IN-SILICO DEVELOPMENTS

The research of Professor Vitaly Chaban is method-centered. We develop mathematical methods to tackle specific real-world problems. These methods are implemented as computer codes. Furthermore, numerous automation and acceleration routines based on the procedures of artificial intelligence have been developed. These are used to enhance the pace of research and development. 

If we see that our method suits these or that problem in materials science we correspondingly widen a group of chemical entities which we simulate. The academic publications and technological developments by Professor Vitaly Chaban and coworkers deal with very different classes of chemical compounds.

Over one hundred standalone computer codes have been developed. They include scientific simulation tools, such as PM7MD, GMSEARCH, and MDCNT, to conduct big pieces of research. They include straightforward but extremely powerful utilities to accelerate data conversion and exchange. They include interfaces to jointly apply general-purpose simulation toolkits, such as MOPAC, GROMACS, GAMESS, QUANTUM ESPRESSO, etc. They include add-ons to extract specific physicochemical and atomistic properties from the externally generated data sets. 

We do not do programming for the sake of programming. All developed codes are employed either in everyday life to manage in-house databases or upon teaching, or to boost commercial developments, or for the published research. Look through the methodology sections of our publications to explore the functionalities of our developments.

Accelerated Electronic Structure for Molecular Dynamics

PM7MD represents a program package to conduct and analyze semiempirical PM7-MD simulations. To obtain forces on atoms, the main program calls MOPAC2012 or MOPAC2016 which have been developed by Stewart Computational Chemistry. The PM7 parametrization within the NDDO approximation is one of the most underestimated tools in today's computational chemistry despite its huge capabilities. The outstanding work on enhancing pseudodiagonalization in semiempirical methods opens great perspectives to simulate thousands of atoms using modern computer architectures. PM7MD is useful to walk along free energy minima for significantly large systems for which ab initio MD is resource-prohibited. By adding boundary conditions, temperature and pressure coupling procedures as well as several technical enhancements, PM7MD adds the possibility to introduce thermal (entropic) effects. PM7MD is continuously upgrading to meet our evolving research needs.

Energy Landscapes in Molecular Kingdoms

GMSEARCH represents a program to find the most stable atomistic configuration of a molecule or nanoparticle or arbitrary atomistically precise ensemble. The program makes robust use of the functions of the SciPy library, with modifications where necessary. The interfaces to some electronic structure algorithms to conduct repetitive single-point calculations have been coded. Those of particular relevance are periodic plane-wave-basis-set DFT and non-periodic atom-centered-basis-set DFT. GMSEARCH has been vigorously applied to globally optimize geometries of metal structures, quantum dots,  solvated multi-charged cations and anions, poly-component mixtures, and many other physicochemical systems.

Potential-based Molecular Dynamics for Carbonaceous Nanoscale Tubes

MDCNT represents the program package to perform molecular dynamics simulations with nanotubes and nanopores. MDCNT was employed in the simulations that constituted the doctoral dissertation of Prof. Vitaly Chaban. The program package includes multiple auxiliary utilities: to generate ideal geometries of nanotubes of various chiralities; to build simulation boxes efficiently; to analyze the solvent inside and outside nanotubes, etc. The codes are in Python and C++ and can be relatively smoothly compiled/executed both under Linux and Windows.

In-Silico Chemical Reactor 

The simulation of chemical reactions is a paramount of a computational chemist worldwide. Apart from a programmed transition state search, it is essential to get hints regarding the possible reaction mechanisms for given chemical compositions (constant ‘N’ ensembles). We have developed a code that reveals possible transformation pathways and rates their relative probabilities by applying the PM7-MD loop empowered with certain momentum perturbation schemes. In this way, the system acquires a gradually higher chemical flexibility. The unraveled reaction routes can, next, be systematized kinetically and thermodynamically. A coveted level of theory can be readily applied to refine and characterize the vibrational properties of the transition states and pertaining activation barriers. The software is particularly useful for investigating poorly understood cases and dealing with many-component systems. The chemical evolution on the Earth can be efficiently simulated. 

ASE Library

In our in-house developments, we strive to maintain compatibility with the global trends in computational chemistry by incorporating and modifying algorithms previously coded in external libraries. The Atomic Simulation Environment (ASE) is provided to the scientific community as a free-of-charge and open-source Python-compatible library for performing atomistic simulations. ASE simplifies interfacing the existing program codes, which perform computationally demanding procedures. The available functions written in Python2/3 provide high-level links to a variety of electronic structure codes. Furthermore, the tools for manipulating and analyzing atomic structures have been natively coded in ASE. Overall, the resources of the Atomistic Simulation Environment substantially enhance the pace of in-house development and broaden the outlook of possible solutions. We simulate a wide range of physical systems, including materials, molecules, and even surfaces while combining electronic-structure methods with the configurations space sampling methods outside of quantum chemistry. PM7-MD and GMSEARCH employ numerous modified and non-modified functions of ASE, which brought their continuous development on a qualitatively different level of generality and extensibility.

SciPy Library

SciPy is a free and open-source Python library for scientific computing with an accent on standard numerical procedures to boost programming. It contains modules for optimization, linear algebra, integration, interpolation, special functions, FFT, signal and image processing, ordinary differential equation solvers, etc. SciPy may be seen as an extension of NumPy, which contains high-performance multidimensional array objects and mathematical functions to operate on them. SciPy implements optimization algorithms and signal processing tools. SciPy is useful for coding the optimization of given functions, and finding maxima and minima points. Furthermore, the linear algebra routines are frequently of use in atomic modeling. Linear algebraic algorithms, such as matrix multiplication, inversion, and eigenvalue decomposition, are omnipresent in quantum chemical modeling. SciPy provides a variety of functions for numerical integration. This is useful for calculating the area under a curve, finding the center of mass of an object, solving differential equations, etc. The availability of the fast Fourier transform functions is sometimes greatly acknowledged. We gratefully employ, extend, and modify (fix errors) SciPy to analyze our molecular ensembles and trial new method development ideas.

Concluding Notes

The in-house developed codes are proprietary to Professor Vitaly Chaban but some of them can be shared in the framework of collaboration.