Computer cost modeling

Chasman and co-workers derived the time complexity of each combination of Hamiltonian evaluation and time integration method for both serial and parallel computers. Table 1 shows their results for systems of two and three dimensions. This time complexity can be used to develop a computational cost model of their application. A full analysis of the communication properties of their application could be developed to derive a complete performance model. This, in turn, could be used to determine the time to completion for a given problem on a given computer or, alternatively, the performance of a given problem on a new parallel supercomputer. 

Figure 7.9 The Xp parameter avoids steep slopes on the Fp curves, whereas minimum Fp does not. (Reprinted from Ahmad, Linnhoff, and Smith, Cost Optimum Heat Exchanger Networks II. Targets and Design for Detailed Capital Cost Models, Computers Chem, Engg., 7 751, 1990 with permission from Elsevier Science, Ltd.)...

In Table 1 the CPU time required by the two methods (LFV and SISM) for 1000 MD integration steps computed on an HP 735 workstation are compared for the same model system, a box of 50 water molecules, respectively. The computation cost per integration step is approximately the same for both methods so that th< syieed up of the SISM over the LFV algorithm is deter-... 

Traditional electronic structure energy calculations consist of a single job. However, a calculation at a very accurate level of theory can take a very long time to complete. In an effort to achieve high accuracy results at less computational cost, several new model chemistries have been defined as a series of calculations to be run and a... 

Each cell in the chart defines a model chemistry. The columns correspond to differcni theoretical methods and the rows to different basis sets. The level of correlation increases as you move to the right across any row, with the Hartree-Fock method jI the extreme left (including no correlation), and the Full Configuration Interaction method at the right (which fuUy accounts for electron correlation). In general, computational cost and accuracy increase as you move to the right as well. The relative costs of different model chemistries for various job types is discussed in... 

Choosing a model chemistry almost always involves a trade-off between accuracy and computational cost. More accurate methods and larger basis sets make jobs run longer. We ll provide some specific examples of these effects throughout the chapters in this part of the book. 

The following table summarizes some recommended research-level model chemistries for predicting total energies for molecular systems of various sizes (the details of the reasons behind these choices will unfold along with the chapters in this part of the book), arranged in order of increasing computational cost ... 

Chapter 6, Selecting an Appropriate Theoretical Method, discusses the model chemistry concept introduced in Chapter 1 in detail. It covers the strengths, computational cost and limitations of a variety of popular methods, beginning with semi-empirical models and continuing through Hartree-Fock, Density Functional Theory, and electron correlation methods. 

There are several different CBS methods, each having its own set of prescriptions and resulting computational cost and accuracy, they are known under the acronyms CBS-4, CBS-q, CBS-Q and CBS-APNO. As an explicit example, we will take the CBS-Q model, " which is computationally similar to the G2(MP2) method. 

The mixed solvent models, where the first solvation sphere is accounted for by including a number of solvent molecules, implicitly include the solute-solvent cavity/ dispersion terms, although the corresponding tenns between the solvent molecules and the continuum are usually neglected. Once discrete solvent molecules are included, however, the problem of configuration sampling arises. Nevertheless, in many cases the first solvation shell is by far the most important, and mixed models may yield substantially better results than pure continuum models, at the price of an increase in computational cost. 

This paper will discuss the formulation of the simulator for the filament winding process which describes the temperature and extent of cure in a cross-section of a composite part. The model consists of two parts the kinetic model to predict the curing kinetics of the polymeric system and the heat transfer model which incorporates the kinetic model. A Galerkin finite element code was written to solve the specially and time dependent system. The program was implemented on a microcomputer to minimize computer costs. 

Because this method avoids iterative calculations to attain the SCF condition, the extended Lagrangian method is a more efficient way of calculating the dipoles at every time step. However, polarizable point dipole methods are still more computationally intensive than nonpolarizable simulations. Evaluating the dipole-dipole interactions in Eqs. (9-7) and (9-20) is several times more expensive than evaluating the Coulombic interactions between point charges in Eq. (9-1). In addition, the requirement for a shorter integration timestep as compared to an additive model increases the computational cost. 

---