arXiv subjects

P. J. Elahi

Publications and source records attributed to P. J. Elahi.

8 recordsLinked to original sources

arXiv2026

Accelerating cosmological simulations on GPUs: a portable approach using OpenMP

In this work we present the porting to Graphics Processing Units (GPUs, using OpenMP target directives) and optimization of a key module within the cosmological {\pinocchio} code, a Lagrangian Perturbation Theory (LPT)-based framework widely used for generating dark matter (DM) halo catalogs. Our optimization focuses on a specific segment of the code responsible for calculating the collapse time of each particle involved in the simulation. Due to the embarrassingly parallel nature of this computation, it represents an ideal candidate for GPU offloading. As part of the porting process, we developed fully GPU-native implementations of both cubic spline and bilinear interpolation routines, required for evaluating collapse times. Since GNU Scientific Library (GSL) does not support GPU offloading, these custom implementations run entirely on the GPU and achieve residuals of only $\sim0.003\%$ when compared to the CPU-based implementation of GSL. Comparative benchmarking on the LEONARDO (NVIDIA-based) and SETONIX (AMD-based) supercomputers reveals notable portability and performance, with speedups of~\textit{4x} and up to~\textit{8x}, respectively. While collapse time calculation is not a primary bottleneck in the overall workflow, the acceleration reduces full production runs by $\sim 100$ seconds each leading to a cumulative saving of $\sim 160000$ Standard-h ($\sim28$ hours wall time) across thousands of simulations. Roofline analysis confirms that our GPU porting achieves over 80\% of the theoretical FP64 peak performance, confirming efficient compute-bound execution. This work demonstrates that OpenMP directives offer a portable, effective strategy for accelerating large-scale cosmological simulations on heterogeneous hardware.

astro-ph.IM↗

arXiv2026

The Southern-sky MWA Rapid Two-metre (SMART) pulsar survey--IV. Survey update and an atlas of 205 non-recycled southern pulsars

The SMART survey is an ambitious effort to conduct sensitive searches for pulsars and fast transients at 140-170 MHz. The novelty of voltage recording, long dwell times and the high-time and -frequency resolutions exchange a large survey speed for high computational cost. The survey covers the entire sky south of +30 degree in declination through a series of dedicated campaigns, accumulating 4 PB of data. The large volumes of data necessitate processing to be approached in multiple phases, and the initial searches focused on a shallow survey of parts of the skies, as reported in earlier papers. These data are also processed for re-detections of hundreds of known pulsars in the southern sky, many of which are also the first detections below 400 MHz. This paper is motivated by the need to address the inherent difficulties in handling large amounts of voltage data and software/processing challenges for routine pulsar detections, and also by the fast-evolving landscape of the SKA Observatory (SKAO). With the construction ramping up towards the full-scale SKA-Low, a low-frequency catalogue of detectable pulsars in the southern sky will prove to be a valuable reference for science verification. A growing sample of low-frequency detections and measurements will also prove invaluable in a variety of science applications including population studies, survey simulations and emission beam models, refining interstellar medium models for electron densities and the spatial distribution of turbulence, and also for forecasting the detection prospects and survey yield from pulsar surveys planned with SKA-Low. We present various data products, including pulse profiles, time series and multi-channel folded archives, along with the measurements of dispersion and rotation measures, and mean flux densities, and this will be periodically updated as more detections flow on from the ongoing data processing.

astro-ph.HE↗

arXiv2019

nIFTy Galaxy Cluster simulations VI: The dynamical imprint of substructure on gaseous cluster outskirts

Galaxy cluster outskirts mark the transition region from the mildly non-linear cosmic web to the highly non-linear, virialised, cluster interior. It is in this transition region that the intra-cluster medium (ICM) begins to influence the properties of accreting galaxies and groups, as ram pressure impacts a galaxy's cold gas content and subsequent star formation rate. Conversely, the thermodynamical properties of the ICM in this transition region should also feel the influence of accreting substructure (i.e. galaxies and groups), whose passage can drive shocks. In this paper, we use a suite of cosmological hydrodynamical zoom simulations of a single galaxy cluster, drawn from the nIFTy comparison project, to study how the dynamics of substructure accreted from the cosmic web influences the thermodynamical properties of the ICM in the cluster's outskirts. We demonstrate how features evident in radial profiles of the ICM (e.g. gas density and temperature) can be linked to strong shocks, transient and short-lived in nature, driven by the passage of substructure. The range of astrophysical codes and galaxy formation models in our comparison are broadly consistent in their predictions (e.g. agreeing when and where shocks occur, but differing in how strong shocks will be); this is as we would expect of a process driven by large-scale gravitational dynamics and strong, inefficiently radiating, shocks. This suggests that mapping such shock structures in the ICM in a cluster's outskirts (via e.g. radio synchrotron emission) could provide a complementary measure of its recent merger and accretion history.

astro-ph.CO↗

arXiv2019

The HI Velocity Function: a test of cosmology or baryon physics?

Accurately predicting the shape of the HI velocity function of galaxies is regarded widely as a fundamental test of any viable dark matter model. Straightforward analyses of cosmological $N$-body simulations imply that the $Λ$CDM model predicts an overabundance of low circular velocity galaxies when compared to observed HI velocity functions. More nuanced analyses that account for the relationship between galaxies and their host haloes suggest that how we model the influence of baryonic processes has a significant impact on HI velocity function predictions. We explore this in detail by modelling HI emission lines of galaxies in the SHARK semi-analytic galaxy formation model, built on the SURFS suite of $Λ$CDM $N$-body simulations. We create a simulated ALFALFA survey, in which we apply the survey selection function and account for effects such as beam confusion, and compare simulated and observed HI velocity width distributions, finding differences of $\lesssim 50$%, orders of magnitude smaller than the discrepancies reported in the past. This is a direct consequence of our careful treatment of survey selection effects and, importantly, how we model the relationship between galaxy and halo circular velocity - the HI mass-maximum circular velocity relation of galaxies is characterised by a large scatter. These biases are complex enough that building a velocity function from the observed HI line widths cannot be done reliably.

astro-ph.GA↗

arXiv2019

Dark Matter Annihilation Feedback in Cosmological Simulations II: The Influence on Gas and Halo Structure

We present new cosmological hydrodynamic simulations that incorporate Dark Matter Annihilation Feedback (DMAF), whereby energy released from the annihilation of dark matter particles through decay channels such as photon or positron-electron pairs provide additional heating sources for local baryonic material. For annihilation rates comparable to WIMP-like particles, we find that the key influence of DMAF is to inhibit gas accretion onto halos. Such diminished gas accretion early in the lifetimes of halos results in reduced gas fractions in smaller halos, and the delayed halo formation times of larger structures, suggesting that DMAF could impact the stellar age distribution in galaxies, and morphology of dwarfs. For a dark matter particle mass of $m_χ\sim10$~MeV, there is a `critical halo mass' of $\sim10^{13}$ M$_{\odot}$ at $z=0$, below which there are large differences when compared to $Λ$CDM, such as a reduction in the abundance of halo structures as large as 25 percent, reduced gas content by 50 percent and central gas densities reduced down to 10 percent within halos of mass $\sim10^{12}$ M$_{\odot}$ but with increasing effects in smaller halos. Higher dark matter particle mass models have a smaller `critical halo mass'. For a $m_χ\sim100$~MeV model, we find differences start appearing below halo masses of $\sim10^{12}$ M$_\odot$ and a $m_χ\gtrsim 1$~GeV model, this mass scale lies below the resolution of our simulations, though we still observe changes in the morphology of dwarf galaxies.

astro-ph.GA↗

arXiv2012

Exploring Galaxy Formation Models and Cosmologies with Galaxy Clustering

Using N-body simulations and galaxy formation models, we study the galaxy stellar mass correlation and the two-point auto-correlation. The simulations are run with cosmological parameters from the WMAP first, third and seven year results, which mainly differ in the perturbation amplitude of σ_{8}. The stellar mass of galaxies are determined using either a semi-analytical galaxy formation model or a simple empirical abundance matching method. Compared to the SDSS DR7 data at z=0 and the DEEP2 results at z=1, we find that the predicted galaxy clusterings from the semi-analytical model are higher than the data at small scales, regardless of the adopted cosmology. Conversely, the abundance matching method predicts good agreement with the data at both z=0 and z=1 for high σ_8 cosmologies (WMAP1 & WMAP7), but the predictions from a low σ_8 cosmology (WMAP3) are significantly lower than the data at z=0. We find that the excess clustering at small-scales in the semi-analytical model mainly arises from satellites in massive haloes, indicating that either the star formation is too efficient in low-mass haloes or tidal stripping is too inefficient at high redshift. Our results show that galaxy clustering is strongly affected by the models for galaxy formation, thus can be used to constrain the baryonic physics. The weak dependence of galaxy clustering on cosmological parameters makes it difficult to constrain the WMAP1 and WMAP7 cosmologies.

astro-ph.CO↗

arXiv2009

Can substructure in the Galactic Halo explain the ATIC and PAMELA results?

Recently, ATIC and PAMELA measured an anomalously large flux of leptonic cosmic rays which may arise from dark matter self-annihilation. While the annihilation signal predicted for a smooth halo is 10^2-10^3 times smaller than the measured excess, the signal can be boosted by the presence of subhalos. We investigate the feasibility of large boost factors using a new Monte Carlo calculation technique that is constrained by previous simulation work on halo substructure. The model accounts for the observed decrease in the amount of substructure with decreasing halo mass and the scatter in halo structural parameters such as the density concentration parameter. Our results suggest that boost factors of the order of 10^2 are ruled out at more than 14 sigma. We conclude that substructure alone, at least with commonly assumed annihilation cross-sections, cannot explain the anomalous flux measured by ATIC and PAMELA.

astro-ph.HE↗

arXiv2009

Subhaloes in Scale-Free Cosmologies

We explore the dependence of the subhalo mass function on the spectral index n of the linear matter power spectrum using scale-free Einstein-de Sitter simulations with n=-1 and n=-2.5. We carefully consider finite volume effects that may call into question previous simulations of n<-2 power spectra. Subhaloes are found using a 6D friends-of-friends algorithm in all haloes originating from high-sigma peaks. For n=-1, we find that the cumulative subhalo mass function is independent of the parameters used in the subhalo finding algorithm and is consistent with the subhalo mass function found in LCDM simulations. In particular, the subhalo mass function is well fit by a power-law with an index of alpha=-0.9, that is the mass function has roughly equal mass in subhaloes per logarithmic interval in subhalo mass. Conversely, for n=-2.5, the algorithm parameters affect the subhalo mass function since subhaloes are more triaxial with less well defined boundaries. We find that the index alpha is generally larger with alpha>=-0.75. We infer that although the subhalo mass function appears to be independent of n so long as n>=-2, it begins to flatten as n->-3. Thus, the common practice of using alpha=-1.0 may greatly overestimate the number of subhaloes at the smallest scales in the CDM hierarchy.

astro-ph↗