The definition of thermal states I'm used to is:

$$\tau_{\beta} = \frac{1}{Z}\,e^{-\beta H}$$

where $Z$ is the partition function defined as $Z= \mathrm{Tr}(e^{-\beta H})$, $\beta$ the inverse temperature and H the system's Hamiltonian. I think if the Hamiltonian is not time dependent $H=\sum_n \varepsilon_n |n\rangle\langle n|$ I can rewrite the thermal state

$$\tau_{\beta_1} = \frac{1}{Z}e^{-\beta H} = \frac{1}{Z}\exp\left(-\beta \sum_n \varepsilon_n |n\rangle\langle n|\right) = \frac{1}{Z}\sum_ne^{-\beta \varepsilon_n} |n\rangle\langle n| .$$

Is this calculation correct?

In case it is correct, I just came across another version of it (I believe)

$$\tau_{\beta_2} = (1-e^{-\beta \omega}) \sum_n e^{-n\beta\omega}| n\rangle\langle n|$$

where $\omega$ is the frequency of some harmonic oscillator.

Can someone explain to me how these two definitions $(\tau_{\beta_1}$ and $\tau_{\beta_2})$ relate to each other?


I would call this the (normalized) density operator (or density matrix), in which case, yes, I think the expressions are correct.

For the quantum harmonic oscillator, in units where $\hbar=1$, and measuring energies from the ground state energy, the energy levels are $\varepsilon_n=n\omega$ for $n=0,1,2,\ldots$ and the partition function is $$ Z=\sum_{n=0}^\infty \exp(-n\beta\omega) = \frac{1}{1-\exp(-\beta\omega)} . $$ So everything is fine: the second formula is a particular example of the first.


Your second version of the density operator is the specific case in which the Hamiltonian is that of a simple quantum harmonic oscillator, namely


In terms of this Hamiltonian, the energy levels are (setting $\hbar=1$) $\varepsilon_n=\omega(n+1/2)$. Thus, the partition function is

$$Z=\sum_{n\geq 0}e^{-\beta\omega(n+1/2)}=\frac{e^{-\beta\omega/2}}{1-e^{-\beta\omega}}.$$

Thus, the density operator is given by

$$\tau_{\beta}=\frac{1}{Z}\sum_{n\geq 0}e^{-\beta\omega(n+1/2)}\,|n\rangle\langle n|=(1-e^{-\beta\omega})\sum_{n\geq 0}e^{-\beta\omega n}\,|n\rangle\langle n|,$$

as stated.


Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.