Daniel Gruner

SciNet: Lessons Learned from Building a Power-efficient Top-20 System and Data Centre

SciNet, one of seven regional HPC consortia operating under the Compute Canada umbrella, runs Canadas first and third fastest computers (as of June 2010) in a state-of-the-art, highly energy-efficient datacentre with a Power Usage Effectiveness (PUE) design-point of 1.16. Power efficiency, computational bang for the buck and system capability for a handful of flagship science projects were important criteria in choosing the nature of the computers and the data centre itself. Here we outline some of the lessons learned in putting together the systems and the data centre that hosts Canadas fastest computer to date.

Efficient evaluation of harmonic polyatomic Franck-Condon factors

Abstract An efficient computational method for obtaining large numbers of Franck-Condon factors between polyatomic harmonic oscillator vibrational states is presented.

A genetic algorithm approach to fitting polyatomic spectra via geometry shifts

Abstract Discrepancies between experimental and theoretical fluorescence/absorption spectra are minimized using a genetic algorithm to shift the calculated excited electronic surface position. The method depends on the extreme sensitivity of polyatomic Franck-Condon factors to the geometrical shift that molecules undergo upon electronic excitation and on the power of genetic algorithms to rapidly locate the required origin shift. Examples of harmonic alkylbenzene Franck-Condon spectra are used to illustrate the method.

Theoretical study of the S1↔S0 spectroscopy of anthracene

Dispersed fluorescence spectra for S0 to S1 vibronic transitions in anthracene are computed using the semiempirical QCFF/PI approach plus corrections resulting from fits to experimental spectral intensities. Extensive comparisons are made with the experimental jet spectra. The nature and assignment of low lying levels are clarified within the normal mode picture, whereas higher energy spectra are shown to be considerably broadened by low order anharmonic corrections.

Intramolecular vibrational redistribution in alkylbenzenes. I. Normal modes and their energy distribution

Normal modes for the ground and first excited singlet state are computed for ethylbenzene, pentylbenzene, and hexylbenzene and are assigned in terms of zero‐order mode descriptors. The energy distribution, i.e., the partitioning of energy between the ring and chain, is computed for the normal modes and the concept of ‘‘system modes,’’ i.e., modes localized on the ring, is examined. Several of the basic assumptions made in traditional interpretations of intramolecular vibrational redistribution in molecules with a ‘‘ring+tail’’ structure are examined in light of the computed normal mode characteristics.

Intramolecular vibrational redistribution in alkylbenzenes. II. Spectroscopy and dynamics

Dispersed fluorescence spectra from 00, 6b1, and 121 excitation of ethylbenzene, pentylbenzene, and hexylbenzene are computed within the normal mode approximation and compared with experiment. The harmonic model is shown to be capable of describing the appearance of the ‘‘new shifted spectrum,’’ originally identified as a ‘‘relaxed spectrum,’’ which becomes increasingly dominant with increasing chain length. Thus, this feature cannot be a consequence of intramolecular vibrational redistribution (IVR). By contrast, the appearance of substantial spectral broadening can only be accounted for through the introduction of anharmonic coupling. An assessment of standard assumptions in the analysis of IVR in ‘‘ring+tail’’ molecules is also provided. Results show that some crucial assumptions, e.g., that electronic transitions are not accompanied by changes in chain excitation, are unjustified.

Nucleation, domain growth, and fluctuations in a bistable chemical system

Phase separation and nucleation processes are investigated for a bistable chemical system. The study utilizes a reactive lattice‐gas cellular automaton model to provide a mesoscopic description of the dynamics. Simulations of steady‐state structure, wave propagation, and critical nucleus size using this model are compared with results based on the deterministic equations of motion. The dynamic structure factor is computed for evolution from the unstable state and the effects of correlations are examined for early and late times. The study provides insight into these processes in a fluctuating, extended medium and also provides a test of the ability of the reactive lattice‐gas method to describe the fluctuations in the system.