Registration Number
0213U003717,
0112U005153
, R & D reports
Title
Hardware-software system for modeling and analysis of natural and synthetic nano-systems in a Grid environment.
popup.stage_title
Head
Yatsenko Leonid Petrovich,
Registration Date
27-02-2013
Organization
Institute of physics NASU
popup.description2
Grid technologies are widely used for modeling of natural and man-made nanosystems, in particular biological macromolecules and artificial nanostructures. The most widely used methods of molecular dynamics and quantum chemistry are allowed to model the structure and dynamics of complex supramolecular systems and obtain the trajectories of all atoms of the system over time. The size of a typical trajectory of molecular dynamics for the development of computational power is growing and now is the tens of gigabytes. The processing of such paths can take hours or even days. This makes the data analysis stage rather complicated technically and numerically intensive. Accordingly there is a need to analyze the trajectories of moving tasks in a grid environment. Integration analysis of molecular dynamics trajectories in the grid environment requires several purposes: 1) the creation of new software for the analysis of the trajectories, 2) optimization of its operation on the individual computer clusters, and 3) providing access to local and remote resources, which are physically stored trajectory, 4) to create an interface for user interaction analysis. At this stage of the project, we were continuing on the development of libraries for molecular modeling and analysis Pteros. Implementation of asynchronous reading of large MD trajectory with parallel dispatching of reading any number of frames between the analytical procedures that are performed by independent subsidiaries flows was developed. The buffer with reference counted which collected and kept read frames until they are processed at least one of analytic flows and remove them after use was Realized. Implemented single universal API to read any data formats MD, which allows for a very simple and transparent to add new file formats in the library. A wrapper for plugins IO program VMD molfile plugins compatible with universal API IO Pteros was implementated. Binding with programming language Python was greatly improved. We developed an additional system of interpretation of the startup analytical programs. Web-based library was implemented. The transfer of development on more advanced distributed version control system GIT was initiated. Implementation of the algorithm grid search was radically redesigned, also full support for the periodic boundary conditions was added. The documentation at http://pteros.sourceforge.net/. was updated. Clusters consisting of a small number of atoms - less than a few hundred, have a significantly different structure and properties than the bulk material. For example, well-known buckyballs are fundamentally different from the structure of graphite and diamond, as well as the newly discovered embedded-shell nanoparticles (CdSe)33, (CdSe)34 does not have the wurtzite structure of bulk CdSe .. We have previously published the results of-flight mass spectroscopy of clusters ZnO, obtained by laser ablation of powder zinc peroxide (ZnO2), and shows that from the proposed precursor can be formed much greater variety of clusters. Additional points were detected clusters of high stability - "magic clusters" - (ZnO)34, (ZnO)60 and (ZnO)78, and found a way to increase their relative amount by adding alkylamines. Our approach to solving the problem of establishing the structure of small atomic clusters is as follows. As a basis we take the experimental data, namely, the mass spectra of these compounds. Next, choose the portion of the spectrum that contains information about the prevalence of (stability) of some characteristic (magic, if any) of the small size clusters. Then construct all possible isomers of the selected clusters and perform a preliminary selection of the "best" of isomers, i.e. those that have the lowest possible total energy. According to the results of quantum chemical calculations, performed the final selection of isomers for these clusters. Based on a comparison of the results of calculations (energy cluster communication) with experimental data (prevalence clusters) we draw conclusions about the realism of selected calculation method. If the results are satisfactory - is performed quantum-chemical calculations of large clusters selected method. We believe that the results thus obtained the calculation can be trusted. At this stage, a complex set of calculations, where n = 12, 13 was carried out by using specially designed software and standard software packages for the quantum-chemical calculations (Firefly) and the energetically most favorable structures of atomic clusters (ZnO)n and (ZnS)n were established. Based on the comparison of the calculation results with experimental data (mass spectra corresponding clusters by laser ablation of a solid target) the following conclusion on the rational choice of calculation method and basis functions for the investigated compounds were proposed: for ZnO a simple basis in calculations from first principles can be used, while such calculations for ZnS poorly agreed with the experimental data, and the semi-empirical AM1 method gives a satisfactory agreement between the calculation results and experiments for both compounds. Also the conclusions on the feasibility to carry out computation of the (ZnO)n clusters stability in the grid environment were formulated : semi-empirical method is efficient to use for clusters with n> 60, ab initio methods - for n> 2 ? 5 depending on the basis used and the level of theory.
Product Description
popup.authors
Єсилевський Семен Олександрович
Березецька Наталя Михайлівна
Блонський Іван Васильович
Дмитрук Ігор Миколайович
Дмитрук Андрій Миколайович
Рябцев Андрій Володимирович
Харкянен Валерій Миколайович
Шинкаренко Євген Вікторович
Яценко Леонід Петрович
popup.nrat_date
2020-04-02
