1: \begin{abstract}

2: We study the problem of the approach to equilibrium in a macroscopic quantum system in an abstract setting.

3: We   prove that, for a typical choice of ``nonequilibrium subspace'', any initial state (from the energy shell)  thermalizes, and in fact does so very quickly, on the order of the Boltzmann time $\tauB:=h/(k_\mathrm{B}T)$.

4: This apparently unrealistic, but mathematically

5: rigorous, conclusion has the important physical implication that the moderately slow decay

6: observed in reality is not typical in the present setting.

7: 

8: The fact that macroscopic systems approach thermal equilibrium may seem puzzling, for example, because it may  seem to conflict with the time-reversibility of the microscopic dynamics.

9: According the present result, what needs to be explained is, not that macroscopic systems approach equilibrium, but that they do so slowly.

10: 

11: Mathematically our result is based on an interesting property of the maximum eigenvalue of the Hadamard product of a positive semi-definite matrix and a random projection matrix.

12: The recent exact formula by Collins for the integral with respect to the Haar measure of the unitary group plays an essential role in our proof.

13: \end{abstract}

14: