Gaussian Processes for Data Fulfilling Linear Differential Equations

@article{Albert2019GaussianPF,
  title={Gaussian Processes for Data Fulfilling Linear Differential Equations},
  author={Christopher G. Albert},
  journal={Proceedings},
  year={2019}
}
  • C. Albert
  • Published 8 September 2019
  • Mathematics, Computer Science
  • Proceedings
A method to reconstruct fields, source strengths and physical parameters based on Gaussian process regression is presented for the case where data are known to fulfill a given linear differential equation with localized sources. The approach is applicable to a wide range of data from physical measurements and numerical simulations. It is based on the well-known invariance of the Gaussian under linear operators, in particular differentiation. Instead of using a generic covariance function to… 

Figures from this paper

Gaussian Process Regression for Data Fulfilling Linear Differential Equations with Localized Sources †

Specialized Gaussian process regression is presented for data that are known to fulfill a given linear differential equation with vanishing or localized sources, and it generates only physically possible solutions, and estimated hyperparameters represent physical properties.

Gaussian Process Priors for Systems of Linear Partial Differential Equations with Constant Coefficients

The Ehrenpreis-Palamodov fundamental principle is applied, which works like a non-linear Fourier transform, to construct GP kernels mirroring standard spectral methods for GPs, and can infer probable solutions of linear PDE systems from any data such as noisy measurements, or pointwise defined initial and boundary conditions.

Physics-Informed Gaussian Process Regression Generalizes Linear PDE Solvers

The results enable the seamless integration of mechanistic models as modular building blocks into probabilistic models by blurring the boundaries between numerical analysis and Bayesian inference.

Bayesian Uncertainty Quantification with Multi-Fidelity Data and Gaussian Processes for Impedance Cardiography of Aortic Dissection

The fully Bayesian uncertainty quantification method is devised in a notation following the tradition of E.T. Jaynes and found that generalization to an arbitrary number of levels of fidelity and parallelisation becomes rather easy.

A connection between probability, physics and neural networks

The central limit theorem suggests that NNs can be constructed to obey a physical law by choosing the activation functions such that they match a particular kernel in the infinite-width limit.

InfPolyn, a Nonparametric Bayesian Characterization for Composition-Dependent Interdiffusion Coefficients

The proposed InfPolyn, Infinite Polynomial, a novel statistical framework to characterize the component-dependent interdiffusion coefficients is proposed, a generalization of the commonly used polynomial fitting method with extended model capacity and flexibility and it is combined with the numerical inversion-based Boltzmann–Matano method for the interDiffusion coefficient estimations.

Expectation-Maximization Algorithm for the Calibration of Complex Simulator Using a Gaussian Process Emulator

Both the variance and bias of the estimates obtained from the proposed EM algorithm are smaller than those from the existing calibration methods.

References

SHOWING 1-10 OF 17 REFERENCES

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.

Kriging for processes solving partial differential equations

Physical laws are often expressed in terms of partial differential equations. However these laws completely determine the observable process only, when sufficiently many boundary conditions are

Inferring solutions of differential equations using noisy multi-fidelity data

Physics-informed transfer path analysis with parameter estimation using Gaussian processes

Gaussian processes regression is applied to augment experimental data of transfer-path analysis (TPA) by known information about the underlying physical properties of the system under investigation.

Linear Operators and Stochastic Partial Differential Equations in Gaussian Process Regression

An extension to Gaussian process (GP) regression models, where the measurements are modeled as linear functionals of the underlying GP and the estimation objective is a general linear operator of the process is discussed.

Probabilistic Numerical Methods for Partial Differential Equations and Bayesian Inverse Problems ∗

A probabilistic numerical method for solution of partial differential equations (PDEs) and application of that method to PDE-constrained inverse problems, which enables the solution of challenging inverse problems whilst accounting for the impact of discretisation error due to numerical solution of the PDE.

Bayesian inference in the numerical solution of Laplace's equation

Inference is not unrelated to numerical analysis: given partial information about a mathematical problem, one has to estimate the unknown "true solution" and uncertainties. Many methods of

Probabilistic Meshless Methods for Partial Differential Equations and Bayesian Inverse Problems

A probabilistic numerical method for solution of partial differential equations (PDEs) and application of that method to PDE-constrained inverse problems, which enables the solution of challenging inverse problems whilst accounting for the impact of discretisation error due to numerical solution of the PDE.

Generalized Hermite interpolation via matrix-valued conditionally positive definite functions

In this paper, we consider a broad class of interpolation problems, for both scalarand vector-valued multivariate functions subject to linear side conditions, such as being divergence-free, where the

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.