No abstract uploaded!

No abstract uploaded!

No abstract uploaded!

Among approaches to defining the Feynman integral, various sequential methods are probably the most popular ones. Here we consider that one based on product formulae, the original idea of which belongs to Nelson71'. This method is powerful in the sense that it enables to prove the strict version of Feynman's basic dynamical formula for a wide class of poten tials^^/including time-dependent ones/4/in comparison with other known methods (see, eg, refs./5, 6/). 0n the other, hand, the method does not correspond exactly to Feynman's heuristic prescription because it replaces the exact action along poly gonal paths by its Riemannian approximation. Moreover, one can regard as an unsatisfactory feature that it employs the" equidistant" time-interval partitions only. In the present paper we show that the last mentioned sour ce of arbitrariness can be removed: the product formula for perturbations of propagators by

We investigate spectral properties of the Laplacian in <i>L</i>² (<i>Q</i>), where <i>Q</i> is a tubular region in ³ of a fixed cross section, and the boundary conditions combined a Dirichlet and a Neumann part. We analyze two complementary situations, when the tube is bent but not twisted, and secondly, it is twisted but not bent. In the first case we derive sufficient conditions for the presence and absence of the discrete spectrum showing, roughly speaking, that they depend on the direction in which the tube is bent. In the second case we show that a constant twist raises the threshold of the essential spectrum and a local slowndown of it gives rise to isolated eigenvalues. Furthermore, we prove that the spectral threshold moves up also under a sufficiently gentle periodic twist.