0
$\begingroup$

Suppose say that I have a quantum state $\vert\psi\rangle$ at time $t = 0$, which is now evolved by a hamiltonian $H$

$$e^{-iHt}\vert\psi\rangle$$. I can ask the question, how much of initial state is still available in the evolved state, this is called survival probability.

$$P(t) = \vert\langle \psi\vert e^{-iHt}\vert\psi\rangle\vert^2$$. Is there a circuit implementation for this survival probability? Any idea how to implement this or source to any link which discusses about this circuit implementation is appreciable.

$\endgroup$
3
  • 1
    $\begingroup$ One may approximate the time evolution operator $e^{-iHt}$ using a quantum circuit, however, further assumptions on $H$ are usually needed. The keyword is "Hamiltonian simulation". If you're interested in a specific (type of) Hamiltonian, I recommend that you specialize your question such that a concrete answer is possible. Otherwise, I suggest that you read up a bit on the literature. arxiv.org/abs/2101.07808 contains a decent overview over the available methods (but there is no review paper AFAIK) $\endgroup$ Commented Dec 6, 2022 at 9:50
  • $\begingroup$ @MarkusHeinrich I was actually thinking of Trotterizing the evolution part (assuming the evolution is Trotterizable) and do a swap test between initial state and evolved state, performing a proper measurement on ancilla qubit, i can find the mod square of overlap between these states. But I don't know whether this will work. Just wondering, is there a more efficient way to find this probability. I am trying to simulate Rabi oscillations using quantum circuit, lets say. $\endgroup$ Commented Dec 6, 2022 at 10:01
  • $\begingroup$ Is $\psi$ unknown or why do you want to do a swap test? You would also need two copies for each run. If $\psi$ is known and a "simple" state, I'd rather rotate the basis measurement accordingly. Anyway, without more details, you may not get a useful answer. $\endgroup$ Commented Dec 6, 2022 at 12:04

1 Answer 1

3
$\begingroup$

We can calculate this survival probability $P(t) = |\langle \psi | e^{-i \mathcal{H}t} | \psi \rangle|^2$ in two ways. The first corresponds to the so-called Hadamard test, as shown in the figure below,

enter image description here

where $H = \frac{1}{\sqrt{2}} \begin{pmatrix} 1 & 1 \\ 1 & -1 \end{pmatrix}$ is the Hadamard gate and $|\psi \rangle$ is some $n$-qubit state. The probability of measuring the ancillary qubit in state $|0\rangle$ is $\frac{1 + |\langle \psi | e^{-i \mathcal{H}t} | \psi \rangle|^2}{2} \equiv \frac{1 + P(t)}{2}$.

Alternatively, we can avoid the use of an ancillary qubit and the corresponding control of the time evolution provided that we have access to the circuit $U$ such that $U |0\rangle = |\psi \rangle$. In such case, we can simply use the following ancilla-free circuit

enter image description here

By measuring all $n$ qubits, the probability of measuring all qubits in state $|0 \rangle$ (i.e., retrieving the fiducial state) is $|\langle 0 | U^{\dagger} e^{-i \mathcal{H} t} U | 0 \rangle|^2 = | \langle \psi | e^{-i \mathcal{H} t} | \psi \rangle |^2 \equiv P(t)$.

Of course, the remaining question is how to implement the time-evolution operator $e^{-i \mathcal{H}t}$, which is the Hamiltonian Simulation problem mentioned by @MarkusHeinrich. The simplest option is Trotterization, but more sophisticated approaches such as truncated Taylor series with linear combination of unitaries method or qubitization are also available.

$\endgroup$
1
  • $\begingroup$ Yeah, I already figured out the way and added them in comments. Thanks for the answer. $\endgroup$ Commented Dec 6, 2022 at 18:28

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.