arXiv subjects

Romain Teyssier

Publications and source records attributed to Romain Teyssier.

At least 127 records · Page 7Linked to original sources

arXiv2016

PKDGRAV3: Beyond Trillion Particle Cosmological Simulations for the Next Era of Galaxy Surveys

We report on the successful completion of a 2 trillion particle cosmological simulation to z=0 run on the Piz Daint supercomputer (CSCS, Switzerland), using 4000+ GPU nodes for a little less than 80h of wall-clock time or 350,000 node hours. Using multiple benchmarks and performance measurements on the US Oak Ridge National Laboratory Titan supercomputer, we demonstrate that our code PKDGRAV3, delivers, to our knowledge, the fastest time-to-solution for large-scale cosmological N-body simulations. This was made possible by using the Fast Multipole Method in conjunction with individual and adaptive particle time steps, both deployed efficiently (and for the first time) on supercomputers with GPU-accelerated nodes. The very low memory footprint of PKDGRAV3 allowed us to run the first ever benchmark with 8 trillion particles on Titan, and to achieve perfect scaling up to 18000 nodes and a peak performance of 10 Pflops.

astro-ph.IM↗

arXiv2016

nIFTy galaxy cluster simulations V: Investigation of the Cluster Infall Region

We examine the properties of the galaxies and dark matter haloes residing in the cluster infall region surrounding the simulated $Λ$CDM galaxy cluster studied by Elahi et al. (2016) at z=0. The $1.1\times10^{15}h^{-1}\text{M}_{\odot}$ galaxy cluster has been simulated with eight different hydrodynamical codes containing a variety of hydrodynamic solvers and subgrid schemes. All models completed a dark-matter only, non-radiative and full-physics run from the same initial conditions. The simulations contain dark matter and gas with mass resolution $m_{\text{DM}}=9.01\times 10^8h^{-1}\text{M}_{\odot}$ and $m_{\text{gas}}=1.9\times 10^8h^{-1}\text{M}_{\odot}$ respectively. We find that the synthetic cluster is surrounded by clear filamentary structures that contain ~60% of haloes in the infall region with mass ~$10^{12.5} - 10^{14} h^{-1}\text{M}_{\odot}$, including 2-3 group-sized haloes ($> 10^{13}h^{-1}\text{M}_{\odot}$). However, we find that only ~10% of objects in the infall region are subhaloes residing in haloes, which may suggest that there is not much ongoing preprocessing occurring in the infall region at z=0. By examining the baryonic content contained within the haloes, we also show that the code-to-code scatter in stellar fraction across all halo masses is typically ~2 orders of magnitude between the two most extreme cases, and this is predominantly due to the differences in subgrid schemes and calibration procedures that each model uses. Models that do not include AGN feedback typically produce too high stellar fractions compared to observations by at least ~1 order of magnitude.

astro-ph.GA↗

arXiv2016

Cosmic Dawn (CoDa): the First Radiation-Hydrodynamics Simulation of Reionization and Galaxy Formation in the Local Universe

Cosmic reionization by starlight from early galaxies affected their evolution, thereby impacting reionization, itself. Star formation suppression, for example, may explain the observed underabundance of Local Group dwarfs relative to N-body predictions for Cold Dark Matter. Reionization modelling requires simulating volumes large enough [~(100Mpc)^3] to sample reionization "patchiness", while resolving millions of galaxy sources above ~10^8 Msun , combining gravitational and gas dynamics with radiative transfer. Modelling the Local Group requires initial cosmological density fluctuations pre-selected to form the well-known structures of the local universe today. Cosmic Dawn ("CoDa") is the first such fully-coupled, radiation-hydrodynamics simulation of reionization of the local universe. Our new hybrid CPU-GPU code, RAMSES-CUDATON, performs hundreds of radiative transfer and ionization rate-solver timesteps on the GPUs for each hydro-gravity timestep on the CPUs. CoDa simulated (91Mpc)^3 with 4096^3 particles and cells, to redshift 4.23, on ORNL supercomputer Titan, utilizing 8192 cores and 8192 GPUs. Global reionization ended slightly later than observed. However, a simple temporal rescaling which brings the evolution of ionized fraction into agreement with observations also reconciles ionizing flux density, cosmic star formation history, CMB electron scattering optical depth and galaxy UV luminosity function with their observed values. Photoionization heating suppressed the star formation of haloes below ~2 x 10^9 Msun , decreasing the abun- dance of faint galaxies around MAB_1600 = [-10,-12]. For most of reionization, star formation was dominated by haloes between 10^10 - 10^11 Msun , so low-mass halo suppression was not reflected by a distinct feature in the global star formation history. (Abridged)

astro-ph.GA↗

arXiv2016

Matter power spectrum and the challenge of percent accuracy

Future galaxy surveys require one percent precision in the theoretical knowledge of the power spectrum over a large range including very nonlinear scales. While this level of accuracy is easily obtained in the linear regime with perturbation theory, it represents a serious challenge for small scales where numerical simulations are required. In this paper we quantify the precision of present-day $N$-body methods, identifying main potential error sources from the set-up of initial conditions to the measurement of the final power spectrum. We directly compare three widely used $N$-body codes, Ramses, Pkdgrav3, and Gadget3 which represent three main discretisation techniques: the particle-mesh method, the tree method, and a hybrid combination of the two. For standard run parameters, the codes agree to within one percent at $k\leq1$ $h\,\rm Mpc^{-1}$ and to within three percent at $k\leq10$ $h\,\rm Mpc^{-1}$. We also consider the bispectrum and show that the reduced bispectra agree at the sub-percent level for $k\leq 2$ $h\,\rm Mpc^{-1}$. In a second step, we quantify potential errors due to initial conditions, box size, and resolution using an extended suite of simulations performed with our fastest code Pkdgrav3. We demonstrate that the simulation box size should not be smaller than $L=0.5$ $h^{-1}\rm Gpc$ to avoid systematic finite-volume effects (while much larger boxes are required to beat down the statistical sample variance). Furthermore, a maximum particle mass of $M_{\rm p}=10^{9}$ $h^{-1}\rm M_{\odot}$ is required to conservatively obtain one percent precision of the matter power spectrum. As a consequence, numerical simulations covering large survey volumes of upcoming missions such as DES, LSST, and Euclid will need more than a trillion particles to reproduce clustering properties at the targeted accuracy.

astro-ph.CO↗

arXiv2016

RHAPSODY-G simulations II - Baryonic growth and metal enrichment in massive galaxy clusters

We study the evolution of the stellar component and the metallicity of both the intracluster medium and of stars in massive ($M_{\rm vir}\approx 6\times 10^{14}$ M$_{\odot}/h$) simulated galaxy clusters from the Rhapsody-G suite in detail and compare them to observational results. The simulations were performed with the AMR code RAMSES and include the effect of AGN feedback at the sub-grid level. AGN feedback is required to produce realistic galaxy and cluster properties and plays a role in mixing material in the central regions and regulating star formation in the central galaxy. In both our low and high resolution runs with fiducial stellar yields, we find that stellar and ICM metallicities are a factor of two lower than in observations. We find that cool core clusters exhibit steeper metallicity gradients than non-cool core clusters, in qualitative agreement with observations. We verify that the ICM metallicities measured in the simulation can be explained by a simple "regulator" model in which the metallicity is set by a balance of stellar yield and gas accretion. It is plausible that a combination of higher resolution and higher metal yield in AMR simulation would allow the metallicity of simulated clusters to match observed values; however this hypothesis needs to be tested with future simulations. Comparison to recent literature highlights that results concerning the metallicity of clusters and cluster galaxies might depend sensitively on the scheme chosen to solve the hydrodynamics.

astro-ph.GA↗

arXiv2016

nIFTy Galaxy Cluster simulations IV: Quantifying the Influence of Baryons on Halo Properties

Building on the initial results of the nIFTy simulated galaxy cluster comparison, we compare and contrast the impact of baryonic physics with a single massive galaxy cluster, run with 11 state-of-the-art codes, spanning adaptive mesh, moving mesh, classic and modern SPH approaches. For each code represented we have a dark matter only (DM) and non-radiative (NR) version of the cluster, as well as a full physics (FP) version for a subset of the codes. We compare both radial mass and kinematic profiles, as well as global measures of the cluster (e.g. concentration, spin, shape), in the NR and FP runs with that in the DM runs. Our analysis reveals good consistency (<= 20%) between global properties of the cluster predicted by different codes when integrated quantities are measured within the virial radius R200. However, we see larger differences for quantities within R2500, especially in the FP runs. The radial profiles reveal a diversity, especially in the cluster centre, between the NR runs, which can be understood straightforwardly from the division of codes into classic SPH and non-classic SPH (including the modern SPH, adaptive and moving mesh codes); and between the FP runs, which can also be understood broadly from the division of codes into those that include AGN feedback and those that do not. The variation with respect to the median is much larger in the FP runs with different baryonic physics prescriptions than in the NR runs with different hydrodynamics solvers.

astro-ph.GA↗

arXiv2016

nIFTY galaxy cluster simulations III: The Similarity & Diversity of Galaxies & Subhaloes

We examine subhaloes and galaxies residing in a simulated LCDM galaxy cluster ($M^{\rm crit}_{200}=1.1\times10^{15}M_\odot/h$) produced by hydrodynamical codes ranging from classic Smooth Particle Hydrodynamics (SPH), newer SPH codes, adaptive and moving mesh codes. These codes use subgrid models to capture galaxy formation physics. We compare how well these codes reproduce the same subhaloes/galaxies in gravity only, non-radiative hydrodynamics and full feedback physics runs by looking at the overall subhalo/galaxy distribution and on an individual objects basis. We find the subhalo population is reproduced to within $\lesssim10\%$ for both dark matter only and non-radiative runs, with individual objects showing code-to-code scatter of $\lesssim0.1$ dex, although the gas in non-radiative simulations shows significant scatter. Including feedback physics significantly increases the diversity. Subhalo mass and $V_{max}$ distributions vary by $\approx20\%$. The galaxy populations also show striking code-to-code variations. Although the Tully-Fisher relation is similar in almost all codes, the number of galaxies with $10^{9}M_\odot/h\lesssim M_*\lesssim 10^{12}M_\odot/h$ can differ by a factor of 4. Individual galaxies show code-to-code scatter of $\sim0.5$ dex in stellar mass. Moreover, strong systematic differences exist, with some codes producing galaxies $70\%$ smaller than others. The diversity partially arises from the inclusion/absence of AGN feedback. Our results combined with our companion papers demonstrate that subgrid physics is not just subject to fine-tuning, but the complexity of building galaxies in all environments remains a challenge. We argue even basic galaxy properties, such as the stellar mass to halo mass, should be treated with errors bars of $\sim0.2-0.4$ dex.

astro-ph.GA↗

arXiv2016

A new method to quantify the effects of baryons on the matter power spectrum

Future large-scale galaxy surveys have the potential to become leading probes for cosmology provided the influence of baryons on the total mass distribution is understood well enough. As hydrodynamical simulations strongly depend on details in the feedback implementations, no unique and robust predictions for baryonic effects currently exist. In this paper we propose a baryonic correction model that modifies the density field of dark-matter-only $N$-body simulations to mimic the effects of baryons from any underlying adopted feedback recipe. The model assumes haloes to consist of 4 components: 1- hot gas in hydrostatical equilibrium, 2- ejected gas from feedback processes, 3- central galaxy stars, and 4- adiabatically relaxed dark matter, which all modify the initial dark-matter-only density profiles. These altered profiles allow to define a displacement field for particles in $N$-body simulations and to modify the total density field accordingly. The main advantage of the baryonic correction model is to connect the total matter density field to the observable distribution of gas and stars in haloes, making it possible to parametrise baryonic effects on the matter power spectrum. We show that the most crucial quantities are the mass fraction of ejected gas and its corresponding ejection radius. The former controls how strongly baryons suppress the power spectrum, while the latter provides a measure of the scale where baryonic effects become important. A comparison with X-ray and Sunyaev-Zel'dovich cluster observations suggests that baryons suppress wave modes above $k\sim0.5$ h/Mpc with a maximum suppression of 10-25 percent around $k\sim 2$ h/Mpc. More detailed observations of the gas in the outskirts of groups and clusters are required to decrease the large uncertainties of these numbers.

astro-ph.CO↗

arXiv2015

nIFTy galaxy cluster simulations II: radiative models

We have simulated the formation of a massive galaxy cluster (M$_{200}^{\rm crit}$ = 1.1$\times$10$^{15}h^{-1}M_{\odot}$) in a $Λ$CDM universe using 10 different codes (RAMSES, 2 incarnations of AREPO and 7 of GADGET), modeling hydrodynamics with full radiative subgrid physics. These codes include Smoothed-Particle Hydrodynamics (SPH), spanning traditional and advanced SPH schemes, adaptive mesh and moving mesh codes. Our goal is to study the consistency between simulated clusters modeled with different radiative physical implementations - such as cooling, star formation and AGN feedback. We compare images of the cluster at $z=0$, global properties such as mass, and radial profiles of various dynamical and thermodynamical quantities. We find that, with respect to non-radiative simulations, dark matter is more centrally concentrated, the extent not simply depending on the presence/absence of AGN feedback. The scatter in global quantities is substantially higher than for non-radiative runs. Intriguingly, adding radiative physics seems to have washed away the marked code-based differences present in the entropy profile seen for non-radiative simulations in Sembolini et al. (2015): radiative physics + classic SPH can produce entropy cores. Furthermore, the inclusion/absence of AGN feedback is not the dividing line -as in the case of describing the stellar content- for whether a code produces an unrealistic temperature inversion and a falling central entropy profile. However, AGN feedback does strongly affect the overall stellar distribution, limiting the effect of overcooling and reducing sensibly the stellar fraction.

astro-ph.CO↗

arXiv2015

Rhapsody-G simulations I: the cool cores, hot gas and stellar content of massive galaxy clusters

We present the Rhapsody-G suite of cosmological hydrodynamic AMR zoom simulations of ten massive galaxy clusters at the $M_{\rm vir}\sim10^{15}\,{\rm M}_\odot$ scale. These simulations include cooling and sub-resolution models for star formation and stellar and supermassive black hole feedback. The sample is selected to capture the whole gamut of assembly histories that produce clusters of similar final mass. We present an overview of the successes and shortcomings of such simulations in reproducing both the stellar properties of galaxies as well as properties of the hot plasma in clusters. In our simulations, a long-lived cool-core/non-cool core dichotomy arises naturally, and the emergence of non-cool cores is related to low angular momentum major mergers. Nevertheless, the cool-core clusters exhibit a low central entropy compared to observations, which cannot be alleviated by thermal AGN feedback. For cluster scaling relations we find that the simulations match well the $M_{500}-Y_{500}$ scaling of Planck SZ clusters but deviate somewhat from the observed X-ray luminosity and temperature scaling relations in the sense of being slightly too bright and too cool at fixed mass, respectively. Stars are produced at an efficiency consistent with abundance matching constraints and central galaxies have star formation rates consistent with recent observations. While our simulations thus match various key properties remarkably well, we conclude that the shortcomings strongly suggest an important role for non-thermal processes (through feedback or otherwise) or thermal conduction in shaping the intra-cluster medium.

astro-ph.CO↗

arXiv2015

Distribution of streaming rates into high-redshift galaxies

We study the accretion along streams from the cosmic web into high-redshift massive galaxies using three sets of AMR hydro-cosmological simulations. We find that the streams keep a roughly constant accretion rate as they penetrate into the halo centre. The mean accretion rate follows the mass and redshift dependence predicted for haloes by the EPS approximation, dM / dt is proportional to Mvir^{1.25} (1 + z)^{2.5}. The distribution of the accretion rates can well be described by a sum of two Gaussians, the primary corresponding to "smooth inflow" and the secondary to "mergers". The same functional form was already found for the distributions of specific star formation rates in observations. The mass fraction in the smooth component is 60 - 90 %, insensitive to redshift or halo mass. The simulations with strong feedback show clear signs of reaccretion due to recycling of galactic winds. The mean accretion rate for the mergers is a factor 2 - 3 larger than that of the smooth component. The standard deviation of the accretion rate is 0.2 - 0.3 dex, showing no trend with mass or redshift. For the smooth component it is 0.12 - 0.24 dex.

astro-ph.CO↗

arXiv2015

Rhapsody-G simulations: galaxy clusters as baryonic closed boxes and the covariance between hot gas and galaxies

Within a sufficiently large cosmic volume, conservation of baryons implies a simple `closed box' view in which the sum of the baryonic components must equal a constant fraction of the total enclosed mass. We present evidence from Rhapsody-G hydrodynamic simulations of massive galaxy clusters that the closed-box expectation may hold to a surprising degree within the interior, non-linear regions of haloes. At a fixed halo mass, we find a significant anti-correlation between hot gas mass fraction and galaxy mass fraction (cold gas + stars), with a rank correlation coefficient of -0.69 within $R_{500c}$. Because of this anti-correlation, the total baryon mass serves as a low-scatter proxy for total cluster mass. The fractional scatter of total baryon fraction scales approximately as $0.02 (Δ_c/100)^{0.6}$, while the scatter of either gas mass or stellar mass is larger in magnitude and declines more slowly with increasing radius. We discuss potential observational tests using cluster samples selected by optical and hot gas properties; the simulations suggest that joint selection on stellar and hot gas has potential to achieve 5% scatter in total halo mass.

astro-ph.CO↗

arXiv2015

PHEW: a parallel segmentation algorithm for three-dimensional AMR datasets - application to structure detection in self-gravitating flows

We introduce PHEW (Parallel HiErarchical Watershed), a new segmentation algorithm to detect structures in astrophysical fluid simulations, and its implementation into the adaptive mesh refinement (AMR) code ramses. PHEW works on the density field defined on the adaptive mesh, and can thus be used on the gas density or the dark matter density after a projection of the particles onto the grid. The algorithm is based on a "watershed" segmentation of the computational volume into dense regions, followed by a merging of the segmented patches based on the saddle point topology of the density field. PHEW is capable of automatically detecting connected regions above the adopted density threshold, as well as the entire set of substructures within. Our algorithm is fully parallel and uses the MPI library. We describe in great detail the parallel algorithm and perform a scaling experiment which proves the capability of phew to run efficiently on massively parallel systems.

astro-ph.IM↗

arXiv2015

Black hole evolution: I. Supernova-regulated black hole growth

The growth of a supermassive black hole (BH) is determined by how much gas the host galaxy is able to feed it, which in turn is controlled by the cosmic environment, through galaxy mergers and accretion of cosmic flows that time how galaxies obtain their gas, but also by internal processes in the galaxy, such as star formation and feedback from stars and the BH itself. In this paper, we study the growth of a 10^12 Msun halo at z=2, which is the progenitor of al group of galaxies at z=0, and of its central BH by means of a high-resolution zoomed cosmological simulation, the Seth simulation. We study the evolution of the BH driven by the accretion of cold gas in the galaxy, and explore the efficiency of the feedback from supernovae (SNe). For a relatively inefficient energy input from SNe, the BH grows at the Eddington rate from early times, and reaches self-regulation once it is massive enough. We find that at early cosmic times z>3.5, efficient feedback from SNe forbids the formation of a settled disc as well as the accumulation of dense cold gas in the vicinity of the BH and starves the central compact object. As the galaxy and its halo accumulate mass, they become able to confine the nuclear inflows provided by major mergers and the BH grows at a sustained near-to-Eddington accretion rate. We argue that this mechanism should be ubiquitous amongst low-mass galaxies, corresponding to galaxies with a stellar mass below <10^9 Msun in our simulations.

astro-ph.GA↗

arXiv2015

Galaxy Evolution: Modeling the Role of Non-thermal Pressure in the Interstellar medium

Galaxy evolution depends strongly on the physics of the interstellar medium (ISM). Motivated by the need to incorporate the properties of the ISM in cosmological simulations we construct a simple method to include the contribution of non-thermal components in the calculation of pressure of interstellar gas. In our method we treat three non-thermal components - turbulence, magnetic fields and cosmic rays - and effectively parametrize their amplitude. We assume that the three components settle into a quasi-steady-state that is governed by the star formation rate, and calibrate their magnitude and density dependence by the observed Radio-FIR correlation, relating synchrotron radiation to star formation rates of galaxies. We implement our model in single cell numerical simulation of a parcel of gas with constant pressure boundary conditions and demonstrate its effect and potential. Then, the non-thermal pressure model is incorporated into RAMSES and hydrodynamic simulations of isolated galaxies with and without the non-thermal pressure model are presented and studied. Specifically, we demonstrate that the inclusion of realistic non-thermal pressure reduces the star formation rate by an order of magnitude and increases the gas depletion time by as much. We conclude that the non-thermal pressure can prolong the star formation epoch and achieve consistency with observations without invoking artificially strong stellar feedback.

astro-ph.CO↗

arXiv2015

Galaxies that Shine: radiation-hydrodynamical simulations of disk galaxies

Radiation feedback is typically implemented using subgrid recipes in hydrodynamical simulations of galaxies. Very little work has so far been performed using radiation-hydrodynamics (RHD), and there is no consensus on the importance of radiation feedback in galaxy evolution. We present RHD simulations of isolated galaxy disks of different masses with a resolution of 18 pc. Besides accounting for supernova feedback, our simulations are the first galaxy-scale simulations to include RHD treatments of photo-ionisation heating and radiation pressure, from both direct optical/UV radiation and multi-scattered, re-processed infrared (IR) radiation. Photo-heating smooths and thickens the disks and suppresses star formation about as much as the inclusion of ("thermal dump") supernova feedback does. These effects decrease with galaxy mass and are mainly due to the prevention of the formation of dense clouds, as opposed to their destruction. Radiation pressure, whether from direct or IR radiation, has little effect, but for the IR radiation we show that its impact is limited by our inability to resolve the high optical depths for which multi-scattering becomes important. While artificially boosting the IR optical depths does reduce the star formation, it does so by smoothing the gas rather than by generating stronger outflows. We conclude that although higher-resolution simulations, and potentially also different supernova implementations, are needed for confirmation, our findings suggest that radiation feedback is more gentle and less effective than is often assumed in subgrid prescriptions.

astro-ph.GA↗

arXiv2015

Milking the spherical cow: on aspherical dynamics in spherical coordinates

Galaxies and the dark matter halos that host them are not spherically symmetric, yet spherical symmetry is a helpful simplifying approximation for idealised calculations and analysis of observational data. The assumption leads to an exact conservation of angular momentum for every particle, making the dynamics unrealistic. But how much does that inaccuracy matter in practice for analyses of stellar distribution functions, collisionless relaxation, or dark matter core-creation? We provide a general answer to this question for a wide class of aspherical systems; specifically, we consider distribution functions that are "maximally stable", i.e. that do not evolve at first order when external potentials (which arise from baryons, large scale tidal fields or infalling substructure) are applied. We show that a spherically-symmetric analysis of such systems gives rise to the false conclusion that the density of particles in phase space is ergodic (a function of energy alone). Using this idea we are able to demonstrate that: (a) observational analyses that falsely assume spherical symmetry are made more accurate by imposing a strong prior preference for near-isotropic velocity dispersions in the centre of spheroids; (b) numerical simulations that use an idealised spherically-symmetric setup can yield misleading results and should be avoided where possible; and (c) triaxial dark matter halos (formed in collisionless cosmological simulations) nearly attain our maximally-stable limit, but their evolution freezes out before reaching it.

astro-ph.GA↗

arXiv2014

Baryonic and dark matter distribution in cosmological simulations of spiral galaxies

We study three cosmological hydrodynamical simulations of Milky Way(MW)-sized halos including a comparison with the dark matter(DM)-only counterparts. We find one of our simulated galaxies with interesting MW-like features. Thanks to a consistently tuned star formation rate and supernovae feedback we obtain an extended disk and a flat rotation curve with a satisfying circular velocity and a reasonable DM density in the solar neighbourhood. Mimicking observational methods, we re-derive the stellar mass and obtain stellar-to-halo mass ratios reduced by more than 50\%. We show the interaction between the baryons and the dark matter which is first contracted by star formation and then cored by feedback processes. Indeed, we report an unprecedentedly observed effect in the DM density profile consisting of a central core combined with an adiabatic contraction at larger galactic radii. The cores obtained are typically $\sim$ 5 kpc large. Moreover, this also impacts the DM density at the solar radius. In our simulation resembling most to the MW, the density is raised from 0.23 GeV/cm$^3$ in the dark matter only run to 0.36 GeV/cm$^3$ (spherical shell) or 0.54 GeV/cm$^3$ (circular ring) in the hydrodynamical run. Studying the subhalos, the dark matter within luminous satellites is also affected by baryonic processes and exihibits cored profiles whereas dark satellites are cuspy. We find a shift in mass compared to DM-only simulations and obtain, for halos in the lower MW mass range, a distribution of luminous satellites comparable to the MW spheroidal dwarf galaxies.

astro-ph.GA↗