Main Page | Class Hierarchy | Class List | File List | Class Members | File Members | Related Pages

Motivations

In the design of computational algorithms for quantum chemical problems a very important and probably most time-consuming part is the design of a computer code, which implement this algorithm. A wide use of ab initio and semiempirical methods for electronic struc- ture calculations in a great part due to the availability of su.ciently convinient Quantum Chemical programs. "Black Box" program packages GAUSSIAN, SPARTAN, GAMESS, MOPAC, HYPERCHEM and others allowed increasingly wide community of scientists to get useful theoretical insights without bothering about fine details of computational algo- rithms and without spending too much time for learning the program operation. A creation of a quantum chemical package require very substantial programming efforts.For example a popular GAUSSIAN package has been developed for more than 30 years with more than 50 developers contributed to the code. While designing new computational algorithms it is reasonable to try to maximally reuse the code already existed and to implement a new method as a module of a some large package to join to the efforts of other researchers and developers. However for several reasons we decided for our purposes to develop our own program package while still trying maximally integrate our code with existing molecular simulation software. To indicate our main research target properties of macromolecules we used the acronym HARLEM (HAmiltonians to Research LargE Molecules) for the name of our program. Our goals in creating HARLEM package were the following:

  1. We aim to overcome shortcomings of the architecture of the existing quantum chemical software. The divide-and-conquer methods for electronic structure calculations we develop differ substantially from the standard quantum chemical methods which make the popular quantum chemical packages like GAUSSIAN to be inadequate for their implementation. For example the concepts of the bond, chemical group are actively used in the fragmen- tation algorithms while GAUSSIAN consider molecules just as a set of atoms placed in the space. Thus we wanted our program to be "more smart" about the chemical concepts, which require a more sophisticated organization of the structural data and the methods to handle them then is typically done in existing quantum chemisry software.

  2. We want to achieve a high degree of the program usability, so it can be a valuable to a wide range of scientists, not neccessary experts in the field of theoretical chemistry. For this purpose the program was designed with a sophisticated Graphical User Interface (GUI). GUI permits fast learning of the program functionality, reduces the time to setup calcula- tions and helps to avoid errors in setup of input parameters. Graphical presentation of the results of the calculations in many cases make them much easier to understand and often give valuable insights that is usually a goal of a scientific study.

  3. The program should have a modular structure which permit easy modifications and extensibility of the program functionality. Many developers with different expertize and goals may add and modify the code. The program architecture should be extensible enough to allow gradual evolution of the program, so the different changes to the code will typically not effect each other. Also chemical and biological problems are rarely can be classified as pure electronic. Usually one would need to perform many different types of calculations such as quantum chemical calculations and molecular dynamics simulations for a compre- hesive theoretical analysis of the system. Therefore we target in the program design to put "under the same roof" different types of molecular modeling. All these properties we tried to achieve by an object-oriented design of the program.

  4. The methods for theoretical analysis of chemical and biological problems proliferate very rapidly and the functionality of the program we develop will always cover only a small fraction of the possible theoretical approaches. We tried therefore to insure that our program will interact smoothly with other modeling software, permiting easy exchange of the data and perform a particular task on the particular software package which does it best

Text.


Generated on Tue Feb 17 02:03:24 2004 for harlem by doxygen 1.3.6