Scattering phase asymptotics with fractal remainders (Q380327)

The authors study the distribution of eigenvalues, in the sense of the Weyl law, of the Laplace-Beltrami operator on Riemannian manifolds \((M,g)\) isometric to the Euclidean space outside of compacts and such that the trapped set is of Liouville measure zero. Such distribution can be recovered by confronting the trace of the Laplacian on \((M,g)\) with the trace of the Laplacian on the Euclidean space. The difference between the two traces can be expressed through a function (defined through a suitable integration) called spectral shift function. If the Hausdorff dimension of the trapped set is of size \( 2 \delta + 1 \) then it is possible to show that the spectral shift function has an asymptotic expansion with remainder of order \(z^{\delta}\). This result is obtained by using the commutator method of \textit{D. Robert} [Ann. Sci. Éc. Norm. Supér. (4) 25, No. 2, 107--134 (1992; Zbl 0801.35100)] and a previous article by the same authors [Ann. Sci. Éc. Norm. Supér. (4) 47, No. 2, 371--448 (2014; Zbl 1297.58007)]. In a nutshell, by using the ciclicity of the trace and suitable cutoff functions one can find an equivalent expression for the spectral shift function for large \(z\), which can be recast with a scaling parameter \(1/h\). Such expression can then be studied by semiclassical analysis methods in that the pseudodifferential operators involved are well behaved. Given the same hypotheses as above on \((M,g)\), it is also shown that if the number of resonances of the Laplacian for \(z\) large, in a ball of suitable radius, is of order \(z^{\beta}\) for some small \(\beta\), then the spectral shift has a remainder of order \(z^{\beta}\). The proof is obtained by following closely an argument by \textit{R. B. Melrose} [Commun. Partial Differ. Equations 13, No. 11, 1431--1439 (1988; Zbl 0686.35089)]; in particular, it is possible to write a Poisson like formula for an implicit polynomial approximation of the spectral shift function and to exploit the properties of the resonances in the Fourier analysis. Last the authors consider cocompact hyperbolic manifolds. In this context they first define a similar spectral shift function (one has to change the reference framework and take into consideration a Laplacian of hyperbolic nature); then, by using the Selberg zeta function, they show that the remainder of this spectral function, when the trapped set is of Hausdorff dimension \( 2 \delta + 1 \) is of order \(z^{\delta}\), as in the other cases. The proofs use classical number theoretic arguments and builds up on ideas by \textit{L. Guillopé} and \textit{M. Zworski} [Ann. Math. (2) 145, No. 3, 597--660 (1997; Zbl 0898.58054)]. The interested reader could benefit greatly from reading the previously cited article by the same authors, since this article is a nice application of the ideas developed there.