Jonathan L. Bentz

Influence of geometry on light harvesting in dendrimeric systems

The exact analytic expression for the mean time to trapping (or mean walklength) for a particle (electron/exciton) performing a random walk on a finite dendrimer lattice with a trap at the center of the dendrimer was obtained. Exact analytic expressions have also been obtained for articulated/extended dendrimeric systems. The full dynamical behavior was determined for each case studied via numerical solution of the stochastic master equation, and the results obtained were shown to be a direct consequence of the structural properties of the dendrimeric system. These studies are linked to the behavior observed in experiments on light harvesting in dendrimeric supermolecules.

Integrating Performance Tools with Large-Scale Scientific Software

Modern performance tools provide methods for easy integration into an application for performance evaluation. For a large-scale scientific software package that has been under development for decades and with developers around the world, several obstacles must be overcome in order to utilize modern performance tools and explore performance bottlenecks. In this paper, we present our experience in integrating performance tools with one popular computational chemistry package. We discuss the difficulties we encountered and the mechanisms developed to integrate performance tools into this code. With performance tools integrated, we show one of the initial performance evaluation results, and discuss what other challenges we are facing to conduct performance evaluation for large-scale scientific packages.

Efficiency of encounter-controlled reaction between diffusing reactants in a finite lattice: topology and boundary effects

The role of dimensionality (Euclidean vs. fractal), spatial extent, boundary effects and system topology on the efficiency of diffusion-reaction processes involving two simultaneously diffusing reactants is analyzed. We present numerically exact values for the mean time to reaction, as gauged by the mean walklength before reactive encounter, obtained via application of the theory of finite Markov processes, and via Monte Carlo simulation. As a general rule, we conclude that for sufficiently large systems, the efficiency of diffusion-reaction processes involving two synchronously diffusing reactants (two-walker case) relative to processes in which one reactant of a pair is anchored at some point in the reaction space (one-walker plus trap case) is higher, and is enhanced the lower the dimensionality of the system. This differential efficiency becomes larger with increasing system size and, for periodic systems, its asymptotic value may depend on the parity of the lattice. Imposing confining boundaries on the system enhances the differential efficiency relative to the periodic case, while decreasing the absolute efficiencies of both two-walker and one-walker plus trap processes. Analytic arguments are presented to provide a rationale for the results obtained. The insights afforded by the analysis to the design of heterogeneous catalyst systems is also discussed.

Synchronous vs. asynchronous dynamics of diffusion-controlled reactions

An analytical method based on the classical ruin problem is developed to compute the mean reaction time between two walkers undergoing a generalized random walk on a 1d lattice. At each time step, either both walkers diffuse simultaneously with probability p (synchronous event) or one of them diffuses while the other remains immobile with complementary probability (asynchronous event). Reaction takes place through same site occupation or position exchange. We study the influence of the degree of synchronicity p of the walkers and the lattice size N on the global reactions efficiency. For odd N, the purely synchronous case (p=1) is always the most effective one, while for even N, the encounter time is minimized by a combination of synchronous and asynchronous events. This new parity effect is fully confirmed by Monte Carlo simulations on 1d lattices as well as for 2d and 3d lattices. In contrast, the 1d continuum approximation valid for sufficiently large lattices predicts a monotonic increase of the efficiency as a function of p. The relevance of the model for several research areas is briefly discussed.

Parallelization of general matrix multiply routines using OpenMP

An application programmer interface (API) is developed to facilitate, via OpenMP, the parallelization of the double precision general matrix multiply routine called from within GAMESS [1] during the execution of the coupled-cluster module for calculating physical properties of molecules. Results are reported using the ATLAS library and the Intel MKL on an Intel machine, and using the ESSL and the ATLAS library on an IBM SP.