Semiparametric modeling: correcting low-dimensional model error in parametric models

DOI10.1016/J.JCP.2015.12.043zbMATH Open1351.62175arXiv1502.07766OpenAlexW1554866327MaRDI QIDQ729465FDOQ729465

Authors: T. Berry, John Harlim 

Publication date: 20 December 2016

Published in: Journal of Computational Physics (Search for Journal in Brave)

Abstract: Semiparametric forecasting and filtering are introduced as a method of addressing model errors arising from unresolved physical phenomena. While traditional parametric models are able to learn high-dimensional systems from small data sets, their rigid parametric structure makes them vulnerable to model error. On the other hand, nonparametric models have a very flexible structure, but they suffer from the curse-of-dimensionality and are not practical for high-dimensional systems. The semiparametric approach loosens the structure of a parametric model by fitting a data-driven nonparametric model for the parameters. Given a parametric dynamical model and a noisy data set of historical observations, an adaptive Kalman filter is used to extract a time-series of the parameter values. A nonparametric forecasting model for the parameters is built by projecting the discrete shift map onto a data-driven basis of smooth functions. Existing techniques for filtering and forecasting algorithms extend naturally to the semiparametric model which can effectively compensate for model error, with forecasting skill approaching that of the perfect model. Semiparametric forecasting and filtering are a generalization of statistical semiparametric models to time-dependent distributions evolving under dynamical systems.

Full work available at URL: https://arxiv.org/abs/1502.07766

Recommendations

• On parameter estimation for semi-linear errors-in-variables models
• Estimation in a semiparametric partially linear errors-in-variables model
• Dimension reduction in a semiparametric regression model with errors in covariates
• Semiparametric errors-in-variables models. A Bayesian approach
• scientific article; zbMATH DE number 4054800
• Approximate Standard Errors in Semiparametric Models

zbMATH Keywords

Kalman filterdiffusion mapsnonparametric modelingmodel errorsemiparametric modelingdiffusion forecast

Mathematics Subject Classification ID

Bayesian inference (62F15) Nonparametric estimation (62G05) Inference from stochastic processes and prediction (62M20)

Cites Work

• Diffusion maps
• Variable kernel density estimation
• Nonparametric and semiparametric models.
• Deterministic Nonperiodic Flow
• Embedology
• Heterogeneous multiscale methods: a review
• Title not available (Why is that?)
• Bayesian inference for nonlinear multivariate diffusion models observed with error
• Time-scale separation from diffusion-mapped delay coordinates
• Physics constrained nonlinear regression models for time series
• Linear theory for filtering nonlinear multiscale systems with model error
• Variable bandwidth diffusion kernels
• Principle of Averaging for Parabolic and Elliptic Differential Equations and for Markov Processes with Small Diffusion
• Delay embeddings for forced systems. I: Deterministic forcing
• Nonparametric Uncertainty Quantification for Stochastic Gradient Flows
• An ensemble Kalman filter for statistical estimation of physics constrained nonlinear regression models
• New perspectives on superparameterization for geophysical turbulence
• Adaptive error covariances estimation methods for ensemble Kalman filters
• Approaches to adaptive filtering
• Estimating sudden changes of biases in linear dynamic systems
• Multiscale couplings in prototype hybrid deterministic/stochastic systems. II: stochastic closures
• Multiscale couplings in prototype hybrid deterministic/stochastic systems. I: Deterministic closures
• Nonlinear Laplacian spectral analysis for time series with intermittency and low-frequency variability

Cited In (5)

• Semi-parametric adjustment to computer models
• Diffusion forecasting model with basis functions from QR-decomposition
• Forecasting turbulent modes with nonparametric diffusion models: learning from noisy data
• Modeling of missing dynamical systems: deriving parametric models using a nonparametric framework
• Machine learning for prediction with missing dynamics

Uses Software

• CRCP

This page was built for publication: Semiparametric modeling: correcting low-dimensional model error in parametric models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q729465)