UDC 519.6
1 V.M. Glushkov Institute of Cybernetics, National Academy of Sciences of Ukraine, Kyiv, Ukraine
khimich505@gmail.com
|
2 V.M. Glushkov Institute of Cybernetics, National Academy of Sciences of Ukraine, Kyiv, Ukraine
alex50popov@gmail.com
|
A PARALLEL ALGORITHM FOR SOLVING THE PARTIAL EIGENVALUE PROBLEM
FOR BLOCK-DIAGONAL BORDERED MATRICES
Abstract. A hybrid algorithm of the iteration method for the subspace of solution of a partial generalized eigenvalue problem
for symmetric positive definite sparse matrices of block-diagonal structure with bordering on hybrid computers with graphic processors is proposed, efficiency coefficients of the algorithm are obtained, and the algorithm is tested against test and practical problems.
Keywords: algebraic eigenvalue problem, computer of hybrid architecture, hybrid algorithm, subspace iteration method, efficiency of parallel algorithm.
FULL TEXT
REFERENCES
- Pissanecki S. Sparse matrix technology [Russian translation]. Moscow: Mir, 1988. 410 p.
- Parlett B. Symmetric eigenvalue problem [Russian translation]. Moscow: Mir, 1983. 382 p.
- Boreskov A.V., Kharlamov A.A. Basics of working with CUDA technology [in Russian]. Moscow: Press, 2010. 232 p.
- NetLib. 2015. URL: http://www.netlib.org/.
- cuBLAS. URL: https://developer.nvidia.com/cublas.
- cuSparse Library. URL: http://docs.nvidia.com/cuda/cuSPARSE/.
- MAGMA. URL: http://icl.cs.utk.edu/magma/.
- AMD. URL: http://www.amd.com/en-gb.
- Math Kernel Library. URL: https://software.intel.com./en-us/mkl.
- SLEPc. 2015. URL: http://slepc.upv.es/.
- LIS. 2015. URL: http://www.ssisc.org/lis/.
- Khimich A.N., Molchanov I.N., Popov A.V., Chistyakova T.V., Yakovlev M.F. Parallel algorithms for solving problems of computational mathematics [in Russian]. Kiev: Nauk. dumka, 2008. 247 p.
- Khimich O.M., Sidoruk V.A. A hybrid algorithm for solving linear systems with sparse matrices based on the block LLT-method. Komp’yuterna matematyka. 2015. N 1 P. 67–74.
- Khimich A.N., Popov A.V., Chistyakov A.V. Hybrid algorithms for solving the algebraic eigenvalue problem with sparse matrices. Kibernetika i sistemnyj analiz. 2017. Vol. 53, N 6. P. 132–146.
- Khimich O.M., Sidoruk V.A. Small-tile hybrid algorithm for factorization of a sparse matrix. Proceedings of the All-Ukrainian scientific-practical conference with international participation "Informatics and Systems Sciences (CCI-2016)" (Poltava, March 19-21, 2016).P. 326–328.
- Velikoivanenko E.A., Milenin A.S., Popov A.V., Sidoruk V.A., Khimich A.N. Methods and technologies of high-performance computing for mathematical modeling of the stress-strain state of structures taking into account ductile fracture. International scientific and technical journal " Problemy upravleniya i informatiki". 2014. N 6. C. 42–52.
- Sergienko I.V., Deineka V.S. Solving combined inverse problems for multicomponent parabolic distributed systems. Cybernetics and Systems Analysis. 2007. Vol. 43, N 5. P. 655–674.
- Velikoivanenko E.A., Milenin A.S., Popov A.V., Sidoruk V.A., Khimich A.N. Methods of numerical forecasting of serviceability of welded structures on computers of hybrid architecture. Cybernetics and Systems Analysis. 2019. Vol. 53, N 1. P. 117–127.
- Baranov A.Yu., Slobodyan Ya.E., Popov A.V., Khimich A.N. Mathematical modeling of the strength of building structures on hybrid computing systems. International scientific and technical journal " Problemy upravleniya i informatiki". 2017. N 4. P. 68–81.
- Popov O.V. Computer study of the validity of solutions of a generalized algebraic problem of eigenvalues. Komp’yuterna matematyka. 2012. N. 1. P. 52–59.
- The SuiteSparse Matrix Collection. URL: https://cise.ufl.edu/research/sparse/matrices/.
- Supercomputer complex SKIT. URL: http://icybcluster.org.ua.
- Khimich A.N., Decret V.A., Popov A.V., Chistyakov A.V. Numerical study of the stability of composite materials on computers of hybrid architecture. International scientific and technical journal " Problemy upravleniya i informatiki". 2018. N 4. P. 73–88.
- Bystrov V.M., Decret V.A., Zelensky V.S. Numerical investigation of the stability of a laminated composite material under compression by a surface load. Problemy obchyslyuvalʹnoyi mekhaniky i mitsnosti konstruktsiy. 2018. Iss. 28. P. 23–33.
- MATLAB for deep learning. URL: https://mathworks.com/.