Dresden 2017 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 46: Modelling and Data Analysis
DY 46.4: Vortrag
Donnerstag, 23. März 2017, 15:45–16:00, HÜL 186
Estimability of model parameters and state variables from observed time series — •Ulrich Parlitz1,2, Jan Schumann-Bischoff1, and Stefan Luther1,2,3 — 1Max Planck Institute for Dynamics and Self-Organization, Am Fassberg 17, 37077 Göttingen, Germany — 2Institute for Nonlinear Dynamics, Georg-August-Universität Göttingen, Friedrich-Hund-Platz 1, 37077 Göttingen, Germany — 3Department of Pharmacology, University Medical Center, Robert-Koch-Str. 40, 37075 Göttingen, Germany
For many physical processes a mathematical model exists but not all state variables and parameters can be measured directly. Instead, their values have to be estimated from observed time series and the question arises whether unique results can be expected for the estimates and how reliable the obtained results are. To answer these questions we propose to analyze the null space of the linearized delay coordinates map [1]. This approach to estimability analysis is illustrated and generalized to multivariate time series.
[1] J. Schumann-Bischoff, S. Luther, U. Parlitz, Phys. Rev. E 94, 032221 (2016).