Title: High Performance computing in Particle Physics
1- High Performance computing in Particle Physics
Cosmology - Application to Neutrino Parameters Correlations
- Nidal CHAMOUN
- Department of Physics,
- HIAST,
- Damascus, Syria
HIAST
2Table of Contents
- Basics
- Need for Powerful Computing in High Energy
Physics - Art of Cosmological Simulations
- The LHC Grid
- Neutrino Parameters model with many free
parameters
3Powerful Computing in High Energy Physics
- Theoretical particle physics is an integral part
of the world-wide activities to search for new
physics beyond the Standard Model of High Energy
Physics. - In order to distinguish signs of non-standard
physics from our present description of
elementary particle interactions it is mandatory
to have theoretical predictions originating from
the underlying theory alone from a priori
computations and without further approximations -
4Powerful Computing in High Energy Physics
- Several applications in computational particle
physics are still far beyond the reach of
state-of-the-art computers, whenever following
the evolution of even simple dynamics equations
responsible for very complex behavior requires
inordinately long execution times. - For instance, In the approach of Lattice Gauge
Theory the continuum of nature is replaced with a
discrete lattice of space-time points.
5Powerful Computing in High Energy Physics
- This lattice approximation allows for numerical
simulations on massively parallel computer
architectures. - Furnished with this conceptual tool, the high
precision experimental data as expected from the
newly planned accelerators can be interpreted in
a clean manner not suffering from any built-in
approximations. - However, despite its space-time economy, the
lattice needs the power of the world's largest
supercomputers to perform the calculations that
are required in the complex problem of solving
the complicated equations describing elementary
particle interactions.
6Powerful Computing in High Energy Physics
- This is the case when we integrate, say, over
configuration space of a three-dimensional
lattice system of too many sites, requiring up to
Tera Monte Carlo steps, which is still an
untreatable task. - Extensive use of parallelism is the main avenue
to boost computer performance.
7Art of Cosmological Simulations
- During the last 10 years new extensive
observations of the Universe were made using both
ground-based telescopes and space instruments. - The huge observational progress has been
accompanied by considerable effort in our
theoretical understanding of the formation of
different components of the observed structure of
the Universe galaxies and their satellites,
clusters of galaxies, and super-clusters. - the standard
- cosmological model
8Art of Cosmological Simulations
- A substantial part of this theoretical progress
is due to the improvement of numerical methods
and models, which mimic structure formation on
different scales using a new generation of
massive parallel supercomputers. - The nonlinear evolution of cosmological
fluctuations can be studied only numerically. The
details of galaxy formation must be followed
using hydrodynamic simulations. - However, many features can already be studied by
semi-analytical methods
9Art of Cosmological Simulations
- Modern astrophysics and cosmology are
characterized by dealing with complex problems
whose dynamical range covers extended space-time
scales, ranging from stability of solar systems
to physics of quasars and large scale
structures. . - Thus, the requirements for modern cosmological
simulations are extreme - a very large dynamical range for force resolution
and many millions of particles are needed. -
10Art of Cosmological Simulations
- Case study Helium and deuterium abundances as
a test for the time variation of the fine
structure constant and the Higgs vacuum
expectation value, - J. Phys. G Nucl. Part. Phys. 34 (2007) 163176,
by Chamoun, Mosquera, Landau Vucetich
We used semi-analytical methods (1991 Astrophys.
J. 378, 50418) to calculate the abundances of
helium and deuterium produced during Big Bang
nucleosynthesis assuming the fine structure
constant and the Higgs vacuum expectation value
may vary in time
11Time Variation of Fundamental Constants ??
Nucleosynthesis
12Time Variation of Fundamental Constants ??
Nucleosynthesis
- We assumed that the discrepancy between SBBN
estimation for 4He and D and their observational
data is due to a change in time for the
fundamental constants the Higgs vev v, the fine
structure constant a. - We analysed the dependence of the 4He and D
abundances on these fundamental constants within
perturbation theory and on deviations with
respect to the mean value of the baryonic density
13Time Variation of Fundamental Constants ??
Nucleosynthesis
- The calculation of the heavier elements
abundance requires much more computations and
coupled equations to be solved.
14Table of Contents
- Basics
- Need for Powerful Computing in High Energy
Physics - Art of Cosmological Simulations
- The LHC Grid
- Neutrino Parameters model with many free
parameters
15The Grid What is it?
processing power on demandlike electrical power
a virtual metacomputer for the seamless access to
dispersed resources
coordinated resource sharing and problem solving
in dynamic, multi-institutional, virtual
organizations
16The GridVirtual Organizations
VO2
VO1
group of individuals or institutes who are
geographically distributed but appear to
function as one single unified organization
Internet
O1
O3
O2
17The Grid What is it?
reality is catching up fast with the dream
- Distributed computing a method of computer
processing in which different parts of a program
run simultaneously on two or more computers that
are communicating with each other over a network.
It is a type of segmented or parallel computing.
It also requires that the division of the program
take into account the different environments on
which the different sections of the program will
be running.
18What the Grid can do ?
What type of applications will the Grid be used
for?
the first big-time users of the Grid will
probably be scientists with challenging
applications that are simply too difficult to do
on just one set of computers.
19Computational problems
- Degree of parallelism ability to split into
many smaller sub-problems that can be worked on
by different processors in parallel - Degree of granularity dependence of each
sub-problem on the result of other sub-problems.
- As a rule of thumb, fine-grained calculations are
better suited to big, monolithic supercomputers - On the other hand, parallel calculations
(high-throughput computing) are ideal for a more
loosely-coupled network of computers.
20Large Hadron Collider Computing Grid project at
CERN
- To study the fundamental properties of subatomic
particles and forces
Brookhaven National Laboratory (US) Fermi
National Accelerator Laboratory (Fermilab)
(US) Forschungszentrum Karlsruhe (Germany)
Rutherford Appleton Laboratory (UK)
CCIN2P3 (France) INFN-CNAF (Italy) SARA/NIKHEF
(Netherlands)
21Grid _at_ CERN
- CERN has a reputation for being at the forefront
of networking technology - "where the Web was
born" is the lab's motto. When it comes to Grid
technology, this is particularly true CERN is
leading some of the most ambitious Grid projects
in the world - The Large Hadron Collider (LHC), to run fully
in autumn 2009, will smash particles, protons to
protons with nearly the speed of light to create
conditions that occurred a few seconds after the
Big Bang.
22Need for Grid Computing in LHC
- These collisions will happen at an unprecedented
energy of 14 trillion electron volts. The beam
collisions would reveal physics beyond the
Standard Model. - Once operational we shall collect data of the
order of 15 Petabytes (15 million Gigabytes)
every year. That is more than 1000x the amount of
information in book form printed every year
around the world , and nearly 1 of all
information that humans produce on the planet
each year - including digital images, photos - This data shall be used by thousands of
scientists from all around the world.
23The LHC Computing Grid
- One of the challenges that this scenario shall
pose would be to build and maintain data storage
and provide analysis infrastructure for the
entire high energy physics community. - The current model adopted at the LHC includes a
four-tier Grid Structure which shall distribute
data worldwide. Formal access to the NWIC Grid is
requested, in support of these activities. - This LHC Computing Grid, launched on October 3,
2008, is a distribution network where The data
stream from the detectors provides approximately
300 GB/s, filtered for "interesting events",
resulting in a "raw data" stream of about 300
MMB/s. - The CERN computer center, considered the "Tier
0" has a dedicated 10 Gb/s connection to the
counting room
24Large Hadron Collider Computing Grid project at
CERN
- the LHC at Geneva produces roughly 15 Petabytes
(15 million Gigabytes) of data annually. This
data are accessed by the high energy physics
community throughout the world
multiple copies of data and automatic reassigning
of computational tasks to available resources
ensures load balancing of resources and
facilitates access to the data
5000 scientists from about 500 research
institutes and universities
25- Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â
- Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â
- MINSP (Short for Minimization of String
potential) - country Syria
- author Nizar Alhafez
- institute HIAST
- domain Theoretical Physics
- contacts nzhafez_at_hiast.edu.sy
- description We seek a minimum for a potential
function (coming from the physics of string
theory), involving many parameters. One run (when
fixing different parameters) lasts for between 3
to 10 hours, and hence the application needs
(when we change parameters in the given regions)
more than 15 years thinking to run the
application on one machine. - requirements The application requires
AXIONsoftware. It has been installed on EUMEDGRID
e-Science
26Table of Contents
- Basics
- Need for Powerful Computing in High Energy
Physics - Art of Cosmological Simulations
- The LHC Grid
- Neutrino Parameters model with many free
parameters
27Neutrino Physics
- Neutrinos (meaning "Small neutral ones") are
elementary particles that often travel close to
the speed of light, lack an electric charge, are
able to pass through ordinary matter almost
undisturbed and are thus extremely difficult to
detect. Neutrinos have a minuscule, but nonzero
mass.
- The establishment of a non-vanishing neutrino
mass in - neutrino oscillation experiments is one of the
major new - Achievements in theoretical physics in the last
decade.
28Neutrino experimental constraints
29A Model for Neutrino Mass Matrix
- Zero minors of the neutrino mass matrix,
PHYSICAL REVIEW D 78, 073002 (2008), by Lashin
Chamoun - We examine the possibility that a certain class
of neutrino mass matrices, namely, those with two
independent vanishing minors in the flavor basis,
regardless of being invertible or not, is
sufficient to describe current data.
- Strategy spanning the free parameters in their
accepted ranges, - and test whether or not there are acceptable
choices meeting the - current data constraints.
30Neutrino Mass Matrix
31Neutrino Mass Matrix
32Neutrino Mass Matrix
33Neutrino Model with two vanishing minors
34What about One Vanishing Minor Neutrino Mass
Matrix
- A work in progress
- Less constraints ?More free parameters
- Need to span a larger space ? need a more
powerful computer - Can deduce correlations among the different
parameters
- Strategy Again, the nature of nested loops
suggests the possibility - of using a distributed computing program which is
split up into parts - that run simultaneously on multiple computers
communicating over - a network .
35Neutrino Model with 1-vanishing minor
- We obtain acceptable points (here, \rho
\sigma) for fixed - acceptable choices of given parameters (here
\delta). However, I - Need to span \delta over its whole admissible
range.
36Table of Contents
- Basics
- Need for Powerful Computing in High Energy
Physics - Art of Cosmological Simulations
- The LHC Grid
- Neutrino Parameters model with many free
parameters
37Discussion