scientific article; zbMATH DE number 4132496

From MaRDI portal
Publication:3034008





Elaborating on Itô's idea at the 5th Berkeley Symposium, the Feynman integral (or generalized measure) is viewed as an element of a certain separating set \(\Gamma\) of linear continuous functionals on a \(\sigma\)- bounded Mackey topological algebra, A. The latter consists of functions on \(C[0,T],\) and is the Fourier image of a Banach space of complex Radon measures of bounded variation on the weak Borel \(\sigma\)-algebra of \(C[0,T]',\) having a certain uniform decrease property outside bounded sets. A sequence \(e_ n\in A,\) which is fundamental in the ``secondary topology determined by \(\Gamma\) on A, such that \(<\gamma,e_ na>\to <\gamma,a>\) for all \(a\in A,\) is called an approximate unit of A w.r.t. \(\gamma \in \Gamma.\) A nontrivial system of subsets of \(C[0,T],\) to which the Feynman generalized measure can be restricted (approximately total sets) is constructed. Such subsets appear as complements of the zero sets of approximate units. The results are of relevance when judging whether a certain finite-dimensional approximation affects or not the limit value of a Feynman integral, e.g. explain the observation that errors of order \(| x_{t+\tau}-x_ t| /\tau^{1/2-\epsilon}\) are irrelevant.












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 Q3034008)