Activities and osmotic coefficients of tropospheric aerosols: (NH4)2SO4(aq) and NaCl(aq)

Moonis R. Ally, Simon L. Clegg, Jerry Braunstein, John M. Simonson
2001 Journal of Chemical Thermodynamics  
Moonis Raza Ally earned a Doctor of Philosophy Degree in Chemical Engineering at the University of Pittsburgh. He is currently a Senior Research Staff Member at Oak Ridge National Laboratory. Dr. Ally has worked mostly in the area of statistical mechanics. Most of his publications are in the area of computational science. Dr. Ally has held a number of leadership positions at Oak Ridge National Laboratory and has had a number of academic responsibilities at the University of Tennessee at
more » ... e. Dr. Ally has much experience evaluating the merit of research and development projects. He has done this both as a member of Management Committees and in developing and completing his own work. A Comparison of Electrolyte Mean Ionic Activity Coefficients in Concentrated Aqueous Solutions as Calculated with Brunauer-Emmett-Teller Model and With Other Models. Overview of the Program The purpose of the minor in Computational Science is to train a new generation of practitioners in to apply current and emerging technologies both in industry and research. Students develop software in traditional languages such as C, C++ or FORTRAN. Throughout their studies students will have access to a wide range of state-of-the-art HPC platforms and technologies. The material can be divided into two sets of modules, namely, core and advanced. The core modules provide a broad-based coverage of the fundamentals of High Performance Computing (HPC) and parallel computing while the advanced modules will focus in-depth on a number of important topics that are relevant to computational science as a whole. Students need to complete the core courses with a C or better grade and may take the advanced courses as major electives. Core Module 1: Software Development: Methodologies, Tools and Techniques Students need to understand the steps involved in software development for the effective use of HPC facilities. This module focuses on the tools and techniques required for rapid code development, code documentation, code debugging, and profiling. Students will be required to develop and maintain high quality portable code for computational science problems in FORTRAN, C or C++ and develop valuable software skills that are vital in the fields of computational science. Topics covered: ! Basic Programming techniques ! Software design techniques ! Software debugging ! Software testing ! Advanced programming techniques and tools such as revision control, makefiles and compilers etc Core Module 2: Fundamental Concepts of High Performance Computing This module provides the required introductory material for all other courses. The objective of this module is to provide a broad overview of tools and techniques used in HPC. On completing this module, students will be able to analyze a problem and compose a parallel solution using domain or functional decomposition. Topics covered: o Computational science o Scalability issues o Discretization techniques Core Module 3: Principles of Parallel Programming: Theory and Practice Parallel programs can be implemented using message passing, shared memory or specialized programming models such as data parallel models. This module will focus on each one of these techniques. Message passing model Parallel programming by definition involves decomposition of a problem into several processes to solve a common task and the cooperation among these processes. Also, the hardware plays an important role in the execution of these processes. The programmer has to not only define these processes but also has to define the allocation of these processes to the available processors. Also, programmers have to specify how these processes will synchronize with each other and exchange data with one another. In message passing model, processes accomplish the above by sending messages to each other. This module will use the de facto standard for message passing, the Message Passing Interface (MPI). Topics covered: ! Point-to-point communications ! Non-blocking communications ! Derived datatypes ! Process topologies ! Collective communications 3.2 Shared memory model Parallel machines also make use of shared memory architecture in which each processor has access to a global memory and processors communicate with one another by accessing the shared memory. The OpenMP standard provides a standard interface to shared memory architectures so that these parallel programs can be ported across various architectures. This module looks at the techniques for parallelizing programs using OpenMP. Topics covered: ! Parallel regions ! Worksharing directives ! Synchronization ! Performance tuning and advanced topics Data-parallel model The data-parallel model originates from vector programming where the programmer writes their applications in terms of highly optimized vector operations. In order to write data-parallel programs, a programming language should provide parallel intrinsics and built-in procedures that operate on data known to all processes. In its simplest form this model extends a sequential programming language such as FORTRAN with parallel constructs for handling large aggregates of data such as arrays. This module will use High Performance FORTRAN (HPF), the de facto standard for data parallel languages. Topics covered: ! HPF ! Data distribution ! Parallel features of HPF ! Passing data to subroutines ! Intrinsic functions and advanced topics At low-energies the projectile polarizes the target. In this case, the most problematic region is inside the charge cloud of the target. The potential should be velocity dependent and the many-body effects become increasingly dominant. The way we have discussed earlier to incorporate the correlation/polarization effect many not be very accurate at very low energies. We know that inside the charge cloud, the scattering electron is indistinguishable from the bound electrons. The basis of the independent particle model on which the adiabatic potential rests, ceases to be valid. The resulting many-body effect yields a short-range non-local, bound-free correlation potential. Inclusion of these effects in practical calculation is extremely difficult. Our implementation, on the other hand, is based on a less rigorous prescription originally proposed by Temkin [9]. In this approach the polarization potential is weakened in the short-range region by neglecting the two-particle bound-free Coulomb interaction whenever the coordinate of the projectile is less than the that of one of the bound electrons. This yields a multipole expansion of which we have retained only the dipole term. In addition to our initial target H 2, we will also investigate e-N 2 , e-CO and e-CO 2 systems. The N 2 problem has long been a battle ground for both theory and experiment. In addition to its fundamental relevance, the system is very attractive for its rich structure and the low-energy cross sections, dominated mainly by 2.4 eV shape resonance. The interplay between resonant and non-resonant effect, has initiated much theoretical work and unfortunately none of them has reproduced the whole observed spectrum of the cross sections. We propose to fill up this gap. The outcome of this proposal will (a) form a data base of cross sections for ro-vibrational excitation of sufficient accuracy to serve as bench-mark for resolution of the controversy between various experimental determinations of cross sections, (b) evaluate the accuracy of the local polarization potentials used in the first stage, (c) calibrate the earlier calculations with our PDE results, (d) generalize the computer code for massive parallel environments, and (e) most importantly enhance our understanding of the complicated collision dynamics. INFORMATION ABOUT PRINCIPAL INVESTIGATORS/PROJECT DIRECTORS(PI/PD) and co-PRINCIPAL INVESTIGATORS/co-PROJECT DIRECTORS
doi:10.1006/jcht.2000.0825 fatcat:fcxdkwqsgrdpnn5lv2ty2lt7jm