miércoles, 3 de septiembre de 2008

Researchers from the Universities of Extremadura and Vigo break the electromagnetic simulation world record at CESGA

English Press Release
They practically double the previous record after having solved a problem with 150 million of unknowns.

The new development allows immediate applications in the naval, aerospatial and telecommunications industries among others.

A mixed group of researchers from the Universities of Extremadura and Vigo and technicians from the Supercomputing Center of Galicia (CESGA) has succeeded in solving an electromagnetic problem with 150 million unknowns using supercomputer Finis Terrae, installed at CESGA, considerably far from the previous world record of 85 million unknowns by a Turkish research group. The scientific work to achieve this challenge continues, with the perspective of achieving 250 million unknowns this winter, also thanks to the support of CESGA's supercomputer.

The application used in the challenge (HEMCUVE++) is the result of a continuous development process by the research group, that began in the year 1999 through a project funded by NAVANTIA. The objective was accomplished thanks to the close collaboration among the researchers and the technicians in charge of Finis Terrae. In order to tackle this simulation, only 64 of its 142 nodes were needed; in other words, 1024 processing nodes and 5,4 Terabytes of RAM memory. The calculus were completed in 6 hours, 3 and a half of which were used to initialize the algorithm.

The researcher from the University of Vigo, Fernando Obelleiro, highlights that “Finis Terrae is a fantastic machine for this kind of challenge. Although there are more powerful machines, Finis Terrae's architecture is suitable for our application”.

Practical applications of the new development

There are many research groups all around the world competing to solve more and more complicated electromagnetic problems but, according to Luis Landesa, member of the team from the University of Extremadura, “what is interesting about this competition is to take full advantage of the resources available, what makes us think a lot and show our most scientific side”.

Despite the fact that the work to tackle this challenge is part of the basic research field, “the computational code we are developing will serve us for the applied uses we are working on”, explains Landesa.

The applied uses he is talking about are the solving of electromagnetic problems in the design of big structures equipped with radio, such as boats, planes and land vehicles. “The electromagnetic phenomena can be simulated computationally”, explains Obelleiro.

Thus, his work is focused on finding the most efficient way to simulate the electromagnetic behavior of the structures in the design stage in order to be able to verify and guarantee the fulfillment of the electromagnetic requirements imposed by the industry or the administrations, minimizing thus the problems after the manufacturing process. “When you build a boat you may have interference problems among its antennas. It would be very expensive to build it and then wait to see what happens, and that is the reason for the simulation, which allows us to do virtual designs of the boat and to observe simulatedly as well the reactions and interactions among its different elements, such as the antennas and radars”, details the researcher form the University of Vigo.

The key is in the new algorithm

The dimension of the problems that must be solved in order to do these simulations is huge. After having spent a lot of time using approximate techniques which gave a rough idea of the electromagnetic behavior of the structures, the evolution of the computers and the algorithmic development allowed the analysis of these problems by means of some very rigorous techniques. “We are talking about millions of unknowns that will enable us to analyze and even predict the electromagnetic behavior of structures with great electrical dimensions with a wealth of detail that had been unapproachable until now, and this will undoubtedly mean important competitive advantages for the industries with an access to this technology”, explains Landesa.

The algorithm developed by the research group allows scalabilities that were unthinkable until now. The code was executed on 1024 processing nodes with an efficiency rate close to 100%, while the previous attempts carried out by other teams only got to use 48 processors, showing a very low efficiency in their parallelization. This fact is of a vital importance in order to take full advantage of the capacity of the new generations of supercomputers, equipped with more and more processors.

This record also opens the door to the application of the computational electromagnetics in fields such as biomedicine (radar images for tumor detection, influence of mobile devices on the human body, etc), metamaterials design (artificial materials with unusual electromagnetic properties, with great importance in the creation of superlens or invisibility cloacks), or the development of ground penetrating radars for the detection of anti-personnel mines, geological structures in the subsoil, etc.

About the research team

The team is composed of the researchers Fernando Obelleiro and José Luis Rodríguez from the University of Vigo, and Luis Landesa and José Manuel Taboada from the University of Extremadura.

The team have been working coordinately for some years for the Armada and the company Navantia developing systems that allow them to tackle studies of electromagnetic compatibility by means of computing in supercomputers such as Finis Terrae, aiming to detect interference among antennas, to predict dangerous radiation levels, to study its radar equivalent surface, etc.

About CESGA

The Supercomputing Center of Galicia (CESGA) is the center for high-performance computing, communications and advanced services used by the Scientific Community of Galicia, the University academic system and the Higher Council for Scientific Research (CSIC).

CESGA houses the supercomputer Finis Terrae, recognized as Singular Scientific and Technological Infrastructure by the Ministry of Science and Innovation.


No hay comentarios: