Information Technology Reference
In-Depth Information
algorithm is used, typically it requires large-scale computing resources for simu-
lations of relevant physical systems.
In the Serbian condensed matter physics community (CMPC), the most prom-
inent use of computing resources is related to the three applications developed at the
national level: SPEEDUP, QSPEEDUP, and GP-SCL.
SPEEDUP code (Balaz 2012-1) uses Monte Carlo-based path integral algorithms
for the calculation of quantum mechanical transition amplitudes for 1D models. It is
based on the use of higher-order short-time effective actions in the time of prop-
agation (Monte Carlo time step), which substantially improves the convergence of
discretized amplitudes to their exact continuum values.
QSPEEDUP code (Vudrag 2010) presents a quasi-Monte Carlo extension of the
SPEEDUP code. The extended algorithm uses Sobol
s set of quasi-random num-
bers for generation of trajectories relevant for calculation of transition amplitudes in
the path integral formalism. Both applications use identical algorithm, supplied with
different random number generators, which allows their uni
'
cation into a single
application designated as (Q)SPEEDUP.
GP-SCL (Vudrag 2012) is a set of codes parallelized using the OpenMP
approach for calculating the dynamics and ground states of quantum fluids (such as
Bose
Einstein condensates and superfluids). Quantum fluids represent macroscopic
quantum phenomena where large numbers of atoms or molecules behave coher-
ently, allowing special properties to emerge. In the mean-field regime, such systems
can be described by a nonlinear Schroedinger equation, usually called the
Gross
-
Pitaevskii equation (Pethick 2008). GP-SCL codes solve the time-(in)
dependent Gross
-
Pitaevskii nonlinear partial differential equation in one, two, and
three space dimensions in an arbitrary trapping potential using imaginary-time and
real-time propagation. The equation is solved using the semi-implicit split-step
Crank
-
Nicolson method by discretizing space and time, as described in Murug
(2009). The discretized equation is then propagated in imaginary or real time over
small time steps.
Applications have been developed by scientists from the Scienti
-
c Computing
Laboratory (SCL 2014) of the Institute of Physics Belgrade (IPB 2014) and are used
by an increasing number of collaborators within Serbia and also from other
countries in Europe, as well as from Brazil, India, and China. The applications are
deployed on the computing infrastructure provided by the Academic and Educa-
tional Grid Initiative of Serbia (AEGIS 2014), which is part of the European Grid
Infrastructure (EGI 2014).
To increase the potential user base of the applications, we have decided to create
the AEGIS CMPC science gateway (SG) in the framework of the SCI-BUS project
(SCI-BUS 2014), to provide seamless access to the software and the data generated
by the applications. This science gateway also hides the complexity of use of the
Grid and applications, which was a barrier for many scientists.
Search WWH ::
Custom Search