Background: There is a single qubit quantum gate of the form $$ \frac{1}{\sqrt{2}}\begin{pmatrix} 1 & -i \\ 1 & i \end{pmatrix}. $$ As far as I know, this was originally introduced as the $T$ gate in Gottesman's thesis https://arxiv.org/abs/quant-ph/9705052 (see equation A.5). In https://arxiv.org/abs/quant-ph/0403025 the authors call a similar version of this gate the $T$ gate.

However, as we know, this eventually fell out of favor since the $T$ gate almost universally means the $\pi/8$ rotation gate (i.e., $T = S^{1/2} = Z^{1/4}$).

Question: Since $T$ is off limits, what should we call this gate? Or does this gate already have a very standard name that isn't $T$?

Things to consider:

  • In https://arxiv.org/abs/1702.06990 the authors call this the $M_3$ gate (they seem to claim Gottesman calls it that in his thesis but I am not sure I can find where he calls it anything other than $T$?).

  • In our recent paper https://arxiv.org/abs/2305.07023 we call it the $M$ gate but admittedly I am asking this question specifically because I think that is a poor name (for starters it literally doesn't stand for anything related to any of its properties).

  • In https://threeplusone.com/pubs/on_gates.pdf the author cites Craig Gidney's stim python package and calls this (and the ones like it) the "axis-cycling gates". The name used is $C$ but I personally am not a fan since this usually denotes the Clifford group or Clifford hierarchy (or even the two qubit canonical gates).

  • There is a twitter post https://twitter.com/CraigGidney/status/1393069730785857536 where some other choices are listed. One choice therein seems to be: ``face" gates denoted by $F$ since this gate always maps the faces of a octahedron to another face.

My thoughts: I personally think ``axis cycling gates" above is pretty descriptive since there are $16$ of these in total and each is defined by their conjugation action: cycling the axes up to a factor of $-1$. The gate above is sort of the canonical version as it acts as $X \to Y$, $Y \to Z$, and $Z \to X$. Maybe $A$ would be a good name?

Another property we point out in our above paper is this gate's importance to the binary tetrahedral group $2T$. In particular, $2T$ contains the the $8$ element Pauli group (from $SU(2)$) and the 16 axis cycling gates and that's it. In other words, the (special) Paulis plus the above gate generate $2T$. However, the obvious choice of name $T$ is out but maybe there is some other letter/notation related to this property.

Edit (7/26/23) In the 3rd version of our paper, we ended up going with $F$ for this gate and calling it the "facet" gate. This was inspired by what Mark said below, however, "face" and "phase" are near-homophones and so we thought facet made more sense (besides it is technically more accurate). This coincides with a comment made by Markus Heinrich in this question. Compare with figure 3 here.

  • $\begingroup$ "What should we call this gate" sounds pretty opinion-based. A better question would be whether there are standard names for this gate. $\endgroup$ Commented May 22, 2023 at 21:39
  • $\begingroup$ @MarkS Fixed. Sorry about that. $\endgroup$ Commented May 23, 2023 at 0:02
  • $\begingroup$ @NorbertSchuch I reorganized to make it clear I am asking both questions. I think it's important to hear preferences but I agree that the primary aim of the question should be to find out if there is some notation already in wide use. $\endgroup$ Commented May 23, 2023 at 0:16
  • $\begingroup$ ... also, the aim of StackExchange is not to collect opinions. -- On a more opinion-based note, I don't think everything needs its own name/letter, and conversely, it happens all the time that several things have the same name. $\endgroup$ Commented May 23, 2023 at 7:56
  • $\begingroup$ F^24 gives the Identity matrix, maybe this could help $\endgroup$ Commented Aug 6, 2023 at 14:58

2 Answers 2


I advocate for calling your $M$ gate by the letter $F$ for face, for the reasons depicted in Ciaran's picture.

One slight problem with $A$ is that $|A\rangle$ usually means (for surface code people at least) the magic state that when injected gives the $T$ gate (this usage was in place for the Bravyi-Kitaev Magic state paper if not before). Wherever possible I think it's nice to call the magic state and the gate it gives by the same name.

I have to admit that when I tried to use $F$ for the face gate, I was prevented by the fact that we were already using $F$ for Fidelity (see eq 104 of https://arxiv.org/abs/1808.00128), but I don't think those two concepts clash too regularly.


In stim, this gate is called C_XYZ and its inverse is called C_ZYX. This was the result of that twitter poll you referenced. The 'C' sort of collides with 'Clifford' and also with 'controlled', which is not ideal.

The 'C_XYZ' Gate

Right handed period 3 axis cycling gate, sending X -> Y -> Z -> X.

Stabilizer Generators:

X -> +Y
Z -> +X

Bloch Rotation:

Axis: +X+Y+Z
Angle: 120 degrees

Unitary Matrix:

[+1-i, -1-i]
[+1-i, +1+i] / 2

Decomposition (into H, S, CX, M, R):

# The following circuit is equivalent (up to global phase) to `C_XYZ 0`
S 0
S 0
S 0
H 0

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.