Comparison of the Target Functions in the Prony’s Problem of Measurement Data Approximation
DOI:
https://doi.org/10.14529/cmse220202Keywords:
measurement data approximation, Prony’s problem, C. Lanczos example of separation of exponentials, local stability, least squares method, variational methodAbstract
In the article, we compare two objective functions in the Prony’s problem of approximation of measurement data by solutions of a linear differential equation of a given order with constant coefficients. The target functions differ in the type of dependence of the gradient on the coefficients of the equation (linear or with complex nonlinearity) and are 1) the norm of the residual of the equation (linear least squares method) or 2) the norm of the approximation error according to A. Householder (variational identification method). In the latter case, the coefficients of the differential equation and the initial conditions of the solution are jointly optimized. For the considered objective functions, the local stability constants of the solution to the Prony’s problem are calculated using local expansions of the dependencies of the optimal coefficients of the equation as implicit functions of the data with the condition that the gradient of the objective function is identically equal to zero. On this basis, a method is proposed for determining the permissible error in the data to ensure a given level of deviation of the solution from the true value. We use the example of K. Lanczos of calculating the exponents given observations of the sum of three exponents with rounding errors to confirm a significant advantage (in terms of the allowable errors in the data) of using the variational objective function. The adequacy of the used local stability indices for considerable perturbations is verified by numerical experiment.References
Marple S.L. Digital spectral analysis: with applications. USA: Prentice-Hall, 1986. 492 p.
Berdyshev V.I., Petrak L.V. Function approximation, compression of numerical information, applications. Ekaterinburg: UB RAN, 1999. 296 p. (in Russian)
Householder A.S. On Prony’s method of fitting exponential decay curves and multiple-hit survival curves. Oak Ridge National Lab. Report ORNL–455. 1950. Oak Ridge, Tennessee. URL: http://www.technicalreports.org/trail/detail/11105 (accessed: 15.12.2021).
Egorshin A.O. Least squares method and the “fast” algorithms in variational problems of identification and filtration (VI method). Avtometriya. 1988. No. 1. P. 30–42. (in Russian). URL: http://www.iae.nsk.su/images/stories/5_Autometria/5_Archives/1988/1/30-42.pdf (accessed: 15.12.2021).
Fuller W.A. Measurement Error Models. New York: Wiley, 1987. 440 p.
Lomov A.A. On Quantitative A Priori Measures of Identifiability of Coefficients of Linear Dynamic Systems. Journal of Computer and System Sciences International. 2011. Vol. 50, no. 1. P. 1–13. DOI: 10.1134/S106423071101014X.
Lomov A.A., Fedoseev A.V. Comparison of Parameter Identification Methods for Linear Dynamic Systems Under Mixed Noise. Springer Journal of Mathematical Sciences. 2021. Vol. 253, no. 3. P. 407–418. DOI: 10.1007/s10958-021-05238-0.
Lanczos C. Applied analysis. USA: Prentice Hall, 1956. 539 p.
Lomov A.A. On Convergence of Computational Algorithms for a Variational Problem of Identifying the Coefficients of Difference Equations. Journal of Applied and Industrial Mathematics. 2020. Vol. 14, no. 3. P. 541–554. DOI: 10.1134/S1990478920030138.
de Prony G. Essai éxperimental et analytique: sur les lois de la dilatabilité de fluides élastique et sur celles de la force expansive de la vapeur de l’alkool, à différentes températures. Journal de l’école Polytechnique. 1795. Vol. 1, no. 22. P. 24–76. URL: http://users.polytech.unice.fr/~leroux/PRONY.pdf (accessed: 15.12.2021).
Mitrofanov G.M., Priimenko V.I. Basics and Applications of the Prony filtering. Tehnologii seismorazvedki. 2011. No. 3. P. 93–108. (in Russian)
Kolomeytseva A.V., Mishugova G.V., Mool A.P., Ryabykh G.Yu. Application of the wavelet transform and the Prony method for the identification of biogenic signals. Vestnik DGTU. 2010. Vol. 10, no. 4. P. 455–465. (in Russian)
Bjorck A. Numerical methods for least squares problems. USA: SIAM, 1996. 425 p. DOI: 10.1137/1.9781611971484.ch1.
Keller I., Plonka G. Modifications of Prony’s Method for the Recovery and Sparse Approximation with Generalized Exponential Sums. Approximation Theory XVI / ed. by G.E. Fasshauer, M. Neamtu, L.L. Schumaker. Cham: Springer International Publishing, 2021. P. 123–152. DOI: 10.1007/978-3-030-57464-2_7.
Osborne M.R. A class of nonlinear regression problems. Data representation / ed. by R.S. Anderssen, M.R. Osborne. St. Lucia: University of Queensland Press, 1970. P. 94–101.
Egorshin A.O., Budyanov V.P. Smoothing of signals and estimation of dynamic parameters in automatic systems using a digital computer. Avtometriya. 1973. No. 1. P. 78–82. (in Russian) URL: https://www.iae.nsk.su/images/stories/5_Autometria/5_Archives/1973/1/78-82.pdf (accessed: 15.12.2021).
Bryson A.E., Ho Y.-C. Applied Optimal Control: Optimization, Estimation and Control. Waltham, MA: Blaisdell. 1969. 481 p.
Egorshin A.O. On discretization of linear differential equations. Bulletin of the South Ural State University. Series: Mathematiсal Modelling, Programming & Computer Software. 2012. Vol. 40(299), no. 14. P. 59–72. (in Russian) URL: https://mmp.susu.ru/article/en/179 (accessed: 15.12.2021).
