Lesson01
Single qubit
\[
\begin{align*}
\ket\psi &= \cos\dfrac\theta2 \ket0 + e^{i\phi}\sin\dfrac\theta2 \ket1 \\
\theta&\in[0, \pi] \\
\phi&\in [0,2\pi]
\end{align*}
\]
, or in matrix representation: \(\ket\psi = \begin{bmatrix} \cos\frac\theta2 \\ e^{i\phi}\sin\frac\theta2 \end{bmatrix}\).
State evolution
Equation of motion:
\[
i\hbar \frac{\partial}{\partial t}\ket\psi = \hat H \ket\psi
\]
Solution to this E.O.M.:
\[
\begin{align*}
\ket{\psi(t)} &= \mathcal U(t)\ket{\psi(0)} \\
&= e^{-i\hat Ht/\hbar} \ket{\psi(0)} \\
&= e^{-i\hat Ht} \ket{\psi(0)}
\end{align*}
\]
- the Evolution Operator: \(\mathcal U(t) = e^{-i\hat Ht/\hbar} = e^{-i\hat Ht}\), and people often let \(\hbar = 1\) for convenience.
Pauli Matrices
The Hamiltonian in the above E.O.M. can always be decomposed as
\[
\hat H = \frac12 \sum_{j=0,1,2,3} \omega_j\sigma_j = \frac12(\omega_0I + \vec \omega\cdot\vec\sigma)
\]
- \(\omega_0I\) term : only gives rise to global phase (shift energy equally for both levels)
- \(\vec \omega\cdot\vec\sigma\) term : is relevant to dynamics
Properties
\[
\sigma_j^2 = I \newline
\{\sigma_j, \sigma_k\} = 2I\delta_{jk} \newline
[\sigma_j, \sigma_k] = 2i\sigma_l\epsilon_{jkl}
\]
Thus, we can rewrite the evolution operator as:
\[
\begin{align*}
\mathcal U(t) &= e^{-iHt} = \exp(-i\cdot\frac12\vec \omega\cdot\vec\sigma\cdot t) \\
&= \exp(-i\cdot\frac{1}{2}\vert{\vec \omega}\vert \cdot \hat\sigma_n\cdot t) \\
&= \left(\cos\frac{\vert{\vec \omega}\vert t}2\right) \hat I - i \left(\sin\frac{\vert{\vec \omega}\vert t}{2}\right) \hat\sigma_n
\end{align*}
\]
where we’ve used the fact of the unit vector \(\hat\sigma_n\) (in Pauli basis) that
\[
(\hat\sigma_n)^2 = \sum_{i,j}n_i\sigma_i n_j\sigma_j = \frac12\sum_{i,j}n_in_j \{\sigma_i, \sigma_j\} \newline
=\frac12\sum_{i,j}n_in_j\cdot2I\delta_{ij} = \sum_i n_i^2I = I.
\]
Precession
If \(\hat\sigma_n = \hat z\), then
\[
\mathcal U(t) =
\begin{bmatrix}
e^{-i\omega t/2} & 0 \\
0 & e^{i\omega t/2}
\end{bmatrix}
\]
, which makes state evolutes \(\ket{\psi(0)}\mapsto\ket{\psi(t)}\) as \(\begin{bmatrix} \cos\frac\theta2 \\ e^{i\phi}\sin\frac\theta2 \end{bmatrix} \mapsto e^{-i\omega t/2} \begin{bmatrix} \cos\frac\theta2 \\ e^{i(\phi+\omega t)}\sin\frac\theta2 \end{bmatrix}\).
So in the Bloch vector picture, \(\theta\mapsto\theta\) and \(\phi\mapsto\phi+\omega t\).
For general \(\hat\sigma_n\), the Bloch vector undergoes a precession about axis \(\hat\sigma_n\), with angular velocity \(\vec \omega\). (the detailed proof is in the next lesson)
Rotating Wave Approximation
framework
-
\(H(t)\) is still an arbitrary Hamiltonian, but now it could be time-dependent.
(It's called the Hamiltonian in the interaction picture.)
-
Introduce a rotating frame-related quantity \(T(t) = e^{iAt}\), here \(A\) is time-independent and Hermitian.
Now we want to transform from the resting frame into the “rotating frame” (which is characterized by \(A\)), and the logic is:
\[
\begin{align*}
i\frac{∂}{∂t} \ket\psi &= H(t) \ket\psi \\
i\frac{∂}{∂t} (T\ket\psi) &= H_I(t)(T\ket\psi), \\
\ket\psi &\mapsto T\ket\psi \\
H(t) &\mapsto H_I(t).
\end{align*}
\]
How do we express \(H_I(t)\) in terms of the original \(H(t)\) and \(T\)?
Notice that:
\[
\begin{align*}
(THT^\dagger)(T\ket\psi) &= TH(T^\dagger T\ket\psi) \\
&= T i\frac{∂}{∂t}(T^\dagger T\ket\psi) \\
&= T i\left[ \frac{∂T^\dagger}{∂t}\cdot (T\ket\psi) + T^\dagger\cdot\frac{∂}{∂t}(T\ket\psi) \right] \\
&= iT\frac{∂T^\dagger}{∂t} \cdot (T\ket\psi) + i \frac{∂}{∂t}(T\ket\psi) \\
\implies H_I(t) &= (THT^\dagger) - iT\frac{∂T^\dagger}{∂t}
\end{align*}
\]
application to RWA
Following the previous definitions for \(H\), \(A\) and \(T\), one specific case that's considered widely is
\[
\begin{cases}
&H = \boxed{\alpha H_0} + H_1 = \boxed{\frac{\omega_0}{2}\sigma_z} + \Omega\cos(\omega t + \phi)\sigma_x \\
&A = \boxed{\beta H_0} = \boxed{\frac{\omega}{2}\sigma_z}
\end{cases}
\]
often we consider the case where \(\omega\gg\Omega\) and \(\omega\gg|\omega-\omega_0|\).
Then, according to the changing rule for \(H(t) \mapsto H_I(t)\) just derived above,
\[
\begin{align*}
H_I(t)
&= (THT^\dagger) - iT\frac{∂T^\dagger}{∂t} \\
&= e^{iAt} (\alpha H_0 + H_1) e^{-iAt} - ie^{iAt}\frac{∂e^{-iAt}}{∂t} \\
&= \boxed{e^{iAt} (\alpha H_0) e^{-iAt}} + e^{iAt} (H_1) e^{-iAt} - A \\
&= \boxed{(\alpha H_0)} + e^{iAt} (\Omega\cos(\omega t + \phi)\sigma_x) e^{-iAt} - A \\
&= \frac{\omega_0}{2}\sigma_z + \Omega\cos(\omega t + \phi) e^{i\omega\sigma_zt/2} \sigma_x e^{-i\omega\sigma_zt/2} - \frac{\omega}{2}\sigma_z \\
&= \frac{\omega_0-\omega}{2}\sigma_z + \Omega\cos(\omega t + \phi)
\begin{bmatrix}
e^{i\omega t/2} & 0 \\
0 & e^{-i\omega t/2}
\end{bmatrix}
\begin{bmatrix}
0 & 1 \\
1 & 0
\end{bmatrix}
\begin{bmatrix}
e^{-i\omega t/2} & 0 \\
0 & e^{i\omega t/2}
\end{bmatrix} \\
&= \frac{\omega_0-\omega}{2}\sigma_z + \Omega\cos(\omega t + \phi)
\begin{bmatrix}
0 & e^{i\omega t} \\
e^{-i\omega t} & 0
\end{bmatrix} \\
&= \frac{\omega_0-\omega}{2}\sigma_z + \frac\Omega 2
\begin{bmatrix}
0 & \textcolor{red}{e^{2i\omega t+i\phi}} + e^{-i\phi} \\
e^{i\phi} + \textcolor{red}{e^{-2i\omega t-i\phi}} & 0
\end{bmatrix}
\end{align*}
\]
(note that the boxed \(\boxed{e^{iAt} (\alpha H_0) e^{-iAt}}\) term can be reduced to \(\boxed{(\alpha H_0)}\) because all terms only involve \(\sigma_z\), hence commute with each other.)
We can ignore the term \(\textcolor{red}{\cancel{e^{2i\omega t+i\phi}}}\) and \(\textcolor{red}{\cancel{e^{-2i\omega t-i\phi}}}\) if we are only interested in the system's long-time behavior, and this leads to:
\[
H_I(t) \approx \frac{\omega_0-\omega}{2}\sigma_z + \frac\Omega 2 (\sigma_x\cos\phi + \sigma_y\sin\phi),
\]
which is the effective Hamiltonian in the rotating frame \((\omega)\).