Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

A multicore solution to Block-Toeplitz linear systems of equations

  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Vidal Maciá, Antonio Manuel
  • Argüelles, Daniel
  • Ranilla, José
There exist algorithms, also called "fast" algorithms, which exploit the special structure of Toeplitz matrices so that, e.g., allow to solve a linear system of equations in O(n2) flops. However, some implementations of classical algorithms that do not use this structure (O(n3) flops) highly reduce the time to solution when several cores are available. That is why it is necessary to work on "fast" algorithms so that they do not lose track of the benefits of new hardware/software. In this work, we propose a new approach to the Generalized Schur Algorithm, a very known al- gorithm for the solution of Toeplitz systems, to work on a Block-Toeplitz matrix. Our algorithm is based on matrix-matrix multiplications, thus allowing to exploit an efficient implementation of this operation if it exists. Our algorithm also makes use of the thread level parallelism featured by multicores to decrease execution time., PROMETEO/2009/013, Generalitat Valenciana. Projects TEC2009-13741, TIN2010-14971 and TIN2011-15734-E of the Ministerio Espanol de Ciencia e Innovacion, and TEC2012-38142-C04 of the Ministerio Espanol de Economia y Competitividad.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Non-linear parallel solver for detecting point sources in CMB maps using Bayesian techniques

  • Alonso, P.
  • Argüeso, F.
  • Cortina, R.
  • Ranilla, J.
  • Vidal Maciá, Antonio Manuel
Abstract In this work we present a suitable computational tool to deal with large matrices and solve systems of non-linear equations. This technique is applied to a very interesting problem: the detection and flux estimation of point sources in Cos- mic Microwave Background (CMB) maps, which allows a good determination of CMB primordial fluctuations and leads to a better knowledge of the chemistry at the early stages of the Universe. The method uses previous information about the statis- tical properties of the sources, so that this knowledge is incorporated in a Bayesian scheme. Simulations show that our approach allows the detection of more sources than previous non-Bayesian techniques, with a small computation time, This work was financially supported by the Spanish Ministerio de Ciencia e Innovacion and by FEDER Projects (TEC2012-38142-C04, TIN2010-14971, TEC2009-13741 and CAPAP-H4 TIN2011-15734-E) Universitat Politecnica de Valencia through Programa de Apoyo a la Investigacion y Desarrollo (PAID-05-11) and Generalitat Valenciana through project PROMETEO/2009/013. We thank Diego Herranz for his help with the CMB simulations.