Sixth Annual DiRAC Science Day
→
Europe/London
Lecture Theatre A (James Clerk Maxwell Building, The King's Building)
Lecture Theatre A
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road
Edinburgh
EH9 3FD
United Kingdom
Antonin Portelli
(The University of Edinburgh)
Description
DiRAC is the integrated supercomputing facility for theoretical modelling and HPC-based research in particle physics, nuclear physics, astronomy and cosmology.
This September, The University of Edinburgh is hosting the sixth Annual DiRAC Science Day. Researchers will present scientific results achieved with this facility, and technical support teams will meet with industrial leaders working to supply leading-edge technical solutions. The day will be splitted in a plenary session in the morning and two parallel sessions (science & technical) in the afternoon. The event will be closed by a poster session. There will be 2 prizes given to posters judged to be 'the best'.
Attendees are encouraged to submit abstracts for the science parallel session and posters. Please register by using the registration form in the left side menu. If you wish to contribute to the event, please submit an abstract using the form in the same menu.
Registration is open until the 2nd of September, abstract submission is closed.
Limited funding for PhD student expenses is provided courtesy of our sponsor Mellanox. All the funding available for student travel is now gone.
Accommodation in Edinburgh is easy to find on standard booking websites. You might also want to try the Pollock Halls owned by the university at http://www.edinburghfirst.co.uk/for-accommodation/pollock-halls/.
Local organising committee: Peter Boyle, Luigi Del Debbio, Antonin Portelli (chair).
The DiRAC Science Day is supported by:
Additional sponsors: Cray, DDN, Lenovo, Mellanox, OCF, Seagate.
Participants
- Adrian Jenkins
- Adrian Tate
- Alan Real
- Andrew Dean
- Andrew Lytle
- Andrew Richardson
- Andrew Turner
- Antonin Portelli
- Azusa Yamaguchi
- Benjamin Lewis
- Chris Bouchard
- Chris Weeden
- Christine Davies
- Clare Jenner
- Colin Bridger
- Colin Smythe
- Craig McNeile
- Darren Harkins
- David Coughlin
- David Taylor
- Derek Burke
- Eduardo Martinez
- Eric Aulagne
- Eric Fauvet
- Eric Tittley
- Eugene Lim
- Felix Sainsbury-Martinez
- Georgina Ellis
- Gerardo Ramon Fox
- Guido Cossu
- James Willis
- James Wurster
- Jeremy Yates
- Jim Roche
- John Swinburne
- Jonathan Chardin
- Julia Kettle
- Justus Tobias Tsang
- Laura Keating
- Mark Hannam
- Mark Richardson
- Mark Rothwell
- Mark Wilkinson
- Matthew Bate
- Matthew Smith
- Matthew Wingate
- Matthieu Schaller
- Michael Jullien
- Michael Watson
- Monique A. Henson
- Mueller Bernhard
- Nicolas Garron
- Nobuya Nishimura
- Paul Brook
- Paul Shellard
- Peter Boyle
- Ricarda Beckmann
- Richard Booth
- Richard Bower
- Richard Regan
- Richard Rollins
- Rod Mcallister
- Roger Horsley
- Roland Young
- Sergio Martin-Alvarez
- Shaoran Hu
- Stephen McLaughlin
- Steve Davey
- Steve Smith
- Stuart Rankin
- Stuart Wilson
- Taysun Kimm
- Terry Rush
- Thomas Helfer
- Tom Goffrey
- Tony Arber
- Torben Kling Petersen
- Wardrope Ian
- William Lucas
-
-
10:00
→
10:30
Coffee & registration 30m Outside the lecture theatres
Outside the lecture theatres
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United Kingdom -
10:30
→
12:45
Plenary session Lecture Theatre A
Lecture Theatre A
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United KingdomConvener: Prof. Peter Boyle (Edinburgh)-
10:30
Opening 20mSpeaker: Dr Jeremy Yates (DiRAC)
-
10:50
DiRAC discussion 25mSpeaker: Prof. Christine Davies (University of Glasgow)
-
11:15
Simulating realistic galaxy clusters 30mSpeaker: Ms Monique A. Henson (Jodrell Bank Centre for Astrophysics, University of Manchester)
-
11:45
UKQCD review 30mHigh performance computing is now essential in research, even for the most fundamental sciences. In particle physics, the forces that bind the quark together are so strong that non-perturbative methods are required. The only way to define the theory properly is through numerical simulations, called lattice QCD (Quantum Chromo Dynamics is the theory of the strong nuclear interaction). Lattice QCD requires both large scale numerical simulations and state-of-the-art algorithms. Most of the computer resources for Lattice QCD in the UK are obtained though the DiRAC facility. I will give an overview of the various projects which use the Dirac resources and are relevant to theoretical particle physics.Speaker: Dr Nicolas Garron (University of Liverpool)
-
12:15
Gravitational waves, numerical relativity, and measuring black holes 30mThe recent first direct detection of gravitational waves relied on accurate theoretical models of the signals produced by colliding black holes. Current models were tuned to numerical solutions of Einstein's equations for the last orbits and merger of two black holes, and a large number of follow-up numerical-relativity simulations were performed to cross-check the results and test for systematic errors. DiRAC was used for many of these simulations. I will summarise the state of the field and the prospects for the future.Speaker: Prof. Mark Hannam (Cardiff University)
-
10:30
-
12:45
→
14:00
Lunch & poster session Lecture Theatre A
Lecture Theatre A
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United Kingdom-
12:45
$D$ and $D_s$ decay constants from Domain Wall Fermions with physical pion masses 1h 15mI will present the status of RBC/UKQCDs charmed meson decay constant study. I will also highlight RBC/UKQCDs wider charm physics program and outline current studies.Speaker: Mr Justus Tobias Tsang (University of Southampton)
-
12:45
3D Hydrodynamic Simulations of the Carbon Shell in a Massive Star 1h 15mStellar models are important for many areas of astrophysics, for example nucleosynthesis yields, supernova progenitor models and understanding physics at extreme conditions. One of the longest standing problems with stellar evolution models is the treatment of convection. To study convection and turbulent motions in stellar interiors, detailed 3D hydrodynamical simulations are needed. Simulations for a resolution study on the carbon shell of massive stars will be presented. Initial conditions are taken from a 15 solar mass stellar model evolved using the Geneva Stellar Evolution code (Eggenberger et. al., 2008, Ap&SS, 316, 43-54). Implicit Large Eddy simulations are computed using the hydrodynamics code PROMPI (Meakin and Arnett, 2007, ApJ, 667, 448-475). These simulations have been run on the Data Centric cluster (COSMA) within the STFC DiRAC facility. The results are analysed within the framework of one-dimensional Reynolds-Averaged Navier-Stokes equations (Chassaing et. al., 2002), where the fields within these equations are decomposed into mean and fluctuating components. We investigate the process of turbulent entrainment at convective boundaries. We find that our models are within the equilibrium entrainment regime (Fedorovich et. al., 2004, J. Atmos. Sci., 61, 281), and also partially agree with the well known meteorological entrainment law (Garcia and Mellado, 2014, J. Atmos. Sci., 1935-1955). By comparing with stellar evolution codes we hope to gain valuable insight into new prescriptions for convective boundary mixing within massive stellar models. This research will be valuable to the astronomy community as the carbon shell has never been studied in this context before.Speaker: Mr Andrea Cristini (Keele University)
-
12:45
COSMOS Intel Parallel Computing Centre: from algorithms to architectures 1h 15mPoster describing the work of the Intel Parallel Computing Centre associated with the COSMOS@DiRAC SMP Facility hosted in Cambridge. This describes the unique hybrid architecture of the COSMOS system, including shared-memory capabilities and accelerators (Xenon Phi) and radical optimisations that have been undertaken for new many-core processors. Highlights include refactoring of the Modal Planck pipeline (evaluating higher-order correlators in the Planck CMB data) which has been the subject of several publications and the HPCwire award at SC ’15 for Best HPDA application. Ongoing work with Intel on highly parallel in-situ visualisation is also summarised.Speaker: Prof. Paul Shellard (University of Cambridge)
-
12:45
Dynamical perturbations on cosmic string solutions in full general relativity 1h 15mWhile static axis-symmetric solutions of gravitating cosmic strings have been found numerically, dynamical solutions are much harder propositions and little has been done in this aspect. We developed a solver for local U(1) together with Einstein field equations. Using the solutions of strings with perturbations from Vachaspati and Garfinkle, we find initial data for our string and construct dynamical solutions of gravitating cosmic string in numerical relativity. This will allow us to investigate various physical scenarios like the collision of wave-trains.Speaker: Mr Thomas Helfer (Kings College London)
-
12:45
Exploring Large Scale Flows and Molecular Cloud Formation in Galaxy Simulations 1h 15mF. G. Ramón Fox (1) and Ian Bonnell (1) (1)University of St. Andrews Galactic scale gas flows feed the growth of molecular clouds, the regions where stars form in high-density cores. These flows also play a role in driving the internal dynamics of these clouds, which affects their overall stability and star formation activity. The triggering of star formation involves a connection between large and small-scale dynamical processes in galaxies, which can be explored using high-resolution hydrodynamical simulations. We present results of our ongoing work in high-resolution N-body and Smoothed Particle Hydrodynamics simulations of a galaxy model developing a realistic spiral arm structure. Our model has a mass resolution in the gas component down to 45 solar masses per gas particle. This allows us to study gas flows in a self-consistent galaxy and understand their role in the formation of molecular clouds. The high resolution allows us to follow individual molecular clouds and study their properties in different environments of a galaxy. The model galaxy also allows us to explore the effects of spiral arms on large scale flows. These simulations aid in understanding molecular cloud growth and star formation from a galactic perspective.Speaker: Mr Gerardo Ramon Fox (University of St. Andrews)
-
12:45
Galactic Feedback and the Circumgalactic Medium 1h 15mThe circumgalactic medium (CGM) encompasses gas that extends from the outskirts of a galaxy to the edge of its halo, mediating gas accretion and outflows. Observations of absorption in background quasar spectra reveal a wealth of information about the structure of the CGM, in particular that it is an environment polluted by metals and cool gas from galactic outflows. By using simulations of galaxies forming in a cosmological volume we can best understand the evolution of these outflows, and put constraints on their subgrid model parameters through careful comparison to observations. We present a suite of 10 pc resolution cosmological zoom simulations, some with 100 pc resolution enforced in the CGM, where we vary the prescriptions for star formation, supernova feedback, and/or radiative transfer, and show how these different models affect the make-up of the CGM. We also discuss our efforts comparing outputs with observations, highlighting the strengths and weaknesses of our models.Speaker: Dr Mark Richardson (University of Oxford)
-
12:45
GRChombo: Numerical relativity with adaptive mesh refinement 1h 15mGRChombo is a new numerical relativity code specifically designed to efficiently and easily simulate the disparate length scales encountered in GR using fully adaptive mesh refinement and the latest parallel computing techniques. Applications range from gravitational wave signals produced from black hole mergers through to critical phenomena for high-dimensional black holes.Speakers: Dr Eugene Lim (King's College London), Mr Thomas Helfer (Kings College London)
-
12:45
Grid: A next generation data parallel C++ QCD library 1h 15mIn this proceedings we discuss the motivation, implementation details, and performance of a new physics code base called Grid. It is intended to be more performant, more general, but similar in spirit to QDP++[ 6 ]. Our approach is to engineer the basic type system to be consistently fast, rather than bolt on a few optimised routines, and we are attempt to write all our optimised routines directly in the Grid framework. It is hoped this will deliver best known practice performance across the next generation of supercomputers, which will provide programming challenges to traditional scalar codes. We illustrate the programming patterns used to implement our goals, and advances in productivity that have been enabled by using new features in C++11Speaker: Dr Azusa Yamaguchi (Software Architect)
-
12:45
Impacts of nuclear physics uncertainty on nucleosynthesis on a Monte-Carlo Method 1h 15mA variety of elements has been synthesised in the evolution of stars and supernovae since the Big-Bang. The thermonuclear nuclear reaction is the physical process responsible for the production of elements, and also drives nuclear-burnings in stellar interiors. We need to know accurate reaction rates to help understand the cosmic origin of elements. In our project, we try to identify “key” nuclear reactions producing heavy nuclei baryon iron, based on large-scale Monte-Carlo simulations. I will show recent results on heavy element nucleosynthesis by the s-process and p-process.Speaker: Dr Nobuya Nishimura (Keele University)
-
12:45
Magnetic fields in galaxies: the small-scale dynamo in a cosmological context 1h 15mRegardless of being acknowledged as a relevant factor in several astrophysical processes, the study of magnetic fields has been one of the most elusive areas of cosmology and galaxy evolution up to this day. The reason for it are both the complications associated with their observation and measurements, and the intricate difficulty to model them. However, the resources required to tackle them are becoming available and new discoveries are starting to take place. On the observational side, recent results suggest that the magnetic field was already on the order of microGauss in galaxies at high redshifts. Meanwhile on the numerical side, computers start to possess the required power and memory to model these magnetic fields efficiently, allowing various amplification mechanisms to be tested. One of the major puzzles to be explained is precisely the rapid dynamo mechanism that transforms the weak values of the magnetic field expected at the beginning of the reionization era into the strong fields found in galaxies. Amongst the most favoured explanations is a fast amplification by small-scale dynamos, due to the low e-folding time characteristic of turbulence eddies. To shed some light on this question, we present in this study the first magnetohydrodynamical cosmological zoom-in simulations of a Milky-Way-like galaxy with enough resolution to capture the small-scale dynamo (i.e. down to ~10pc cell size) operating on the galactic Interstellar Medium (ISM). The simulations, performed with the code RAMSES, also include realistic turbulent star formation and dynamical supernovae (SNe) feedback. Starting from various physically motivated initial strengths for the magnetic field, we find that amplification does take place in a cosmological context, where the processes of interaction and accretion of pristine gas are not strong enough to halt it. This amplification typically follows an exponential law, where the magnetic energy per unit mass grows as E_(mag,0) e^(Gamma t). After an initial epoch of fast increase during the initial collapse where Gamma ~ 5 Gyr^-1, the process slows down to Gamma=1.2 Gyr^-1. This value of Gamma is still in the same order of magnitude as the galaxy’s angular rotation frequency Omega ~ 5. We also find that the kinematic energy spectrum confirms the presence of Kolmogorov turbulence and the magnetic energy density spectrum display different periods of activity of the dynamo. The magnetic field profiles show a tangled structure around the disk plane, with the toroidal and the radial components dominating along the z-axis away from the centre. This is in agreement with previous studies whenever feedback is included. The radial profile however, displays a more organised structure, and in the central region of the galaxy the magnetic field is dominated by the gas kinematics. The star formation is also affected by strong magnetic fields, as the presence of a strong initial comoving magnetic seed (B_0 = 10^(-12) Gauss) triggers a reduction of the stellar content by 16% with respect to a weak initial comoving magnetic concordant with a Biermann battery seeded field (B_0 = 10^(-20) Gauss) after 1 Gyr. This study takes the first steps to study the evolution of the magnetic field in galaxies within a cosmological context and finds that amplification is already occurring with ~10pc resolution and turbulent star formation and dynamical SNe feedback models.Speaker: Mr Sergio Martin-Alvarez (University of Oxford)
-
12:45
Measuring non-Gaussianity in galaxy surveys: a new window on the Universe 1h 15mWe apply highly efficient modal methods (developed during the Planck analysis) to large-scale structure distributions in order to optimally evaluate the bispectrum in three dimensions. Currently, deployed on dark matter distributions from N-body codes, this tool is intended for estimating higher-order correlation functions from huge galaxy surveys like Euclid.Speaker: Prof. Paul Shellard (University of Cambridge)
-
12:45
Models of Fully-Convective Stars: Differential Rotation, Meridional Circulation and Residual Entropy 1h 15mLow-Mass stars are typically fully convective, and as such their dynamics may differ significantly from sun-like stars. Here we present a series of 3D anelastic simulations of fully convective stars, designed to investigate how the meridional circulation, the differential rotation, and entropy are affected by varying stellar parameters, such as the luminosity or the rotation rate. We also investigate, more specifically, a theoretical model in which isorotation contours and residual entropy ($S' = S - S_{r}$) are intrinsically linked via the thermal wind equation (as proposed in the Solar context by Balbus 2009). We have selected our simulation parameters in such as way as to span the transition between Solar-like differential rotation (fast equator + slow poles) and "anti-Solar" differential rotation (slow equator + fast poles), as characterised by the convective Rossby number and $\Delta\Omega$. We illustrate the transition from single-celled to multi-celled MC profiles, and from positive to negative latitudinal entropy gradients. We find generally poor correlation between the residual entropy profile and the isorotation contours, suggesting that the link between these quantities is not as strong as suggested by some previous models.Speaker: Mr Felix Sainsbury-Martinez (University of Exeter)
-
12:45
Multigrid for Chiral Fermions 1h 15mWe present a new class of multigrid solver algorithm suitable for the solution of 5d chiral fermions such as Domain Wall fermions and the Continued Fraction overlap. Unlike HDCG, arXiv:1402.2585, the algorithm works directly on a nearest neighbour fine operator. The fine operator used is Hermitian indefinite, for example Γ5Ddwf, and convergence is achieved with an indefinite matrix solver such as outer iteration based on conjugate residual. As a result coarse space representations of the operator remain nearest neighbour, giving an 8 point stencil rather than the 81 point stencil used in HDCG. It is hoped this may make it viable to recalculate the matrix elements of the little Dirac operator in an HMC evolution.Speaker: Prof. Peter Boyle (Edinburgh)
-
12:45
Radiation hydrodynamics of mini-haloes and their contribution to reionisation 1h 15mReionisation in the early universe is likely driven by LyC photons from dwarf galaxies. Using high-resolution, cosmological radiation-hydrodynamic simulations, we study the escape of LyC photons from mini-haloes with the mass Mhalo < 1e8 Msun. Our simulations include a new thermo-turbulent star formation model, non-equilibrium photo-chemistry, and important stellar feedback processes (photo-ionisation by young massive stars, radiation pressure from UV and IR photons, and mechanical supernova explosions). We find that the photon number-weighted mean escape fraction in mini-haloes is higher than that in atomic-cooling haloes, although the instantaneous fraction in individual haloes varies significantly. Because star formation is very stochastic and dominated by a single or a few gas clumps, the escape fraction is generally determined by radiation feedback, rather than supernova explosions. Interestingly, the resulting stellar mass of the proto-galaxies in the mini-halos is found to follow the empirical stellar mass-to-halo mass relation in the local Universe. We discuss the importance of the photons from mini-haloes for reionisation of the universe.Speaker: Dr Taysun Kimm (KICC and IoA, University of Cambridge)
-
12:45
Radiative transfer simulations of the lyman alpha forest: Does the QSOs dominate the UVB at z~6 ? 1h 15mLyman-alpha forest data probing the post-reionization Universe shows surprisingly large opacity fluctuations over rather large ( 50 comoving Mpc/h) spatial scales. We model these fluctuations using a hybrid approach utilizing the large volume Millennium simulation to predict the spatial distribution of QSOs combined with smaller scale full hydrodynamical simulation performed with RAMSES and post-processed with the radiative transfer code ATON. We produce realictic mock absorption spectra that account for the contribution of galaxies and QSOs to the ionising UV background. This improved models confirm our earlier findings that a significant (>50%) contribution of ionising photons from QSOs can explain the large reported opacity fluctuations on large scales. The inferred QSO luminosity function is thereby consistent with recent estimates of the space density of QSOs at this redshift. Our simulations still somewhat struggle, however, to reproduce the very long (110 comoving Mpc/h) high opacity absorption through observed in ULAS J0148+0600, perhaps suggesting an even later end of reionization than assumed in our previously favoured model. Medium-deep/medium area QSO surveys as well as targeted searches for the predicted strong transverse QSO proximity effect whould illuminate the origin of the observed large scale opacity fluctuations. They would allow to substantiate whether UV fluctuations due to QSO are indeed primarily responsible, or whether significant contributions from other recently proposed mechanisms such as large scale fluctuations in temperature and mean free path (even in the absence of rare bright sources) are required.Speaker: Dr Jonathan Chardin (Institute of Astronomy (IOA) and Kavli Institute for Cosmology Cambridge (KICC), University of Cambridge)
-
12:45
Recent Results from the Multi-dimensional Stellar Implicit Code 1h 15mStellar evolution is influenced by a wide range of complex, multi-dimensional phenomena. The MUlti-dimensional Stellar Implicit Code, MUSIC, is a new code developed in Exeter with international collaborators that is designed to study such processes. MUSIC is a fully compressible, time-implicit code, which utilises realistic opacity and equation of state data, and is capable of simulating a wide range of Mach number flows simultaneously. We present recent results using MUSIC to study a young solar type star. In the first study the non-local aspect of stellar convection, and its influence on key quantities such as the convective velocity and overshooting layer width is discussed. In the second study, we present the first calculations describing multi-dimensional structure of accreting young stars. We discuss how MUSIC simulations of accretion can be used to refine an implementation of accretion processes within one-dimensional stellar evolutionary models. Ultimately insight from multi-dimensional simulations will be used to build and improve models of stellar evolution. Finally we discuss ongoing studies of shear driven mixing, rotation, and chemical mixing across convective boundaries using MUSIC.Speaker: Dr Tom Goffrey (University of Exeter)
-
12:45
Semi-leptonic B−decays 1h 15mIn the Standard Model of elementary particle physics, b-quarks are the heaviest quarks forming hadronic bound states. Their large mass allows for many decay modes and by that tests of the Standard Model. Using numerical lattice QCD simulations we calculate decays of B-mesons, quark-antiquark pairs with one constituent a b-quark, into one other hadron and leptons. These calculations lead to determinations of CKM matrix elements and form factors which can be compared to experimental results.Speaker: Dr Oliver Witzel (University of Edinburgh)
-
12:45
Shaken and stirred - the role of turbulence, rotation, magnetism, and radiation in the formation of protostars 1h 15mWe expand on Lewis, et al. (2015) and (2016, submitted), which considered only ideal MHD simulations of the collapse of molecular cloud cores without turbulence, and use radiation magnetohydrodynamical calculations to explore how the physics of turbulence, rotation, magnetism and radiation influences the formation of protostars. The inclusion of a flux limited diffusion radiative transfer (FLD R-T) scheme in the SPH calculations (i.e. an SPRMHD scheme) promotes the formation of larger discs, as opposed to the very small and dense discs produced by the MHD only calculations. We also find that the gravitational collapse proceeds in a very different manner in cores with transonic turbulence compared with subsonic cores across a wide variety of field strengths and geometries. Cores with subsonic - and in particular very subsonic - turbulence still contain a bipolar jet and (depending on the magnetic field) form a pseudo-disc, albeit without the symmetry seen for laminar cores. Transonic (i.e. ~ Mach 1) cores are highly disrupted by the turbulent motion, which acts to suppress the formation of a pseudo-disc and hence a bipolar outflow. However, increasing the initial angular momentum of the core so that the rotational and turbulent energies are approximately equal allows even turbulent cores to produce jets and outflows.Speaker: Mr Benjamin Lewis (University of Exeter)
-
12:45
SWIFT - Vectorisation 1h 15mSWIFT - Smoothed-Particle Hydrodynamics With Interdependent Fine-grained Tasking. We will describe the methods and techniques used to vectorise SWIFT and present the improvement to performance due to vectorisation.Speaker: Mr James Willis (ICC, Durham University)
-
12:45
The intergalactic medium at the end of reionization 1h 15mObservations of absorption lines in the spectra of QSOs out to redshift 7 provide an important probe of the intergalactic medium (IGM) at the tail end of reionization. In this talk I will discuss my work using high-resolution cosmological simulations to model the enrichment and ionization state of the high-redshift IGM. I will present results from cosmological simulations run with different hydrodynamic solvers and different feedback models. I will show results from a large boxes and zoom-in runs, some of which also include radiative transfer. I will demonstrate how some common observational probes of the neutral fraction of the IGM, such as the extent of the highly ionized regions around quasars, are affected by the environment of the host halo. I will further show that observations of metal-line absorbers allow us to test the effect of galactic feedback on the IGM and to place constraints on commonly used wind prescriptions. Finally, I will discuss how these absorption lines may also help us understand the nature of the sources responsible for reionization.Speaker: Ms Laura Keating (Institute of Astronomy, University of Cambridge)
-
12:45
The MACSIS Project: Simulating the giants of the Universe 1h 15mAs the largest gravitationally bound structures in the Universe, galaxy clusters offer a crucial opportunity to study both the physics of structure formation and cosmology. There has been a wealth of work characterising the structure and properties of galaxy clusters in numerical simulations, however these have been focussed on dark matter only simulations or have been restricted to small samples of high mass (>10^15 solar masses) clusters. The studies in the latter category are typically limited in their scope due to their small sample size or since they lack essential baryonic physics, for example feedback from Active Galactic Nuclei (AGN). Since galaxy clusters are a sensitive probe of the late time evolution of the Universe, further study of the properties of massive galaxy clusters in simulations with realistic baryonic physics is needed. The MACSIS project is a set of hydrodynamical simulations of massive galaxy clusters which extends the BAHAMAS sample of galaxy groups and clusters to higher masses. These simulations use the sub-grid physics models developed in the BAHAMAS project, which includes feedback from supernovae and active galactic nuclei. Taken with the BAHAMAS sample, the simulations form a population which covers almost two orders of magnitude in mass, with more than 250 clusters with masses greater than 10ˆ15 solar masses at z=0. In this poster I will show that the MACSIS project reproduces key cluster observables and also makes predictions for how cluster properties evolve over time. Since these simulations consist of both dark matter only and hydrodynamic simulations, I will also present how we have used these simulations to determine the impact of baryons on cluster structure and observables.Speaker: Ms Monique A. Henson (Jodrell Bank Centre for Astrophysics, University of Manchester)
-
12:45
The Spin of the Proton 1h 15mThe proton consists of two valence up quarks, one down quark together with a `sea' of quark anti-quark pairs and gluons. How each constituent contributes to the total spin of the proton has remained a mystery for many years. In particular the quark contribution is much smaller than expected. We discuss here our lattice QCD determination of the quark contribution, using a novel technique, based on a field theoretic application of the Feynman-Hellmann theorem.Speaker: Dr Roger Horsley (University of Edinburgh)
-
12:45
Using the AREPO code to study the ISM 1h 15mWe show how the AREPO code can be used to follow the ISM in spiral type galaxy simulations at unprecedented scales. The resolution extends from 6 kpc down to sub-pc scales and includes important physical processes such as supernova feedback, time-dependent chemistry and self-shielding, and gas self gravity. Such resolution means that we can resolve sub-structure within individual molecular cloud and therefore do not need a "clumping factor" in the chemical model. This allows predictive models to be made of the HI, CO, and C+ emission from such galaxies. Moreover, this also allows the individual molecular clouds to be extracted from the simulations for studying star formation at higher resolution.Speaker: Dr Rowan Smith (University of Manchester)
-
12:45
-
14:00
→
15:30
Science parallel session Lecture Theatre B
Lecture Theatre B
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United KingdomConvener: Dr Matthew Wingate (University of Cambridge)-
14:00
THE HVP contribution to the anomalous magnetic moment of the muon 30mWe report on the HPQCD calculation of the u/d HVP contribution to a_mu, discussed in arXiv:1601.03071. This allows us to obtain a total HVP contribution from u, d, s and c quarks and including an estimate of disconnected pieces and QED and isospin effects of 666(6)(12) x 10^{-10}. Our result implies a discrepancy between the experimental determination of a_mu and the Standard Model of 3 sigma. We discuss prospects for improvements to this calculation underway with the MILC and Fermilab Lattice Collaborations.Speaker: Prof. Christine Davies (University of Glasgow)
-
14:30
Robustness of Inflation to Inhomogenous Inflation 20mWe investigate the robustness of inflation to inhomogenous initial conditions using the numerical relativity package GRCHOMBO. We show that small field inflation is generally less robust than large field inflation. Furthermore, we show that while increasing inhomogeneities can lead to the formation of black holes during inflation, the black holes will not achieve the Nariai mass and hence will never form dominant black hole spacetimes that can stop successful inflation. Finally, we investigate a class of initial conditions with a mix of expanding and collapsing regions, and find that inflation is successful as long as the spacetime is initially open.Speaker: Dr Eugene Lim (King's College London)
-
14:50
A critical view of the interstellar medium modelling in cosmological simulations 20mThe accurate modelling of stellar feedback from massive stars to the interstellar medium (ISM) is of crucial importance for understanding galaxy formation. Failure to include these processes in hydrodynamic simulations leads to the catastrophic overproduction of stars relative to observations. Supernovae (SNe) and winds from massive stars are a major source of stellar feedback but correctly capturing the evolution of stellar-driven outflows requires a very high resolution. Typically, this problem is countered with the adoption of sub-grid recipes for feedback, injecting energy and/or momentum as well as mass and metals into the ISM. Such recipes can involve the simple thermal or kinetic injection of the canonical 10^51 ergs per SN into surrounding gas, delaying of cooling or a more complicated scheme where the correct energy and momentum is injected according to the stage of the SN blastwave resolved. Simulation results can be highly sensitive to the numerical details of the sub-grid recipe employed, thus questioning the accuracy and reliability of these models. We present a novel implementation of star formation and stellar feedback in the moving-mesh code AREPO and use it to study systematically the effectiveness of several feedback schemes commonly found in the literature. We carry out a variety of simulations of both isolated systems and cosmological 'zoom in' simulations to unpick the dependence of results on resolution and choice of sub-grid recipe with the aim of developing an accurate model of SN feedback and the ISM in general.Speaker: Mr Matthew Smith (Institute of Astronomy, Cambridge)
-
15:10
Magnetic fields in galaxies: the small-scale dynamo in a cosmological context 20mRegardless of being acknowledged as a relevant factor in several astrophysical processes, the study of magnetic fields has been one of the most elusive areas of cosmology and galaxy evolution up to this day. The reason for it are both the complications associated with their observation and measurements, and the intricate difficulty to model them. However, the resources required to tackle them are becoming available and new discoveries are starting to take place. On the observational side, recent results suggest that the magnetic field was already on the order of microGauss in galaxies at high redshifts. Meanwhile on the numerical side, computers start to possess the required power and memory to model these magnetic fields efficiently, allowing various amplification mechanisms to be tested. One of the major puzzles to be explained is precisely the rapid dynamo mechanism that transforms the weak values of the magnetic field expected at the beginning of the reionization era into the strong fields found in galaxies. Amongst the most favoured explanations is a fast amplification by small-scale dynamos, due to the low e-folding time characteristic of turbulence eddies. To shed some light on this question, we present in this study the first magnetohydrodynamical cosmological zoom-in simulations of a Milky-Way-like galaxy with enough resolution to capture the small-scale dynamo (i.e. down to ~10pc cell size) operating on the galactic Interstellar Medium (ISM). The simulations, performed with the code RAMSES, also include realistic turbulent star formation and dynamical supernovae (SNe) feedback. Starting from various physically motivated initial strengths for the magnetic field, we find that amplification does take place in a cosmological context, where the processes of interaction and accretion of pristine gas are not strong enough to halt it. This amplification typically follows an exponential law, where the magnetic energy per unit mass grows as E_(mag,0) e^(Gamma t). After an initial epoch of fast increase during the initial collapse where Gamma ~ 5 Gyr^-1, the process slows down to Gamma=1.2 Gyr^-1. This value of Gamma is still in the same order of magnitude as the galaxy’s angular rotation frequency Omega ~ 5. We also find that the kinematic energy spectrum confirms the presence of Kolmogorov turbulence and the magnetic energy density spectrum display different periods of activity of the dynamo. The magnetic field profiles show a tangled structure around the disk plane, with the toroidal and the radial components dominating along the z-axis away from the centre. This is in agreement with previous studies whenever feedback is included. The radial profile however, displays a more organised structure, and in the central region of the galaxy the magnetic field is dominated by the gas kinematics. The star formation is also affected by strong magnetic fields, as the presence of a strong initial comoving magnetic seed (B_0 = 10^(-12) Gauss) triggers a reduction of the stellar content by 16% with respect to a weak initial comoving magnetic concordant with a Biermann battery seeded field (B_0 = 10^(-20) Gauss) after 1 Gyr. This study takes the first steps to study the evolution of the magnetic field in galaxies within a cosmological context and finds that amplification is already occurring with ~10pc resolution and turbulent star formation and dynamical SNe feedback models.Speaker: Mr Sergio Martin-Alvarez (University of Oxford)
-
14:00
-
14:00
→
15:30
Technical parallel session Lecture Theatre A
Lecture Theatre A
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United KingdomConvener: Dr Antonin Portelli (The University of Edinburgh)-
14:00
Emerging technologies 30mSpeaker: Prof. Peter Boyle (Edinburgh)
-
14:30
Grid, a next generation data parallel C++ library 20mWe discuss the development of a new data parallel library targeted to any problem requiring solving partial differential equations on a regular grid. The main motivation in the development is the proliferation of the hierarchies of parallelism in modern architectures that requires constant adaptations of large parts of the codebases for each new architecture like the new Intel Knights Landing. Our code is mainly using high level C++, localising the architecture dependence in a small part of the code and guaranteeing performance by expression templates and a organization of memory structures that automatically follows the underlying vector units. We review the status and design choices with a particular stress on the lattice QCD performance of the code.Speaker: Dr Guido Cossu (Edinburgh University)
-
14:50
An Overview of the DiRAC-3 Benchmark Suite 20mIn preparation for the procurement of their next generation of supercomputers, DiRAC have developed a suite of benchmarks based on the software in use on their current systems. Working with vendors, this will allow us to characterise the performance of key particle physics and astrophysics algorithms on newly available hardware. This talk will provide an overview of the design and development process and highlight some of the code optimisations we were able to achieve on behalf of researchers.Speaker: Dr Richard Rollins (DiRAC / University of Manchester)
-
15:10
SWIFT - Scaling on next generation architectures 20mThe next-generation cosmological code SWIFT has been demonstrated to outperform current codes by more than an order of magnitude on ordinary x86-based clusters such as the cosma system in DiRAC. This has been achieved thank to the use of better algorithm, task-based parallelism and asynchronous communications. However, the next generation of machines that will contain many more cores per node, clocked at lower frequencies, brings more challenges. To recover the same performance than on a regular Xeon-based system, a code has to scale twice as well on a KNL architecture. This implies that all parts of the code have to be well-suited to this paradigm and have to be written in parallel. In this talk we present some improvements to the SWIFT code aimed at tackling these challenges. Parallelisation of the task scheduler and of most of the operations performed in-between time-steps is now necessary and brings significant improvements to the code scalability on KNL and as a knock-on effect on regular Xeon chips.Speaker: Mr James Willis (ICC, Durham University)
-
14:00
-
15:30
→
16:00
Coffee break & poster prize announcement 30m Outside the lecture theatres
Outside the lecture theatres
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United Kingdom -
16:00
→
17:00
Science parallel session Lecture Theatre B
Lecture Theatre B
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United KingdomConvener: Dr Matthew Wingate (University of Cambridge)-
16:00
From seed to supermassive: growing black holes at high redshift 20mThe growth of supermassive black holes (SMBHs) powering quasars at redshift z=6 and above is still poorly understood. Efforts so far have either concentrated on black hole (BH) seed formation mechanisms, or approximated the SMBHs using massive seeds in low resolution simulations. The work presented here will bridge the gap, following the growth of a SMBH from its origins as a stellar mass seed at high redshift, to its massive representation at redshift z=6. The growth of this object is studied using RAMSES with a custom zoom-within-zoom refinement scheme that allows the gas to be tracked over many orders of magnitude, and offers new insights into the accretion patterns of high redshift black holes.Speaker: Ms Ricarda Beckmann (University of Oxford)
-
16:20
Planet formation and disc evolution 20mPlanet formation occurs in accretion discs that surround young stars. Using DiRAC we have been able to characterize the dynamics of these discs throughout the various stages of their evolution from massive, self-gravity dominated discs through to their eventual dispersal through photo-evaporation. I will present new insights that have arisen out of these simulations on topics including turbulence, stability and fragmentation of both gas in the self-gravitating phase; along with the implications for planet formation. I will also discuss new results on the conditions under which interactions with external stars can trigger fragmentation and the role that nearby massive stars play in disc dispersal.Speaker: Dr Richard Booth (Institute of Astronomy, Cambridge, UK)
-
16:40
Modelling the birth of a star 20mNumerical simulations are required to model the complicated physical processes that occur during the formation of a star. As we learn more about the star formation process, we begin to include even more complex processes, which can become numerically expensive. Thus, not only is the new physics required, but so are novel algorithms and HPC facilities. I will discuss our research on star formation, in which our simulations start from a rotating spherical gas cloud. We find that, using realistic magnetic field strengths, discs do not form in models using ideal magnetohydrodynamics (MHD), which contradicts observations. However, discs do form if non-ideal MHD is included - and only if the Hall effect is included and a particular initial geometry is used. Although more realistic, modelling non-ideal MHD is a numerical challenge since the timestep required to resolve these processes can be hundreds of times smaller than required by other conditions (e.g. the Courant condition). These numerical difficulties will be discussed in concert with the astrophysical results.Speaker: Dr James Wurster (University of Exeter)
-
16:00
-
16:00
→
17:00
Technical parallel session Lecture Theatre A
Lecture Theatre A
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United KingdomConvener: Dr Antonin Portelli (The University of Edinburgh)-
16:00
Co-design of Cray Software Components 30mHPC Co-design is a set of interactive, dependent design activities used as a vehicle for simultaneous improvement of HPC Software and Hardware over a longer period. We describe the co-design model that is used by the Cray EMEA Research Lab and also the way that co-design informs technical project management. We then describe some specific co-design activities in the area of memory hierarchy design, heterogeneous systems and workflow management and show how those projects might affect future hardware specifications.Speaker: Adrian Tate (Cray)
-
16:30
AAAI, Cloud & Data Managment; DiRAC in the National E-Infrastructure 30mThe National e-Infrastructure is currently (re)forming in the UK. It's key theme is access to the appropriate resources. It's core consituents are: - Authentication, Authorisation and Accounting Infrastructure - Use of Cloud technologies to allow academic researchers and business to run workflows using the appropriate IT services - Data Movement services that allow the researcher to find, select and combine data and move it to an appropriate place/service This talk will discuss how DiRAC will play its part in rolling out these services.Speaker: Dr Jeremy Yates (DiRAC)
-
16:00
-
17:00
→
18:00
Drink reception 1h Outside the lecture theatres
Outside the lecture theatres
James Clerk Maxwell Building, The King's Building
Peter Guthrie Tait Road Edinburgh EH9 3FD United Kingdom
-
10:00
→
10:30