Scattering fidelity in elastodynamics

The recent introduction of the concept of scattering fidelity causes us to revisit the experiment by Lobkis and Weaver [Phys. Rev. Lett. 90, 254302 (2003)]. There, the "distortion" of the coda of an acoustic signal is measured under temperature changes. This quantity is, in fact, the negative logarithm of scattering fidelity. We reanalyze their experimental data for two samples, and we find good agreement with random matrix predictions for the standard fidelity. Usually, one may expect such an agreement for chaotic systems, only. While the first sample may indeed be assumed chaotic, for the second sample, a perfect cuboid, such an agreement is surprising. For the first sample, the random matrix analysis yields perturbation strengths compatible with semiclassical predictions. For the cuboid, the measured perturbation strengths are by a common factor of 5/3 too large. Apart from that, the experimental curves for the distortion are well reproduced.