Finite element computations of shallow water flows can be applied to many practical problems: design of river, coastal and offshore structures, disaster prediction, and other applications related to hydrodynamics, thermal and chemical transport behavior in oceans, lakes and rivers.

In this lecture, recent advances in finite element modeling of shallow water flows were explained and two finite element methods were presented. In addition, numerical examples were used to demonstrate that the strategies presented are applicable to large-scale computations of various shallow water flow problems.

Finite element computations of shallow water flows can be applied to many practical problems: design of river, coastal and offshore structures, disaster prediction, and other applications related to hydrodynamics, thermal and chemical transport behavior in oceans, lakes and rivers.

The utility of quantum chemistry calculations in the parameterization of atomistic force fields for molecular mechanics applications has been thoroughly demonstrated for relatively simple molecules such as alkanes and ethers. For aromatic and other large conjugated molecules, it is a computationally demanding task. Quantum chemistry codes developed specifically for workstation platforms have been successfully employed for deriving force fields for large molecules. Several of these systems, and issues involved with large-scale quantum chemistry calculations, including use of parallel architectures, were discussed in this seminar.

Fast networks have made it possible to aggregate distributed central processing units, memory, and storage resources to provide the potential for application performance superior to that achievable on a single system. These aggregate resources enable the implementation of problems which could not previously be solved and constitute a new generation of high-performance programs.

However, achieving the performance potential of such applications can be difficult. Since resources of the metasystem often reside in distinct administrative domains, and are under the control of distinct local schedulers, it is not feasible to rely on a single system scheduler to deliver application performance. Metacomputing applications must be scheduled carefully using application-specific and dynamic system information in order to leverage heterogeneity, minimize the effects of contention for shared resources, and satisfy the developer's performance criteria.

The speaker introduced application-level scheduling, a scheduling paradigm in which all resources in the metasystem are evaluated in terms of their impact on an application. The principles of application-level scheduling are being used to guide the development of AppLeS Application-Level Scheduling agents for individual metacomputing applications. The AppLeS architecture and description of experiments with AppLeS prototypes was also outlined.

This talk began with an investigation of instability and the transition to three-dimensionality in cylinder wakes of different cross-sections. Results from Floquet analysis and direct numerical simulations on cylinders of circular, elliptic, and square cross-sections were presented. These results show that different modes of instability are present and that their interaction can be quite complex, resulting in significantly distorted three-dimensional flow even at modest Reynolds numbers. The possibility of a period doubling scenario arising from spanwise subharmonic instability was also discussed. Consequences of the appearance of three-dimensionality were elucidated.

Recent developments in networking make it feasible to construct high-performance computations that couple geographically distributed computing, information, and display resources. The I-WAY was an ATM-based wide area computing environment designed to support experimentation in this area. Its use was demonstrated with over 60 applications at the Supercomputing conference in December 1995. This researcher described the I-WAY experiment and the lessons that have been learned from it. Also discussed was ongoing research that seeks to develop a software infrastructure, called Globus, to support applications such as those deployed on the I-WAY.

Under many circumstances a material solidifies (crystallizes) from its melt in a complex geometrical pattern-called a dendrite-which separates the liquid and solid regions during the cooling process. This microscopic tree-like structure occurs because it efficiently disposes of the heat of solidification for pure materials, and excess solute for alloys. Recently, considerable progress has been made with the simulation of these patterns using the phase-field method. The formulation of the model for pure materials and alloys was described in this talk. Results showing the development of segregation patterns within the solid dendrites were highlighted as well.

Photosynthesis is the process by which light energy is converted into chemical energy to synthesize sugars and starches from atmospheric carbon dioxide. It is the principal source of the biomass on Earth and is responsible for creating an oxidizing atmosphere above the surface of the Earth.

This seminar described the electronic spectroscopy of chlorophyll and explain why chlorophyll is a particularly good choice to be the central molecule in this process. It also described the excited states of the "special pair," a dimer of chlorophyll, and showed why this again is a unique structure, guaranteeing that the energy is transferred from the antenna chlorophyll to the reaction center.

Conflict in the interactions on short and long length scales in biomolecules leads topological frustration. A consequence of topoligical frustration is that biomolecular folding takes place by distinct mechanisms which can be succinctly summarized in terms of the kinetic partitioning mechanism (KPM).

In this seminar, estimates of the various dominant time scales in the KPM were provided. Experimental evidence of KPM for kinetics of refolding of proteins and RNA was also provided.