{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,14]],"date-time":"2026-07-14T17:31:58Z","timestamp":1784050318585,"version":"3.55.0"},"reference-count":20,"publisher":"MDPI AG","issue":"2","license":[{"start":{"date-parts":[[2020,1,27]],"date-time":"2020-01-27T00:00:00Z","timestamp":1580083200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100009318","name":"Helmholtz Association","doi-asserted-by":"publisher","award":["ZT-I-0010"],"award-info":[{"award-number":["ZT-I-0010"]}],"id":[{"id":"10.13039\/501100009318","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Entropy"],"abstract":"<jats:p>Specialized Gaussian process regression is presented for data that are known to fulfill a given linear differential equation with vanishing or localized sources. The method allows estimation of system parameters as well as strength and location of point sources. It is applicable to a wide range of data from measurement and simulation. The underlying principle is the well-known invariance of the Gaussian probability distribution under linear operators, in particular differentiation. In contrast to approaches with a generic covariance function\/kernel, we restrict the Gaussian process to generate only solutions of the homogeneous part of the differential equation. This requires specialized kernels with a direct correspondence of certain kernel hyperparameters to parameters in the underlying equation and leads to more reliable regression results with less training data. Inhomogeneous contributions from linear superposition of point sources are treated via a linear model over fundamental solutions. Maximum likelihood estimates for hyperparameters and source positions are obtained by nonlinear optimization. For differential equations representing laws of physics the present approach generates only physically possible solutions, and estimated hyperparameters represent physical properties. After a general derivation, modeling of source-free data and parameter estimation is demonstrated for Laplace\u2019s equation and the heat\/diffusion equation. Finally, the Helmholtz equation with point sources is treated, representing scalar wave data such as acoustic pressure in the frequency domain.<\/jats:p>","DOI":"10.3390\/e22020152","type":"journal-article","created":{"date-parts":[[2020,1,27]],"date-time":"2020-01-27T11:41:57Z","timestamp":1580125317000},"page":"152","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":17,"title":["Gaussian Process Regression for Data Fulfilling Linear Differential Equations with Localized Sources"],"prefix":"10.3390","volume":"22","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4773-416X","authenticated-orcid":false,"given":"Christopher G.","family":"Albert","sequence":"first","affiliation":[{"name":"Max-Planck-Institut f\u00fcr Plasmaphysik, Boltzmannstr. 2, 85748 Garching, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Katharina","family":"Rath","sequence":"additional","affiliation":[{"name":"Max-Planck-Institut f\u00fcr Plasmaphysik, Boltzmannstr. 2, 85748 Garching, Germany"},{"name":"Department of Statistics, Ludwig-Maximilians-Universit\u00e4t M\u00fcnchen, Ludwigstr. 33, 80539 Munich, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2020,1,27]]},"reference":[{"key":"ref_1","unstructured":"Albert, C.G. (July, January 30). Gaussian Processes for Data Fulfilling Linear Differential Equations. Proceedings of the 39th International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering, Munich, Germany."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"237","DOI":"10.1007\/978-94-015-6844-9_17","article-title":"Kriging Variables that Satisfy the Partial Differential Equation \u0394Z = Y","volume":"4","author":"Dong","year":"1989","journal-title":"Geostatistics"},{"key":"ref_3","unstructured":"van den Boogaart, K.G. (2001, January 10\u201312). Kriging for Processes Solving Partial Differential Equations. Proceedings of the IAMG2001, Cancun, Mexiko."},{"key":"ref_4","unstructured":"Graepel, T. (2003, January 21\u201324). Solving Noisy Linear Operator Equations by Gaussian Processes: Application to Ordinary and Partial Differential Equations. Proceedings of the Twentieth International Conference on Machine Learning, Washington, DC, USA."},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"S\u00e4rkk\u00e4, S. (2011, January 14\u201317). Linear Operators and Stochastic Partial Differential Equations in Gaussian Process Regression. Proceedings of the 21st International Conference on Artificial Neural Networks, Espoo, Finland.","DOI":"10.1007\/978-3-642-21738-8_20"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"736","DOI":"10.1016\/j.jcp.2017.01.060","article-title":"Inferring Solutions of Differential Equations Using Noisy Multi-Fidelity Data","volume":"335","author":"Raissi","year":"2017","journal-title":"J. Comput. Phys."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"683","DOI":"10.1016\/j.jcp.2017.07.050","article-title":"Machine Learning of Linear Differential Equations Using Gaussian Processes","volume":"348","author":"Raissi","year":"2017","journal-title":"J. Comput. Phys."},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Yang, X., Tartakovsky, G., and Tartakovsky, A. (2018). Physics-Informed Kriging: A Physics-Informed Gaussian Process Regression Method for Data-Model Convergence. arXiv.","DOI":"10.1016\/j.jcp.2019.06.041"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"72","DOI":"10.1063\/1.3703622","article-title":"Bayesian Inference in the Numerical Solution of Laplace\u2019s Equation","volume":"1443","author":"Mendes","year":"2012","journal-title":"AIP Conf. Proc."},{"key":"ref_10","unstructured":"Cockayne, J., Oates, C., Sullivan, T., and Girolami, M. (2016). Probabilistic Numerical Methods for Partial Differential Equations and Bayesian Inverse Problems. arXiv."},{"key":"ref_11","unstructured":"Albert, C. (2019, January 9\u201313). Physics-Informed Transfer Path Analysis With Parameter Estimation Using Gaussian Processes. Proceedings of the 23rd International Congress on Acoustics, Aachen, Germany."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"33","DOI":"10.1016\/0010-4655(76)90008-4","article-title":"Computation of Ideal MHD Equilibria","volume":"12","author":"Lackner","year":"1976","journal-title":"Comput. Phys. Commun."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"205","DOI":"10.1016\/0955-7997(95)00062-3","article-title":"The Method of Fundamental Solutions for Poisson\u2019s Equation","volume":"16","author":"Golberg","year":"1995","journal-title":"Eng. Anal. Bound. Elem."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"543","DOI":"10.1017\/S0962492906270016","article-title":"Kernel Techniques: From Machine Learning to Meshless Methods","volume":"15","author":"Schaback","year":"2006","journal-title":"Acta Numer."},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Rasmussen, C.E., and Williams, C.K.I. (2006). Gaussian Processes for Machine Learning, MIT Press.","DOI":"10.7551\/mitpress\/3206.001.0001"},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1111\/j.2517-6161.1978.tb01643.x","article-title":"Curve Fitting and Optimal Design for Prediction","volume":"40","year":"1978","journal-title":"J. R. Stat. Soc. Ser. B"},{"key":"ref_17","unstructured":"Hansen, N., Akimoto, Y., and Baudis, P. (2019, December 13). CMA-ES\/Pycma on Github. Available online: https:\/\/doi.org\/10.5281\/zenodo.2559634."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"661","DOI":"10.1090\/S0025-5718-1994-1254147-6","article-title":"Generalized Hermite Interpolation via Matrix-Valued Conditionally Positive Definite Functions","volume":"63","author":"Narcowich","year":"1994","journal-title":"Math. Comput."},{"key":"ref_19","unstructured":"Mac\u00eado, I., and Castro, R. (2019, December 13). Learning Divergence-Free and Curl-Free Vector Fields with Matrix-Valued Kernels. Available online: http:\/\/preprint.impa.br\/FullText\/Macedo__Thu_Oct_21_16_38_10_BRDT_2010\/macedo-MVRBFs.pdf."},{"key":"ref_20","doi-asserted-by":"crossref","unstructured":"Cobb, A.D., Everett, R., Markham, A., and Roberts, S.J. (2018, January 19\u201323). Identifying Sources and Sinks in the Presence of Multiple Agents with Gaussian Process Vector Calculus. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD \u201918), London, UK.","DOI":"10.1145\/3219819.3220065"}],"container-title":["Entropy"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1099-4300\/22\/2\/152\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,13]],"date-time":"2025-10-13T13:20:44Z","timestamp":1760361644000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1099-4300\/22\/2\/152"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2020,1,27]]},"references-count":20,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2020,2]]}},"alternative-id":["e22020152"],"URL":"https:\/\/doi.org\/10.3390\/e22020152","relation":{},"ISSN":["1099-4300"],"issn-type":[{"value":"1099-4300","type":"electronic"}],"subject":[],"published":{"date-parts":[[2020,1,27]]}}}