Jose E. Roman

SLEPc: A scalable and flexible toolkit for the solution of eigenvalue problems

The Scalable Library for Eigenvalue Problem Computations (SLEPc) is a software library for computing a few eigenvalues and associated eigenvectors of a large sparse matrix or matrix pencil. It has been developed on top of PETSc and enforces the same programming paradigm.The emphasis of the software is on methods and techniques appropriate for problems in which the associated matrices are sparse, for example, those arising after the discretization of partial differential equations. Therefore, most of the methods offered by the library are projection methods such as Arnoldi or Lanczos, or other methods with similar properties. SLEPc provides basic methods as well as more sophisticated algorithms. It also provides built-in support for spectral transformations such as the shift-and-invert technique. SLEPc is a general library in the sense that it covers standard and generalized eigenvalue problems, both Hermitian and non-Hermitian, with either real or complex arithmetic.SLEPc can be easily applied to real world problems. To illustrate this, several case studies arising from real applications are presented and solved with SLEPc with little programming effort. The addressed problems include a matrix-free standard problem, a complex generalized problem, and a singular value decomposition. The implemented codes exhibit good properties regarding flexibility as well as parallel performance.

Parallel Arnoldi eigensolvers with enhanced scalability via global communications rearrangement

This paper presents several new variants of the single-vector Arnoldi algorithm for computing approximations to eigenvalues and eigenvectors of a non-symmetric matrix. The context of this work is the efficient implementation of industrial-strength, parallel, sparse eigensolvers, in which robustness is of paramount importance, as well as efficiency. For this reason, Arnoldi variants that employ Gram-Schmidt with iterative reorthogonalization are considered. The proposed algorithms aim at improving the scalability when running in massively parallel platforms with many processors. The main goal is to reduce the performance penalty induced by global communications required in vector inner products and norms. In the proposed algorithms, this is achieved by reorganizing the stages that involve these operations, particularly the orthogonalization and normalization of vectors, in such a way that several global communications are grouped together while guaranteeing that the numerical stability of the process is maintained. The numerical properties of the new algorithms are assessed by means of a large set of test matrices. Also, scalability analyses show a significant improvement in parallel performance.

SLEPc: scalable library for eigenvalue problem computations

The eigenvalue problem is one of the most important problems in numerical linear algebra. Several public domain software libraries are available for solving it. In this work, a new petsc-based package is presented, which is intended to be an easy-to-use yet efficient object-oriented parallel framework for the solution of standard and generalised eigenproblems, either in real or complex arithmetic. The main objective is to allow the solution of real world problems in a straightforward way, especially in the case of large software projects.

Parallel Computation of 3-D Soil-Structure Interaction in Time Domain with a Coupled FEM/SBFEM Approach

This paper introduces a parallel algorithm for the scaled boundary finite element method (SBFEM). The application code is designed to run on clusters of computers, and it enables the analysis of large-scale soil-structure-interaction problems, where an unbounded domain has to fulfill the radiation condition for wave propagation to infinity. The main focus of the paper is on the mathematical description and numerical implementation of the SBFEM. In particular, we describe in detail the algorithm to compute the acceleration unit impulse response matrices used in the SBFEM as well as the solvers for the Riccati and Lyapunov equations. Finally, two test cases validate the new code, illustrating the numerical accuracy of the results and the parallel performances.

Multi-dimensional gyrokinetic parameter studies based on eigenvalue computations

Abstract Plasma microinstabilities, which can be described in the framework of the linear gyrokinetic equations, are routinely computed in the context of stability analyses and transport predictions for magnetic confinement fusion experiments. The GENE code, which solves the gyrokinetic equations, has been coupled to the SLEPc package for an efficient iterative, matrix-free, and parallel computation of rightmost eigenvalues. This setup is presented, including the preconditioner which is necessary for the newly implemented Jacobi–Davidson solver. The fast computation of instabilities at a single parameter set is exploited to make parameter scans viable, that is to compute the solution at many points in the parameter space. Several issues related to parameter scans are discussed, such as an efficient parallelization over parameter sets and subspace recycling.

Memory-efficient Arnoldi algorithms for linearizations of matrix polynomials in Chebyshev basis

Novel memory-efficient Arnoldi algorithms for solving matrix polynomial eigenvalue problems are presented. More specifically, we consider the case of matrix polynomials expressed in the Chebyshev basis, which is often numerically more appropriate than the standard monomial basis for a larger degree d. The standard way of solving polynomial eigenvalue problems proceeds by linearization, which increases the problem size by a factor d. Consequently, the memory requirements of Krylov subspace methods applied to the linearization grow by this factor. In this paper, we develop two variants of the Arnoldi method that build the Krylov subspace basis implicitly, in a way that only vectors of length equal to the size of the original problem need to be stored. The proposed variants are generalizations of the so-called quadratic Arnoldi method and two-level orthogonal Arnoldi procedure methods, which have been developed for the monomial case. We also show how the typical ingredients of a full implementation of the Arnoldi method, including shift-and-invert and restarting, can be incorporated. Numerical experiments are presented for matrix polynomials up to degree 30 arising from the interpolation of nonlinear eigenvalue problems, which stem from boundary element discretizations of PDE eigenvalue problems. Copyright (C) 2013 John Wiley & Sons, Ltd.

Towards the availability of Java bindings in open MPI

We present an ongoing effort to provide Java bindings in Open MPI [4]. We advocate including the bindings in the MPI distribution, rather than using a standalone, pure Java implementation.

Strategies for spectrum slicing based on restarted Lanczos methods

In the context of symmetric-definite generalized eigenvalue problems, it is often required to compute all eigenvalues contained in a prescribed interval. For large-scale problems, the method of choice is the so-called spectrum slicing technique: a shift-and-invert Lanczos method combined with a dynamic shift selection that sweeps the interval in a smart way. This kind of strategies were proposed initially in the context of unrestarted Lanczos methods, back in the 1990’s. We propose variations that try to incorporate recent developments in the field of Krylov methods, including thick restarting in the Lanczos solver and a rational Krylov update when moving from one shift to the next. We discuss a parallel implementation in the SLEPc library and provide performance results.

Parallel implementation of the MAGPACK package for the analysis of high-nuclearity spin clusters

Abstract Molecular clusters are formed by a finite number of exchange-coupled paramagnetic centers and they are model systems between molecules and extended solids. In order to simulate their properties and extrapolate to solids, the size of the systems to be treated should be as large as possible. In this context, the use of efficient parallel codes is essential. We present the parallel programs ParAni and ParIso , for anisotropic and isotropic models, that enable the calculation of large energy matrices in parallel and the subsequent computation of the relevant spectral information. The evaluation of the matrix elements is based on the serial package Magpack that uses the irreducible tensor operators technique and takes into account all kinds of anisotropic and isotropic magnetic interactions. To obtain the eigenvalues, the energy matrix is partially diagonalized by means of the SLEPc library. The calculation of eigenvalues and eigenvectors of these spin clusters enables us to evaluate the bulk magnetic properties (magnetic susceptibility and magnetization) as well as the spectroscopic properties (inelastic neutron scattering spectra). The results are encouraging in terms of parallel efficiency and open the way to address very challenging problems.

3D Alpha Modes of a Nuclear Power Reactor

This work is focused on the determination of the dominant α-modes corresponding to three time-eigenvalue problems. A nodal collocation method is used for the discretization of the differential eigenvalue problem and a restarted Krylov method has been used to solve the algebraic eigenvalue problem. Also, the relation between the λ-modes and α-modes is studied through the concept of prompt neutron generation time, which has several interpretations. To present numerical results of the different α-modes, we have studied a benchmark reactor and a configuration associated with Ringhals NPP.