Center for Computational Physics LPR MIPT

We are engaged in atomic modeling of industrial liquids: solvents, fuels, oils and lubricants. We are interested in the predictive power of atomistic modeling. Such calculations are necessary for conducting multi-scale modeling of physical processes: fluid flow in mechanisms (lubricant), in the pore space (fuel extraction), through membranes (lithium extraction). We participate in Competitions for calculating the properties of industrial liquids, conducted by industry leaders in the United States (Army Research Lab, Dow Chemical, NIST, etc.). In 2018, our team took second place in the competition, beating teams from NIST, Imperial College London, Shanghai Jiao Tong Univ. In 2019— he took the first place among participants from the Army Research Lab, John Hopkins Univ., Caltech and Shanghai Jiao Tong Univ. The contestants competed to accurately predict the properties of the lubricating fluid at extreme pressures. In 2023, we also won in terms of the accuracy of predicting the viscosity of binary mixtures.

The use of modern methods and GPU technologies makes it possible to speed up the calculation processes of third-party implementations tenfold and reduce energy costs significantly, compared with approaches using only the CPU. Our team has leading competencies in optimizing and profiling computational codes for specific hybrid architectures. We have access to Russia's leading supercomputers. Experience in the application, evaluation and acceleration of various methods and software implementations: StarCCM, Ansys CFX, Fluent, FlowVision, OpenFOAM, Trilinos PETSc, AMGX, Hypre, amgcl, GROMACS, LAMMPS, OpenMM… A highly efficient header-only OpenSource library created by the team for implementing high-performance computing methods 50+ articles related to Scopus/WOS

Development of mathematical models of the optimized process, taking into account the necessary constraints and parameters, by changing which optimization needs to be carried out, building the objective function, finding the parameter values at which the objective function takes the maximum (minimum) value. On this topic, our team has: 10 years in comprehensive optimization 5 commercial projects 20+ related articles in Scopus

We develop multiphysical simulators, software that simulates physico-chemical processes of various nature: flows in porous media, filtration of impurities, thermophysics, modeling of elastic properties and destruction of materials, modeling of bulk media. Our solutions allow us to reproduce the details of the technical process under conditions and parameters that are not available in the experiment on all scales of interest.

We are developing and applying direct methods and piggyback models to describe multiphase filtration in porous media. In practice, such models are being implemented by our team in the oil and gas industry, hydrology and soil science, and materials science.