Pasar al contenido principal

EXTENSION DE LA LIBRERIA SLEPC PARA POLINOMIOS MATRICIALES, FUNCIONES MATRICIALES Y ECUACIONES MATRICIALES EN PLATAFORMAS DE COMPUTACION EMERGENTES

TIN2013-41049-P

•
Nombre agencia financiadora Ministerio de Economía y Competitividad
Acrónimo agencia financiadora MINECO
Programa Programa Estatal de Fomento de la Investigación Científica y Técnica de Excelencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D (Excelencia / Generación de Conocimiento)
Año convocatoria 2013
Plan Estatal 2013-2016 Plan Estatal de Investigación Científica y Técnica y de Innovación
Unidad de gestión Secretaría de Estado de Educación, Formación Profesional y Universidades
Centro beneficiario UNIVERSITAT POLITÈCNICA DE VALÈNCIA
Identificador persistente http://dx.doi.org/10.13039/501100003329

Resultados relacionados

Resultados totales (Incluyendo duplicados): 12
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Capítulo o parte de un libro (BookPart)

GPU implementation of Krylov solvers for block-tridiagonal eigenvalue problems

  • Lamas Daviña, Alejandro
  • Jose E. Roman|||0000-0003-1144-6772
In an eigenvalue problem defined by one or two matrices with block-tridiagonal structure, if only a few eigenpairs are required it is interesting to consider iterative methods based on Krylov subspaces, even if matrix blocks are dense. In this context, using the GPU for the associated dense linear algebra may provide high performance. We analyze this in an implementation done in the context of SLEPc, the Scalable Library for Eigenvalue Problem Computations. In the case of a generalized eigenproblem or when interior eigenvalues are computed with shift-and-invert, the main computational kernel is the solution of linear systems with a block-tridiagonal matrix. We explore possible implementations of this operation on the GPU, including a block cyclic reduction algorithm., This work was partially supported by the Spanish Ministry of Economy and Competitiveness under grant TIN2013-41049-P. Alejandro Lamas was supported by the Spanish Ministry of Education, Culture and Sport through grant FPU13-06655.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Design and implementation of Java bindings in Open MPI

  • Vega Gisbert, Oscar
  • Jose E. Roman|||0000-0003-1144-6772
  • Squyres, Jeffrey M.
This paper describes the Java MPI bindings that have been included in the Open MPI distribution. Open MPI is one of the most popular implementations of MPI, the Message-Passing Interface, which is the predominant programming paradigm for parallel applications on distributed memory computers. We have added Java support to Open MPI, exposing MPI functionality to Java programmers. Our approach is based on the Java Native Interface, and has similarities with previous efforts, as well as important differences. This paper serves as a reference for the application program interface, and in addition we provide details of the internal implementation to justify some of the design decisions. We also show some results to assess the performance of the bindings. (C) 2016 Elsevier B.V. All rights reserved., We are indebted to Siegmar Grog for his exhaustive testing of the Java bindings. We also thank Ralph Castain for helping in the integration of the Java bindings in the Open MPI infrastructure. The NPB-MPJ benchmarks used in Section 5 were kindly provided by Guillermo Lopez Taboada. The first two authors were supported by the Spanish Ministry of Economy and Competitiveness under project number TIN2013-41049-P.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Optimized analysis of isotropic high-nuclearity spin clusters with GPU acceleration

  • Lamas Daviña, Alejandro
  • Ramos Peinado, Enrique|||0000-0002-5960-5501
  • Jose E. Roman|||0000-0003-1144-6772
The numerical simulation of molecular clusters formed by a finite number of exchange-coupled paramagnetic centers is very relevant for many applications, modeling systems between molecules and extended solids. In the context of realistic scenarios, many centers need to be considered, and thus the required computational effort grows very fast. In a previous work (Ramos et al., 2010), a set of parallel programs were presented with standard message-passing parallelization (MPI) for both anisotropic and isotropic systems. In this work, we have further developed the code for isotropic models. On one hand, the computational cost has been significantly reduced by avoiding some of the matrix diagonalizations, corresponding to blocks with negligible contribution for the particular configuration. On the other hand, we have extended the parallelization in order to exploit available graphics processing units (GPUs). The new MPI-GPU paradigm reduces the computational time by at least one additional order of magnitude and enables the resolution of larger problems. © 2016 Elsevier B.V. All rights reserved., This work was partially supported by the Spanish Ministry of Economy and Competitiveness under grant TIN2013-41049-P. Alejandro Lamas Davina was supported by the Spanish Ministry of Education, Culture and Sports through a grant with reference FPU13-06655.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

On low-frequency variability of the midlatitude ocean gyres

  • Shevchenko, I. V.
  • Berloff, P. S.
  • Guerrero López, David
  • Jose E. Roman|||0000-0003-1144-6772
This paper studies the large-scale low-frequency variability of the wind-driven midlatitude ocean gyres and their western boundary currents, such as the Gulf Stream or Kuroshio, simulated with the eddy-resolving quasi-geostrophic model. We applied empirical orthogonal functions analysis to turbulent flow solutions and statistically extracted robust and significant large-scale decadal variability modes concentrated around the eastward jet extension of the western boundary currents. In order to interpret these statistical modes dynamically, we linearized the governing quasi-geostrophic equations around the time-mean circulation and solved for the corresponding full set of linear eigenmodes with their eigenfrequencies. We then projected the extracted decadal variability on the eigenmodes and found that this variability is a multimodal coherent pattern phenomenon rather than a single mode or a combination of several modes as in the flow regimes preceding developed turbulence., The first two authors are thankful to the Natural Environment Research Council for the support of this work through the grant NE/J006602/1 and the use of ARCHER (the UK National Supercomputing Service). We express our gratitude to S. Burbidge and M. Harvey for their help with Imperial College London cluster, as well as to A. Thomas for his help with managing and maintaining the data storage. The last two authors were supported by the Spanish Ministry of Economy and Competitiveness under grant TIN2013-41049-P, and this support is gratefully acknowledged. We would also like to thank unknown referees for valuable comments and suggestions, which helped us to improve the paper.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Improving accuracy of parallel SLICOT model reduction routines for stable systems

  • Guerrero López, David
  • Jose E. Roman|||0000-0003-1144-6772
This paper shows part of the work carried out to develop parallel versions of the SLICOT routines for model reduction of stable systems. In particular, the routines that have been parallelised are those based on the solution of Lyapunov equations. The goal is to be able to work with larger unreduced models and also to obtain better performance in the reduction process. New routines have been developed using standard libraries to improve portability and efficiency. A preliminary version was released previously by the authors, which achieved high performance. However, accuracy improvements have been necessary in order to make the new routines similar to the sequential ones in this aspect. Routines presented in this paper preserve good performance obtained by the previous parallel implementation while maintaining high accuracy of sequential SLICOT routines., This work was partially supported by the Spanish Ministry of Economy and Competitiveness under grant TIN2013-41049-P
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Parallel Krylov Solvers for the Polynomial Eigenvalue Problem in SLEPc

  • Campos, Carmen
  • Jose E. Roman|||0000-0003-1144-6772
Polynomial eigenvalue problems are often found in scientific computing applications. When the coefficient matrices of the polynomial are large and sparse, usually only a few eigenpairs are required and projection methods are the best choice. We focus on Krylov methods that operate on the companion linearization of the polynomial but exploit the block structure with the aim of being memory-efficient in the representation of the Krylov subspace basis. The problem may appear in the form of a low-degree polynomial (quartic or quintic, say) expressed in the monomial basis, or a high-degree polynomial (coming from interpolation of a nonlinear eigenproblem) expressed in a nonmonomial basis. We have implemented a parallel solver in SLEPc covering both cases that is able to compute exterior as well as interior eigenvalues via spectral transformation. We discuss important issues such as scaling and restart and illustrate the robustness and performance of the solver with some numerical experiments., The first author was supported by the Spanish Ministry of Education, Culture and Sport through an FPU grant with reference AP2012-0608.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Development of a Finite Volume Inter-cell Polynomial Expansion Method for the Neutron Diffusion Equation

  • Bernal García, Álvaro
  • Jose E. Roman|||0000-0003-1144-6772
  • Miró Herrero, Rafael|||0000-0003-1012-0869
  • Ginestar Peiro, Damián|||0000-0003-1243-6648
  • Verdú Martín, Gumersindo Jesús|||0000-0001-5098-080X
Heterogeneous nuclear reactors require numerical methods to solve the neutron diffusion equation (NDE) to obtain the neutron flux distribution inside them, by discretizing the heterogeneous geometry in a set of homogeneous regions. This discretization requires additional equations at the inner faces of two adjacent cells: neutron flux and current continuity, which imply an excess of equations. The finite volume method (FVM) is suitable to be applied to NDE, because it can be easily applied to any mesh and it is typically used in the transport equations due to the conservation of the transported quantity within the volume. However, the gradient and face-averaged values in the FVM are typically calculated as a function of the cell-averaged values of adjacent cells. So, if the materials of the adjacent cells are different, the neutron current condition could not be accomplished. Therefore, a polynomial expansion of the neutron flux is developed in each cell for assuring the accomplishment of the flux and current continuity and calculating both analytically. In this polynomial expansion, the polynomial terms for each cell were assigned previously and the constant coefficients are determined by solving the eigenvalue problem with SLEPc. A sensitivity analysis for determining the best set of polynomial terms is performed., This work has been partially supported by the Spanish Ministerio de Eduacion Cultura y Deporte [grant number FPU13/01009]; the Spanish Ministerio de Ciencia e Innovacion [project number ENE2014-59442-P], [project number ENE2012-34585]; the Generalitat Valenciana [project number PROMETEOII/2014/008]; the Universitat Politecnica de Valencia [project number UPPTE/2012/118]; and the Spanish Ministerio de Economia y Competitividad [project number TIN2013-41049-P].
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Problema de autovalores de la ecuación de la difusión neutrónica discretizada con el Método de Volúmenes Finitos en un PWR

  • Bernal García, Álvaro
  • Jose E. Roman|||0000-0003-1144-6772
  • Miró Herrero, Rafael|||0000-0003-1012-0869
  • Verdú Martín, Gumersindo Jesús|||0000-0001-5098-080X
En este trabajo, se ha desarrollado un algoritmo para aplicar el Método de Volúmenes Finitos (MVF) a la ecuación de la difusión neutrónica, que calcula varios autovalores y se ha incorporado al algoritmo la lectura de secciones eficaces en función de variables termohidráulicas. Para su validación, se ha simulado un reactor PWR, cuyas secciones eficaces se han modelado con la metodología SIMTAB para las condiciones termohidráulicas de Begin of Cycle (BOC) y Hot Full Power (HFP). Se ha realizado un análisis de sensibilidad de la discretización de la geometría, utilizando 3 mallas de hexaedros y otras 3 de tetraedros. En cada caso, se calculan 5 autovalores y sus correspondientes potencias., Este trabajo ha sido financiado parcialmente por el Ministerio de Educación Cultura y Deporte bajo la beca FPU13/01009, el Ministerio de Ciencia e Innovación bajo los proyectos ENE2014-59442-P y ENE2012-34585, la Generalitat Valenciana bajo el proyecto PROMETEOII/2014/008, la Universitat Politècnica de València bajo el proyecto UPPTE/2012/118, y el Ministerio de Economía y Competitividad bajo el proyecto TIN2013-41049-P.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Parallel iterative refinement in polynomial eigenvalue problems

  • Campos, Carmen
  • Jose E. Roman|||0000-0003-1144-6772
Methods for the polynomial eigenvalue problem sometimes need to be followed by an iterative refinement process to improve the accuracy of the computed solutions. This can be accomplished by means of a Newton iteration tailored to matrix polynomials. The computational cost of this step is usually higher than the cost of computing the initial approximations, due to the need of solving multiple linear systems of equations with a bordered coefficient matrix. An effective parallelization is thus important, and we propose different approaches for the message-passing scenario. Some schemes use a subcommunicator strategy in order to improve the scalability whenever direct linear solvers are used. We show performance results for the various alternatives implemented in the context of SLEPc, the Scalable Library for Eigenvalue Problem Computations., This work was partially supported by the Spanish Ministry of Economy and Competitiveness under grant TIN2013-41049-P. Carmen Campos was supported by the Spanish Ministry of Education, Culture and Sport through an FPU grant with reference AP2012-0608. The computational experiments of Section 5 were carried out on the supercomputer Tirant at Universitat de Valencia.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Control rod drop transient analysis with the coupled parallel code pCTF-PARCSv2.7

  • Ramos Peinado, Enrique|||0000-0002-5960-5501
  • Jose E. Roman|||0000-0003-1144-6772
  • Miró Herrero, Rafael|||0000-0003-1012-0869
  • Abarca Giménez, Agustín
  • Bermejo, Juan A.
In order to reduce the response time when simulating large reactors in detail, a parallel version of the thermal–hydraulic subchannel code COBRA-TF (CTF) has been developed using the standard Message Passing Interface (MPI). The parallelization is oriented to reactor cells, so it is best suited for models consisting of many cells. The generation of the Jacobian matrix is parallelized, in such a way that each processor is in charge of generating the data associated with a subset of cells. Also, the solution of the linear system of equations is done in parallel, using the PETSc toolkit. With the goal of creating a powerful tool to simulate the reactor core behavior during asymmetrical transients, the 3D neutron diffusion code PARCSv2.7 (PARCS) has been coupled with the parallel version of CTF (pCTF) using the Parallel Virtual Machine (PVM) technology. In order to validate the correctness of the parallel coupled code, a control rod drop transient has been simulated comparing the results with the real experimental measures acquired during an NPP real test. 2015 Elsevier Ltd. All rights reserved., This work has been partially supported by the Universitat Politecnica de Valencia under Projects COBRA_PAR (PAID-05-11-2810) and OpenNUC (PAID-05-12), and by the Spanish Ministerio de Economa y Competitividad under Projects SLEPc-PFE (TIN2013-41049-P) and NUC-MULTPHYS (ENE2012-34585). The authors would like to acknowledge the technical support provided by CNAT and IBERDROLA GENERACION S.A. for the realization of this work.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Assembly Discontinuity Factors for the Neutron Diffusion Equation discretized with the Finite Volume Method. Application to BWR

  • Bernal-Garcia, Alvaro
  • Jose E. Roman|||0000-0003-1144-6772
  • Miró Herrero, Rafael|||0000-0003-1012-0869
  • Verdú Martín, Gumersindo Jesús|||0000-0001-5098-080X
The neutron flux spatial distribution in Boiling Water Reactors (BWRs) can be calculated by means of the Neutron Diffusion Equation (NDE), which is a space- and time-dependent differential equation. In steady state conditions, the time derivative terms are zero and this equation is rewritten as an eigenvalue problem. In addition, the spatial partial derivatives terms are transformed into algebraic terms by discretizing the geometry and using numerical methods. As regards the geometrical discretization, BWRs are complex systems containing different components of different geometries and materials, but they are usually modelled as parallelepiped nodes each one containing only one homogenized material to simplify the solution of the NDE. There are several techniques to correct the homogenization in the node, but the most commonly used in BWRs is that based on Assembly Discontinuity Factors (ADFs). As regards numerical methods, the Finite Volume Method (FVM) is feasible and suitable to be applied to the NDE. In this paper, a FVM based on a polynomial expansion method has been used to obtain the matrices of the eigenvalue problem, assuring the accomplishment of the ADFs for a BWR This eigenvalue problem has been solved by means of the SLEPc library. (C) 2016 Elsevier Ltd. All rights reserved., This work has been partially supported by the Spanish Ministerio de Eduacion Cultura y Deporte under the grant FPU13/01009, the Spanish Ministerio de Ciencia e Innovacion under projects ENE2014-59442-P, the Spanish Ministerio de Economia y Competitividad and the European Fondo Europeo de Desarrollo Regional (FEDER) under project ENE2015-68353-P (MINECO/FEDER), the Generalitat Valenciana under projects PROMETEOII/2014/008, the Universitat Politecnica de Valencia under project UPPTE/2012/118, and the Spanish Ministerio de Economia y Competitividad under the project TIN2013-41049-P.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Restarted Q-Arnoldi-type methods exploiting symmetry in quadratic eigenvalue problems

  • Campos, Carmen
  • Jose E. Roman|||0000-0003-1144-6772
We investigate how to adapt the Q-Arnoldi method for the case of symmetric quadratic eigenvalue problems, that is, we are interested in computing a few eigenpairs of with M, C, K symmetric matrices. This problem has no particular structure, in the sense that eigenvalues can be complex or even defective. Still, symmetry of the matrices can be exploited to some extent. For this, we perform a symmetric linearization , where A, B are symmetric matrices but the pair (A, B) is indefinite and hence standard Lanczos methods are not applicable. We implement a symmetric-indefinite Lanczos method and enrich it with a thick-restart technique. This method uses pseudo inner products induced by matrix B for the orthogonalization of vectors (indefinite Gram-Schmidt). The projected problem is also an indefinite matrix pair. The next step is to write a specialized, memory-efficient version that exploits the block structure of A and B, referring only to the original problem matrices M, C, K as in the Q-Arnoldi method. This results in what we have called the Q-Lanczos method. Furthermore, we define a stabilized variant analog of the TOAR method. We show results obtained with parallel implementations in SLEPc., This work was supported by the Spanish Ministry of Economy and Competitiveness under Grant TIN2013-41049-P. Carmen Campos was supported by the Spanish Ministry of Education, Culture and Sport through an FPU Grant with reference AP2012-0608.