Ion Gh. Rosca

Evaluating Java performance for linear algebra numerical computations

Abstract In this paper we evaluate the performance of a Java library in the context of designing numerical solutions for linear algebra problems. Nowadays there are a number of libraries like LAPACK or ATLAS available in C or FORTRAN for solving such problems. Java is a relatively new object-oriented language and is almost universally recognized as a very good programming language for writing portable programs. Despite these advantages, it is a common believe that Java still lags behind C/C++ or Fortran performance especially for computational intensive numerical applications. We developed a Java library for matrix computations and show that using a set of optimization techniques Java can achieve performance comparable with other libraries developed in C or Fortran.

Parallel algorithms for large scale econometric models

Abstract In this paper we have developed algorithms to solve macroeconometric models with forward-looking variables based on Newton method for nonlinear systems of equations. The most difficult step for Newton methods represents the resolution of a large linear system for each iteration. Thus, we compare the performances resulted by solving this linear system using two iterative methods and the direct method. We’ve also described an implementation of the parallel versions of such algorithms using a software package. Our experiments confirm that the iterative methods have a low computational complexity and storage requirements, but the parallel versions of direct methods show a superior speedup.