Technology of Balanced Identification for Selection of Pine Transpiration Mathematical Model
Sokolov A.V.1, Bolondinsky V.K.2, Voloshinov V.V.1
1Institute for Information Transmission Problems (Kharkevich institute), Moscow, Russia
2Forest Research Institute, Karelian Research Centre, RAS, Petrozavodsk, Russia
Abstract. The application of numerical technology for evaluation the correspondence of a mathematical model and experimental data via the balanced (optimization) identification method is demonstrated with comparing various models of pine transpiration. A quantitative measure of the model evaluation is the cross-validation error. Current implementation of the technology allow the researcher to formulate the computing task in a text file, which contains: mathematical model formulas (including differential and/or integration equations); declarations of parameters and/or functions to be identified; data source (with experimental measurements) and additional settings of the numerical method. As a result, the software package returns unknown parameters, functions, and modeling errors. This technology is successfully used to various models in biology, medicine, physics, etc.
Key words: mathematical modeling, balanced identification, regularization, distributed computing, Everest platform, pine transpiration.
