Speaker
Describe the added value of the Grid for the scientific/technical activity you (plan to) do on the Grid. This should include the scale of the activity and of the potential user community and the relevance for other scientific or business applications
Continuous improvement of a nuclear code is essential for its
predicting power and
reliability, obviously the most important aspect being the
underlying physical models
and parameters but another key issue is the computational time
performance. This
performance is affected by the complexity of the actual physical
model which in turn
dictates the level of approximations employed to the algorithm
used. The
simplifications are most of the time satisfactory but with time
they become too
general for user requirements and tend to restrict the
applicability of the
algorithm, thus new techniques or models are needed.
Therefore it is desirable to exploit modern hardware and software
platforms in order
to perform faster and with better precision the necessary
calculations and thus
improving overall the process of evaluating the nuclear data.
The paper presents the techniques used to improve the performance
of low-energy
nuclear reaction codes with respect to computational time, namely
the porting process
of EMPIRE-II and SCAT2MIN nuclear computer codes to grid
environment [4], method that
provides the highest achievable performance at a reduced cost and
a higher security.
Report on the experience (or the proposed activity). It would be very important to mention key services which are essential for the success of your activity on the EGEE infrastructure.
Porting the EMPIRE-II [5] nuclear reaction code to computational
GRIDs was done by
integrating it on the GILDA [6] Grid testbed [7] during
ICTP/INFM-Democritos workshop
on “Porting Scientific Applications on Computational GRIDs” [8].
The purpose was to
reduce the time used to systematically evaluate in one run all
reaction channels of
all stable isotopes corresponding to a given element,
and possibly neighboring elements using local and/or regional
parameters [6].
Also a modified version of SCAT-2 [9] nuclear code that searches
"best fit" nuclear
optical model parameters was ported to the parallel environment
using OpenMPI [10]
library, the code being prepared to run on the in the EGEE FUSION
VO Grid[9]. The
performance tests show that the code is very scalable with the
number of processors
and thus perfect for the parallel and distributed computational
environments.
With a forward look to future evolution, discuss the issues you have encountered (or that you expect) in using the EGEE infrastructure. Wherever possible, point out the experience limitations (both in terms of existing services or missing functionality)
Porting the nuclear-reaction computer code EMPIRE-II to GRID
infrastructure and the
graphical integration into Genius web portal proved successful.
However, there are
still remaining issues regarding the MySQL EXFOR experimental
database as well as an
eventual implementation of a facility for plotting online the
calculated results in
comparison with the experimental data.
The code SCAT-2 proved to be a highly scalable one, suitable for
distributed
architectures. It is thus supporting the next step of a
concurrent study by taking
into account all known experimental elastic-scattering angular
distributions for a
given isotope/element. The further use of dynamic chunks of data
may improve the load
balancing on heterogeneous environments, at the cost of the
overhead of communication.
The reliability and cost effectiveness of future facilities
relies on improved
nuclear reaction models as well as faster computer codes and
finally the covariance
evaluation of nuclear models. In spite of it being a
long-standing subject, only
recent investigations in this respect have been motivated by the
computational power
of present computers. This work has showed that it is rather
straightforward to
integrate actual nuclear computer codes into GRID computational
environment for
assessment of large data sets. Following the achievement of this
goal concurrently
and also performing large scale computations with MPI enabled
nuclear codes, the
related outcome will be firsthand the reduced computational time,
and also one of the
first steps towards nuclear models uncertainties generation.
Describe the scientific/technical community and the scientific/technical activity using (planning to use) the EGEE infrastructure. A high-level description is needed (neither a detailed specialist report nor a list of references).
There was always a stringent need for high accuracy nuclear data
due to nuclear
fusion and fission projects, and not only, that have intensified
in the last years
due to new foreseen facilities. The experiments are becoming more
complex, expensive
and lack the power to measure all the required quantities. The
comparison of the
measured and calculated cross sections for nuclear reactions
showed in several
occasions noticeable differences between nuclear computer codes
predictions above the
energy regions with experimental data and also a significant
deviation from unity of
calculated-to-experimental ratio [1], moreover these evaluations
being reported
without estimates of uncertainties. Recent efforts have begun to
address this
uncertainty issue through the generation of covariance files for
nuclear model
calculations [2], a task that requires large scale computations.
It is thus justified
the effort made to improve the actual status of the nuclear
computer codes which in
turn translates in reliable, safer and cost effective modern
facilities [3].
