When deriving a Lindblad equation (for example Breuer chapter 3), one crucial assumption is that $\tau_b$, the reservoir correlation function decay time, is (in short) the smallest relevant time scale. What I cannot understand is why these correlation decay at all.
If the bath is composed, for example, of uncoupled harmonic oscillators, and the system is so much smaller than the environment to the point that we assume its state is constant, wouldn't these correlation functions simply oscillate? If not, what determines the decay time?
My calculations of what I understand are the bath correlations in this case might illustrate my misunderstanding. With the interaction to the system being $B_\omega = (a+a^\dagger)/2 \equiv X$:
$$ B_\omega(t) = X_\omega (t) = \cos(\omega t) X + \sin(\omega t) P $$ $$ \langle B_\omega (t) B_\omega(0)\rangle = \cos(\omega t ) (\langle a^\dagger a \rangle + 1/2) +\kappa = \cos(\omega t ) (n(\omega,T) + 1/2) $$ where $\kappa$ are terms with zero average that I didn't write for clarity, and $n$ is the Bose Einstein distribution.