
Measurements of Floquet code plaquette stabilizers
James R. Wootton
IBM Quantum, IBM Research Europe – Zurich
(Dated: October 25, 2022)
The recently introduced Floquet codes have already inspired several follow up works in terms of
theory and simulation. Here we report the first preliminary results on their experimental imple-
mentation, using IBM Quantum hardware. Specifically, we implement the stabilizer measurements
of the original Floquet code based on the honeycomb lattice model, as well as the more recently
introduced Floquet Color code. The stabilizers of these are measured on a variety of systems, with
the rate of syndrome changes used as a proxy for the noise of the device.
I. INTRODUCTION
Floquet codes are a concept that has recently emerged
in quantum error correction [1]. The first explicit ex-
ample was the honeycomb Floquet code [1–5], based on
Kitaev’s honeycomb lattice model [6]. The concept was
then extended with two additional codes. First was the
e↔mautomorphism code [7], based on a generalization
of honeycomb Floquet code. Next was the Floquet Color
code [8–10], which can be derived from Color codes [11].
Both the original honeycomb Floquet code and the
Floquet Color code share similarities in implementation.
Specifically, both are based on the measurement of two-
body observables corresponding to edges in a hexagonal
lattice. This makes them both particularly well suited
to the heavy-hexagon architecture of IBM Quantum pro-
cessors [12], in which qubits reside on both the vertices
and links of a hexagonal lattice. These can then be used
as the required code qubits and auxiliary qubits respec-
tively, where the latter are used within the circuits to
measure two-body observables (see Fig. 1).
Most of the available hardware is not sufficient to per-
form a sensible analysis of the logical subspace. The 27
qubit Falcon devices, for example allow only two plaque-
ttes to be realized. Nevertheless, the hardware is suf-
ficient to measure the stabilizers, and assess how they
detect noise on the devices. Here we will perform this for
the honeycomb Floquet code and Floquet Color codes.
II. HONEYCOMB FLOQUET CODES
Kitaev’s honeycomb lattice is a well-studied model in
both condensed matter and quantum information [6]. It
is defined on a hexagonal lattice with qubits on the ver-
tices, as shown in Fig. 2. The links are labelled x,yand
zdepending on orientation. Each is assigned a two-body
link operator σα⊗σαon the two adjacent vertex qubits,
where α∈ {x, y, z}is the link type.
Important to any study of this model are the plaquette
operators, since they commute with all the link operators.
Each is defined as the product of link operators around
a corresponding plaquette. They can be expressed as
W=σx
0σy
1σz
2σx
3σy
4σz
5(1)
FIG. 1. Circuits to measure operators for (a) x-links, (b) y-
links and (c) z-links. The two vertex qubits for which the
parity is measured are at the top and bottom of each circuit.
The auxiliary is in the middle. This should be initialized in
the |0istate. If it remains in the output state of a previous
measurement, the end result will be the XOR of the two.
using the numbering of Fig. 2(c).
The most obvious means to convert this model to a
quantum error correcting code is arguably to perform
the two-body parity measurements corresponding to the
link operators, treating them as the gauge operators of
a subsystem code. The plaquettes then form the corre-
sponding stabilizer generator. Unfortunately, such a code
has a trivial logical subspace [13]. However, two separate
means to carve out a non-trivial subspace were intro-
duced in the last year. One of these was the honeycomb
Floquet code, and the other was an approach based on
matching codes [14]. In both, a particular order is used
for the measurement of the gauge operators in order to
effectively create a logical subspace.
For the honeycomb Floquet codes, the measurement
schedule is defined using a tri-colouring of the plaquettes
and links of the hexgaonal lattice. This is done such that
arXiv:2210.13154v1 [quant-ph] 24 Oct 2022