Software package for dynamic simulation of valence atoms interactions in chemical systems applicable for parallel calculation in a flexible size network of computers.

Research proposal
by Leonid Sakharov

Concept.

A practical application of the virtual experimentation of behavior of chemical systems is always a compromise between a priory advantage of the usage of the most detailed defined models and constraints due to limited computation power at the current stage of technical evolution. The proposal is based on the following premises about current and near future technical abilities for numerical simulation of multi atom (chemical) systems:

• A direct calculation on the base of Quantum Mechanical Theory is yet out of practical implementation for the routine research and probably can find a limited usage for verification of the realism of models in the very local part of examined systems.
• A common used atomic (molecular) dynamic concept based on the empirical potential of atom interaction gives a too generalized picture of behavior of multi atoms systems and can be significantly improved by incorporating concept of valence of the atoms and developing empirical atoms interaction potentials matched in for the areas of applicability with measurable physical constants and fundamental physical laws.
• The relatively independent blocking architecture of a software application can help scale up their usability from testing of stability of small molecules depending on environment and temperature to large biochemical system behavior that will need an application of number of computers connected into network of flexible size.

Chemical model.

The following model of the chemical system will be implemented in the proposed software package:

• Atoms in the system are interacting obeying laws of mechanics and bind forces between of each pair of atoms are determining on the base of the energy potential.
• The potential of interaction for the same pair of atoms can have more then one functional dependence reflecting availability of covalent bonding for the specific situation. Difference values of minimum energy between these potentials have a sense of the energy of chemical bonding between these atoms.
• The potential of atom interactions will be constructed in the way to match measurable properties of compounds in crystalline form such as module of elasticity, coefficient of thermal expansion, melting temperature and temperature dependence of thermocapacity, law of electrostatic force for small distances, Van der Waals force for average distances and goes to the zero for non neighbors.
• The criteria for the choice between using covalent or "physical" potential of interaction for each pair of atoms will be done with the respect of valence limit for each of atoms. For example, the four closest hydrogen atoms to carbon will given a covalent potential, all other will considered as having "physical" potential.
• The system will be bordered with walls that have idealistic, like electrostatic, potential of interactions with atoms.

Software architecture.

The module architecture will be implemented for the proposed software package. It will permit inherent implementation of a parallel calculation via standard network of data exchange. On the planning stage of the project executable files with following functionality supposed to be developed:

• Simulation of atoms movements.
• Calculation of neighborhood of atoms.
• Presentation in 3D view atoms positions and movements including stereovision mode.
• Analysis of stable structures (molecules) in the system and frequency of molecular reactions (events establishing or braking covalent bonds) for possible usage for Monte Carlo approach in the future that is much more fast but needs these data as initial input parameters.
• Memory handler and tasks coordinator for the parallel calculations mode.

The process of virtual experiment in general terms can be described as following:

• User sets initial situation in chemical system. For example, for the model verification, it can be atom coordinates known from crystal structure of compounds or random positions for gas phase.
• Program calculates a function of interaction potential in form of arrays with indexes correlated to the distance between atoms for covalent and Van der Waals like cases. Also arrays of indexes of near neighbors for each atom will be created.
• The new positions and velocities of the atoms after time steps will be produced on the base laws of mechanics for atoms moving under influence given by potentials of interaction forces.
• For atoms that will move far enough from initial positions an array of indexes new neighbors will be defined. In parallel to modeling user will be able to observe the image of atoms in the system with ability to move his point of view inside of virtual system.
• The parameters of the system as whole and statistical characteristics of elementary reactions (events of changing valence binds of atom) will be registered.

The second version of the proposed software package has to work in the network of the computers with ability flexible scaling up calculating power depending on complexity of the task. One of the possible solutions for it is to hold current information about the atoms positions and movements on the public server in form of files assessable via FTP protocol. Independent computers with identical software will read a file of available tasks that will be dynamically created by coordinator program and perform a part of calculations. Such independent tasks can be a calculation of movements of part of atoms in system on one time step or calculation of indexes of neighbor. Other parallel computers can be used for observation only. Implementation of such distributed calculations can be used in different administrative schemes such as: Beowulf computer clusters, in Ethernet o the one organization especially overnight, in Internet communities of scientists.

Areas of usage.

After verification of adequacy of behavior of the developed virtual model on the relative simple objects like crystal and simple gas mixtures one can image following areas of applications of the proposed software (in order of increasing difficulty):

• Verification of stability and chemical activity of different isomers and modifications of known molecules in different chemical environments to find prospect and to filter obviously hopeless leads.
• Collection of statistical information about probabilities of molecular reactions in different configurations of atoms to establish a database that later can be used for Monte Carlo simulations.
• Modeling of special important biochemical reactions like a transportation of molecules via cell membranes and processes with participation of ferments and DNA.

Proposal Team Experience

On the primary stage of developing of the proposed project the only one member of the team, Dr. Leonid Sakharov, is considered as principal developer. Dr. Sakharov has advanced degree in Chemistry and in the same time an extensive experience as a scientific software developer. In the scope of proposed project he has self-sufficient skills and library of code solutions to perform all stages of the development completely on its own in the limited period of time. Depending on available funding a parts of the project could be helpful to delegate on the contract base to the professionals in specialized areas of science and software engineering whose personification can not be defined on this stage of planning.

Dr. Leonid Sakharov

The detailed resume with full list of publications of Dr. Sakharov can be open at resume He has PhD in Chemistry with education and extensive experience in X-ray crystallography, kinetic of crystal growth and nucleation, epitaxial thin films deposition, chromatography software development.

Here are his achievements that relate directly to the proposed project:

• A theoretical study "The temperature dependence of theoretical strength of high melted materials. Sakharov L.G. - Neorg. Mater. (rus.), 1989, v. 25, N 5, p. 768-772." with the approach of approximation of atoms interaction potential to defined theoretical strength of crystals gives a basis for extension this concept for usage in atomic dynamic simulation.
• Code solutions in the developed last year's scientific software applications by Dr. Sakharov under roof LeoKrut can be directly or with small modifications incorporated into proposed software.
• LeoCrystal - application for Monte Carlo modeling of the processes of crystal growth is implemented a solution of optimization of calculation process by extensive usage of preliminary created array of results for variety possible situations that permits dramatic increase in productivity of calculations in real time. This concept with appropriate modifications is supposed to be implemented in the proposed software as well.

The developed and implemented library of code solutions: for stereo presentation of objects in three dimensional space; NetCDF (scientific format independent to operation system) files reading and writing; FTP file exchange; calculation mathematical expression in form of string; sharing operative memory between several applications and others can be directly used in proposed application.

The preliminary code work in this direction was already initiated but set on hold with clear understanding that such project has to have outside funding support. From the moment when such funding on the level of professional consulting will be established a first working pilot version of the application can be presented in scope of several months. On the base of its performance a decision about further development should be done including estimation of realistic areas of usage, commercial perceptiveness and correspondingly a level of further funding and structure of future team of developers.