I would like to calculate the Pauli spin operator rotation $$ U^{\dagger } \overset{\rightharpoonup }{\sigma } U$$

where $$\overset{\rightharpoonup }{\sigma }=\sigma _x \overset{\rightharpoonup }{x}+\overset{\rightharpoonup }{y} \sigma _z+\sigma _z \overset{\rightharpoonup }{z}$$


$$U = e^{i t \omega \sigma _z}$$

I know that this operation can be considered a rotation of the eigenvalues of the operators:

$$ U^{\dagger } \sigma _x U = \underset{n}{\Sigma } e_n U^{\dagger } |n'\rangle \langle n'| U=\underset{n}{\Sigma } e_n |n'\rangle \langle n'|$$

and these eigenvalues represent the rotation axis of the overall matrix $U^{\dagger } \overset{\rightharpoonup }{\sigma } U$ (or Hamiltonian in my case of interest).

In other words, this should be really straightforward to compute without need a lot of algebra or Baker–Campbell–Hausdorff formulas. However I don't quite see how to do it most straightforwardly; any thoughts?


$ {\rm U}\left(t\right) \equiv \exp\left({\rm i}\,\omega t\,\sigma_{z}\right) $ satisfies $\ddot{\rm U}\left(t\right) + \omega^{2}{\rm U}\left(t\right) = 0$ with ${\rm U}\left(0\right) = 1$ and $\dot{{\rm U}}\left(0\right) = {\rm i}\omega\sigma_{z}$. Then

\begin{align} {\rm U}\left(t\right) &= \exp\left({\rm i}\,\omega t\,\sigma_{z}\right) = \cos\left(\omega t\right) - {\rm i}\sin\left(\omega t\right)\,\sigma_{z} \\[3mm] {\rm U}^{\dagger}\left(t\right) &= \exp\left(-{\rm i}\,\omega t\,\sigma_{z}\right) = \cos\left(\omega t\right) + {\rm i}\sin\left(\omega t\right)\,\sigma_{z} \end{align}

Define $A\left(t\right) \equiv {\rm U}^{\dagger}\left(t\right)A{\rm U}\left(t\right)$ such that ${\rm i}\dot{A} = \omega\left[\sigma_{z},A\left(t\right)\right]$.

$$ {\rm i}\dot{\sigma}_{x}\left(t\right) = \omega\left[{\rm i}\sigma_{y}\left(t\right)\right]\,, \qquad {\rm i}\dot{\sigma}_{y}\left(t\right) = \omega\left[-{\rm i}\sigma_{x}\left(t\right)\right]\,; \qquad {{\rm d}\left[\sigma_{x}\left(t\right) + {\rm i}\sigma_{y}\left(t\right)\right] \over {\rm d}t} = -{\rm i}\omega \left[\sigma_{x}\left(t\right) + {\rm i}\sigma_{y}\left(t\right)\right] $$

\begin{align} \sigma_{x}\left(t\right) + {\rm i}\sigma_{y}\left(t\right) &= \left(\sigma_{x} + {\rm i}\sigma_{y}\right)\exp\left(-{\rm i}\omega t\right) \\[3mm] \sigma_{x}\left(t\right) - {\rm i}\sigma_{y}\left(t\right) &= \left(\sigma_{x} - {\rm i}\sigma_{y}\right)\exp\left({\rm i}\omega t\right) \end{align} Then $$ \sigma_{x}\left(t\right) = \sigma_{x}\cos\left(\omega t\right) + \sigma_{y}\sin\left(\omega t\right)\,, \qquad \sigma_{y}\left(t\right) = -\sigma_{x}\sin\left(\omega t\right) + \sigma_{y}\cos\left(\omega t\right) $$

  • $\begingroup$ I'm familiar with the Euler expansion for the exponentiated pauli operator, but I dont immediately see how it follows from the differential equation statement? Or is it not a constructive derivation, in other words, are you just saying that both, exponentials and sin/cos will satisfy the 2nd Order ODE. And then you match boundary conditions for the two ways of expressing the particular solution? $\endgroup$ – AimForClarity Sep 25 '13 at 2:02
  • $\begingroup$ @AimForClarity That works since $\sigma_{i}^{2} = 1\,, i = x, y, z$. $\endgroup$ – Felix Marin Sep 25 '13 at 2:04
  • $\begingroup$ oh ok so you're Taylor expanding the exponential? $\endgroup$ – AimForClarity Sep 25 '13 at 2:10

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.