• Corpus ID: 15175048

On solving Ordinary Differential Equations using Gaussian Processes

  title={On solving Ordinary Differential Equations using Gaussian Processes},
  author={David Barber},
  • D. Barber
  • Published 17 August 2014
  • Computer Science
  • ArXiv
We describe a set of Gaussian Process based approaches that can be used to solve non-linear Ordinary Differential Equations. We suggest an explicit probabilistic solver and two implicit methods, one analogous to Picard iteration and the other to gradient matching. All methods have greater accuracy than previously suggested Gaussian Process approaches. We also suggest a general approach that can yield error estimates from any standard ODE solver. 

Figures from this paper

Inferring Non-linear State Dynamics using Gaussian Processes

A novel method that uses Gaussian processes to infer the dynamics of states governed by non-linear ordinary differential equations (ODE’s) and provides a proxy to the intractable posterior distribution over states using mean-field approximation, thus simplifying previous approaches.

Probabilistic solvers for partial differential equations

A probabilistic solver suitable for linear partial differential equations (PDE) with mixed (Dirichlet and Neumann) boundary conditions defined on arbitrary geometries is developed.

Bayesian Smooth-and-Match strategy for ordinary differential equations models that are linear in the parameters

This work focuses on the class of techniques that use smoothing to avoid direct integration and, in particular, on a Bayesian Smooth-and-Match strategy that allows to obtain the ODEs' solution while performing inference on models that are linear in the parameters.

Bayesian smooth‐and‐match inference for ordinary differential equations models linear in the parameters

This work develops a Bayesian smooth‐and‐match strategy that approximates the ODE solution while performing Bayesian inference on the model parameters and incorporates in the strategy two main sources of uncertainty: the noise level of the measured observations and the model approximation error.

A probabilistic model for the numerical solution of initial value problems

A new view of probabilistic ODE solvers as active inference agents operating on stochastic differential equation models that estimate the unknown initial value problem (IVP) solution from approximate observations of the solution derivative, as provided by the ODE dynamics is provided.



Gaussian Processes for Bayesian Estimation in Ordinary Differential Equations

A Gaussian process model is proposed that directly links state derivative information with system observations, simplifying previous approaches and improving estimation accuracy in coupled ordinary differential equations.

Solving Noisy Linear Operator Equations by Gaussian Processes: Application to Ordinary and Partial Differential Equations

The problem of solving stochastic linear operator equations in a Bayesian Ganssian process (GP) framework is formulated in the spirit of a collocation method based on noisy evaluations of the target function at randomly drawn or deliberately chosen points.

Accelerating Bayesian Inference over Nonlinear Differential Equations with Gaussian Processes

This work presents an accelerated sampling procedure which enables Bayesian inference of parameters in nonlinear ordinary and delay differential equations via the novel use of Gaussian processes (GP).

Probabilistic Solutions to Differential Equations and their Application to Riemannian Statistics

This work studies a probabilistic numerical method for the solution of both boundary and initial value problems that returns a joint Gaussian process posterior over the solution that permits marginalising the uncertainty of the numerical solution such that statistics are less sensitive to inaccuracies.

Bayesian Solution of Ordinary Differential Equations

In the numerical solution of ordinary differential equations, a function y(x) is to be reconstructed from knowledge of the functional form of its derivative: dy/dx = f (x, y), together with an

ODE parameter inference using adaptive gradient matching with Gaussian processes

The present paper discusses a method based on nonparametric Bayesian statistics with Gaussian processes due to Calderhead et al. (2008), and shows how inference in this model can be substantially improved by consistently inferring all parameters from the joint dis- tribution.

Bayesian Uncertainty Quantification for Differential Equations

A fully Bayesian inferential framework to quantify uncertainty in models defined by general systems of analytically intractable differential equations, and probabilistically characterises the solution uncertainty introduced when models are chaotic, ill-conditioned, or contain unmodelled functional uncertainty.

Exponential integrators

The main intention in this article is to present the mathematics behind these methods, and derive error bounds that are independent of stiffness or highest frequencies in the system.

Gaussian Processes for Machine Learning

The treatment is comprehensive and self-contained, targeted at researchers and students in machine learning and applied statistics, and deals with the supervised learning problem for both regression and classification.

An Introduction to Ordinary Differential Equations