Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Research › peer-review
Integrated computational environment for grid generation parallel technologies. / Il’in, Valery.
Parallel Computational Technologies - 14th International Conference, PCT 2020, Revised Selected Papers. ed. / Leonid Sokolinsky; Mikhail Zymbler. Springer Gabler, 2020. p. 58-68 (Communications in Computer and Information Science; Vol. 1263 CCIS).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Research › peer-review
}
TY - GEN
T1 - Integrated computational environment for grid generation parallel technologies
AU - Il’in, Valery
N1 - Publisher Copyright: © Springer Nature Switzerland AG 2020.
PY - 2020/1/1
Y1 - 2020/1/1
N2 - This paper is devoted to the conception and general structure of the integrated computational environment for constructing multi-dimensional large grids (with 1010 nodes and more) for high-performance solutions of interdisciplinary direct and inverse mathematical modelling problems in computational domains with complicated geometrical boundaries and contrast material properties. This includes direct and inverse statements which are described by the system of differential and/or integral equations. The constructed computational grid domain consists of subdomains featuring a grid, which may be of different types (structured or non-structured); discretization at the internal boundaries can be consistent or non-consistent. The methodology of such quasi-structured meshes makes it possible to use various algorithms and codes in the subdomains, as well as different data structure formats and their conversion. The proposed technologies include grid quality control, the generation of dynamic grids adapted to singularities of input geometric data of structures and multigrid approaches with local refinements, taking into account information about the solution to be obtained. The balanced grid domain decomposition, based on hybrid programming at the heterogeneous clusters with distributed and hierarchical shared memory, supports scalable parallelization. In addition, the paper outlines the technological requirements to provide a successful long-life cycle for the proposed computational environment. In a sense, the considered development presents a stable software ecosystem (integrated grid generator DELAUNAY) for supercomputing modelling in the epoch of big data and artificial intellect.
AB - This paper is devoted to the conception and general structure of the integrated computational environment for constructing multi-dimensional large grids (with 1010 nodes and more) for high-performance solutions of interdisciplinary direct and inverse mathematical modelling problems in computational domains with complicated geometrical boundaries and contrast material properties. This includes direct and inverse statements which are described by the system of differential and/or integral equations. The constructed computational grid domain consists of subdomains featuring a grid, which may be of different types (structured or non-structured); discretization at the internal boundaries can be consistent or non-consistent. The methodology of such quasi-structured meshes makes it possible to use various algorithms and codes in the subdomains, as well as different data structure formats and their conversion. The proposed technologies include grid quality control, the generation of dynamic grids adapted to singularities of input geometric data of structures and multigrid approaches with local refinements, taking into account information about the solution to be obtained. The balanced grid domain decomposition, based on hybrid programming at the heterogeneous clusters with distributed and hierarchical shared memory, supports scalable parallelization. In addition, the paper outlines the technological requirements to provide a successful long-life cycle for the proposed computational environment. In a sense, the considered development presents a stable software ecosystem (integrated grid generator DELAUNAY) for supercomputing modelling in the epoch of big data and artificial intellect.
KW - Adaptive quasi-structured grids
KW - Data structures
KW - Grid computational domain
KW - Grid generation methods
KW - Multi-dimensional boundary value problems
KW - Scalable parallelization
UR - http://www.scopus.com/inward/record.url?scp=85089315276&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-55326-5_5
DO - 10.1007/978-3-030-55326-5_5
M3 - Conference contribution
AN - SCOPUS:85089315276
SN - 9783030553258
T3 - Communications in Computer and Information Science
SP - 58
EP - 68
BT - Parallel Computational Technologies - 14th International Conference, PCT 2020, Revised Selected Papers
A2 - Sokolinsky, Leonid
A2 - Zymbler, Mikhail
PB - Springer Gabler
T2 - 14th International Scientific Conference on Parallel Computational Technologies, PCT 2020
Y2 - 27 May 2020 through 29 May 2020
ER -
ID: 24954780