Fits with effective field theory gain much strength from priors drawn from mock catalogs. Those mocks are most often built with a halo occupation distribution. That is a recipe that puts galaxies into dark matter halos. Ivanov shows that for the standard seven-parameter recipe, the priors can be worked out by formula. The key point is that halo settings rest only on peak height, a gauge of how rare a halo is.
He draws formulas for the fixed settings from the halo model. The random settings need a split plan based on physics, since the halo model fails there. The formula-based priors match simulations very well. They cut the time to work them out to near zero.
Catalog record
| CREATOR | Mikhail M. Ivanov |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 72 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-07270
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Simulation-Based Priors without Simulations: an Analytic Perspective on EFT Parameters of Galaxies Mikhail M. Ivanov Abstract Effective field theory (EFT)-based full-shape analysis with simulation-based priors (SBPs) is a novel approach to galaxy clustering data analysis, which significantly boosts the constraining power by efficiently incorporating field-level simulation information from small scales. So far, SBPs have been mostly extracted from a large set of mock catalogs generated with the Halo Occupation Distribution (HOD) approach. We show that given a halo mass function model and assuming that the EFT parameters of halos depend only on the peak height, HOD-based priors can be computed analytically for the standard 7-parameter HOD model. We derive the relevant expressions for deterministic EFT parameters from the halo model. The halo model, however, fails to accurately describe stochastic EFT parameters, for which we use a physically motivated phenomenological prescription. We compare our analytic priors with simulations and find an excellent agreement. Our approach provides analytic insights into the physics behind the EFT parameters of galaxies, and allows one to reduce the computation time of SBPs to virtually nothing. As an application, we produce a set of 100,000 EFT parameters by sampling both HOD and cosmological models, and explicitly demonstrate that any cosmology dependence can be completely absorbed by relatively minor shifts of the HOD parameters. Our approach can be generalized to other variants of the HOD and cosmological models beyond Λ \Lambda CDM. Preprint: MIT-CTP/5851 1 Introduction Effective field theory (EFT) for Large-Scale Structure [ 1 , 2 , 3 ] is a systematic analytic approach to structure formation that provides unmatched accuracy and flexibility for quasi-linear clustering observables. It has been successfully applied to various kinds of cosmological data in the perturbative regime, e.g. galaxy clustering in redshift-space ( [ 4 , 5 , 6 , 7 , 8 ] , see also [ 9 ] for the recent application of this technique by the Dark Energy Survey Instrument collaboration), projected galaxy clustering [ 10 ] , and the Lyman- \alpha forest [ 11 , 12 , 13 ] . One important feature of EFT is the presence of free EFT parameters (a.k.a. “nuisance parameters”), which are not calculable within the EFT and need to be fitted from data or numerical simulations. The standard approach is to determine the EFT parameters from the data itself. While this approach produces conservative results (provided that the priors on EFT parameters are chosen to be appropriately wide), it leads to a significant degradation of constraining power, see [ 14 , 15 , 15 ] for detailed discussions. 1 1 1 Note that effect is different from the so-called prior projection effects, originally studied in [ 4 , 16 , 7 ] , see also [ 17 , 18 ] for recent discussions. The simulation-based priors eliminate these projection effects [ 19 ] . In contrast, matching EFT parameters from the numerical simulations leads to a dramatic improvement of constraining power [ 20 , 21 , 19 ] . This approach is known as the EFT-based full-shape analysis with the simulation-based priors (SBPs). (See also [ 22 , 23 ] for similar ideas in the context of other clustering models, and more recent works [ 24 , 25 ] .) The significant information gain can be achieved if the SBPs are calibrated at the field level [ 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 ] , which allows one to break the degeneracies present in the correlation functions [ 19 ] . In addition, the EFT-based full-shape analysis with SBPs allows one to efficiently incorporate small scale information beyond the commonly used two-point clustering observables. Until now, SBPs were calibrated using a combination of N-body simulations and the Halo Occupation Distribution (HOD) framework [ 35 , 36 ] to map galaxies onto the N-body simulation output (see [ 37 ] for a review). Ref. [ 38 ] extended this to hydrodynamical simulations. Given a dependence of SBPs on computationally expensive simulations, it is desirable to develop an analytic approach to SBPs, which could reduce their computational cost. In addition, such an approach should provide insights into the physics behind the EFT parameters, increasing the interpretability of the EFT-SBP results. This is especially relevant in light of the nominal power of this approach in constraining cosmological parameters [ 21 , 19 ] . We address these questions in our note. Our main aim is to use the halo model formalism [ 39 , 40 , 41 ] to predict the relations between the EFT parameters of dark matter halos and galaxies. These relations are well known for the galaxy bias parameters [ 42 ] . In this work we critically revisit the derivation of these expressions, and generalize these calculations to the case of redshift-space and stochasticity EFT parameters. Let us briefly introduce the key ingredients of our analytic approach using the known example of the galaxy bias parameters. Galaxy and halo bias are functions encapsulating the dependence of the galaxy and halo overdensity \delta_{h/g} on the underlying dark matter field \delta_{m} (see [ 43 ] for a review). The bias expansion is formulated in terms of operators \mathcal{O}_{a}(\bm{x}) built out of the dark matter observables such as the tidal field. Schematically, the galaxy and halo bias expressions can be written as \begin{split}&\delta_{h}(\bm{x})=\sum_{a=1}b_{\mathcal{O}_{a}}^{h}\mathcal{O}_{a}(\bm{x})\,,\quad\delta_{g}(\bm{x})=\sum_{a=1}b_{\mathcal{O}_{a}}^{g}\mathcal{O}_{a}(\bm{x})\,,\end{split} (1) where we have omitted the explicit time-dependence for brevity. The halo model predicts that the bias parameters of galaxies b_{\mathcal{O}_{a}}^{g} are given by the bias parameters of halos weighted with the halo mass function \bar{n}(M) , and the mean halo occupation distribution \langle N_{g}\rangle(M) , b_{\mathcal{O}_{a}}^{g}=\underbrace{\int d\mathop{\rm ln}\nolimits M\bar{n}(M)}_{\text{halo mass function}}~\underbrace{\langle N_{g}\rangle(M)}_{\text{galaxy-DM connection}}~\underbrace{b_{\mathcal{O}_{a}}^{h}(M)}_{\text{Halo bias}}~\,, (2) where the integral above is over the halo mass M . If we were to compute the bias parameters analytically, we would need three key ingredients. The first ingredient is the halo mass function. Thanks to decades of intense theoretical efforts, many analytic results are readily available in the literature [ 44 , 45 , 46 ] . The second ingredient is the halo occupation distribution \langle N_{g}\rangle(M) which captures the galaxy-dark matter connection. The basic theoretical HOD models such as [ 36 ] provide analytic formulas for \langle N_{g}\rangle(M) , which can be used to do the integral ( 2 ) numerically. Finally, the last ingredient in the above formula is the bias parameter of halos as a function of the halo mass, b_{\mathcal{O}_{a}}^{h}(M) , or a related quantity of the halo model known as the “peak height” [ 39 ] . Since there are no available analytic results for all of the relevant EFT parameters for dark matter halos, in practice one has to extract b_{\mathcal{O}_{a}}^{h}(M) from N-body simulations. However, these dependencies have to be calibrated from the simulations only once. All in all, with the three ingredients described above, the halo mass function model, the analytic expression for the HOD function, and the halo bias as a function of the halo mass, one can compute the HOD-based galaxy bias parameters analytically. Our goal here is to carry out this computation explicitly, and extend this argument to other EFT parameters. Finally, we will compare the results against numerical simulations. The analytic approach