---
title: 'Qutrit Random Cubic Codes: 3D CSS Models'
url: https://www.emergentmind.com/topics/qutrit-random-cubic-codes
type: topic
---

# Qutrit Random Cubic Codes: 3D CSS Models

Searching arXiv for recent papers directly relevant to qutrit random cubic codes and closely related qutrit/cubic-code literature.
Searching arXiv for "qutrit random cubic codes".
Qutrit random cubic codes are a family of three-dimensional local qutrit Calderbank-Shor-Steane stabilizer Hamiltonians built on the cube-check geometry of Haah’s Code 1 but with spatially varying stabilizer coefficients constrained by local commutation and a global torus relation [2606.19873]. In this primary sense, they are non-translation-invariant qutrit stabilizer models on a cubic lattice that rigorously retain the no-string property and numerically exhibit logical-operator and degeneracy behavior distinct from translation-invariant fracton codes. In a looser but related usage, the same phrase also refers to random-looking large qutrit stabilizer or CSS codes whose magic-state distillation map for the qutrit strange state has cubic noise suppression [2408.00436].

## 1. Definition and lattice realization

The system is defined on the periodic cubic lattice
\[
\Lambda_L=(\mathbb Z/L\mathbb Z)^3,
\]
with two qutrits per lattice site, \((r,1)\) and \((r,2)\), so the total number of physical qutrits is
\[
n=2L^3.
\]
A qutrit has computational basis \(\{\lvert a\rangle:a\in\mathbb F_3\}\), with Pauli operators
\[
X\lvert a\rangle=\lvert a+1\rangle,\qquad Z\lvert a\rangle=\omega^a\lvert a\rangle,\qquad \omega=e^{2\pi i/3},
\]
and
\[
Z^zX^x=\omega^{zx}X^xZ^z,\qquad x,z\in\mathbb F_3.
\]
For CSS operators, commutation reduces to
\[
\mathbf z\cdot \mathbf x=\sum_j z_jx_j=0\pmod 3.
\]

The local geometry uses the eight corners
\[
O=(0,0,0),\quad A=(1,0,0),\quad B=(0,1,0),\quad C=(0,0,1),
\]
\[
D=(0,1,1),\quad E=(1,0,1),\quad F=(1,1,0),\quad G=(1,1,1).
\]
At each cube anchored at \(r\), one places a \(Z\)-type stabilizer \(A_r\) and an \(X\)-type stabilizer \(B_r\). The \(Z\)-check has no support at \(O\) and two-qutrit support at \(G\), while the \(X\)-check is reflected: it has two-qutrit support at \(O\) and no support at \(G\). The other six corners each carry a single-qutrit operator [2606.19873].

The stabilizers are
\[
A_r=
Z_{r+G,1}^{w_{1,r}}
Z_{r+G,2}^{w_{2,r}}
Z_{r+D,1}^{u_{D,r}}
Z_{r+E,1}^{u_{E,r}}
Z_{r+F,1}^{u_{F,r}}
Z_{r+A,2}^{v_{A,r}}
Z_{r+B,2}^{v_{B,r}}
Z_{r+C,2}^{v_{C,r}},
\]
with exponent vector
\[
\mathbf a_r=(w_{1,r},w_{2,r},u_{D,r},u_{E,r},u_{F,r},v_{A,r},v_{B,r},v_{C,r})\in(\mathbb F_3^*)^8,
\]
and
\[
B_r=
X_{r+O,1}^{t_{1,r}}
X_{r+O,2}^{-t_{2,r}}
X_{r+A,2}^{r_{A,r}}
X_{r+B,2}^{r_{B,r}}
X_{r+C,2}^{r_{C,r}}
X_{r+D,1}^{s_{D,r}}
X_{r+E,1}^{s_{E,r}}
X_{r+F,1}^{s_{F,r}},
\]
with
\[
\mathbf b_r=(t_{1,r},t_{2,r},r_{A,r},r_{B,r},r_{C,r},s_{D,r},s_{E,r},s_{F,r})\in(\mathbb F_3^*)^8.
\]

The Hamiltonian is
\[
H=-\sum_{r\in\Lambda_L}\left(A_r+A_r^\dagger+B_r+B_r^\dagger\right).
\]
Since qutrit stabilizers are order \(3\),
\[
\Pi(S)=\frac{I+S+S^2}{3},
\]
and the ground space is the common \(+1\) eigenspace of all \(A_r\) and \(B_r\) [2606.19873].

## 2. Admissibility, constrained randomness, and relation to Haah-type models

The code is not defined by arbitrary spatially varying coefficients. Admissibility requires both a global \(Z\)-check relation and local CSS commutation. On the torus,
\[
\prod_{r\in\Lambda_L}A_r=I.
\]
At the coefficient level this becomes the sitewise constraints
\[
w_{1,r}+u_{D,r+x}+u_{E,r+y}+u_{F,r+z}=0,
\]
\[
w_{2,r}+v_{A,r+y+z}+v_{B,r+x+z}+v_{C,r+x+y}=0,
\]
all in \(\mathbb F_3\) [2606.19873].

Local commutation requires every \(A_r\) to commute with every \(B_{r'}\). The construction introduces the ratios
\[
\alpha_D(r)=\frac{u_{D,r+y+z}}{w_{2,r+y+z}},\qquad
\alpha_E(r)=\frac{u_{E,r+x+z}}{w_{2,r+x+z}},\qquad
\alpha_F(r)=\frac{u_{F,r+x+y}}{w_{2,r+x+y}},
\]
\[
\beta_A(r)=\frac{v_{A,r+x}}{w_{1,r+x}},\qquad
\beta_B(r)=\frac{v_{B,r+y}}{w_{1,r+y}},\qquad
\beta_C(r)=\frac{v_{C,r+z}}{w_{1,r+z}},
\]
\[
\rho(r)=\frac{w_{1,r+x+y+z}}{w_{2,r+x+y+z}}.
\]
Once one chooses \(t_{1,r}\in\mathbb F_3^*\), the \(X\)-check coefficients are fixed by
\[
r_{A,r}=-\alpha_D(r)t_{1,r},\qquad
r_{B,r}=-\alpha_E(r)t_{1,r},\qquad
r_{C,r}=-\alpha_F(r)t_{1,r},
\]
\[
s_{D,r}=\rho(r)\beta_A(r)t_{1,r},\qquad
s_{E,r}=\rho(r)\beta_B(r)t_{1,r},\qquad
s_{F,r}=\rho(r)\beta_C(r)t_{1,r},
\]
\[
t_{2,r}=\rho(r)t_{1,r}.
\]
The remaining commutation equations reduce to constraints purely on the \(Z\)-field. A defining feature of the model is therefore that the randomness is highly constrained: one chooses an admissible \(Z\)-field, and the \(X\)-field is then locally determined [2606.19873].

The family is best understood as a qutrit, spatially nonuniform analog of Haah’s Code 1. The similarities are the same cube-check support geometry, the same CSS structure, and the same local pattern of a \(Z\)-check and reflected \(X\)-check. The differences are qutrits instead of qubits, coefficients in \(\mathbb F_3\) with nonzero values \(1\) and \(2\), and the breaking of translation invariance through spatially varying admissible coefficients [2606.19873].

A convenient uniform benchmark is the spatially constant qutrit version with
\[
w_1=w_2=u_D=v_A=1,\qquad u_E=u_F=v_B=v_C=2,
\]
and corresponding \(X\)-coefficients
\[
t_1=t_2=r_B=r_C=s_D=1,\qquad r_A=s_E=s_F=2.
\]
The data indicate that this uniform model behaves much more like standard Haah-type fracton codes, whereas constrained randomness alters both the logical-operator structure and the finite-size degeneracy [2606.19873].

## 3. No-string theorem and local proof structure

A central rigorous result is that every locally admissible qutrit random cubic code has no fixed-width logical strings. In Haah’s sense, a logical string segment consists of a finite Pauli operator \(P\) together with two congruent anchor cubes \(\Omega_1,\Omega_2\), such that all stabilizer checks away from the anchors commute with \(P\). The segment is nontrivial if every equivalent representative remains connected and trivial if stabilizer multiplication can disconnect it [2606.19873].

The theorem states that for any locally admissible coefficient field on \(\mathbb Z^3\),
\[
\phi(w)<\infty\qquad\text{for every finite }w,
\]
with the explicit conservative bound
\[
\phi(w)\le 27(w+4)+48.
\]
On the periodic torus, the same holds for string segments contained in a contractible no-wraparound region. The theorem does not rule out noncontractible plane-supported logical operators [2606.19873].

The proof is a qutrit-weighted extension of Haah’s original argument for Code 1 and uses three local mechanisms: corner erasing, good-edge erasing, and a confusing constraint on exposed edges. For corner erasing, a corner with local \(Z\)-vector \(z\in\mathbb F_3^2\) imposes
\[
z\cdot \xi(p)=0
\]
on a local \(X\)-exponent \(\xi(p)=(a,b)\). For good-edge erasing, the local commutation condition on an \(X\)-type operator across a good edge is
\[
\alpha b+\beta c=0
\]
for nonzero \(\alpha,\beta\in\mathbb F_3^*\). After flattening to thickness-one rectangles, the exposed edge obeys the triangular recurrence
\[
\alpha_j b_j+\beta_j a_{j+1}=0,\qquad \gamma_j b_{j+1}=0,
\]
with all coefficients nonzero, implying
\[
b_{j+1}=0,\qquad a_{j+2}=0,\qquad b_{j+2}=0.
\]
Iterating this row by row creates an empty interval that disconnects the support [2606.19873].

The significance of the theorem is structural rather than merely finite-size. The proof is purely local and deterministic: it does not use translation symmetry, the \([111]\) line symmetry used in numerics, or statistical randomness. It uses only the cube support geometry, the reflected CSS structure, the local commutation equations, and the requirement that all coefficients are nonzero in \(\mathbb F_3\) [2606.19873].

## 4. Degeneracy, membrane logicals, and charge-push diagnostics

Finite-size numerics were carried out for \(10\le L\le 25\) in a subfamily of admissible fields invariant along the \([111]\) direction,
\[
\mathbf a_{r+\ell(x+y+z)}=\mathbf a_r.
\]
The ground-state degeneracy is summarized by
\[
k=\log_3\mathrm{GSD}.
\]
Among sampled admissible models, the smallest observed degeneracy exponent follows the parity rule
\[
k=
\begin{cases}
2,& L\ \text{odd},\\
4,& L\ \text{even}.
\end{cases}
\]
The retained models satisfy
\[
\rank(H_Z)=\rank(H_X)=L^3-\frac{k}{2}.
\]
By contrast, the translation-invariant benchmark exhibits strong arithmetic finite-size effects, including
\[
k=54\quad\text{at }L=18.
\]
This contrast is one of the clearest numerical distinctions between the random and uniform qutrit cubic models [2606.19873].

A second numerical result is that noncontractible plane-supported logical operators span the entire logical space in all curated random samples studied. For a plane support \(S\), the logical content in the \(X\)-sector is
\[
k_X(S)=\rank\begin{pmatrix} H_X\\ P_X(S)\end{pmatrix}-\rank(H_X),
\]
and similarly in the \(Z\)-sector,
\[
k_Z(S)=\rank\begin{pmatrix} H_Z\\ P_Z(S)\end{pmatrix}-\rank(H_Z).
\]
The reported outcome is
\[
k_X=k_Z=k
\]
for all \(374\) curated samples. The smallest plane-supported logical operator weights obey
\[
w_X^{\min}(L)=w_Z^{\min}(L)=
\begin{cases}
L^2/2,& L\ \text{even},\\
L^2,& L\ \text{odd}.
\end{cases}
\]
The paper explicitly notes that this is not a full code-distance proof; it is a minimum over plane-supported representatives rather than a global optimization over all possible Pauli supports [2606.19873].

The same study searched for axis-aligned tube logical operators with support \(w\times w\times L\). The diagnostic
\[
d_X(S)=|S|-\rank H_Z[:,S]-\rank(H_X)+\rank H_X[:,C],
\]
and the analogous \(d_Z(S)\), vanish in all scanned cases:
\[
d_X(S)=d_Z(S)=0.
\]
This strengthens the conclusion that the logical operators observed numerically are membrane-like rather than wide-string-like [2606.19873].

Charge-push diagnostics were used to probe the presence or absence of self-similar fractal mechanisms. Starting from a single local \(XI\) operator, one repeatedly cancels moving charges using only \(XI\) or \(X^2I\) moves, with update exponents
\[
t=-q/w_{1,r}
\]
for the first-qutrit channel and
\[
t=-q/w_{2,r}
\]
for the second-qutrit channel. In the translation-invariant qutrit Haah reference, the process shows exact power-of-three recurrence,
\[
N(1)=N(3)=N(9)=N(27)=4.
\]
For representative random models, this recurrence disappears. At layer \(9\), the charge count is distributed between \(7\) and \(12\), with mean
\[
\langle N(9)\rangle=9.28,
\]
and no sample returns to the four-charge pattern. At layer \(27\),
\[
N^{\rm avg}(27)\approx 17.84,\qquad N^{\rm min}(27)=14,
\]
whereas the uniform model again returns to \(4\) [2606.19873].

These numerical results do not prove the absence of every possible irregular fractal support. They do, however, provide the paper’s main evidence that the canonical self-similar fractal mechanism of the translation-invariant Haah-type model is absent in the representative random qutrit codes that were studied [2606.19873].

## 5. Distillation-oriented random cubic qutrit codes

In a distinct but related line of work, the phrase “qutrit random cubic codes” is used for random-looking large qutrit stabilizer or CSS codes whose magic-state distillation map suppresses noise to third order for the qutrit strange state [2408.00436]. The motivating question is whether distillation of a contextual qutrit state becomes more generic as code length grows, and whether thresholds can approach the contextuality boundary.

The target magic state is the qutrit strange state
\[
\lvert S\rangle=\frac{1}{\sqrt{2}}(\lvert 1\rangle-\lvert 2\rangle).
\]
After Clifford twirling, noisy copies are described by
\[
\hat{\rho}_S(\epsilon)=(1-\epsilon)\lvert S\rangle\langle S\rvert+\epsilon \frac{1}{3},
\]
and the state is contextual for \(\epsilon<3/4\). The paper treats the strange state as a canonical test case for the question of whether contextuality is sufficient for universal qutrit quantum computation [2408.00436].

A central theorem links projection success onto a stabilizer code to the complete weight enumerator of the dual code. For a single-qudit mixed state with Wigner function \(W(\hat \rho;x,z)\), the probability for successful projection onto the trivial-syndrome eigenspace of \(\mathcal S\) is
\[
\left(\hat{\Pi}^0_{\mathcal S}\rho^n\right)
=
w(\mathcal S^\perp;\{W(\hat \rho;i,j)\})
=
\sum_{(\vec u|\vec v)\in\mathcal S^\perp}\prod_{i=1}^n W(\hat \rho;u_i,v_i).
\]
For the strange state, the Wigner function takes only two values, reducing the complete weight enumerator to a simple weight enumerator. The output noise parameter obeys
\[
\epsilon' = 3\frac{3A(z(\epsilon))+B(z(\epsilon))}{4B(z(\epsilon))},
\qquad
z(\epsilon)=\frac{3-\epsilon}{8\epsilon-6},
\]
with
\[
A(z)=w(\mathcal S;1,z),\qquad B(z)=w(\mathcal S^\perp;1,z).
\]
Cubic suppression means
\[
\epsilon'=O(\epsilon^3)
\]
for small \(\epsilon\) [2408.00436].

The search covered all \( [[n,1]]_3 \) stabilizer codes for \( n\le 9 \), all \( [[11,1]]_3 \) codes obtainable by shortening a \( [[12,0,6]]_3 \) stabilizer state, and odd-length CSS codes \( [[n,1]]_3 \) with a complete set of transversal Clifford gates for odd \(n\le 23\). The CSS search used maximal self-orthogonal ternary codes \([n,\lfloor n/2\rfloor]_3\) via
\[
H=\left(\begin{array}{c|c}
G_c & 0\\
0 & G_c
\end{array}\right).
\]
The restriction to codes with transversal \((\hat H^2)^{\otimes n}\) allows one to search only the trivial syndrome sector [2408.00436].

The main empirical findings are sharply length dependent. For all searched codes with \(n<23\), no better than linear suppression was found except for the 11-qutrit Golay code. For \(n=23\), the study found over \(600\) CSS codes with cubic noise suppression; more precisely, \(646\) inequivalent indecomposable \(23\)-qutrit CSS codes. Their thresholds range from about \(0.063\) to \(0.318\), and none surpass the Golay threshold. The 11-qutrit Golay benchmark is an \( [[11,1,5]]_3 \) CSS code with
\[
A(z)=1+528 z^6+7920 z^8+11000 z^9+23760 z^{10}+15840 z^{11},
\]
for which
\[
\epsilon'\approx \frac{55}{18}\epsilon^3+O(\epsilon^4),\qquad \epsilon_*\approx 0.387.
\]
The highest-threshold new \(23\)-qutrit code is a \( [[23,1,5]]_3 \) code with
\[
\epsilon'\approx \frac{73}{18}\epsilon^3+O(\epsilon^4).
\]
This suggests that cubic distillation of the strange state may be somewhat generic for sufficiently large structured qutrit CSS codes, while still leaving open whether any family approaches the contextuality boundary \(\epsilon=3/4\) [2408.00436].

## 6. Terminology, adjacent constructions, and unresolved questions

A recurrent ambiguity is the word “cubic,” which does not have a single meaning across the relevant literature.

| Usage | Meaning | Source |
|---|---|---|
| Qutrit random cubic codes | Cube-check 3D local qutrit CSS stabilizer Hamiltonians with spatially varying admissible coefficients | [2606.19873] |
| Random cubic codes in distillation | Random-looking large qutrit stabilizer/CSS codes with cubic noise suppression for \(\lvert S\rangle\) | [2408.00436] |
| Ternary cubic codes | Quasi-cyclic ternary codes of co-index \(3\) arising as Gray images | [1612.00914] |

In coding theory, “cubic code” can mean a quasi-cyclic code of co-index \(3\). A representative construction works over
\[
R=\mathbb F_3+u\mathbb F_3+u^2\mathbb F_3,\qquad u^3=1,
\]
with extension
\[
\mathcal R=\mathbb F_{3^m}+u\mathbb F_{3^m}+u^2\mathbb F_{3^m},
\]
trace map \(\operatorname{Tr}\), evaluation code
\[
C(m)=\{\operatorname{ev}(a):a\in\mathcal R\},
\]
and Gray map
\[
\phi(a+ub+u^2c)=(a,b,c).
\]
The ternary image \(\phi(C(m))\) is quasi-cyclic of co-index \(3\), and the paper derives three-weight and two-weight families, proves optimality in the two-weight cases via the ternary Griesmer bound, and analyzes dual distance and secret sharing consequences [1612.00914]. This is a different notion of “cubic” from the cube-check Hamiltonian usage.

A separate but adjacent strand constructs qutrit stabilizer codes from quasi-cyclic codes over \(\mathbb F_9\) with large Hermitian hulls, using quantum Construction X. For an \([n,k]_{q^2}\)-linear code \(C\),
\[
e:=k-\dim(C\cap C^{\perp H}),
\]
and there exists an
\[
[n+e,\; n-2k+e,\; d(Q)]_q
\]
quantum stabilizer code with
\[
d(Q)\ge \min\{d(C^{\perp H}),\; d(C+C^{\perp H})+1\}.
\]
In the qutrit case, this systematic construction yields many best-known codes and an optimal
\[
[17,7,5]_3
\]
code improving the previous \([17,7,4]_3\) [1906.04964]. This literature supplies qutrit stabilizer-code context but does not study qutrit random cubic codes in the cube-check sense.

The nearest antecedent on the Hamiltonian side is the study of 3D local qupit stabilizer codes without string logical operators. That work considers prime local dimension \(d=p\), one stabilizer per cube, and gives a sufficient algebraic condition for the absence of string logical operators. The abstract states that the minimal prime dimension satisfying the sufficient condition is \(5\), while for \(d=3\) the paper reports numerical indications that the maximum string length may still be bounded, without a proof [1202.0052]. In this lineage, the qutrit random cubic codes are notable because they establish a fully qutrit, cube-check, no-string family with spatially varying stabilizers [2606.19873].

Several unresolved questions remain explicit in the literature. For the three-dimensional Hamiltonian family, the plane-logical and charge-push data do not constitute a full code-distance proof, and the absence of every possible irregular fractal support is not established [2606.19873]. For the distillation-oriented family, the existence of many \(23\)-qutrit cubic-suppressing CSS codes does not show that thresholds can approach the contextuality boundary \(\epsilon=3/4\), nor does it establish that all contextual qutrit states are distillable [2408.00436]. Taken together, these results suggest broader classes of spatially nonuniform stabilizer phases and qutrit code families than those captured by the standard translation-invariant topological or fracton templates, but they stop short of a complete classification.

Source: https://www.emergentmind.com/topics/qutrit-random-cubic-codes