Skip to main content

Stack Exchange Network

Stack Exchange network consists of 183 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange
Asked
Modified today
Viewed 167 times
4
$\begingroup$

Consider SWAP operator for qubits. It is written in terms of Pauli operators as:

$ SWAP= \frac{1}{2}(\mathbf{I} + \sum_{a\in\{x,y,z\}} \sigma^{a}_{1}\sigma^{a}_{2}) $

I have "verified" that above representation is correct. Basically what I did was consider basis states as: $| \uparrow \uparrow \rangle, | \uparrow \downarrow \rangle, | \downarrow \uparrow \rangle,| \downarrow \downarrow \rangle $ and used the action of SWAP opertor on these basis states. For eg. $SWAP | \downarrow \uparrow \rangle = | \uparrow \downarrow \rangle$ (i.e. it swaps the two qubits).

I now want to generalise it for the qutrit case (they are also called clock-states). For this case one will have analogous $\sigma^{z}$ and $\sigma^{x}$ operators as \begin{equation} Z=\left(\begin{array}{ccc} 1 & 0 & 0 \\ 0 & \omega & 0 \\ 0 & 0 & \omega^2 \end{array}\right), \quad X=\left(\begin{array}{ccc} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 1 & 0 & 0 \end{array}\right) . \end{equation} The above operators are unitary and have eigenvalues $1, \omega, \omega^{2}$; $\omega= e^{2 \pi i /3}$.

To be noticed that I can easily derive the matrix form of the SWAP operator for this case as well by considering the basis states. What I want is an analogous formula for the operator in terms of $Z$ and $X$ operators. To understand it, I was trying to understand how the form of the SWAP operator could be derived for the qubit case (I could only verify it and not derive it). Any help in this matter is appreciated, maybe some references where I can get more information related to this.

New contributor
Erosannin is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
$\endgroup$

2 Answers 2

2
$\begingroup$

The formula is actually well-known and the general case is such that the Pauli operators on the two systems are inverses of each other.

To see this, let us define the $Z$ and $X$ operators on the computational basis $\{|x\rangle\}_{x\in\mathbb{Z}_d}$ of $\mathbb{C}^d$ as $X|x\rangle = |x\oplus 1\rangle$ and $Z|x\rangle = \omega^x |x\rangle$ where $\oplus$ is addition modulo $d$ and $\omega = e^{2\pi i/d}$ is a primitive $d$-th root of unity. The operators $Z^z X^x$ form an orthogonal basis of the space of linear operators $L(\mathbb{C}^d)$ and $Z^z X^x = \omega^{zx} X^x Z^z$.

Computing the expansion of SWAP in this basis can be obtained by evaluation Hilbert-Schmidt inner products, using $\mathrm{SWAP}^\dagger=\mathrm{SWAP}$ and the SWAP trick $\mathrm{tr}(\mathrm{SWAP} A \otimes B) = \mathrm{tr}(AB)$: $$ (\mathrm{SWAP}, Z^z X^x\otimes Z^u X^v) = \mathrm{tr}(\mathrm{SWAP}^\dagger Z^z X^x\otimes Z^u X^v) = \mathrm{tr}(\mathrm{SWAP} Z^z X^x\otimes Z^u X^v) = \mathrm{tr}(Z^z X^x Z^u X^v) = \omega^{ux}\mathrm{tr}(Z^{z\oplus u} X^{x\oplus v}) = d\,\omega^{ux} \delta_{z,-u} \delta_{x,-v} \,. $$ In the last line, we used that the $Z$ and $X$ operators are traceless, thus the only way the trace does not vanish is if the operators are raised to the power of zero (and thus $Z^z X^x$ and $Z^u X^v$ are inverses of each other).

Hence the formula is $$ \mathrm{SWAP} = \frac{1}{d}\sum_{z,x} \omega^{-zx} Z^z X^x\otimes Z^{-z} X^{-x} \,. $$ However, we typically choose differently normalized operators, namely so-called Weyl operators, defined as $$ W(z,x) = \tau^{-zx} Z^z X^x \, $$ where $\tau = (-1)^d e^{i\pi/d}$ is such that $\tau^2 = \omega$ for all $d$ (note that $\tau=i$ for $d=2$ and we obtain the standard Pauli operators). With this convention, $$ \mathrm{SWAP} = \frac{1}{d}\sum_{z,x} W(z,x) \otimes W(-z,-x) \,. $$

$\endgroup$
2
$\begingroup$

There is a unique way to write any matrix (not just the SWAP gate) on the two-qutrit system as a linear combination of matrices $$ X_1^aZ_1^bX_2^cZ_2^d, ~~~a,b,c,d \in \{0,1,2\}. $$

This is because they form an orthonormal basis (up to a scalar factor) in the space of all matrices on $\mathbb{C}^3 \otimes \mathbb{C}^3$ under the Hilbert-Schmidt inner product defined by $$ \langle A, B \rangle_{\rm HS} = \text{Tr}(B^\dagger A). $$

To find coefficients you can simply compute HS inner products with the SWAP gate, as $$ x = \sum \langle x, e_i\rangle e_i $$ for any vector $x$ and any orthonormal basis $\{e_i\}$ (assuming the inner product is anti-linear on the second argument).

I suspect the formula will be similar, i.e. only elements $X_1^aZ_1^bX_2^aZ_2^b$ will have non-zero coefficients in the sum.

$\endgroup$

Your Answer

Post as a guest

Required, but never shown

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

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.

Morty Proxy This is a proxified and sanitized view of the page, visit original site.