1) Rayleigh-Ritz variational characterization of Laplace eigenvalues. See Friday's notes dec3.pdf You can also characterize the nth eigenvalue as the minimum over all collections of n-dimensional subspaces of C1 functions having zero boundary data of the maximum Rayleigh quotient for such subspaces. (mini-max characterization) Notice this implies that larger domains have lower natural frequencies. An interesting question is the extent to which you can "hear the shape of a drum"... http://en.wikipedia.org/wiki/Hearing_the_shape_of_a_drum 2) The example of a square or rectangle, also on Friday's notes. Notice we can deduce we have a complete eigenbasis because we prove the closure of the span of our collection includes all differentiable functions with zero boundary data...any missing eigenspace would have a unit eigenfunction perpendicular to this closed space, and so wouldn't be approximatable in the L2 norm. Play with square eigenvalues and eigenfunctions via the wave equation here: http://www.falstad.com/membrane   http://www.falstad.com/circosc disk domain. http://www.falstad.com/mathphysics.html more applets 3) Separating variables for heat and wave equations in radial and spherical coordinates, in R2, R3, Rn. This involves the spherical harmonics in the non-radial directions: http://en.wikipedia.org/wiki/Spherical_harmonics http://en.wikipedia.org/wiki/Spherical_harmonics#Higher_dimensions Some of the details are explained further in these notes: dec6.pdf And it involves Bessels differential equation in the radial direction: http://en.wikipedia.org/wiki/Bessel_function |