Pasar al contenido principal

ALGORITMOS PARALELOS Y SOFTWARE PARA CALCULO ESPECTRAL EN APLICACIONES DE FISICA COMPUTACIONAL

PID2022-139568NB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2022
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 5
Encontrada(s) 1 página(s)

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

Correlated volumes for extended wave functions on a random-regular graph

  • Pino, Manuel
  • Jose E. Roman|||0000-0003-1144-6772
[EN] We study the metallic phase of the Anderson model in a random -regular graph, specifically the degree of ergodicity of the high-energy wave functions. We use the multifractal formalism to analyze numerical data for unprecedented large system sizes, obtaining a set of correlated volumes Nq which control the finite -size effects of the wave function q moment. Those volumes grow very fast, ln[ln(Nq)] <^> W, with disorder strength but show no tendency to diverge, at least in an intermediate metallic regime. Close to the Anderson transitions, we characterize the crossover to system sizes much smaller than the first correlated volume. Once this crossover has taken place, we obtain evidence of a scaling in which the derivative of the first fractal dimension behaves critically with an exponent nu = 1., We thank A. Rodriguez for useful discussions. M.P. and J.E.R. acknowledge support by the Spanish MCIN/AEI/10.13039/501100011033 through Grants No. PID2020-114830GB-I00 and No. PID2022-139568NB-I00, respectively. M.P. further acknowledges funding from the European Commission FET-Open project AVaQus GA 899561 and from MCIN/AEI/10.13039/501100011033 together with European Union NextGenerationEU/PRTR project Consolidacion Investigadora CNS2022-136025. The computational experiments were carried out on the supercomputer Tirant III belonging to Universitat de Valencia and SCAYLE, Supercomputacion Castilla y Leon.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Solvers for the Hermitian and the pseudo-Hermitian Bethe-Salpeter equation in the Yambo code: Implementation and performance

  • Milev, Petru
  • Nalabothula, Muralidhar
  • Esquembre-Kucukalic, Ali
  • Filippone, Francesco
  • Molina-Sanchez, Alejandro
  • Wirtz, Ludger
  • Sangalli, Davide
  • Blanca Mellado-Pinto
  • Alvarruiz Bermejo, Fernando|||0000-0001-5957-9561
  • Ramos Peinado, Enrique|||0000-0002-5960-5501
  • Jose E. Roman|||0000-0003-1144-6772
[EN] We analyze the performance of two strategies in solving the structured eigenvalue problem deriving from the Bethe-Salpeter equation (BSE) in condensed matter physics. The BSE matrix is constructed with the Yambo code, and the two strategies are implemented by interfacing Yambo with the ScaLAPACK and ELPA libraries for direct diagonalization, and with the SLEPc library for the iterative approach. We consider both the Hermitian (Tamm-Dancoff approximation) and pseudo-Hermitian forms, addressing dense matrices of three different sizes. A description of the implementation is also provided, with details for the pseudo-Hermitian case. Timing and memory utilization are analyzed on both CPU and GPU clusters. Our results demonstrate that it is now feasible to handle dense BSE matrices of the order of 105., Innovation Study ISOLV-BSE has received funding through the Inno4scale project, which is funded by the European High-Performance Computing Joint Undertaking (JU) under Grant Agreement No 101118139. The JU receives support from the European Union's Horizon Europe Programme. DS acknowledges funding from the MaX "MAterials design at the eXascale" project, co-funded by the European High Performance Computing joint Undertaking (JU) and participating countries (Grant Agreement No. 101093374) . DS also acknowledges funding from the European Union's Horizon Europe research and innovation programme under the Marie Sklodowska-Curie grant agreement 101118915 (project TIMES) . DS and PM acknowledge the CINECA, Italy award under the ISCRA initiative, for the availability of high performance computing resources and support, through the project IsCc6_ISOL-RUN. BM, FA, ER, and JER. were partly supported by grant PID2022-139568NB-I00 funded by MCIN/AEI, Spain/10.13039/501100011033 and by "ERDF/EU". BM was also supported by Universitat Politecnica de Valencia, Spain in its PAID-01-23 programme. NM and LW acknowledge the use of HPC facilities of the University of Luxembourg [68] .
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Improving performance of contour integral-based nonlinear eigensolvers with infinite GMRES

  • Liu, Yuqi
  • Shao, Meiyue
  • Jose E. Roman|||0000-0003-1144-6772
[EN] In this work, the infinite GMRES algorithm, recently proposed by [Jarlebring and Correnty, SIAM J. Matrix Anal. Appl., 43 (2022), pp. 1382--1405] is employed in contour integralbased nonlinear eigensolvers, avoiding the computation of costly factorizations at each quadrature node to solve the linear systems efficiently. Several techniques are applied to make the infinite GMRES memory-friendly, computationally efficient, and numerically stable in practice. More specifically, we analyze the relationship between polynomial eigenvalue problems and their scaled linearizations, and provide a novel weighting strategy which can significantly accelerate the convergence of infinite GMRES in this particular context. We also adopt the technique of TOAR to infinite GMRES to reduce the memory footprint. Theoretical analysis and numerical experiments are provided to illustrate the efficiency of the proposed algorithm., Funding: The first and third authors were partly supported by the National Natural Science Foundation of China under grant 92370105. The second author was supported by grant PID2022139568NB-I00 funded by MCIN/AEI/10.13039/501100011033 and by ERDF/EU, and by grant RED2022-134176-T.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

PHIDE: A Parallel Hybrid Direct-Iterative Eigensolver for Hermitian Eigenvalue Problems

  • Li, Shengguo
  • Wu, Xinzhe
  • Yuan, Ziyang
  • Wang, Ruibo
  • Li, Tiejun
  • Xie, Yi
  • Yang, Bo
  • Chen, Xuguang
  • Jose E. Roman|||0000-0003-1144-6772
[EN] In this paper, we propose a Parallel Hybrid Direct-Iterative Eigensolver for Hermitian Eigenvalue Problems without tridiagonalization, denoted by PHIDE, which combines direct and iterative methods. PHIDE first reduces a Hermitian matrix to banded form, then applies a spectrum slicing algorithm to the banded matrix, and finally computes the eigenvectors of the original matrix via backtransformation. Compared with conventional direct eigensolvers, PHIDE avoids tridiagonalization, which involves many memory-bound operations. In PHIDE, the banded eigenvalue problem is solved using the contour integral method implemented in FEAST, which may yield slightly lower accuracy than tridiagonalization-based approaches. For sequences of correlated Hermitian eigenvalue problems arising in density functional theory (DFT), PHIDE achieves an average speedup of 1.22x over the state-of-the-art direct solver in ELPA when using 1024 processes. Numerical experiments are conducted on dense Hermitian matrices from real applications as well as large sparse matrices from the SuiteSparse and ELSES collections., This work was supported in part by National Key R&D Program of China under Grant 2023YFB3001700, in part by National NaturalScience Foundation of China under Grant U1611261, Grant 61872392, andGrant U1811461. The work of J. E. Roman was supported in part by MICIU/AEI/10.13039/501100011033 under Grant PID2022-139568NB-I00, andin part by ERDF/EU under Grant RED2022-134176-T.
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Correlated volumes for extended wave functions on a random-regular graph

  • Pino, Manuel
  • Roman, Jose E.
9 pags., 8 figs., We study the metallic phase of the Anderson model in a random-regular graph, specifically the degree of ergodicity of the high-energy wave functions. We use the multifractal formalism to analyze numerical data for unprecedented large system sizes, obtaining a set of correlated volumes Nq which control the finite-size effects of the wave function q moment. Those volumes grow very fast, ln[ln(Nq)]∼W, with disorder strength but show no tendency to diverge, at least in an intermediate metallic regime. Close to the Anderson transitions, we characterize the crossover to system sizes much smaller than the first correlated volume. Once this crossover has taken place, we obtain evidence of a scaling in which the derivative of the first fractal dimension behaves critically with an exponent ν=1., M.P. and J.E.R. acknowledge support by the Spanish MCIN/AEI/10.13039/501100011033 through Grants No. PID2020-114830GB-I00 and No. PID2022-139568NB-I00, respectively. M.P. further acknowledges funding from the European Commission FET-Open project AVaQus GA 899561 and from MCIN/AEI/10.13039/501100011033 together with European Union NextGenerationEU/PRTR project Consolidación Investigadora CNS2022-136025. The computational experiments were carried out on the supercomputer Tirant III belonging to Universitat de València and SCAYLE, Supercomputación Castilla y León.