Home
Register
Preliminary list of attendees
Durham University
Visitor Accomodation
Travel to Durham
AMR simulations in Cosmology
Titles and Abstract:
Review talks
Paul Ricker (Illinois)
Dan Sheeler (Chicago)
Brad Gallagher (Chicago)
Brian O'Shea (San Diego)
Contributed talks:
Marcus Bruggen (Bremen)
Alessandro Gardini (Illinois)
Alexander Knebe (Swinburn)
Antonella Maselli (Florence)
Brian O'Shea (San Diego)
Paul Ricker (Illinois)
Yann Rasera (Saclay)
Romain Teyssier (Saclay)
Vicent Quilis (Valencia)
Pablo Araya (Groningen)
Erik-Jan Rijkhorst (Leiden)
Matthias Hoeft (Bremen)
Craig Booth (Durham)
Claudio Dalla Vecchia (Durham)
Adrian Jenkins and Takashi Okamoto (Durham)
Review talks
Paul Ricker (Illinois)
FLASH is a modular, parallel hydrodynamical code developed by the ASC Center for Thermonuclear Flashes at the University of Chicago. Originally developed to study thermonuclear runaway in compact objects, FLASH uses adaptive mesh refinement (AMR) to achieve high resolution in regions of interest and employs a framework designed to make adding new physics straightforward. As one of the original developers of FLASH, I have contributed heavily to FLASH modules that solve the Poisson equation, track particles, and handle comoving coordinates. These modules are now well-tested and give FLASH the ability to solve problems in cosmological structure formation. I will discuss the numerical methods used by FLASH that are relevant to cosmological simulation and present the results of several tests that demonstrate the code's accuracy and performance. FLASH is available for free download from the ASCI center in Chicago . My group at the University of Illinois has recently begun a web site devoted to using FLASH for cosmological simulations.
Top
Dan Sheeler (Chicago)
FLASH 2.3, the current release, represents a fast, portable, modular, well tested multi-physics simulation application. Many improvements in the application architecture have been made since initial development. Having been developed with the constraint of supporting an internal and external user base, some improvements have been implemented inconsistently throughout the code or not at all. FLASH 3.0, the future release, hopes to remedy this situation. This talk will address the plans for the future of FLASH from a code architecture perspective and from an end user point of view, specifically addressing backward compatibility and data interfaces. The talk will also address new capabilities planned for the code.
Top
Brad Gallagher (Chicago)
The Flash code produces a variety of output data. Primarily the data is in the form of Paramesh data structures which include data that describes the AMR grid and physical variables solution data. In Flash we offer the ability to output this data in serial and parallel and provide support for both HDF5 and Parallel NetCDF compatible binary output formats. Performance comparison data between the implementation of these two I/O formats in Flash will be presented. In addition the Flash code can output particle data into separate data files in a number of user defined ways. Visualizing AMR datasets, especially in 3-d is an interesting challenge and an open area of research. Mike Papka and Randy Hudson from the Futures Lab at Argonne National Laboratory have developed scalable, parallel 3d volume rendering and 3d iso-surface generation tools that render the AMR dataset directly. I will discuss how the 3d iso-surface tool, "Flashvis" works and give a demonstration of its use.
Top
Brian O'Shea (San Diego)
In this talk I will discuss Enzo, an AMR cosmology simulation code originally written by Greg Bryan. This code is a mature, fast adaptive mesh code which has been used for a wide variety of applications. I will highlight the methods used in the code and show information on performance and scaling for both adaptive mesh and large unigrid cosmological simulations. In addition, I will present a brief historical overview of previously published work done using Enzo and discuss current and future development work. The AMR code Enzo
and accompanying documentation can be downloaded. .
Top
Contributed talks
Brian O'Shea (San Diego)
Recent theoretical and numerical work suggests that the IMF of the first generation of stars was very top-heavy, tending towards stars with masses in excess of 100 solar masses. Stars in this mass range are copious emitters of ultraviolet radiation and may explode in highly energetic pair-instability supernovae with a distinct enrichment signature. I present results of simulations of the formation and death of this first generation of massive primordial stars performed using the AMR code Enzo
and their effects on their local environment.
Top
Alexander Knebe (Swinburn)
The open source adaptive mesh refinement C-code MLAPM suited for cosmological N-body simulation is going to be introduced. One of the novelties with MLAPM is the way arbitrarily shaped refinements are most efficiently navigated in computer memory. This makes the code extremely fast and allows high-resolution simulations including millions of particles on single CPU workstations in competetive times (even without parallelisation). The latest release version also includes on-the-fly halo analysis at the same accuracy level as the actual simulation. The MLAPM code
and accompanying documentation can be downloaded
Top
Antonella Maselli (Firenze)
CRASH is a grid-based radiative transfer scheme, specifically designed for cosmological simulations. It applies on precomputed arbitrary density distributions, and allows to follow the time evolution of the temperature and ionization structure of the simulated H/He gas. I will present the mean features of the implementation, which has been validated by several tests, and I will give some hints on possible extentions which could further improve the versatility of the code.
Top
Marcus Bruggen (Bremen)
I will discuss various mechanisms by which AGN can heat the intracluster medium. In particular, I will show 3D simulations of wave dissipation inside clusters that have been performed with a modified version of FLASH. One of the challenges that AGN heating models face is the nature of a posssible feedback between the radiative cooling of the ICM and the mechanical luminosity of the AGN. A simple 1D model is presented which demonstrates how a simple feedback mechanism could operate and which results in a quasistatic state for a broad range of parameters.
Top
Alessandro Gardini (Illinois)
We study the heating/cooling mechanisms in galaxy clusters that host a cooling flow by hydrodynamic simulations using FLASH. We pay particular attention to models with cavities filled by hot plasma. In order to contribute to explain the current observations, our aim is to reproduce realistic, while idealized, galaxy clusters, possibly accounting for the majority of physical phenomena that affect the behaviour of the ICM. We plan to perform a set of simulations of the ICM with different properties and including different physical heating/cooling mechanisms. We consider a set of realistic clusters, where density and temperature profiles are similar to those obtained by the Chandra data concerning cluster with a strong cool core. We assume that these clusters are in hydrostatic equilibrium and we will check their stability running pure hydrodynamic simulations. Then we will add gas cooling and compute the corresponding evolution of the ICM. Hence we will add heating mechanisms to counterbalance the cooling: namely either and both radiative transfer and hot plasma bubbles. We expect to test a duty cycle of the AGN activity as well as the possible precession of the jets. In the future more sophisticate approaches to the ICM will be adopted: in particular the magnetic field has to be introduced as well as non-thermal radio emission mechanisms for the hot plasma. We will also check the effects of turbolent mixing that can be generated in the ICM e.g. by the motion of the galaxies or by galactic winds. Moreover, a study of the gas metallicity and of the transfer of the metal rich material from the cluster core to outer regions can also be addressed.
Top
Paul Ricker (Illinois)
We are on the threshold of an era in which large observational samples (> 10^4) of galaxy clusters are available to high redshift in several wavebands. With so much data about to become available, there is corresponding interest in refining theoretical predictions for the statistical properties of clusters in different cosmological models. I will discuss an effort to produce gigaparsec-scale simulations of cluster formation including gas using the adaptive mesh refinement (AMR) code FLASH. These simulations will produce clusters in numbers comparable to those available from the forthcoming surveys while allowing us to self-consistently model cluster observables and determine the systematic effects that gasdynamics has on the statistics of clusters.
Top
Vicent Quilis (Valencia)
A new cosmological multidimensional hydrodynamic and N-body code based on an Adaptive Mesh Refinement scheme is described and tested. The hydro part is based on modern high-resolution shock-capturing techniques, whereas N-body approach is based on the Particle Mesh method. The code has been specifically designed for cosmological applications. Tests including shocks, strong gradients, and gravity have been considered. A cosmological test based on Santa Barbara cluster is also presented. The usefulness of the code is discussed. In particular, this powerful tool is expected to be appropriate to describe the evolution of the hot gas component located inside asymmetric cosmological structures.
Top
Romain Teyssier (Saclay)
Top
Yann Rasera (Saclay)
RAMSES (Teyssier 2002) is an hydrodynamical and N-body AMR code with atomic cooling, UV heating and now, Kennicutt-like star formation. We have performed a series of high resolution cosmological simulations to compute global and halo star formation history in a LCDM universe. At the same time, we developed an analytical model based on Extended Press-Schechter theory and standard cooling recipes. This model predicts simulation results taking into account resolution effects. The cross-validation between analytical model and simulations allows us to investigate the "true" average star formation history after removing resolution effects. I will present both simulation results and analytical predictions."
Top
Pablo Araya (Groningen)
I describe the first results of the application of DTFE to a set of cosmological SPH simulations. The Delaunay Tessellation Field Estimator (DTFE) is a fully self-adaptive method for reconstructing the underlying continuous density, velocity and/or other physical field from a discrete set of spatially irregularly sampled values. The method is based using the Delaunay tessellation of the point set, firstly for obtaining an estimate of the local density and secondly as a natural multidimensional interpolation interval. The major virtues of this self-adaptive and natural routine are its ability to retain the detailed anisotropic patterns in the particle distribution and its capacity to follow its (hierarchical) substructure at every represented level. Here we discuss the structure and intricate detail visible in the DTFE renderings of hydro-dynamical quantities like pressure, entropy and predicted free-free emission of forming clusters in Gadget's simulations. Its potential for a better understanding of cosmological large scale gas distributions will be discussed.
Top
Erik-Jan Rijkhorst (Leiden)
The interacting stellar winds model has been very successful in explaining observed cylindrical and bipolar shapes of planetary nebulae. However, many nebulae have a multi polar or point-symmetric shape. 3D simulations using Flash show that these seemingly enigmatic forms can be reproduced by a two-wind model in which the confining disk is warped. In order to make our models more realistic, the effects of ionisation need to be taken into account. We have recently started developing a radiation transport module for calculating ionisation rates. The implementation of this characteristics based method in a parallel AMR code like Flash raises a number of issues. Besides for planetary nebulae, this module may be of interest for cosmological applications as well.
Top
Matthias Hoeft
Using high-resolution cosmological simulations, focused on void regions, we study the formation and evolution of the galaxy population. We examine luminosities and colors and compare those properties with the results of a cluster galaxy sample. Simulations are performed with Gadget.
Top
Craig Booth (Durham)
Grafic 2 (Bertschinger, 2001. Available for download here) is a tool that allows the generation of multiscale Gaussian random fields for use in cosmological simulations.
The maximum resolution of a Gaussian random field is usually limited by the maximum size of a single FFT that can be performed. This represents a severe limitation for AMR codes, which are able to achieve a very large dynamic range. Grafic 2 overcomes this problem by generating its random numbers in real-space rather than in k-space.
In this talk I will outline the process by which Grafic 2 generates
Gaussian random fields, summarise some of the advantages of using
Grafic over more standard methods of initializing multiscale cosmological
simulations, and finally present some of our additions and extensions
to Grafic 2.
Top
Claudio Dalla Vecchia (Durham)
The observed cooling rate of hot gas in clusters is much lower than that inferred from the gas density profiles. This suggests that the gas is being heated by some source. We use an adaptive-mesh refinement code (FLASH) to simulate the effect of multiple, randomly positioned, injections of thermal energy within 50 kpc of the centre of an initially isothermal cluster with mass M_200=3x10^(14) Msol and kT=3.1 keV. We have performed eight simulations with spherical bubbles of energy generated every 10^8 years, over a total of 1.5 Gyr. Each bubble is created by injecting thermal energy steadily for 10^7 years; the total energy of each bubble ranges from 0.1--3x10^(60) erg, depending on the simulation. We find that 2x10^(60) erg per bubble (corresponding to a average power of 6.3x10^(44) erg/s) effectively balances energy loss in the cluster and prevents the accumulation of gas below kT=1 keV from exceeding the observational limits of 30 Msol/yr. This injection rate is comparable to the radiated luminosity of the cluster, and the required energy and periodic timescale of events are consistent with observations of bubbles produced by central AGN in clusters. The effectiveness of this process depends primarily on the total amount of injected energy and the initial location of the bubbles, but is relatively insensitive to the exact duty cycle of events. astro-ph/0402441
Top
Adrian Jenkins and Takashi Okamoto (Durham)
Gadget 2 is an SPH PM/Tree code parallelised under MPI, developed mainly by Volker Springel, and Naoki Yoshida, Simon White and Lars Hernquist. It is an improved version of Gadget. It uses a Peano-Hilbert bifurcation of space to improve load-balance and communication, a PM mesh, to speed-up gravity and improve accuracy at high redshifts, and has a multiple-time steps.
Takashi performed simple test SPH simulations using individual and
global timesteps. He will briefly describe the implementation of
individual timesteps in Gadget2 and show its performance and drawbacks.
Top