CC BY
4methods: the simple outer method (e.g., Jacobi) performed in double precision and the more robust method
(e.g., BiCGStab + AMG preconditioner) performed in single precision. It is shown that the proper choice of the
inner solver stopping criteria is the key aspect of achieving the appropriate convergence rate, and the overall
convergence can be equal to the basic double precision method configuration, providing the decrease in the
calculation time by a factor of 1.6-1.7.
This work was supported by the Russian Science Foundation (grant 18-71-10075).
References
1. Krasnopolsky B., Medvedev A. Evaluating Performance of Mixed Precision Linear Solvers with Iterative Refinement
// Supercomputing Frontiers and Innovations. 2021 (accepted).
Numerical simulation of propellers rotation under cavitation conditions
O. L. Krutyakova, A. S. Kozelkov, V. V. Kurulin
FSUE �Russian Federal Nuclear Center � All-Russia Research Institute of Experimental Physics� Nizhny Novgorod
Region, Sarov
Email: kurulin@mail.ru
DOI 10.24412/cl-35065-2021-1-01-75
Cavitation influences the design of propellers considerably, so the research on their operation under cavi-
tation conditions is critical [1]. The work describes numerical simulation of cavitation processes when a model
VP1304 propeller is rotating. Volume of Fluid (VOF) method is used for numerical simulation, which is realized
in LOGOS software package. It allows numerical simulation of two-phase problems with free surface [2]. Cavi-
tation is accounted for when the method is supplemented with the account for the phase-to-phase mass ex-
change; cavitation models are used to compute its rates [3, 4]. The work shows a physical-mathematical mod-
el, briefly describes a numerical method used, as well as cavitation models; the models are compared after-
wards. Numerical simulation of the rotating propellers is demonstrated using the problems of flow-past of the
model VP1304 propeller under the conditions of developed cavitation.
References
1. A.N. Ivanov. Hydrodynamics of developed cavitation flows. Leningrad: Sudostroyeniye, 1980.
2. Kurulin V.V., Kozelkov A.S., Efremov V.R., Yatsevich S.V., Tarasova N.V. Application of VOF method to solve
complex problems with free surface // XXV All-Russia Workshop with international participation on jet, separated and
non-stationary flows. Saint-Petersburg, 2018. pp. 144-145.
3. Schnerr G. H., Sauer J. Physical and numerical modeling of unsteady cavitation dynamics // Proceedings of the
Fourth International Conference on Multiphase. Flow, New Orleans, USA, May 27 � June 1, 2001. P.1�12.
4. Zwart P.J., Gerber A.G., Belamri T. A Two-phase flow model for predicting cavitation dynamics // The Fifth
International Conference on Multiphase Flow. Yokohama, Japan. 2004.
Research on the accuracy of numerical simulation of acoustic perturbations in a fluid basing
on Navier � Stokes equations
O. L. Krutyakova, A. S. Kozelkov, V. V. Kurulin
FSUE �Russian Federal Nuclear Center � All-Russia Research Institute of Experimental Physics� Nizhny Novgorod
Region, Sarov
Email: kurulin@mail.ru
DOI 10.24412/cl-35065-2021-1-01-76
The work is devoted to the necessary space and time resolution when simulating the propagation of
acoustic perturbations in a fluid, analysis of the accuracy of the solution as a function of the parameters of the
iteration procedure and numerical discretization of the equations. A popular SIMPLE method [1] together with
finite-volume discretization of the equations on the basis of the home software package LOGOS [2] is used as a
numerical method. A problem of perturbations propagation from a harmonic-oscillations source in a fluid is
described for the assessment [3]. Space and time resolution necessary to provide acceptable accuracy of the
solution is estimated. The produced estimations are validated using the problem of propagation of harmonic
waves from a point source in a fluid.
References
1. Lashkin S.V., Kozelkov A.S., Yalozo A.V., Gerasimov V.Y., Zelensky D.K. Efficiency analysis of the parallel
implementation of the SIMPLE algorithm on multiprocessor computers // Journal Of Applied Mechanics And Technical
Physics, 2017, v.58, Issue 7, p. 1242-1259..
2. Kozelkov A.S., Kurulin V.V., Lashkin S.V., Shagaliev R.M., Yalozo A.V., Investigation of supercomputer capabilities
for the scalable numerical simulation of computational fluid dynamics problems in industrial applications //
Computational mathematics and mathematical physics, 2016, V. 56, Issue 8, P. 1524�1535.
3. Fenton J. D. A Fifth-Order Stokes Theory for Steady Waves // Coastal and Ocean Eng., 1985, v.111, Issue 2, p. 216-
234 [4]Zwart P.J., Gerber A.G., Belamri T. A Two-phase flow model for predicting cavitation dynamics // Fifth International
Conference on Multiphase Flow. Yokohama, Japan. 2004.
Algorithms and implementation of active knowledge in LuNA system
V. Malyshkin, V. Perepelkin
Institute of Computational Mathematics and Mathematical Geophysics SB RAS
Email: malysh@ssd.sscc.ru
DOI 10.24412/cl-35065-2021-1-01-77
Active knowledge development problems in the domain of numerical simulations on supercomputers is
discussed. Basic components of the LuNA system, which supports active knowledge, are concerned. Peculiari-
ties and limitations of the system, as well as its current condition and abilities are discussed. Applications of
LuNA for a series of tests are analyzed.
References
1. Valkovsky, V., Malyshkin, V.: Synthesis of parallel programs and systems on the basis of computational models.
Nauka, Novosibirsk, 1988 (in Russian).
2. Victor Malyshkin. Active Knowledge, LuNA and Literacy for Oncoming Centuries // Springer, LNCS, V. 9465 (2015),
pp. 292-303. DOI: 10.1007/978-3-319-25527-9_19.
3. Victor Malyshkin, Vladislav Perepelkin, and Georgy Schukin. Distributed Algorithm of Data Allocation in the
Fragmented Programming System LuNA // Springer, LNCS, V. 9251 (2015), pp. 80-85. DOI: 10.1007/978-3-319-21909-7_8.
Execution trace based optimization of fragmented programs performance
V. Perepelkin, V. Malyshkin
Institute of Computational Mathematics and Mathematical Geophysics SB RAS
Email: malysh@ssd.sscc.ru
DOI 10.24412/cl-35065-2021-1-01-78
Profiling and tracing of parallel programs execution is a source of information on quantitative characteris-
tics of efficiency, such as computing nodes load over time, communication subsystem load, memory consump-
tion, etc. It is essential, that the characteristics can be measured with connection to source code locations of
the program executed. This information can often be used to significantly improve performance of parallel
