{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,21]],"date-time":"2026-03-21T09:18:28Z","timestamp":1774084708610,"version":"3.50.1"},"reference-count":47,"publisher":"Springer Science and Business Media LLC","issue":"36","license":[{"start":{"date-parts":[[2024,9,25]],"date-time":"2024-09-25T00:00:00Z","timestamp":1727222400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2024,9,25]],"date-time":"2024-09-25T00:00:00Z","timestamp":1727222400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/501100003451","name":"Universidad del Pa\u00eds Vasco","doi-asserted-by":"crossref","id":[{"id":"10.13039\/501100003451","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Neural Comput &amp; Applic"],"published-print":{"date-parts":[[2024,12]]},"abstract":"<jats:title>Abstract<\/jats:title><jats:p>Generalization is a key property of machine learning models to perform accurately on unseen data. Conversely, in the field of scientific machine learning (SciML), generalization entails not only predictive accuracy but also the capacity of the model to encapsulate underlying physical principles. In this paper, we delve into the concept of generalization for Physics-informed neural networks (PINNs) by investigating the consistency of the predictions of a PINN outside of its training domain. Through the lenses of a novel metric and statistical analysis, we study the scenarios in which a PINN can provide consistent predictions outside the region considered for training and hereinafter assess whether the algorithmic setup of the model can influence its potential for generalizing. Our results highlight why overparametrization is not a crucial component in SciML while encouraging overfitting on the training data. Despite being counterintuitive, the outcome of our analysis serves as a guideline for training PINNs for engineering applications.<\/jats:p>","DOI":"10.1007\/s00521-024-10178-2","type":"journal-article","created":{"date-parts":[[2024,9,25]],"date-time":"2024-09-25T04:33:39Z","timestamp":1727238819000},"page":"22677-22696","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":17,"title":["On the generalization of PINNs outside the training domain and the hyperparameters influencing it"],"prefix":"10.1007","volume":"36","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-0062-1187","authenticated-orcid":false,"given":"Andrea","family":"Bonfanti","sequence":"first","affiliation":[]},{"given":"Roberto","family":"Santana","sequence":"additional","affiliation":[]},{"given":"Marco","family":"Ellero","sequence":"additional","affiliation":[]},{"given":"Babak","family":"Gholami","sequence":"additional","affiliation":[]}],"member":"297","published-online":{"date-parts":[[2024,9,25]]},"reference":[{"issue":"6","key":"10178_CR1","doi-asserted-by":"publisher","first-page":"413","DOI":"10.1038\/s42254-022-00441-7","volume":"4","author":"J Thiyagalingam","year":"2022","unstructured":"Thiyagalingam J, Shankar M, Fox G, Hey T (2022) Scientific machine learning benchmarks. Nat Rev Phys 4(6):413\u2013420","journal-title":"Nat Rev Phys"},{"key":"10178_CR2","doi-asserted-by":"publisher","DOI":"10.1016\/j.jcp.2020.109913","volume":"425","author":"L Yang","year":"2021","unstructured":"Yang L, Meng X, Karniadakis GE (2021) B-PINNs: Bayesian Physics-informed neural networks for forward and inverse PDE problems with noisy data. J Comput Phys 425:109913","journal-title":"J Comput Phys"},{"key":"10178_CR3","volume":"67","author":"Y Sun","year":"2022","unstructured":"Sun Y, Sengupta U, Juniper M (2022) Physics-informed deep learning for simultaneous surrogate modelling and PDE-constrained optimization. Bull Am Phys Soc 67:116042","journal-title":"Bull Am Phys Soc"},{"key":"10178_CR4","doi-asserted-by":"publisher","DOI":"10.1016\/j.cma.2023.116401","volume":"417","author":"H Jeong","year":"2023","unstructured":"Jeong H, Batuwatta-Gamage C, Bai J, Xie YM, Rathanayaka C, Zhou Y, Gu Y (2023) A complete Physics-Informed Neural Network-based framework for structural topology optimization. Comput Methods Appl Mech Eng 417:116401","journal-title":"Comput Methods Appl Mech Eng"},{"issue":"3","key":"10178_CR5","doi-asserted-by":"publisher","first-page":"195","DOI":"10.1002\/cnm.1640100303","volume":"10","author":"MWMG Dissanayake","year":"1994","unstructured":"Dissanayake MWMG, Phan-Thien N (1994) Neural network-based approximations for solving partial differential equations. Commun Numer Methods Eng 10(3):195\u2013201","journal-title":"Commun Numer Methods Eng"},{"issue":"5","key":"10178_CR6","doi-asserted-by":"publisher","first-page":"987","DOI":"10.1109\/72.712178","volume":"9","author":"IE Lagaris","year":"1998","unstructured":"Lagaris IE, Likas A, Fotiadis DI (1998) Artificial neural networks for solving ordinary and partial differential equations. IEEE Trans Neural Netw 9(5):987\u20131000","journal-title":"IEEE Trans Neural Netw"},{"key":"10178_CR7","doi-asserted-by":"publisher","first-page":"686","DOI":"10.1016\/j.jcp.2018.10.045","volume":"378","author":"M Raissi","year":"2019","unstructured":"Raissi M, Perdikaris P, Karniadakis GE (2019) Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. J Comput Phys 378:686\u2013707","journal-title":"J Comput Phys"},{"key":"10178_CR8","doi-asserted-by":"publisher","DOI":"10.1016\/j.cma.2019.112623","volume":"358","author":"G Kissas","year":"2020","unstructured":"Kissas G, Yang Y, Hwuang E, Witschey WR, Detre JA, Perdikaris P (2020) Machine learning in cardiovascular flows modeling: predicting arterial blood pressure from non-invasive 4D flow MRI data using Physics-informed neural networks. Comput Methods Appl Mech Eng 358:112623","journal-title":"Comput Methods Appl Mech Eng"},{"issue":"1","key":"10178_CR9","doi-asserted-by":"publisher","first-page":"15475","DOI":"10.1038\/s41598-022-19157-w","volume":"12","author":"B Zapf","year":"2022","unstructured":"Zapf B, Haubner J, Kuchta M, Ringstad G, Eide PK, Mardal K (2022) Investigating molecular transport in the human brain from MRI with Physics-informed neural networks. Sci Rep 12(1):15475","journal-title":"Sci Rep"},{"issue":"2194","key":"10178_CR10","doi-asserted-by":"publisher","first-page":"20200093","DOI":"10.1098\/rsta.2020.0093","volume":"379","author":"K Kashinath","year":"2021","unstructured":"Kashinath K et al (2021) Physics-informed machine learning: case studies for weather and climate modelling. Philos Trans R Soc 379(2194):20200093","journal-title":"Philos Trans R Soc"},{"key":"10178_CR11","doi-asserted-by":"publisher","DOI":"10.1016\/j.ress.2022.108900","volume":"230","author":"Y Xu","year":"2022","unstructured":"Xu Y, Kohtz S, Boakye J, Gardoni P, Wang P (2022) Physics-informed machine learning for reliability and systems safety applications: state of the art and challenges. Reliab Eng Syst Saf 230:108900","journal-title":"Reliab Eng Syst Saf"},{"key":"10178_CR12","doi-asserted-by":"publisher","DOI":"10.1016\/j.csite.2023.103265","volume":"49","author":"JS Goud","year":"2023","unstructured":"Goud JS, Srilatha P, Kumar RSV, Sowmya G, Gamaoun F, Nagaraja KV, Chohan JS, Khan U, Eldin SM (2023) Heat transfer analysis in a longitudinal porous trapezoidal fin by non-Fourier heat conduction model: an application of artificial neural network with Levenberg\u2013Marquardt approach. Case Stud Therm Eng 49:103265","journal-title":"Case Stud Therm Eng"},{"issue":"6","key":"10178_CR13","doi-asserted-by":"publisher","first-page":"422","DOI":"10.1038\/s42254-021-00314-5","volume":"3","author":"GE Karniadakis","year":"2021","unstructured":"Karniadakis GE, Kevrekidis IG, Lu L, Perdikaris P, Wang S, Yang L (2021) Physics-informed machine learning. Nat Rev Phys 3(6):422\u2013440","journal-title":"Nat Rev Phys"},{"key":"10178_CR14","doi-asserted-by":"publisher","DOI":"10.1016\/j.asoc.2023.110328","volume":"142","author":"Z Tang","year":"2023","unstructured":"Tang Z, Dong S, Yang X, Zhang J (2023) Application of a parallel physics-informed neural network to solve the multi-body dynamic equations for full-scale train collisions. Appl Soft Comput 142:110328","journal-title":"Appl Soft Comput"},{"key":"10178_CR15","doi-asserted-by":"publisher","DOI":"10.1016\/j.paerosci.2022.100849","volume":"134","author":"J Li","year":"2022","unstructured":"Li J, Du X, Martins JR (2022) Machine learning in aerodynamic shape optimization. Progress Aerosp Sci 134:100849","journal-title":"Progress Aerosp Sci"},{"key":"10178_CR16","doi-asserted-by":"publisher","first-page":"408","DOI":"10.1108\/HFF-04-2023-0173","volume":"34","author":"SI Abdelsalam","year":"2023","unstructured":"Abdelsalam SI, Magesh A, Tamizharasi P, Zaher AZ (2023) Versatile response of a Sutterby nanofluid under activation energy: hyperthermia therapy. Int J Numer Methods Heat Fluid Flow 34:408","journal-title":"Int J Numer Methods Heat Fluid Flow"},{"issue":"5","key":"10178_CR17","doi-asserted-by":"publisher","DOI":"10.1016\/j.heliyon.2023.e15916","volume":"9","author":"SI Abdelsalam","year":"2023","unstructured":"Abdelsalam SI, Alsharif AM, Abd Elmaboud Y, Abdellateef AI (2023) Assorted kerosene-based nanofluid across a dual-zone vertical annulus with electroosmosis. Heliyon 9(5):e15916","journal-title":"Heliyon"},{"key":"10178_CR18","doi-asserted-by":"publisher","DOI":"10.1016\/j.asej.2022.101811","volume":"14","author":"W Weera","year":"2023","unstructured":"Weera W, Kumar RSV, Sowmya G, Khan U, Prasannakumara BC, Mahmoud EE, Yahia IS (2023) Convective-radiative thermal investigation of a porous dovetail fin using spectral collocation method. Ain Shams Eng J 14:101811","journal-title":"Ain Shams Eng J"},{"key":"10178_CR19","doi-asserted-by":"publisher","DOI":"10.1016\/j.csite.2023.102777","volume":"43","author":"A Abdulrahman","year":"2023","unstructured":"Abdulrahman A, Gamaoun F, Kumar RSV, Khan U, Gill HS, Nagaraja KV, Eldin SM, Galal AM (2023) Study of thermal variation in a longitudinal exponential porous fin wetted with TiO2\u2013SiO2\/hexanol hybrid nanofluid using hybrid residual power series method. Case Stud Thermal Eng 43:102777","journal-title":"Case Stud Thermal Eng"},{"key":"10178_CR20","first-page":"1","volume":"347","author":"E Kharazmi","year":"2021","unstructured":"Kharazmi E, Zhang Z, Karniadakis GE (2021) hp-VPINNs: variational Physics-informed neural networks with domain decomposition. Comput Methods Appl Mech Eng 347:1","journal-title":"Comput Methods Appl Mech Eng"},{"key":"10178_CR21","unstructured":"Moseley B, Markham A, Nissen-Meyer T (2021) Finite Basis Physics-Informed Neural Networks (FBPINNs): a scalable domain decomposition approach for solving differential equations. arXiv preprint arXiv:2107.07871"},{"key":"10178_CR22","doi-asserted-by":"crossref","unstructured":"Hu Z, Jagtap AD, Karniadakis GE, Kawaguchi K (2022) Augmented Physics-Informed Neural Networks (APINNs): a gating network-based soft domain decomposition methodology. arXiv preprint arXiv:2211.08939","DOI":"10.1016\/j.engappai.2023.107183"},{"key":"10178_CR23","unstructured":"Bishop R, Kraus MA (2022) Mixture-of-experts-ensemble meta-learning for physics-informed neural networks. In: Proceedings of 33. Forum Bauinformatik"},{"issue":"2","key":"10178_CR24","doi-asserted-by":"publisher","first-page":"981","DOI":"10.1093\/imanum\/drab032","volume":"42","author":"S Mishra","year":"2022","unstructured":"Mishra S, Molinaro R (2022) Estimates on the generalization error of Physics-informed neural networks for approximating a class of inverse problems for PDEs. IMA J Numer Anal 42(2):981\u20131022","journal-title":"IMA J Numer Anal"},{"key":"10178_CR25","doi-asserted-by":"publisher","first-page":"428","DOI":"10.1016\/j.neunet.2020.08.022","volume":"132","author":"MS Advani","year":"2020","unstructured":"Advani MS, Saxe AM, Sompolinsky H (2020) High-dimensional dynamics of generalization error in neural networks. Neural Netw 132:428\u2013446","journal-title":"Neural Netw"},{"key":"10178_CR26","unstructured":"Olson M, Wyner A, Berk R (2018) Modern neural networks generalize on small data sets. In: Advances in neural information processing systems, vol 31"},{"issue":"4","key":"10178_CR27","doi-asserted-by":"publisher","first-page":"667","DOI":"10.1002\/cpa.22008","volume":"75","author":"S Mei","year":"2022","unstructured":"Mei S, Montanari A (2022) The generalization error of random features regression: precise asymptotics and the double descent curve. Commun Pure Appl Math 75(4):667\u2013766","journal-title":"Commun Pure Appl Math"},{"key":"10178_CR28","unstructured":"Wang Y, Han X, Chang C, Zha D, Braga-Neto U, Hu X (2022) Auto-PINN: understanding and optimizing Physics-informed neural architecture. arXiv:2205.13748"},{"key":"10178_CR29","doi-asserted-by":"crossref","unstructured":"Chaudhari M, Kulkarni I, Damodaran M (2021) Exploring Physics-informed neural networks for compressible flow prediction. In: Proceedings of 16th Asian congress of fluid mechanics","DOI":"10.1007\/978-981-15-5183-3_34"},{"key":"10178_CR30","doi-asserted-by":"publisher","DOI":"10.1016\/j.cma.2022.115671","volume":"403","author":"C Wu","year":"2023","unstructured":"Wu C, Zhu M, Tan Q, Kartha Y, Lu L (2023) A comprehensive study of non-adaptive and residual-based adaptive sampling for Physics-informed neural networks. Comput Methods Appl Mech Eng 403:115671","journal-title":"Comput Methods Appl Mech Eng"},{"key":"10178_CR31","unstructured":"Shama R, Shankar V (2022) Accelerated training of Physics-informed neural networks (PINNs) using meshless discretizations. In: Advances in neural information processing systems, vol 35, pp 1034\u20131046"},{"issue":"1","key":"10178_CR32","doi-asserted-by":"publisher","first-page":"208","DOI":"10.1137\/19M1274067","volume":"63","author":"L Lu","year":"2021","unstructured":"Lu L, Meng X, Mao Z, Karniadakis GE (2021) DeepXDE: a deep learning library for solving differential equations. SIAM Rev 63(1):208","journal-title":"SIAM Rev"},{"key":"10178_CR33","unstructured":"Kingma DP, Ba J (2014) Adam: a method for stochastic optimization. https:\/\/arxiv.org\/abs\/1412.6980"},{"issue":"1","key":"10178_CR34","doi-asserted-by":"publisher","first-page":"503","DOI":"10.1007\/BF01589116","volume":"45","author":"DC Liu","year":"1989","unstructured":"Liu DC, Nocedal J (1989) On the limited memory BFGS method for large scale optimization. Math Program 45(1):503\u2013528","journal-title":"Math Program"},{"key":"10178_CR35","doi-asserted-by":"crossref","unstructured":"De Ryck T, Jagtap AD, Mishra S (2022) Error estimates for Physics-informed neural networks approximating the Navier\u2013Stokes equations. arXiv preprint arXiv:2203.09346","DOI":"10.1007\/s10444-022-09985-9"},{"key":"10178_CR36","first-page":"1","volume":"1","author":"PE McKight","year":"2010","unstructured":"McKight PE, Najab J (2010) Kruskal-Wallis test. Corsini Encycl Psychol 1:1\u201310","journal-title":"Corsini Encycl Psychol"},{"key":"10178_CR37","unstructured":"Bonfanti A, Bruno G, Cipriani C (2024) The challenges of the nonlinear regime for Physics informed neural networks. arXiv preprint arXiv:2402.03864"},{"key":"10178_CR38","volume-title":"NIST handbook of mathematical functions","author":"WP Reinhardt","year":"2010","unstructured":"Reinhardt WP, Walker PL (2010) Jacobian elliptic functions. In: Olver FWJ, Lozier DM, Boisvert RF, Clark CW (eds) NIST handbook of mathematical functions, Chap 22. Cambridge University Press, Cambridge, UK","edition":"22"},{"key":"10178_CR39","first-page":"23","volume":"1050","author":"G Valle-Perez","year":"2018","unstructured":"Valle-Perez G, Camargo CQ, Louis AA (2018) Deep learning generalizes because the parameter-function map is biased towards simple functions. stat 1050:23","journal-title":"stat"},{"key":"10178_CR40","unstructured":"Jagtap A, Karniadakis GE (2021) Extended Physics-informed neural networks (XPINNs): a generalized space-time domain decomposition based deep learning framework for nonlinear partial differential equations. In: AAAI spring symposium: MLPS"},{"key":"10178_CR41","unstructured":"Wang S, Sankaran S, Perdikaris P (2022) Respecting causality is all you need for training Physics-informed neural networks. https:\/\/arxiv.org\/abs\/2203.07404"},{"key":"10178_CR42","doi-asserted-by":"publisher","DOI":"10.1016\/j.jcp.2021.110768","volume":"449","author":"S Wang","year":"2020","unstructured":"Wang S, Yu X, Perdikaris P (2020) When and why PINNs fail to train: a neural tangent kernel perspective. J Comput Phys 449:110768","journal-title":"J Comput Phys"},{"issue":"3","key":"10178_CR43","doi-asserted-by":"publisher","first-page":"1811","DOI":"10.1137\/20M1344883","volume":"59","author":"S Mishra","year":"2021","unstructured":"Mishra S, Rusch TK (2021) Enhancing accuracy of deep learning algorithms by training with low-discrepancy sequences. SIAM J Numer Anal 59(3):1811\u20131834","journal-title":"SIAM J Numer Anal"},{"key":"10178_CR44","unstructured":"Krishnapriyan AS, Gholami A, Zhe S, Kirby RM, Mahoney MW (2021) Characterizing possible failure modes in Physics-informed neural networks. In: Advances in neural information processing systems, vol 34, pp 26548\u201326560"},{"key":"10178_CR45","unstructured":"Katsiaryna H, Alexander I (2023) Improved training of Physics-informed neural networks with model ensembles. arXiv preprint arXiv:2204.05108"},{"key":"10178_CR46","unstructured":"Rohrhofer FM, Posch S, G\u00f6\u00dfnitzer C, Geiger BC (2022) On the role of fixed points of dynamical systems in training physics-informed neural networks. arXiv preprint arXiv:2203.13648"},{"key":"10178_CR47","doi-asserted-by":"publisher","DOI":"10.1016\/j.cma.2021.114424","volume":"389","author":"H Wang","year":"2022","unstructured":"Wang H, Planas R, Chandramowlishwaran A, Bostanabad R (2022) Mosaic flows: a transferable deep learning framework for solving PDEs on unseen domains. Comput Methods Appl Mech Eng 389:114424","journal-title":"Comput Methods Appl Mech Eng"}],"container-title":["Neural Computing and Applications"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s00521-024-10178-2.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s00521-024-10178-2\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s00521-024-10178-2.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,11,26]],"date-time":"2024-11-26T20:04:01Z","timestamp":1732651441000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s00521-024-10178-2"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,9,25]]},"references-count":47,"journal-issue":{"issue":"36","published-print":{"date-parts":[[2024,12]]}},"alternative-id":["10178"],"URL":"https:\/\/doi.org\/10.1007\/s00521-024-10178-2","relation":{},"ISSN":["0941-0643","1433-3058"],"issn-type":[{"value":"0941-0643","type":"print"},{"value":"1433-3058","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,9,25]]},"assertion":[{"value":"24 August 2023","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"1 July 2024","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"25 September 2024","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors declare that there is no conflict of interest regarding the publication of this paper.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}}]}}