Showing papers by "Nicolae Goga published in 2009"

A formal analysis of ISO/IEEE P11073-20601 standard of medical device communication

Abstract: This article presents the formal work done for the ISO/IEEE P11073-20601 Draft Standard for Health informatics - Personal health device communication - Application profile - Optimized exchange protocol. ISO/IEEE 11073 family defines standards for device communication between agents (e.g. blood pressure monitors, weighing scales) that collect information about a person and manager (e.g., cell phone, health appliance, or personal computer) for collection, display, and possible later re-transmission. The particular draft standard ISO/IEEE P11073-20601 defines protocols for data exchange between agents and managers. Although such a system in medical use must be extremely reliable under all circumstances, until now a formal analysis has been used only in analyzing draft IEEE P11073-20601 as a part of developing the protocols in this family of standards. We analyzed the protocols both manually and automatically. For the automated analysis of safety properties we applied model-checking techniques, which is feasible since the number of processes is limited, and because we can abstract from most data in the protocols. We used the language Promela and the tool Spin, since this combination has a good reputation in practical applications and it is well documented. The results of our work are now incorporated in this standard.

A Multiscalling Constant Lambda Molecular Dynamic Gromacs Implementation

Abstract: Molecular dynamics indicates the general process of describing complex chemical systems in terms of a realistic atomic model, with the aim to understand and predict macroscopic properties based on detailed knowledge on an atomic scale Molecular dynamics (MD) is one of the methods used now-a-days by the scientific community to study the properties of polymers [5, 15, 12, 16, 2] Such methods next to other that are developed [3, 10] are used as complementary to the laboratory experiments [7, 13, 14] for advancing the knowledge in the field of plastic materials A molecular system can be described with a finegrained representation that is a detailed, low-level model of it A coarse-grained representation of a molecular system is a model where some of these fine details have been smoothed over or averaged out [17, 18] In MD, coarse graining consists in replacing a fine grained description of a molecular system such as polyethylene, for example, with a lower-resolution coarse-grained model that averages or smooths away fine details The advantage of using coarse grained representations is the fact that it speeds-up the simulations The price paid is the fact that some of the results obtained have a larger error margin than in the case of using FG models or sometimes even some macroscopic phenomena under investigation are not observed in the simulations There are recent efforts [6, 11] to combine the advantages of the two simulations in a single multiscaling simulation In such a simulation a molecular system has a double nature, being modeled in a coarse grained representation while the fine grain details are represented proportionally with a scaling factor λ In the Christen approach, λ is a constant during the simulation while for Praprotnik, λ depends on the coarse grain coordinates of each particle Nevertheless, all the proposed approaches have limitations In the model proposed by Christen for constant lambda, in a pure coarse grained simulation for keeping the fine grained particles together, Christen computes (bonded) forces between the FG particles and the only ones that are not computed being the non-bonded forces between unconnected atoms In this case, next to the addition in the complexity and computational time, pure CG is no longer pure Praprotnik approach for space lambda computes only forces between two particles Its multiscaling model is applicable only for simple systems, such as butane, with only one coarse grain particle corresponding to an assembly of 4 fine grained atoms For more complex systems with more than one CG particle, such as polycarbonate for example, systems of each movements need the computation of the forces with the contribution of 3 particles (angle forces) or 4 (dihedral forces), Praprotnik modeling cannot be applied Moreover, Praprotnik model does not follow a very systematic approach for the force computation that usually starts from the Hamiltonian The MD group of the University of Groningen started a research to improve the existing methods In this paper we summarize the theoretical approach for constant lambda that solves the problem mentioned for Christen approach and we discuss its implementation in Gromacs package for molecular simulation In the final Section we present the related performance issues

New Gromacs Implementations for Multiscaling Space MD

Abstract: Among the methods used nowadays for studying the microscopic properties of polymers, proteins, membranes and other bio-materials is the molecular dynamics. One of the widely used software in molecular dynamics is Gromacs developed at the University of Groningen. From a computational perspective, molecular dynamics requires large computational power and an increased storage capacity. Different physical models were recently defined for trying to reduce the complexity and to make more efficient the computational it molecular models. In this paper we present the Gromacs implementation of a new multiscaling MD model with a space dependent multiscaling parameter and we discuss efficiency measurements.

Multiscaling algorithms for molecular dynamics simulations with GROMACS

Abstract: This article presents the parallel implementation of a new multiscale model that is currently developed in the Molecular Dynamics Group from the University of Groningen. Multiscale methods combine the advantages of two levels of simulation, an atomistic and a coarse-grain one, with a small loss of performance. We designed a parallel implementation of this multiscale approach based on the improved parallel algorithms from the GROMACS MD simulator. The properties of the simulated systems were undisturbed and the errors were kept to a minimum. By having a parallel multiscale simulation, one can take advantage of a reduced simulation time by running it on multiple processors, and in the same time, access to both atomistic and macroscopic details is offered for a better understanding of the desired phenomenon.