Statistics of resonance poles, phase shifts and time delays in quantum chaotic scattering: Random matrix approach for systems with broken time-reversal invariance

Assuming the validity of random matrices for describing the statistics of a closed chaotic quantum system, we study analytically some statistical properties of the S-matrix characterizing scattering in its open counterpart. In the first part of the paper we attempt to expose systematically ideas underlying the so-called stochastic (Heidelberg) approach to chaotic quantum scattering. Then we concentrate on systems with broken time-reversal invariance coupled to continua via M open channels; a = 1,2,...,M. A physical realization of this case corresponds to the chaotic scattering in ballistic microstructures pierced by a strong enough magnetic flux. By using the supersymmetry method we derive an explicit expression for the density of S-matrix poles (resonances) in the complex energy plane. When all scattering channels are considered to be equivalent our expression describes a crossover from the chi(2) distribution of resonance widths (regime of isolated resonances) to a broad power-like distribution typical for the regime of overlapping resonances. The first moment is found to reproduce exactly the Moldauer-Simonius relation between the mean resonance width and the transmission coefficient, Under the same assumptions we derive an explicit expression for the parametric correlation function of densities of eigenphases theta(a) of the S-matrix (taken module 2 pi). We use it to find the distribution of derivatives tau(a) = partial derivative theta(a)/partial derivative E of these eigenphases with respect to the energy (''partial delay times'') as well as with respect to an arbitrary external parameter. We also find the parametric correlations of the Wigner-Smith time delay tau(w)(E)=(1/M) Sigma(a) partial derivative theta(a)/partial derivative E at two different energies E - Omega/2 and E + Omega/2 as well as at two different values of the external parameter. The relation between our results and those following from the semiclassical approach as well as the relevance to experiments are briefly discussed. (C) 1997 American Institute of Physics.