arXiv subjects

Arya Farahi

Publications and source records attributed to Arya Farahi.

At least 37 records · Page 2Linked to original sources

arXiv2025

I-trustworthy Models. A framework for trustworthiness evaluation of probabilistic classifiers

As probabilistic models continue to permeate various facets of our society and contribute to scientific advancements, it becomes a necessity to go beyond traditional metrics such as predictive accuracy and error rates and assess their trustworthiness. Grounded in the competence-based theory of trust, this work formalizes I-trustworthy framework -- a novel framework for assessing the trustworthiness of probabilistic classifiers for inference tasks by linking local calibration to trustworthiness. To assess I-trustworthiness, we use the local calibration error (LCE) and develop a method of hypothesis-testing. This method utilizes a kernel-based test statistic, Kernel Local Calibration Error (KLCE), to test local calibration of a probabilistic classifier. This study provides theoretical guarantees by offering convergence bounds for an unbiased estimator of KLCE. Additionally, we present a diagnostic tool designed to identify and measure biases in cases of miscalibration. The effectiveness of the proposed test statistic is demonstrated through its application to both simulated and real-world datasets. Finally, LCE of related recalibration methods is studied, and we provide evidence of insufficiency of existing methods to achieve I-trustworthiness.

stat.ML↗

arXiv2025

Baryon Pasting the Uchuu Lightcone Simulation

We present the Baryon Pasted (BP) X-ray and thermal Sunyaev-Zel'dovich (tSZ) maps derived from the half-sky Uchuu Lightcone simulation. These BP-Uchuu maps are constructed using more than $75$ million dark matter halos with masses $M_{500c} \geq 10^{13} M_\odot$ within the redshift range $0 \leq z \leq 2$. A distinctive feature of our BP-Uchuu Lightcone maps is their capability to assess the influence of both extrinsic and intrinsic scatter caused by triaxial gaseous halos and internal gas characteristics, respectively, at the map level. We show that triaxial gas drives substantial scatter in X-ray luminosities of clusters and groups, accounting for nearly half of the total scatter in core-excised measurements. Additionally, scatter in the thermal pressure and gas density profiles of halos enhances the X-ray and SZ power spectra, leading to biases in cosmological parameter estimates. These findings are statistically robust due to the extensive sky coverage and large halo sample in the BP-Uchuu maps. The BP-Uchuu maps are publicly available on https://app.globus.org/file-manager?origin_id=cf8dadb7-b6e9-4e2c-abc1-0813877efc13 .

astro-ph.CO↗

arXiv2024

Masses of Sunyaev-Zel'dovich Galaxy Clusters Detected by The Atacama Cosmology Telescope: Stacked Lensing Measurements with Subaru HSC Year 3 data

We present a stacked lensing analysis of 96 galaxy clusters selected by the thermal Sunyaev-Zel'dovich (SZ) effect in maps of the cosmic microwave background (CMB). We select foreground galaxy clusters with a $5σ$-level SZ threshold in CMB observations from the Atacama Cosmology Telescope, while we define background source galaxies for the lensing analysis with secure photometric redshift cuts in Year 3 data of the Subaru Hyper Suprime Cam survey. We detect the stacked lensing signal in the range of $0.1 < R\, [h^{-1}\mathrm{Mpc}] < 100$ in each of three cluster redshift bins, $0.092<z\le0.445$, $0.445<z\le0.695$, and $0.695<z\le1.180$, with 32 galaxy clusters in each bin. The cumulative signal-to-noise ratios of the lensing signal are $14.6$, $12.0$, and $6.6$, respectively. Using a halo-based forward model, we then constrain statistical relationships between the mass inferred from the SZ observation (i.e. SZ mass) and the total mass derived from our stacked lensing measurements. At the average SZ mass in the cluster sample ($2.1-2.4\times10^{14}\, h^{-1}M_\odot$), our likelihood analysis shows that the average total mass differs from the SZ counterpart by a factor of $1.3 \pm 0.2$, $1.6 \pm 0.2$, and $1.6 \pm 0.3$ ($68\%$) in the aforementioned redshift ranges, respectively. Our limits are consistent with previous lensing measurements, and we find that the cluster modeling choices can introduce a $1σ$-level difference in our parameter inferences.

astro-ph.CO↗

arXiv2024

How DREAMS are made: Emulating Satellite Galaxy and Subhalo Populations with Diffusion Models and Point Clouds

The connection between galaxies and their host dark matter (DM) halos is critical to our understanding of cosmology, galaxy formation, and DM physics. To maximize the return of upcoming cosmological surveys, we need an accurate way to model this complex relationship. Many techniques have been developed to model this connection, from Halo Occupation Distribution (HOD) to empirical and semi-analytic models to hydrodynamic. Hydrodynamic simulations can incorporate more detailed astrophysical processes but are computationally expensive; HODs, on the other hand, are computationally cheap but have limited accuracy. In this work, we present NeHOD, a generative framework based on variational diffusion model and Transformer, for painting galaxies/subhalos on top of DM with an accuracy of hydrodynamic simulations but at a computational cost similar to HOD. By modeling galaxies/subhalos as point clouds, instead of binning or voxelization, we can resolve small spatial scales down to the resolution of the simulations. For each halo, NeHOD predicts the positions, velocities, masses, and concentrations of its central and satellite galaxies. We train NeHOD on the TNG-Warm DM suite of the DREAMS project, which consists of 1024 high-resolution zoom-in hydrodynamic simulations of Milky Way-mass halos with varying warm DM mass and astrophysical parameters. We show that our model captures the complex relationships between subhalo properties as a function of the simulation parameters, including the mass functions, stellar-halo mass relations, concentration-mass relations, and spatial clustering. Our method can be used for a large variety of downstream applications, from galaxy clustering to strong lensing studies.

astro-ph.GA↗

arXiv2024

Impact of Property Covariance on Cluster Weak lensing Scaling Relations

We present an investigation into a hitherto unexplored systematic that affects the accuracy of galaxy cluster mass estimates with weak gravitational lensing. Specifically, we study the covariance between the weak lensing signal, $ΔΣ$, and the "true" cluster galaxy number count, $N_{\rm gal}$, as measured within a spherical volume that is void of projection effects. By quantifying the impact of this covariance on mass calibration, this work reveals a significant source of systematic uncertainty. Using the MDPL2 simulation with galaxies traced by the SAGE semi-analytic model, we measure the intrinsic property covariance between these observables within the 3D vicinity of the cluster, spanning a range of dynamical mass and redshift values relevant for optical cluster surveys. Our results reveal a negative covariance at small radial scales ($R \lesssim R_{\rm 200c}$) and a null covariance at large scales ($R \gtrsim R_{\rm 200c}$) across most mass and redshift bins. We also find that this covariance results in a $2-3\%$ bias in the halo mass estimates in most bins. Furthermore, by modeling $N_{\rm gal}$ and $ΔΣ$ as multi-(log)-linear equations of secondary halo properties, we provide a quantitative explanation for the physical origin of the negative covariance at small scales. Specifically, we demonstrate that the $N_{\rm gal}$-$ΔΣ$ covariance can be explained by the secondary properties of halos that probe their formation history. We attribute the difference between our results and the positive bias seen in other works with (mock)-cluster finders to projection effects. These findings highlight the importance of accounting for the covariance between observables in cluster mass estimation, which is crucial for obtaining accurate constraints on cosmological parameters.

astro-ph.CO↗

arXiv2024

Introducing the DREAMS Project: DaRk mattEr and Astrophysics with Machine learning and Simulations

We introduce the DREAMS project, an innovative approach to understanding the astrophysical implications of alternative dark matter models and their effects on galaxy formation and evolution. The DREAMS project will ultimately comprise thousands of cosmological hydrodynamic simulations that simultaneously vary over dark matter physics, astrophysics, and cosmology in modeling a range of systems -- from galaxy clusters to ultra-faint satellites. Such extensive simulation suites can provide adequate training sets for machine-learning-based analyses. This paper introduces two new cosmological hydrodynamical suites of Warm Dark Matter, each comprised of 1024 simulations generated using the Arepo code. One suite consists of uniform-box simulations covering a $(25~h^{-1}~{\rm M}_\odot)^3$ volume, while the other consists of Milky Way zoom-ins with sufficient resolution to capture the properties of classical satellites. For each simulation, the Warm Dark Matter particle mass is varied along with the initial density field and several parameters controlling the strength of baryonic feedback within the IllustrisTNG model. We provide two examples, separately utilizing emulators and Convolutional Neural Networks, to demonstrate how such simulation suites can be used to disentangle the effects of dark matter and baryonic physics on galactic properties. The DREAMS project can be extended further to include different dark matter models, galaxy formation physics, and astrophysical targets. In this way, it will provide an unparalleled opportunity to characterize uncertainties on predictions for small-scale observables, leading to robust predictions for testing the particle physics nature of dark matter on these scales.

astro-ph.GA↗

arXiv2024

Can we constrain warm dark matter masses with individual galaxies?

We study the impact of warm dark matter mass on the internal properties of individual galaxies using a large suite of 1,024 state-of-the-art cosmological hydrodynamic simulations from the DREAMS project. We take individual galaxies' properties from the simulations, which have different cosmologies, astrophysics, and warm dark matter masses, and train normalizing flows to learn the posterior of the parameters. We find that our models cannot infer the value of the warm dark matter mass, even when the values of the cosmological and astrophysical parameters are given explicitly. This result holds for galaxies with stellar mass larger than $2\times10^8 M_\odot/h$ at both low and high redshifts. We calculate the mutual information and find no significant dependence between the WDM mass and galaxy properties. On the other hand, our models can infer the value of $Ω_{\rm m}$ with a $\sim10\%$ accuracy from the properties of individual galaxies while marginalizing astrophysics and warm dark matter masses.

astro-ph.CO↗

arXiv2024

U-Trustworthy Models.Reliability, Competence, and Confidence in Decision-Making

With growing concerns regarding bias and discrimination in predictive models, the AI community has increasingly focused on assessing AI system trustworthiness. Conventionally, trustworthy AI literature relies on the probabilistic framework and calibration as prerequisites for trustworthiness. In this work, we depart from this viewpoint by proposing a novel trust framework inspired by the philosophy literature on trust. We present a precise mathematical definition of trustworthiness, termed $\mathcal{U}$-trustworthiness, specifically tailored for a subset of tasks aimed at maximizing a utility function. We argue that a model's $\mathcal{U}$-trustworthiness is contingent upon its ability to maximize Bayes utility within this task subset. Our first set of results challenges the probabilistic framework by demonstrating its potential to favor less trustworthy models and introduce the risk of misleading trustworthiness assessments. Within the context of $\mathcal{U}$-trustworthiness, we prove that properly-ranked models are inherently $\mathcal{U}$-trustworthy. Furthermore, we advocate for the adoption of the AUC metric as the preferred measure of trustworthiness. By offering both theoretical guarantees and experimental validation, AUC enables robust evaluation of trustworthiness, thereby enhancing model selection and hyperparameter tuning to yield more trustworthy outcomes.

stat.ML↗

arXiv2023

C2-GaMe: Classification of Cluster Galaxy Membership with Machine Learning

We present Classification of Cluster GAlaxy MEmbers (C$^2$-GaMe), a classification algorithm based on a suite of machine learning models that differentiates galaxies into orbiting, infalling, and background (interloper) populations, using phase space information as input. We train and test C$^2$-GaMe with the galaxies from UniverseMachine mock catalog based on Multi-Dark Planck 2 N-body simulations. We show that probabilistic classification is superior to deterministic classification in estimating the physical properties of clusters, including density profiles and velocity dispersion. We propose a set of estimators to get an unbiased estimation of cluster properties. We demonstrate that C$^2$-GaMe can recover the distribution of orbiting and infalling galaxies' position and velocity distribution with $<1\%$ statistical error when using probabilistic predictions in the presence of interlopers in the projected phase space. Additionally, we demonstrate the robustness of trained models by applying them to a different simulation. Finally, adding a specific star formation rate and the ratio of the galaxy's halo mass to the cluster's halo mass as additional features improves the classification performance. We discuss potential applications of this technique to enhance cluster cosmology and galaxy quenching.

astro-ph.CO↗

arXiv2023

Benchmarks and Explanations for Deep Learning Estimates of X-ray Galaxy Cluster Masses

We evaluate the effectiveness of deep learning (DL) models for reconstructing the masses of galaxy clusters using X-ray photometry data from next-generation surveys. We establish these constraints using a catalogue of realistic mock eROSITA X-ray observations which use hydrodynamical simulations to model realistic cluster morphology, background emission, telescope response, and AGN sources. Using bolometric X-ray photon maps as input, DL models achieve a predictive mass scatter of $σ_{\ln M_\mathrm{500c}} = 17.8\%$, a factor of two improvements on scalar observables such as richness $N_\mathrm{gal}$, 1D velocity dispersion $σ_\mathrm{v,1D}$, and photon count $N_\mathrm{phot}$ as well as a $32\%$ improvement upon idealised, volume-integrated measurements of the bolometric X-ray luminosity $L_X$. We then show that extending this model to handle multichannel X-ray photon maps, separated in low, medium, and high energy bands, further reduces the mass scatter to $16.2\%$. We also tested a multimodal DL model incorporating both dynamical and X-ray cluster probes and achieved marginal gains at a mass scatter of $15.9\%$. Finally, we conduct a quantitative interpretability study of our DL models and find that they greatly down-weight the importance of pixels in the centres of clusters and at the location of AGN sources, validating previous claims of DL modelling improvements and suggesting practical and theoretical benefits for using DL in X-ray mass inference.

astro-ph.CO↗

arXiv2023

Fire and Smoke Digital Twin -- A computational framework for modeling fire incident outcomes

Fires and burning are the chief causes of particulate matter (PM2.5), a key measurement of air quality in communities and cities worldwide. This work develops a live fire tracking platform to show active reported fires from over twenty cities in the U.S., as well as predict their smoke paths and impacts on the air quality of regions within their range. Specifically, our close to real-time tracking and predictions culminates in a digital twin to protect public health and inform the public of fire and air quality risk. This tool tracks fire incidents in real-time, utilizes the 3D building footprints of Austin to simulate smoke outputs, and predicts fire incident smoke falloffs within the complex city environment. Results from this study include a complete fire and smoke digital twin model for Austin. We work in cooperation with the City of Austin Fire Department to ensure the accuracy of our forecast and also show that air quality sensor density within our cities cannot validate urban fire presence. We additionally release code and methodology to replicate these results for any city in the world. This work paves the path for similar digital twin models to be developed and deployed to better protect the health and safety of citizens.

cs.CY↗

arXiv2022

A Machine Learning Approach to Enhancing eROSITA Observations

The eROSITA X-ray telescope, launched in 2019, is predicted to observe roughly 100,000 galaxy clusters. Follow-up observations of these clusters from Chandra, for example, will be needed to resolve outstanding questions about galaxy cluster physics. Deep Chandra cluster observations are expensive and follow-up of every eROSITA cluster is infeasible, therefore, objects chosen for follow-up must be chosen with care. To address this, we have developed an algorithm for predicting longer duration, background-free observations based on mock eROSITA observations. We make use of the hydrodynamic cosmological simulation Magneticum, have simulated eROSITA instrument conditions using SIXTE, and have applied a novel convolutional neural network to output a deep Chandra-like "super observation" of each cluster in our simulation sample. Any follow-up merit assessment tool should be designed with a specific use case in mind; our model produces observations that accurately and precisely reproduce the cluster morphology, which is a critical ingredient for determining cluster dynamical state and core type. Our model will advance our understanding of galaxy clusters by improving follow-up selection and demonstrates that image-to-image deep learning algorithms are a viable method for simulating realistic follow-up observations.

astro-ph.CO↗

arXiv2022

Optical selection bias and projection effects in stacked galaxy cluster weak lensing

Cosmological constraints from current and upcoming galaxy cluster surveys are limited by the accuracy of cluster mass calibration. In particular, optically identified galaxy clusters are prone to selection effects that can bias the weak lensing mass calibration. We investigate the selection bias of the stacked cluster lensing signal associated with optically selected clusters, using clusters identified by the redMaPPer algorithm in the Buzzard simulations as a case study. We find that at a given cluster halo mass, the residuals of redMaPPer richness and weak lensing signal are positively correlated. As a result, for a given richness selection, the stacked lensing signal is biased high compared with what we would expect from the underlying halo mass probability distribution. The cluster lensing selection bias can thus lead to overestimated mean cluster mass and biased cosmology results. We show that the lensing selection bias exhibits a strong scale-dependence and is approximately 20 to 60 percent for $ΔΣ$ at large scales. This selection bias largely originates from spurious member galaxies within +/- 20 to 60 Mpc/h along the line of sight, highlighting the importance of quantifying projection effects associated with the broad redshift distribution of member galaxies in photometric cluster surveys. While our results qualitatively agree with those in the literature, accurate quantitative modelling of the selection bias is needed to achieve the goals of cluster lensing cosmology and will require synthetic catalogues covering a wide range of galaxy-halo connection models.

astro-ph.CO↗

arXiv2022

Correlations of Dark Matter, Gas and Stellar Profiles in Dark Matter Halos

Halos of similar mass and redshift exhibit a large degree of variability in their differential properties, such as dark matter, hot gas, and stellar mass density profiles. This variability is an indicator of diversity in the formation history of these dark matter halos that is reflected in the coupling of scatters about the mean relations. In this work, we show that the strength of this coupling depends on the scale at which halo profiles are measured. By analyzing the outputs of the IllustrisTNG hydrodynamical cosmological simulations we report the radial- and mass-dependent couplings between the dark matter, hot gas, and stellar mass radial density profiles utilizing the population diversity in dark matter halos. We find that for the same mass halos the scatters in the density of baryons and dark matter are strongly coupled at large scales ($r>R_{200}$); but the coupling between gas and dark matter density profiles fades near the core of halos ($r < 0.3 R_{200}$). We then show that the correlation between halo profile and integrated quantities induces a radius-dependent additive bias in the profile observables of halos when halos are selected on properties other than their mass. We discuss the impact of this effect on cluster abundance and cross-correlations cosmology with multi-wavelength cosmological surveys.

astro-ph.CO↗

arXiv2022

KLLR: A scale-dependent, multivariate model class for regression analysis

The underlying physics of astronomical systems governs the relation between their measurable properties. Consequently, quantifying the statistical relationships between system-level observable properties of a population offers insights into the astrophysical drivers of that class of systems. While purely linear models capture behavior over a limited range of system scale, the fact that astrophysics is ultimately scale-dependent implies the need for a more flexible approach to describing population statistics over a wide dynamic range. For such applications, we introduce and implement a class of Kernel-Localized Linear Regression (KLLR) models. KLLR is a natural extension to the commonly-used linear models that allows the parameters of the linear model -- normalization, slope, and covariance matrix -- to be scale-dependent. KLLR performs inference in two steps: (1) it estimates the mean relation between a set of independent variables and a dependent variable and; (2) it estimates the conditional covariance of the dependent variables given a set of independent variables. We demonstrate the model's performance in a simulated setting and showcase an application of the proposed model in analyzing the baryonic content of dark matter halos. As a part of this work, we publicly release a Python implementation of the KLLR method.

astro-ph.IM↗

arXiv2022

HSC-XXL : Baryon budget of the 136 XXL Groups and Clusters

We present our determination of the baryon budget for an X-ray-selected XXL sample of 136 galaxy groups and clusters spanning nearly two orders of magnitude in mass ($M_{500}\sim 10^{13}-10^{15}M_\odot$) and the redshift range $0< z < 1$. Our joint analysis is based on the combination of HSC-SSP weak-lensing mass measurements, XXL X-ray gas mass measurements, and HSC and SDSS multiband photometry. We carry out a Bayesian analysis of multivariate mass-scaling relations of gas mass, galaxy stellar mass, stellar mass of brightest cluster galaxies (BCGs), and soft-band X-ray luminosity, by taking into account the intrinsic covariance between cluster properties, selection effect, weak-lensing mass calibration, and observational error covariance matrix. The mass-dependent slope of the gas mass--total mass ($M_{500}$) relation is found to be $1.29_{-0.10}^{+0.16}$, which is steeper than the self-similar prediction of unity, whereas the slope of the stellar mass--total mass relation is shallower than unity, $0.85_{-0.09}^{+0.12}$. The BCG stellar mass weakly depends on cluster mass with a slope of $0.49_{-0.10}^{+0.11}$. The baryon, gas mass, and stellar mass fractions as a function of $M_{500}$ agree with the results from numerical simulations and previous observations. We successfully constrain the full intrinsic covariance of the baryonic contents. The BCG stellar mass shows the larger intrinsic scatter at a given halo total mass, followed in order by stellar mass and gas mass. We find a significant positive intrinsic correlation coefficient between total (and satellite) stellar mass and BCG stellar mass and no evidence for intrinsic correlation between gas mass and stellar mass. All the baryonic components show no redshift evolution.

astro-ph.CO↗

arXiv2021

Galaxy Velocity Bias in Cosmological Simulations: Towards Percent-level Calibration

Galaxy cluster masses, rich with cosmological information, can be estimated from internal dark matter (DM) velocity dispersions, which in turn can be observationally inferred from satellite galaxy velocities. However, galaxies are biased tracers of the DM, and the bias can vary over host halo and galaxy properties as well as time. We precisely calibrate the velocity bias, b_v -- defined as the ratio of galaxy and DM velocity dispersions -- as a function of redshift, host halo mass, and galaxy stellar mass threshold (Mstarsat), for massive halos (M200c > 1e13.5 msun) from five cosmological simulations: IllustrisTNG, Magneticum, Bahamas + Macsis, The Three Hundred Project, and MultiDark Planck-2. We first compare scaling relations for galaxy and DM velocity dispersion across simulations; the former is estimated using a new ensemble velocity likelihood method that is unbiased for low galaxy counts per halo, while the latter uses a local linear regression. The simulations show consistent trends of b_v increasing with M200c and decreasing with redshift and Mstarsat. The ensemble-estimated theoretical uncertainty in b_v is 2-3% but becomes percent-level when considering only the three highest resolution simulations. We update the mass-richness normalization previously estimated by Farahi et al. (2016) for an SDSS redMaPPer cluster sample. The improved accuracy of our b_v estimates reduces the mass normalization uncertainty from 22% to 8%, demonstrating that dynamical estimation techniques can be competitive with weak lensing in calibrating population mean masses. We discuss necessary steps for further improving this precision. Our estimates for b_v(M200c, Mstarsat, z) are made publicly available.

astro-ph.CO↗

arXiv2021

Baryonic Imprints on DM Halos: Population Statistics from Dwarf Galaxies to Galaxy Clusters

In a purely cold dark matter universe, the initial matter power spectrum and its subsequent gravitational growth contain no special mass- or time-scales, and so neither do the emergent population statistics of internal dark matter (DM) halo properties. Using 1.5 million halos from three IllustrisTNG realizations of a LambdaCDM universe, we show that galaxy formation physics drives non-monotonic features ("wiggles") into DM property statistics across six decades in halo mass, from dwarf galaxies to galaxy clusters. We characterize these features by extracting the halo mass-dependent statistics of five DM halo properties -- velocity dispersion, NFW concentration, density- and velocity-space shapes, and formation time -- using kernel-localized linear regression (KLLR). Comparing precise estimates of normalizations, slopes, and covariances between realizations with and without galaxy formation, we find systematic deviations across all mass-scales, with maximum deviations of 25% at the Milky-Way mass of 1e12 Msun. The mass-dependence of the wiggles is set by the interplay between different cooling and feedback mechanisms, and we discuss its observational implications. The property covariances depend strongly on halo mass and physics treatment, but the correlations are mostly robust. Using multivariate KLLR and interpretable machine learning, we show the halo concentration and velocity-space shape are principal contributors, at different mass, to the velocity dispersion variance. Statistics of mass accretion rate and DM surface pressure energy are provided in an appendix. We publicly release halo property catalogs and KLLR parameters for the TNG runs at twenty epochs up to z = 12.

astro-ph.CO↗