Symbolic-numeric algorithm for parameter estimation in discrete-time models with $\exp$
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866916423976091648 |
|---|---|
| author | Berman, Yosef Forrest, Joshua Grote, Matthew Ovchinnikov, Alexey Rueda, Sonia |
| author_facet | Berman, Yosef Forrest, Joshua Grote, Matthew Ovchinnikov, Alexey Rueda, Sonia |
| contents | Dynamic models describe phenomena across scientific disciplines, yet to make these models useful in application the unknown parameter values of the models must be determined. Discrete-time dynamic models are widely used to model biological processes, but it is often difficult to determine these parameters. In this paper, we propose a symbolic-numeric approach for parameter estimation in discrete-time models that involve univariate non-algebraic (locally) analytic functions such as exp. We illustrate the performance (precision) of our approach by applying our approach to two archetypal discrete-time models in biology (the flour beetle 'LPA' model and discrete Lotka-Volterra competition model). Unlike optimization-based methods, our algorithm guarantees to find all solutions of the parameter values up to a specified precision given time-series data for the measured variables provided that there are finitely many parameter values that fit the data and that the used polynomial system solver can find all roots of the associated polynomial system with interval coefficients. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_16220 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Symbolic-numeric algorithm for parameter estimation in discrete-time models with $\exp$ Berman, Yosef Forrest, Joshua Grote, Matthew Ovchinnikov, Alexey Rueda, Sonia Quantitative Methods Symbolic Computation Systems and Control Commutative Algebra Dynamical Systems 92B05, 68W30, 14Q20, 39A60, 13P15 Dynamic models describe phenomena across scientific disciplines, yet to make these models useful in application the unknown parameter values of the models must be determined. Discrete-time dynamic models are widely used to model biological processes, but it is often difficult to determine these parameters. In this paper, we propose a symbolic-numeric approach for parameter estimation in discrete-time models that involve univariate non-algebraic (locally) analytic functions such as exp. We illustrate the performance (precision) of our approach by applying our approach to two archetypal discrete-time models in biology (the flour beetle 'LPA' model and discrete Lotka-Volterra competition model). Unlike optimization-based methods, our algorithm guarantees to find all solutions of the parameter values up to a specified precision given time-series data for the measured variables provided that there are finitely many parameter values that fit the data and that the used polynomial system solver can find all roots of the associated polynomial system with interval coefficients. |
| title | Symbolic-numeric algorithm for parameter estimation in discrete-time models with $\exp$ |
| topic | Quantitative Methods Symbolic Computation Systems and Control Commutative Algebra Dynamical Systems 92B05, 68W30, 14Q20, 39A60, 13P15 |
| url | https://arxiv.org/abs/2401.16220 |