scientific article; zbMATH DE number 1405301

From MaRDI portal
Publication:4938264





The authors discuss the possibility of using artificial neural networks for the solution of differential equations. They suggest that the approximating functions should be of the form \(A(x)+ F(x, N(x,p))\), where \(A\) satisfies the boundary conditions and \(N\) is a single-output feedforward neural network with parameters \(p\) and \(n\) input units fed with the input vector \(x\). A minimization process is performed to get the best approximation. A similar process can be performed for eigenvalue problems. The ideas are illustrated by a number of examples including the non-local Schrödinger equations and the Henon-Heiles problem. It is suggested that the method can be implemented by digital signal processing. For the entire collection see [Zbl 0927.00051].











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4938264)