Chris Loken

SciNet: Lessons Learned from Building a Power-efficient Top-20 System and Data Centre

SciNet, one of seven regional HPC consortia operating under the Compute Canada umbrella, runs Canadas first and third fastest computers (as of June 2010) in a state-of-the-art, highly energy-efficient datacentre with a Power Usage Effectiveness (PUE) design-point of 1.16. Power efficiency, computational bang for the buck and system capability for a handful of flagship science projects were important criteria in choosing the nature of the computers and the data centre itself. Here we outline some of the lessons learned in putting together the systems and the data centre that hosts Canadas fastest computer to date.

The Observational Consequences of Merging Clusters of Galaxies

We present an observational analysis of the numerical simulations of galaxy custer mergers. We identify several observational signatures of recent merger activity and quantitatively assess the uncertainty introduced into cluster mass estimates when invoking the commonly held assumptions of hydrostatic equilibrium, virial equilibrium, spherical symmetry, and isothermality. We find that mergers result in multiple X-ray peaks, long-lived elongation of the X-ray emission, as well as isophotal twisting and centroid shifting to a degree consistent with recent observations. We also find an enlargement of the X-ray core relative to the dark matter core. Mergers result in nonisothermal clusters exhibiting observable inhomogeneities in the emission-weighted X-ray temperature of several keV on linear scales of less than 0.5 Mpc. The resulting gasdynamics are extremely complex, and we present an example of what might be observed by a high resolution X-ray spectograph. We further speculate that the gas dynamics, via shocks, bulk flows, and turbulence, play an important role in the evolution of cluster galaxies and associated radio sources, particularly wide-angle tailed (WAT) sources and radio halos. We find that X-ray based cluster mass estimates made under equilibrium assumptions can be uncertain by 50% or more in the first 2 Gyrs after a merger and by up to 25% after 2 Gyrs depending on the details of the analysis and projection effects. Uncertainties can be considerably larger if the temperature is not well constrained. Similar uncertainties are observed in the X-ray derived baryon mass fractions. Virial mass estimates are typically overestimated because the observed one-dimensional velocity dispersion can be severely contaminated by the infall velocity of the subcluster.

A Universal Temperature Profile for Galaxy Clusters

We investigate the predicted present-day temperature profiles of the hot, X-ray-emitting gas in galaxy clusters for two cosmological models—a current best-guess ΛCDM model and a standard cold dark matter (SCDM) model. Our numerically simulated catalogs of clusters are derived from high-resolution (15 h-1 kpc) simulations which make use of a sophisticated, Eulerian-based, adaptive mesh-refinement code that faithfully captures the shocks that are essential for correctly modeling cluster temperatures. We show that the temperature structure on Mpc scales is highly complex and non-isothermal. However, the temperature profiles of the simulated ΛCDM and SCDM clusters are remarkably similar and drop off as T ∝ (1 + r/ax)-δ, where ax ~ rvir/1.5 and δ ~ 1.6. This decrease is in good agreement with the observational results of Markevitch et al. but diverges, primarily in the innermost regions, from their fit which assumes a polytropic equation of state. Our result is also in good agreement with a recent sample of clusters observed by BeppoSAX, though there is some indication of missing physics at small radii (r 0.2 rvir, our universal temperature profile is consistent with our most recent simulations, which include both radiative cooling and supernovae feedback.

When clusters collide - A numerical Hydro/N-body simulation of merging galaxy clusters

A 3D numerical simulation of two merging clusters of galaxies, using a hybrid Hydro/N-body code, is presented. The hydrodynamics of the code is solved by an Eulerian finite difference method. Initial results disclose that the X-ray emission of the dominant cluster becomes elongated and broadened; heating occurs at the core of the dominant cluster as a result of multiple shocks, and high velocity gas motions within the intracluster medium. It is predicted that clusters which have undergone recent mergers and do not have cooling flows will have high peculiar gas velocities and that the shocks and turbulence generated during the merger may power cluster-wide radio halos. Prolonged high-velocity gas motions through the dominant cluster core possibly play a major role in the formation and shaping of wide-angle tailed radio sources associated with central dominant galaxies. The N-body component of the simulation reveals the subcluster to be dispersed as it passes through the dominant cluster.

The X-Ray Luminosity Function and Gas Mass Function for Optically Selected Poor and Rich Clusters of Galaxies

We present the first X-ray luminosity function (XLF) for an optically selected sample of 49 nearby poor clusters of galaxies and a sample of 67 Abell clusters with z ≤ 0.15. We have extended the measured cluster XLF by more than a factor of 10 in X-ray luminosity. Our poor cluster sample was drawn from an optical catalog of groups with 0.01 ≤ z ≤ 0.03 composed of Zwicky galaxies. The X-ray emission was measured from the ROSAT all-sky survey. About 45% of the poor clusters were detected, with 0.5-2.0 keV luminosities from 1.7 to 65 × 1041 h-2 ergs s-1. These are among the X-ray brightest, optically selected poor clusters in the northern hemisphere. For this sample, the poor cluster XLF was found to be a smooth extrapolation of the rich cluster XLF. A new Hydro/N-body simulation of a hot + cold dark matter model with Ωtotal = 1, Ων = 0.2, and a baryon fraction of 7.5% was used to model and understand our observational selection effects. We found that the observed cluster gas mass function was consistent with our model.

Redshift and Optical Properties for S Statistically Complete Sample of Poor Galaxy Clusters

From the poor cluster catalog of White et al. (1996), we define a sample of 71 optically-selected poor galaxy clusters. The surface-density enhance- ment we require for our clusters falls between that of the loose associations of Turner and Gott (1976) and the Hickson compact groups (Hickson, 1982). We review the selection biases and determine the statistical comleteness of the sample. For this sample, we report new velocity measurements made with the ARC 3.5-m Dual-Imaging spectrograph and the 2.3-m Steward Observatory MX fiber spectrograph. Combining our own measurements with those from the literature, we examine the velocity distributions, velocity dispersions, and 1-d velocity substructure for our poor cluster sample, and compare our results to other poor cluster samples. We find that approximately half of the sample may have significant 1-d velocity substructure. The optical morphology, large-scale environment, and velocity field of many of these clusters is indicative of young, dynamically evolving systems. In future papers, we will use this sample to derive the poor cluster X-ray luminosity function and gas mass function (see astro-ph/9606120), and will examine the optical/X-ray properties of the clusters in more detail.From the poor cluster catalog of White et al. (1996), we define a sample of 71 optically-selected poor galaxy clusters. The surface-density enhancement we require for our clusters falls between that of the loose associations of Turner and Gott (1976) and the Hickson compact groups (Hickson, 1982). We review the selection biases and determine the statistical completeness of the sample. For this sample, we report new velocity measurements made with the ARC 3.5-m Dual-Imaging spectrograph and the 2.3-m Steward Observatory MX fiber spectrograph. Combining our own measurements with those from the literature, we examine the velocity distributions, velocity dispersions, and 1-d velocity substructure for our poor cluster sample, and compare our results to other poor cluster samples. We find that approximately half of the sample may have significant 1-d velocity substructure. The optical morphology, large-scale environment, and velocity field of many of these clusters are indicative of young, dynamically evolving systems. In future papers, we will use this sample to derive the poor cluster X-ray luminosity function and gas mass function, and will examine the optical/X-ray properties of the clusters in more detail. – 2 –

The radio properties of CD galaxies in Abell clusters. II: the VLA sample

The radio properties of a sample of 91 cD galaxies in Abell clusters are presented. The radio data consist of new 6 cm VLA maps with a resolution of 1″-2″ as well as data from the literature. The principal goal of this project was to investigate the possible effects of global and local cluster properties on the radio properties of cDs. The dominant predictor of central radio activity is the presence of x-ray cooling flows, as reported in Paper I [Burns, J. O., AJ, 99, 14 (1990)].

Ram-pressure confinement of a hypersonic jet

The results of a high-resolution, hydrodynamical simulation of an extremely supersonic jet with a Mach number (relative to the undisturbed ambient medium) of 4650 are presented. Both the Mach number and the final length of the jet are greater than those of any previously published simulation. It is found that the hypersonic jet is highly dynamic in nature with varying head velocity and shape; the importance of the back action of the cocoon upon the jet is also clearly seen. Such hypersonically moving jets can account for the observations of luminous extragalactic radio jets, associated with classical double radio sources, which appear to be overpressured with respect to the surrounding ambient medium

The evolution of X-ray clusters in a cold plus hot dark matter universe

We present the first self-consistently computed results on the evolution of X-ray properties of galaxy clusters in a Cold + Hot Dark Matter (CHDM) model. We have performed a hydrodynamic plus N-body simulation for the COBE-compatible CHDM model with standard mass components: Omega(hot) = 0.3, Omega(cold) = 0.6 and Omega(baryon) = 0.1 (h = 0.5). In contrast with the CDM model, which fails to reproduce the observed temperature distribution function dN/dT (Bryan et al. 1994b), the CHDM model fits the observational dN/dT quite well. Our results on X-ray luminosity are less firm but even more intriguing. We find that the resulting X-ray luminosity functions at redshifts z = 0.0, 0.2, 0.4, 0.7 are well fit by observations, where they overlap. The fact that both temperatures and luminosities provide a reasonable fit to the available observational data indicates that, unless we are missing some essential physics, there is neither room nor need for a large fraction of gas in rich clusters: 10% (or less) in baryons is sufficient to explain their X-ray properties. We also see a tight correlation between X-ray luminosity and gas temperature.

Propagation and stability properties of radio jets in cluster cooling flow atmospheres

We present results from the first self-consistent, multidimensional simulations of perturbed radio jets in a realistic, cluster cooling flow atmosphere. The supersonic, steady state cooling flow atmosphere used in this work is unique in that it is evolved numerically from an isothermal, equilibrium state. Jets of varying Mach number are passed through this atmosphere and their ultimate rate is found to depend strongly on Mach number. Low Mach number jets (M j =3) effectively stagnate due to the ram pressure of the cooling flow atmosphere while medium Mach number jets (M j =6-12) become unstable and disrupt in the cooling flow to form amorphous structures. High Mach number jets (M j >50) manage to avoid disruption and are able to propagate through the cooling flow