Metos3D: a marine ecosystem toolkit for optimization and simulation in 3-D – Simulation Package v0.2
Author:
Piwonski J.,Slawig T.
Abstract
Abstract. A general programming interface for parameter identification for marine ecosystem models is introduced. A comprehensive solver software for periodic steady-states is implemented that includes a fixed point iteration (spin-up) and a Newton solver. The software is based on the Portable, Extensible Toolkit for Scientific Computation (PETSc) library and uses transport matrices for efficient off-line simulation in 3-D. In addition to the usage of PETSc's parallel data structures and PETSc's Newton solver, an own load balancing algorithm is implemented. A simple verification is carried out using a well investigated biogeochemical model for phosphate (PO4) and dissolved organic phosphorous (DOP) with 7 parameters. The model is coupled via the interface to transport matrices that correspond to a longitudinal and latitudinal resolution of 2.8125° and 15 vertical layers. Initial tests show that both solvers and the load balancing algorithm work correctly. Further experiments demonstrate the robustness of the Newton solver with respect to parameter variations. Moreover, the numerical tests reveal that, with optimal control settings, the Newton solver converges at least 6 times faster towards a solution than the spin-up. However, additional twin experiments reveal differences between both solvers regarding a derivative-based black-box optimization. Whereas an optimization run with spin-up-based model evaluations is capable to identify model parameters of a reference solution, Newton-based model evaluations result in an inaccurate gradient approximation.
Publisher
Copernicus GmbH
Reference38 articles.
1. Balay, S., Gropp, W. D., McInnes, L. C., and Smith, B. F.: Efficient management of parallelism in object oriented numerical software libraries, in: Modern Software Tools in Scientific Computing, edited by: Arge, E., Bruaset, A. M., and Langtangen, H. P., Birkhäuser Press, Basel, 163–202, 1997. 2. Balay, S., Brown, J., Buschelman, K., Eijkhout, V., Gropp, W. D., Kaushik, D., Knepley, M. G., McInnes, L. C., Smith, B. F., and Zhang, H.: PETSc Users Manual, Tech. Rep. ANL-95/11 – Revision 3.3, Argonne National Laboratory, Lemont, 2012a. 3. Balay, S., Buschelman, K., Gropp, W. D., Kaushik, D., Knepley, M. G., McInnes, L. C., Smith, B. F., and Zhang, H.: PETSc Web page, available at: http://www.mcs.anl.gov/petsc/ (last access: 12 July 2013), 2012b. 4. Bernsen, E., Dijkstra, H. A., and Wubs, F. W.: A method to reduce the spin-up time of ocean models, Ocean Model., 20, 380–392, https://doi.org/10.1016/j.ocemod.2007.10.008, 2008. 5. Boyer, T., Antonov, J., Baranova, O., Coleman, C., Garcia, H., Grodsky, A., Johnson, D., Locarnini, R., Mishonov, A., O'Brien, T., Paver, C., Reagan, J., Seidov, D., Smolyar, I., and Zweng, M.: World Ocean Database 2013, Tech. rep., NOAA Atlas NESDIS 72, s. Levitus, edited by: Mishonov, A., Technical Ed., Silver Spring, 2013.
|
|