1995/07/07 by Baowen Li, Li, Baowen, Marko Robnik +1
Physics and Astronomy · #Chaotic Dynamics (nlin.CD) #Condensed Matter (cond-mat) #FOS: Physical sciences #Quantum Physics (quant-ph) #chao-dyn #cond-mat #nlin.CD #quant-ph
paper · pdf · doi:10.48550/arxiv.chao-dyn/9507002
16 pages, file in plain LaTeX, 6 figures upon request Submitted to J. Phys. A: Math. Gen.
arxiv created 1995/07/07 · arxiv updated 2009/11/30
The boundary integral method (BIM) is a formulation of Helmholtz equation in the form of an integral equation suitable for numerical discretization to solve the quantum billiard. This paper is an extensive numerical survey of BIM in a variety of quantum billiards, integrable (circle, rectangle), KAM systems (Robnik billiard) and fully chaotic (ergodic, such as stadium, Sinai billiard and cardioid billiard). On the theoretical side we point out some serious flaws in the derivation of BIM in the literature and show how the final formula (which nevertheless was correct) should be derived in a sound way and we also argue that a simple minded application of BIM in nonconvex geometries presents serious difficulties or even fails. On the numerical side we have analyzed the scaling of the averaged absolute value of the systematic error ΔE of the eigenenergy in units of mean level spacing with the density of discretization (b = number of numerical nodes on the boundary within one de Broglie wavelength), and we find that in all cases the error obeys a power law <|ΔE|> = A b-α, where α (and also A) varies from case to case (it is not universal), and is affected strongly by the existence of exterior chords in nonconvex geometries, whereas the degree of the classical chaos seems to be practically irrelevant. We comment on the semiclassical limit of BIM and make suggestions about a proper formulation with correct semiclassical limit in nonconvex geometries.