000456491 001__ 456491 000456491 005__ 20211110162956.0 000456491 0247_ $$2INSPIRETeX$$aGuidetti:2021xvb 000456491 0247_ $$2inspire$$ainspire:1851894 000456491 0247_ $$2arXiv$$aarXiv:2103.08662 000456491 0247_ $$2datacite_doi$$a10.3204/PUBDB-2021-01487 000456491 037__ $$aPUBDB-2021-01487 000456491 041__ $$aEnglish 000456491 088__ $$2arXiv$$aarXiv:2103.08662 000456491 088__ $$2DESY$$aDESY-21-037 000456491 1001_ $$0P:(DE-H253)PIP1087239$$aGuidetti, V.$$b0$$udesy 000456491 245__ $$adNNsolve: an efficient NN-based PDE solver 000456491 260__ $$c2021 000456491 3367_ $$0PUB:(DE-HGF)25$$2PUB:(DE-HGF)$$aPreprint$$bpreprint$$mpreprint$$s1616511905_21199 000456491 3367_ $$2ORCID$$aWORKING_PAPER 000456491 3367_ $$028$$2EndNote$$aElectronic Article 000456491 3367_ $$2DRIVER$$apreprint 000456491 3367_ $$2BibTeX$$aARTICLE 000456491 3367_ $$2DataCite$$aOutput Types/Working Paper 000456491 500__ $$a16 pages + 9 pages of appendices, 7 figures, LaTeX, code to be released soon 000456491 520__ $$aNeural Networks (NNs) can be used to solve Ordinary and Partial Differential Equations (ODEs and PDEs) by redefining the question as an optimization problem. The objective function to be optimized is the sum of the squares of the PDE to be solved and of the initial/boundary conditions. A feed forward NN is trained to minimise this loss function evaluated on a set of collocation points sampled from the domain where the problem is defined. A compact and smooth solution, that only depends on the weights of the trained NN, is then obtained. This approach is often referred to as PINN, from Physics Informed Neural Network~\cite{raissi2017physics_1, raissi2017physics_2}. Despite the success of the PINN approach in solving various classes of PDEs, an implementation of this idea that is capable of solving a large class of ODEs and PDEs with good accuracy and without the need to finely tune the hyperparameters of the network, is not available yet. In this paper, we introduce a new implementation of this concept - called dNNsolve - that makes use of dual Neural Networks to solve ODEs/PDEs. These include: i) sine and sigmoidal activation functions, that provide a more efficient basis to capture both secular and periodic patterns in the solutions; ii) a newly designed architecture, that makes it easy for the the NN to approximate the solution using the basis functions mentioned above. We show that dNNsolve is capable of solving a broad range of ODEs/PDEs in 1, 2 and 3 spacetime dimensions, without the need of hyperparameter fine-tuning. 000456491 536__ $$0G:(DE-HGF)POF4-611$$a611 - Fundamental Particles and Forces (POF4-611)$$cPOF4-611$$fPOF IV$$x0 000456491 536__ $$0G:(EU-Grant)647995$$aSTRINGFLATION - Inflation in String Theory - Connecting Quantum Gravity with Observations (647995)$$c647995$$fERC-2014-CoG$$x1 000456491 588__ $$aDataset connected to INSPIRE 000456491 693__ $$0EXP:(DE-MLZ)NOSPEC-20140101$$5EXP:(DE-MLZ)NOSPEC-20140101$$eNo specific instrument$$x0 000456491 7001_ $$0P:(DE-H253)PIP1091885$$aMuia, F.$$b1$$udesy 000456491 7001_ $$0P:(DE-H253)PIP1086156$$aWelling, Yvette Maria$$b2$$eCorresponding author 000456491 7001_ $$0P:(DE-H253)PIP1013212$$aWestphal, Alexander$$b3 000456491 773__ $$p1-25 000456491 8564_ $$uhttps://bib-pubdb1.desy.de/record/456491/files/2103.08662v1.pdf$$yOpenAccess 000456491 8564_ $$uhttps://bib-pubdb1.desy.de/record/456491/files/2103.08662v1.pdf?subformat=pdfa$$xpdfa$$yOpenAccess 000456491 909CO $$ooai:bib-pubdb1.desy.de:456491$$pdnbdelivery$$pec_fundedresources$$pVDB$$pdriver$$popen_access$$popenaire 000456491 9101_ $$0I:(DE-588b)2008985-5$$6P:(DE-H253)PIP1087239$$aDeutsches Elektronen-Synchrotron$$b0$$kDESY 000456491 9101_ $$0I:(DE-588b)2008985-5$$6P:(DE-H253)PIP1091885$$aDeutsches Elektronen-Synchrotron$$b1$$kDESY 000456491 9101_ $$0I:(DE-588b)2008985-5$$6P:(DE-H253)PIP1086156$$aDeutsches Elektronen-Synchrotron$$b2$$kDESY 000456491 9101_ $$0I:(DE-588b)2008985-5$$6P:(DE-H253)PIP1013212$$aDeutsches Elektronen-Synchrotron$$b3$$kDESY 000456491 9130_ $$0G:(DE-HGF)POF3-611$$1G:(DE-HGF)POF3-610$$2G:(DE-HGF)POF3-600$$3G:(DE-HGF)POF3$$4G:(DE-HGF)POF$$aDE-HGF$$bForschungsbereich Materie$$lMaterie und Universum$$vFundamental Particles and Forces$$x0 000456491 9131_ $$0G:(DE-HGF)POF4-611$$1G:(DE-HGF)POF4-610$$2G:(DE-HGF)POF4-600$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$aDE-HGF$$bForschungsbereich Materie$$lMatter and the Universe$$vFundamental Particles and Forces$$x0 000456491 9141_ $$y2021 000456491 915__ $$0StatID:(DE-HGF)0510$$2StatID$$aOpenAccess 000456491 915__ $$0LIC:(DE-HGF)CCBY4$$2HGFVOC$$aCreative Commons Attribution CC BY 4.0 000456491 9201_ $$0I:(DE-H253)T-20120731$$kT$$lTheorie-Gruppe$$x0 000456491 980__ $$apreprint 000456491 980__ $$aVDB 000456491 980__ $$aUNRESTRICTED 000456491 980__ $$aI:(DE-H253)T-20120731 000456491 9801_ $$aFullTexts