Background on Computational Chemistry

In order to balance public domain science with a high quality commercial software product it has been necessary for us to reimplement almost every aspect of computational chemistry embodied in HyperChem. All HyperChem source code is written in C or C-t-t, specified, designed, and implemented by Hyper-Chem s developers. We have stood on the scientific shoulders of giants, but we have not used their FORTRAN code Thus, although we have had access to MOPAC and other public domain codes for testing and other purposes, HyperChem computes MINDO, MNDO, and AMI wave functions, for example, with HyperChem code, not MOPAC code. We have made the effort to implement modern chemical science in a modern software-engineered product. 

The curve above shows that as R—the potential energy approaches a constant, which is the energy of the two individual atoms. Further, there is a global minimum for this potential surface at intermediate distances. At very short distances, the energy rises to +°° as the two atoms repel each other. 

The semi-empirical methods of HyperChem are quantum mechanical methods that can describe the breaking and formation of chemical bonds, as well as provide information about the distribution of electrons in the system. HyperChem s molecular mechanics techniques, on the other hand, do not explicitly treat the electrons, but instead describe the energetics only as interactions among the nuclei. Since these approximations result in substantial computational savings, the molecular mechanics methods can be applied to much larger systems than the quantum mechanical methods. There are many molecular properties, however, which are not accurately described by these methods. For instance, molecular bonds are neither formed nor broken during HyperChem s molecular mechanics computations the set of fixed bonds is provided as input to the computation. 

This difference is shown in the next illustration which presents the qualitative form of a potential curve for a diatomic molecule for both a molecular mechanics method (like AMBER) or a semi-empirical method (like AMI). At large internuclear distances, the differences between the two methods are obvious. With AMI, the molecule properly dissociates into atoms, while the AMBERpoten-tial continues to rise. However, in explorations of the potential curve only around the minimum, results from the two methods might be rather similar. Indeed, it is quite possible that AMBER will give more accurate structural results than AMI. This is due to the closer link between experimental data and computed results of molecular mechanics calculations. 

HyperChem provides three types of potential energy surface sampling algorithms. These are found in the HyperChem Compute menu Single Point, Geometry Optimization, and Molecular Dynamics. 

---

The various methods used in quantum chemistry make it possible to compute equilibrium intermolecular distances, to describe intermolecular forces and chemical reactions too. The usual way to calculate these properties is based on the independent particle model this is the Hartree-Fock method. The expansion of one-electron wave-functions (molecular orbitals) in practice requires technical work on computers. It was believed for years and years that ab initio computations will become a routine task even for large molecules. In spite of the enormous increase and development in computer technique, however, this expectation has not been fulfilled. The treatment of large, extended molecular systems still needs special theoretical background. In other words, some approximations should be used in the methods which describe the properties of molecules of large size and/or interacting systems. The further approximations are to be chosen carefully this caution is especially important when going beyond the HF level. The inclusion of the electron correlation in the calculations in a convenient way is still one of the most significant tasks of quantum chemistry. 

Computational chemistry is of course another technique to obtain theoretical information on perfect crystals at variable temperature. The background for this approach has been introduced in [113] and will not be further discussed here. It is important to stress that cryo-crystallography is not necessarily an experimental science, because predictions or explanations obtained from theoretical modeling are equally important in modem studies. 

Each unit is introduced by a sixty to ninety-minute lecture providing an overview of the method, some necessary background information not otherwise covered in the curriculum, and an oudine of the goals of die experiments and exercises. Thus die total lecture time over die course of the semester is four or five hours. The course is designed to facilitate hands-on exploration and active learning as much as possible. In this context the course cannot and does not provide comprehensive coverage of computational chemistry. 

Quantum Systems in Chemistry and Physics encompasses abroad spectrum of research where scientists of different backgrounds and interestsjointly place special emphasis on quantum theory applied to molecules, molecular interactions and materials. The meeting was divided into several sessions, each addressing a different aspect of the field 1 - Density matrices and density functionals 2 - Electron correlation treatments 3 - Relativistic formulations and effects 4 - Valence theory (chemical bond and bond breaking) 5 -Nuclear motion (vibronic effects and flexible molecules) 6 - Response theory (properties and spectra) 7 - Reactive collisions and chemical reactions, computational chemistry and physics and 8 - Condensed matter (clusters and crystals, surfaces and interfaces). 

Emphasis is laid on the constant-isomer series of benzenoids, not at least on the relevant enumerations in this connection. This is a typical branch of computational chemistry. A substantial amount of new computational results are reported, as well as original contributions to the background theory in the area. 

The section that follows describes basic background concepts and nomenclature. Then a classification of various programming models is outlined. Computational chemistry applications rely on many kinds of linear algebra and on equation-solving techniques that use new computer science algorithms. These implementations are delineated. A partial review of current and planned applications, developed on today s MPP supercomputers for chemistry, is presented. The last section of text gives a summary and our conclusions. Finally, we present a glossary and an appendix that reviews the currently available MPP machines. 

Before we go into some detail for the theoretical background for expert systems, we should take some time to get familiar with some of the concepts of dealing with information in a computer. I decided to use one of the newer developments in chemistry research to focus on a problem that is essential for many aspects in computational chemistry the topic of molecular descriptors. Molecular descriptors describe a molecule in a way that is appropriate for processing in a computer program. Like the term suggests, a molecular descriptor describes certain features of a molecule. 

Information about Reviews in Computational Chemistry is available on the World Wide Web. Background material on the scope and style are provided for potential readers and authors. In addition, the tables of contents of all volumes, guidance for potential authors, and ordering information are included. The Reviews in Computational Chemistry home page is being used as needed to present color graphics, supplementary material, and errata as adjuncts to the chapters. Your Web browser will find Reviews in Computational Chemistry at http //chem.iupui.edu/ boyd/rcc.html. 

---