Imre Szeberényi

Constrained Euler buckling: an interplay of computation and analysis

We consider elastic buckling of an inextensible beam with hinged ends and fixed end displacements, confined to the plane, and in the presence of rigid, frictionless sidewalls which constrain overall lateral displacements. We formulate the geometrically nonlinear (Euler) problem and develop global search and path-following algorithms to find equilibria in various classes satisfying different contact patterns and hence boundary conditions. We derive complete analytical results for the case of line contacts with the sidewalls, and partial results for point contact and mixed cases. The analysis is essential to understanding the numerical results, for in contrast to the unconstrained problem, we find a very rich bifurcation structure, with the cardinality of branches growing exponentially with mode number.

Saleve: simple Web-services based environment for parameter study applications

The goal of the Saleve Project is to develop and evaluate mechanisms and abstractions that may connect the diverse research community of the distributed (mainly the grid) computing to those users, who are not familiar with distributed computing as such, but who would simply like to use the results in their everyday tasks. We show a simple Web-services based, domain-specific computational framework that integrates smoothly into the well-known, traditional user environments and requires learning no new technologies.

Simple Grid Access for Parameter Study Applications

The problem domain of Parameter Study covers a wide range of engineering and scientific tasks which can be easily processed parallel. However, the research community users, who would like to benefit from the supercomputing power of a Grid are not familiar with the required distributed computing techniques. For this reason, the presented Salevedevelopment framework aims to free this limit by providing a general solution for PS tasks by inserting an abstraction layer between the user application and the Grid middleware. We design a new Saleve plugin for the gLite middleware of the EGEE Grid and discuss the authentication issues.

Applying the improved saleve framework for modeling abrasion of pebbles

Saleve is a generic framework for making the development of Parameter Study tasks easy for scientists and engineers not familiar with distributed technologies In this paper we present our lightweight authentication procedure for Saleve to delegate user credentials towards the grid Then we present a detailed statistics of abrasion of pebbles gained with Saleve Finally we make remarks on the makespan of the application and look for ways to reduce it.

Demonstration of P-GRADE Job-Mode for the Grid

The P-GRADE job execution mode will be demonstrated on a small Grid containing 3 clusters from Budapest and London. The first demonstration illustrates the Grid execution of a parallel meteorology application. The parallel program will be on-line monitored remotely in the Grid and locally visualized on the submitting machine. The second demonstration will use a parallel traffic simulation program developed in P-GRADE to show the usage of the P-GRADE job mode for Grid execution. The parallel program will be check-pointed and migrated to another cluster of the Grid. On-line job and execution monitoring will be demonstrated.

Harnessing Wasted Computing Power for Scientific Computing

Nowadays more and more general purpose workstations installed in a student laboratory have a built in multi-core CPU and graphics card providing significant computing power. In most cases the utilization of these resources is low, and limited to lecture hours. The concept of utility computing plays an important role in technological development. In this paper, we introduce a cloud management system which enables the simultaneous use of both dedicated resources and opportunistic environment. All the free workstations are allocated to a resource pool, and can be used like ordinary cloud resources. Our solution leverages the advantages of HTCondor and OpenNebula systems. Modern graphics processing units (GPUs) with many-core architectures have emerged as general-purpose parallel computing platforms that can dramatically accelerate scientific applications used for various simulations. Our business model harnesses the computing power of GPUs as well, using the needed amount of unused machines.

Dictionary compression and information source correction

This paper introduces a method to compress, store, and search a dictionary of a natural language The dictionary can be represented as groups of words derived form a stem We describe how to represent and store a word group in a way that is compact and efficiently searchable The compression efficiency of the used algorithm highly depends on the quality of the information source The currently available tools and data sources contain several mistakes, which can be cleaned by the introduced method The paper also analyzes the efficiency of XML and two binary formats, and proposes two methods: directed acyclic graph transformation and word group regrouping that can be used to increase efficiency.

Parallel, Recursive Computation of Global Stability Charts for Liquid Bridges

We investigate the stability of liquid bridges by using a parallel, recursive algorithm. The core of the recursion is the Parallel Simplex Algorithm introduced in [10]. We discuss the PVM implementation of the algorithm and compare our results with earlier published computations.

Solving Generalized Boundary Value Problems with Distributed Computing and Recursive Programming

A wide field of technical problems can be described by boundary value problems (BVPs) associated with ordinary differential equations (ODEs). We show a numerical method and its distributed parallel implementation for solving BVPs. We generalize our method presented earlier in [5] to a recursive scheme, capable of solving more complex technical problems, such as global search for stability boundaries. In this paper we also present the efficient, PVM based implementation of our recursive algorithm.