2
$\begingroup$

Grover's algorithm circuit may be implemented as follows:

enter image description here

(from here)

It is shown very elegantly by @MartinVesely (How to interpret a 4 qubit quantum circuit as a matrix?) how to translate a 4 qubit circuit into its complete matrix representation.

Building on the previous question, I would like translate the full Grover's ciruit into a single state matrix. I attempted as follows:

Op 1: $H \otimes H \otimes H \otimes H$

Op 2: $X \otimes X \otimes X \otimes X$

Op 3: ?

Op 4: $CNOT \otimes I \otimes I$

Op 5: ?

...

Does anyone know how to correctly translate operation 3 and operation 5 in particular into its state matrix, please?

$\endgroup$
3
  • $\begingroup$ Could you clarify which two qubit gates you are using? I see you are using CX/CNOT. Is the other two qubit gate a controlled Rz or a controlled phase rotation? $\endgroup$
    – Callum
    Jan 3 at 21:17
  • $\begingroup$ @CallumMacpherson the diagram is not mine either. Link to the source is given above. I am not clear myself what gate is Operation 3 using? And how to "skip lines" using some formula similar to what's mentioned in quantumcomputing.stackexchange.com/questions/9614/…? $\endgroup$
    – James
    Jan 3 at 21:20
  • $\begingroup$ Ah okay thanks, looks like its a U1 gate. Can give an answer now. $\endgroup$
    – Callum
    Jan 3 at 21:22

2 Answers 2

2
$\begingroup$

What you have looks like a controlled-U1 gate, which for two qubits has the form:

$$\begin{bmatrix} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & e^{i \theta} \end{bmatrix}$$ Notice that the cu1 gate doesn't distinguish between the control qubit and the target qubit. Both give the same result.

The matrix for a cu1 embedded into multiple qubits is a more complicated diagonal matrix, with several of the diagonal entries having the values $e^{i \theta}$

You can use

qc = QuantumCircuit(4)
qc.cu1(.79, 1, 3)

array_to_latex(Operator(qc).data, max_size=100)

to see one example.

$\endgroup$
2
  • $\begingroup$ thank you. If this controlled-U1 gate "skips over" other lines while connecting its 2 lines, how should the 16x16 matrix be constructed for say operation 3? I am trying to do all the steps using matrix algebra. $\endgroup$
    – James
    Jan 4 at 0:23
  • $\begingroup$ It's probably worth noting that the blog article in the question gives a full listing for the code for the circuit (and, observe that the angle is $\pi/4$ rather than .79). So you could just use the above method to directly compute the unitary for the whole circuit, as well as identify what the individual parts are. $\endgroup$
    – DaftWullie
    Jan 4 at 7:30
1
$\begingroup$

You can figure out 3) and 5) using the formula given by @MartinVesely in the page you linked to.

As was discussed in the post or the matrix of a CU gate with a control and a target separated by $k$ qubits is as follows.

\begin{equation} CU_k = \begin{pmatrix} I_\frac{N}{2}& O_\frac{N}{2} \\ O_\frac{N}{2} &I_\frac{N}{4} \otimes U \end{pmatrix} \end{equation}

Here $N=2^{k+2}$, $O$ is the all-zero matrix and $I$ is the identity. The gates in your diagram seem to be Controlled U1 gates where U1 is represented by the following unitary.

\begin{equation} U1(\lambda) = \begin{pmatrix} 1& 0 \\ 0 &e^{i \lambda} \end{pmatrix} \end{equation}

Now we see there are two qubits between the control and the target qubit of the CU1 gate. Therefore $k=2$ and $N=16$). Substituting in $N=16$ and $U=U1$ we get the following... \begin{equation} CU1_{k=2} = \begin{pmatrix} I_8& O_8 \\ O_8 &I_4 \otimes U1 \end{pmatrix} \end{equation}.

If you have the patience you can write down the full $16 \times 16$ matrix from this by expanding the four entries. Maybe you could factorise this matrix into a tensor product of $2 \times 2$ matrices. I'd have to think about it more as I'm not sure there is a nice form.

Edit: Similarly to Frank I calculated the unitary (I did the first CU1 gate in the circuit). Wouldn't want to do this by hand.

enter image description here

I did this symbolically with pytket. Note the factor of $\pi$ due to different conventions. $(\lambda = \pi \theta)$

from pytket import Circuit, OpType
from pytket.utils import circuit_to_symbolic_unitary
from sympy import Symbol

theta = Symbol("theta") #theta = 1 / 4 in your example
circ = Circuit(4).add_gate(OpType.CU1, [theta], [0, 3])
circuit_to_symbolic_unitary(circ)
$\endgroup$
1
  • $\begingroup$ thank you. yes, the same formula should work for any gate! $\endgroup$
    – James
    Jan 4 at 0:37

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.