Catálogo de publicaciones - libros

Incomplete factorization preconditioners combined with Krylov subspace accelerators are currently among the most effective methods for iteratively solving large systems of linear equations. In this paper we consider the use of a dual threshold incomplete LU factorization (ILUT) preconditioner for the iterative solution of the linear equation systems encountered when performing electronic structure calculations that involve density fitting. Two questions are addressed, how the overall performance of the ILUT method varies as a function of the accuracy of the preconditioning matrix, and whether it is possible to make approximations to the original matrix on which the LU decomposition is based and still obtain a good preconditioner. With respect to overall performance both computational and memory storage requirements are considered, while in terms of approximations both those based on numerical and physical arguments are considered. The results indicate that under the right circumstances the ILUT method is superior to fully direct approaches such as singular value decomposition.

In a joint effort between computer scientists and theoretical chemists new tools have been developed for Grid applications leading to the efficient management of large computational campaigns in the field of quantum chemical calculations. For that purpose, the Vienna Grid Environment (VGE) software has been successfully extended allowing efficient job submission, status control and data retrieval. In addition, the services of the Compchem Virtual Organization of Enabling Grids for E-science (EGEE) Grid environment have been used. Extensive photodynamical simulation runs using the software packages COLUMBUS and NEWTON-X have been performed on the cis-trans isomerization of a model retinal system, aiming at a detailed picture of the primary processes of vision.

Several Molecular Dynamics simulations of the lamellar solid -zirconium phosphate have been performed in order to estimate their proton permeability. To this end we first tested the formulation of the Force Field and then we carried out the molecular dynamics calculations aimed at evaluationg proton mobility.

Decision support systems (DSS) are in the center of today’s experts’ attention, due to their abilities to allow significant increase of the quality of optimal decision selection among a large number of alternatives. In this paper we discuss assessment criteria of delivery quality in the transport logistics applaying methods from non-classical logic.

The essential problem of Distributed Virtual Environment (DVE) is to build scalable communication architecture, on which a large number of objects can communicate each other in a dynamical fashion. In this paper, a new XML-based scalable active interest management approach, which applies active routing technique to interest management, is presented to provide a heuristic method to solve the traditional scalability problem in DVE. The new approach uses XML to describe the interest representation model of objects, and implements active package filtering and transmission by XML routers based on the bi-directional shared multicast infrastructure. We developed the prototype system, and performed experiments in campus network. Experimental results show that this approach can prevent hosts from receiving redundant packets, thus efficiently reducing the total traffic in virtual environment.

This paper is concerned with the design and implementation of the context handlers for context-awareness in a ubiquitous computing environment. The context handlers address design issues of context handling: abstraction of context values, semantic interoperability of context information, logical and semantic interpretation of context information, and structuralization of context information for specifying various situations. To address these design issues, we define a structural context model, a context schema based on ontology, and user-friendly context value ontology. We also use ontology reasoning and rule-based reasoning. This paper focuses on context handling methods in a context-aware system. We expect that these context handling methods will help context-aware system developers to design and implement context handlers in a ubiquitous computing environment.

The workflow model has been successively applied to traditional computing environments such as business processes and distributed computing in order to perform service composition, flow management, parallel execution, and time-driven services. Recently, there have been many studies to adopt the workflow model into ubiquitous computing environments for context-aware and autonomous services. A service in ubiquitous computing environments must be executed according to a user’s situation information, which is generated dynamically from sensors. Such existing workflow systems as FollowMe and uFlow support context-aware services through workflow models. However, when a user’s situation is dynamically changed, the systems don’t have a method to immediately adopt the change into an already on-going service workflow. In this paper, we propose a context-aware workflow system, which can apply changes of user’s service demand or situation information into an on-going workflow without breaking its operation. The suggested workflow system can re-apply the new services into an initial workflow scenario without interrupting or deleting workflow service. To do this, the proposed system represents contexts described in a workflow as an RDF-based DItree (Document Instance tree). The system uses the tree information to recognize an exact position to be changed in the on-going workflow for the user’s situation changes, and to reconstruct only the position under the influence of the changes in the DItree. Therefore, the suggested system can quickly and efficiently apply a change of the user’s new situation into an on-going workflow without much loss of time and space, and can offer a context-aware service continuously according to a new workflow.

The UPnP technology is an important enabler to allow devices to be connected seamlessly in home network. ZigBee is a wireless technology developed as an open global standard to address the unique needs of low-cost, low-power. Since a digital home consists of diverse technologies, integration between various home networking standards is a research issue opened in the field of home networking. In this respect, interoperability between UPnP and ZigBee is no exception. To address it, this paper proposes a software bridge that interconnects ZigBee devices with UPnP networks. The proposed software bridge guarantees seamless interaction by abstracting ZigBee devices as virtual UPnP devices and efficiently manages service information of ZigBee networks by employing a device registry. From experiments on the physical environment, it is shown that it performs well.

This paper presents the design and the automation of a methodology for the creation, execution and administration of large computational and parametric experiments in distributed systems. This methodology is oriented to be a part of a workflow specification framework. These computational experiments involve the execution of a large number of tasks, the grouping of the results and their interpretation. This computational problem is automatically broken down and distributed in a transparent way between local or remote computational resources in a distributed environment. A procedure for specifying experiments is provided, which can work as a part of a general workflow specification procedure. It takes into account the conditions for executing the model, the strategy for executing simultaneously the processes of the computational experiment and provisions for deferred presentation of the results.

A new color image segmentation method is proposed in this paper. The proposed method is based on the human perception that in general human has attention on 3 or 4 major color objects in the image at first. Therefore, to determine the objects, three intensity distributions are constructed by sampling them randomly and sufficiently from three R, G, and B channel images. And three means are computed from three intensity distributions. Next, these steps are repeated many times to obtain three mean distribution sets. Each of these distributions comes to show normal shape based on the central limit theorem. To segment objects, each of the normal distribution is divided into 4 sections according to the standard deviation (section1 below - , section 2 between - and , section 3 between and , and section 4 over ). Then sections with similar representative values are merged based on the threshold. This threshold is not chosen as constant but varies based on the difference of representative values of each section to reflect various characteristics for various images. Above merging process is iterated to reduce fine textures such as speckles remained even after the merging. Finally, segmented results of each channel images are combined to obtain a final segmentation result. The performance of the proposed method is evaluated through experiments over some images.