Consider a qubit-cavity system governed by the usual Jaynes-Cummings Hamiltonian in the dispersive regime $$ H_0 = \omega_q \frac{\sigma_z}{2} + \omega_a a^\dagger a + \chi a^\dagger a \frac{\sigma_z}{2} $$ where $a$ is the annihilation operator of the cavity, and $\sigma_z$ is the z Pauli operator on the qubit. We drive the system with a time-dependent Hamiltonian $$ H_d(t) = \Omega \mathrm{e}^{i t \omega_q} \sigma_+ + \Omega^* \mathrm{e}^{-i t \omega_q} \sigma_- $$ where $\sigma_\pm$ are the Pauli creation and annihilation operators. For $|\Omega| \ll \chi$, since the drive is resonant with the qubit at frequency $\omega_d = \omega_q - 0 \chi$, the system will see Rabi oscillations for the photonless cavity state $|0\rangle$. In other words, if the initial state of the system is $|\psi(t=0)\rangle = |0,g\rangle$, the system will oscillate between $|0, g\rangle$ and $|0, e\rangle$ such that $$ |\langle 0,g| \psi(t) \rangle|^2 = \cos^2\left(\frac{\Omega t}{2}\right) $$ $$ |\langle 0,e| \psi(t) \rangle|^2 = \sin^2\left(\frac{\Omega t}{2}\right) $$

Question: How can I compute the exact form of $|\psi(t)\rangle$ starting from any initial state $|\psi(t=0)\rangle$ under this time-dependent drive?


1 Answer 1


First point: The undriven version of this Hamiltonian is not at all like the Jaynes-Cummings Hamiltonian

Solutions of the Jaynes-Cummings dynamics are usually based on the observation that the interaction term only couples two states. This is not the case for the Hamiltonian given by the OP, since

$$a^\dagger a \sigma_z |n,g\rangle = a^\dagger a(|e\rangle\langle e|-|g\rangle\langle g|)|n,g\rangle = -a^\dagger a |n,g\rangle = -n |n,g\rangle \,$$


$$a^\dagger a \sigma_z |n,e\rangle = a^\dagger a(|e\rangle\langle e|-|g\rangle\langle g|)|n,e\rangle = +a^\dagger a |n,e\rangle = +n |n,e\rangle \,,$$

where $|n,e/g\rangle = |n\rangle_\mathrm{photon} \otimes |e/g\rangle$, with $|n\rangle$ a photon number Fock state and $|e/g\rangle$ the excited/ground state, respectively, of the qubit.

This Hamiltonian therefore only yields a phase evolution of these states. You can check this by writing down the equation of motion $|\dot{\psi}\rangle \propto iH|\psi\rangle$ for these states.

We can also understand this, since the $a^\dagger a$-term simply counts the photon number and the $\sigma_z$-term measures the excitedness/inversion of the qubit. This term is therefore diagonal in the Fock-state basis.

In the standard Jaynes-Cummings model, we instead have the interaction term $(a^\dagger \sigma_- + a \sigma_+)$. This term couples $|n,e\rangle$ to $|n+1,g\rangle$.

Second point: The time-dependent solution can be found in the same way as for the JC model

In the question, Rabi oscillations are obtained through the driving term, which contains $\sigma_\pm$ and therefore couples the states. You can then simply set up the equations of motion $|\dot{\psi}\rangle \propto iH|\psi\rangle$ in the $\{|n,g\rangle, |n,e\rangle\}$-basis and solve the resulting equations of motion for your initial states. This method is the same as for the Jaynes-Cummings model and can be found in any standard textbook on quantum optics. If the time-dependent drive is a problem, one can go to an interaction picture rotating at the drive frequency.

  • $\begingroup$ Thank you for your answer. Concerning your first point, the Hamiltonian I mention is a Jaynes-Cummings, but in the dispersive regime (see for instance equations (3.1) to (3.3) in arxiv.org/abs/0810.1336 for the derivation). $\endgroup$
    – Ronan
    Commented Feb 2, 2021 at 14:18
  • $\begingroup$ Concerning your second point, I have a hard time actually solving in the $|n,g\rangle, |n,e\rangle$ basis, even by going in the frame rotating at $\omega_q$. The driving term $H_d$ couples both the ground and excited states, which actually do not rotate at the same frequency due to the dispersive term in $\chi a^\dagger a \sigma_z / 2$. Do you have any idea on how to deal with this? Many thanks again $\endgroup$
    – Ronan
    Commented Feb 2, 2021 at 14:20
  • $\begingroup$ @Ronan ah I see! In that case, could you include the resulting equations of motion in your question to pinpoint where exactly the problem lies? This would be helpful, since for the standard (non-dispersive) JC model, the solution is rather simple. So it would be good to show why you're stuck for this Hamiltonian and not for the JC model. $\endgroup$ Commented Feb 2, 2021 at 15:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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