TY - JOUR

T1 - Parallel speedup analysis of an adjoint ensemble-based source identification algorithm

AU - Penenko, Alexey

AU - Gochakov, Alexander

N1 - Funding Information: The construction of the scenario for modeling, the adaptation of the algorithm to the Novosibirsk city conditions, and numerical experiments were carried out with the financial support of the Russian Foundation for Basic Research and the Novosibirsk Region Government within the framework of the scientific project No. 19-47-540011. 2GIS Company granted the information about the road traffic intensity. The continuation problem analysis was supported by Russian Foundation for Basic Research project No. 20-01-00560. The development of the algorithm modification with weaker misfit monotonicity and its analysis were supported by State Budget Program No 0315-2019-0004. Publisher Copyright: © 2021 Institute of Physics Publishing. All rights reserved. Copyright: Copyright 2021 Elsevier B.V., All rights reserved.

PY - 2021/1/4

Y1 - 2021/1/4

N2 - In the approach to the inverse problems, based on the sensitivity operators, the inverse problem is transformed into the family of quasi-linear operator equations. The sensitivity operator is constructed of the ensemble of the adjoint problem solutions corresponding to different adjoint problems source functions. Hence, from the theoretical point of view, sensitivity operator-based algorithms can be efficiently parallelized. The paper's objective is to carry out the experimental study of the algorithm computation time for the different number of computation cores.

AB - In the approach to the inverse problems, based on the sensitivity operators, the inverse problem is transformed into the family of quasi-linear operator equations. The sensitivity operator is constructed of the ensemble of the adjoint problem solutions corresponding to different adjoint problems source functions. Hence, from the theoretical point of view, sensitivity operator-based algorithms can be efficiently parallelized. The paper's objective is to carry out the experimental study of the algorithm computation time for the different number of computation cores.

UR - http://www.scopus.com/inward/record.url?scp=85100710970&partnerID=8YFLogxK

U2 - 10.1088/1742-6596/1715/1/012072

DO - 10.1088/1742-6596/1715/1/012072

M3 - Conference article

AN - SCOPUS:85100710970

VL - 1715

JO - Journal of Physics: Conference Series

JF - Journal of Physics: Conference Series

SN - 1742-6588

IS - 1

M1 - 012072

T2 - International Conference on Marchuk Scientific Readings 2020, MSR 2020

Y2 - 19 October 2020 through 23 October 2020

ER -