next up previous contents index
Next: Master Equation II: the Up: Master Equation I: Derivation Previous: Born Approximation   Contents   Index

Explicit Form of Master Equation

The equation of motion Eq.(7.20) is pretty useless unless one specifies at least some more details for the interaction Hamiltonian $ V\equiv H_{SB}$. Denoting system operators by $ S'_j$ and bath operators by $ B_k$, the most general form of $ V$ is
$\displaystyle V=\sum_{jk}c_{jk} S'_j\otimes B_k \equiv \sum_k S_k \otimes B_k,$     (23)

where we have re-defined the sum over $ j$ as a new system operator ( $ \rightarrow$ similarity to Schmid-decomposition).

Remark: Note that $ S_k$ and $ B_k$ need not necessarily be hermitian. Inserting Eq.(7.23) into Eq.(7.22), we have

$\displaystyle \frac{d}{dt} \tilde{\rho}(t)$ $\displaystyle =$ $\displaystyle -i \sum_k{\rm Tr}_B [\tilde{S}_k(t) \tilde{B}_k(t),
R_0 \rho(t=0)]$  
  $\displaystyle -$ $\displaystyle \int_0^t dt'\sum_{kl}{\rm Tr}_B[\tilde{S}_k(t) \tilde{B}_k(t),
[\tilde{S}_l(t') \tilde{B}_l(t'),R_0 \tilde{\rho}(t')]].$  

To simplify things, we will assume
$\displaystyle {\rm Tr}_B \tilde{B}_k(t) R_0 =0$     (24)

from now on. This is no serious restriction. We furthermore introduce the bath correlation functions
$\displaystyle C_{kl}(t,t')\equiv {\rm Tr}_B \left[\tilde{B}_k(t)\tilde{B}_l(t')R_0\right].$     (25)

Assumption 1:
$\displaystyle [R_0,H_B]=0$   bath in equilibrium$\displaystyle .$     (26)

This means
$\displaystyle C_{kl}(t,t')\equiv C_{kl}(t-t').$     (27)

We then have
$\displaystyle \frac{d}{dt} \tilde{\rho}(t)$ $\displaystyle =$ $\displaystyle -\int_0^t dt'\sum_{kl}\Big[C_{kl}(t-t')
\left\{ \tilde{S}_k(t)\ti...
...(t') \tilde{\rho}(t') - \tilde{S}_l(t') \tilde{\rho}(t')
\tilde{S}_k(t)\right\}$  
  $\displaystyle +$ $\displaystyle C_{lk}(t'-t)\left\{
\tilde{\rho}(t')\tilde{S}_l(t') \tilde{S}_k(t) - \tilde{S}_k(t)\tilde{\rho}(t')
\tilde{S}_l(t') \right\}\Big].$ (28)

Assumption 2a (Markov approximation): the bath correlation function $ C_{kl}(\tau)$ is strongly peaked around $ \tau\equiv t-t'=0$ with a peak width $ \delta \tau \ll \gamma^{-1}$, where $ \gamma$ is a `typical rate of change of $ \tilde{\rho}(t')$.' Note that the condition $ \delta \tau \ll \gamma^{-1}$ can only be checked after the equation of motion for $ \tilde{\rho}(t)$ has been solved. In the interaction picture, one then replaces $ \tilde{\rho}(t')\to \tilde{\rho}(t)$ under the integral to obtain
$\displaystyle \frac{d}{dt} \tilde{\rho}(t)$ $\displaystyle =$ $\displaystyle -\int_0^t dt'\sum_{kl}\Big[C_{kl}(t-t')
\left\{ \tilde{S}_k(t)\ti...
..._l(t') \tilde{\rho}(t) - \tilde{S}_l(t') \tilde{\rho}(t)
\tilde{S}_k(t)\right\}$  
  $\displaystyle +$ $\displaystyle C_{lk}(t'-t)\left\{
\tilde{\rho}(t)\tilde{S}_l(t') \tilde{S}_k(t) - \tilde{S}_k(t)\tilde{\rho}(t)
\tilde{S}_l(t') \right\}\Big].$ (29)

The important fact is that this approximation is carried out in the interaction (and not in the original Schrödinger) picture: in the interaction picture, the only relevant time-scale the change of the density matrix is $ \gamma^{-1}$ and not (the usually much faster) timescales from the free evolution with $ H_S$. In fact, one now transforms back into the Schrödinger picture, using Eq.(7.14),
$\displaystyle \frac{d}{dt} \tilde{\rho}(t)$ $\displaystyle =$ $\displaystyle i[H_S,\tilde{\rho}(t)] + e^{iH_S t} \frac{d}{dt}\rho(t) e^{-iH_St}$  
$\displaystyle \leadsto \frac{d}{dt}\rho(t)$ $\displaystyle =$ $\displaystyle -i[H_S,{\rho}(t)] + e^{-iH_S t} \frac{d}{dt}\tilde\rho(t) e^{iH_St}$ (30)

which leads to
$\displaystyle \leadsto \frac{d}{dt}\rho(t)$ $\displaystyle =$ $\displaystyle -i[H_S,{\rho}(t)]$  
  $\displaystyle -$ $\displaystyle \int_0^t dt'\sum_{kl}\Big[C_{kl}(t-t')
\left\{ e^{-iH_S t}\tilde{...
...ilde{\rho}(t) - \tilde{S}_l(t') \tilde{\rho}(t)
\tilde{S}_k(t)\right\}e^{iH_St}$  
  $\displaystyle +$ $\displaystyle C_{lk}(t'-t)\left\{e^{-iH_S t}
\tilde{\rho}(t)\tilde{S}_l(t') \tilde{S}_k(t) - \tilde{S}_k(t)\tilde{\rho}(t)
\tilde{S}_l(t') \right\}e^{iH_St}\Big]$  
  $\displaystyle =$ $\displaystyle -i[H_S,{\rho}(t)]$  
  $\displaystyle -$ $\displaystyle \int_0^t dt'\sum_{kl}\Big[C_{kl}(t-t')
\left\{ {S}_k\tilde{S}_l(t'-t) {\rho}(t) - \tilde{S}_l(t'-t) {\rho}(t)
{S}_k\right\}$  
  $\displaystyle +$ $\displaystyle C_{lk}(t'-t)\left\{
{\rho}(t)\tilde{S}_l(t'-t) {S}_k - {S}_k{\rho}(t)
\tilde{S}_l(t'-t) \right\}\Big].$ (31)

Assumption 2b (Markov approximation): the integral over $ t'$ can be carried out to $ t=\infty$. This in fact is completely consistent with assumption 2a (see above). Defining
$\displaystyle D_k\equiv \lim_{t\to\infty}\int_0^{t} d\tau\sum_lC_{kl}(\tau)\til...
...iv \lim_{t\to\infty}\int_0^{t} d\tau\sum_lC_{lk}(-\tau)\tilde{S}_l(-\tau),\quad$     (32)

we can write
$\displaystyle \fbox{$ \begin{array}{rcl} \displaystyle \frac{d}{dt}\rho(t) &=& ...
... {\rho}(t)
{S}_k+
{\rho}(t)E_k {S}_k - {S}_k{\rho}(t)
E_k \Big].\end{array}$\ }$     (33)


next up previous contents index
Next: Master Equation II: the Up: Master Equation I: Derivation Previous: Born Approximation   Contents   Index
Tobias Brandes 2004-02-18