Projective error models: Stabilizer codes, Clifford codes, and weak stabilizer codes

1. Problem Statement

Quantum error correction (QEC) aims to protect quantum information from noise by encoding it into a specific subspace (code space) of a Hilbert space. The central challenges are:

1. Identifying which noise models a given code space can correct.
2. Determining the largest code space capable of correcting a specific noise model.

While the Knill-Laflamme conditions provide a general criterion for error correction, they are often difficult to apply directly to specific algebraic structures. Standard approaches focus heavily on Stabilizer Codes (based on abelian subgroups of the Pauli group). However, broader classes of codes, such as Clifford Codes (introduced by Knill), exist but lack a unified mathematical framework connecting them to stabilizer codes and general error models.

The paper addresses the need for a rigorous mathematical framework using projective representation theory to:

• Define and analyze Projective Error Models.
• Systematically compare three classes of codes: Stabilizer Codes, Clifford Codes, and a newly introduced class called Weak Stabilizer Codes.
• Determine the obstructions that prevent a code from being a stabilizer code and characterize when Clifford codes coincide with weak stabilizer codes.

2. Methodology

The author employs tools from finite group theory and projective representation theory to model quantum errors and code spaces.

• Projective Error Models: Instead of modeling errors as the full unitary group, the paper defines an error model as a pair $(G, \pi)$, where $G$ is a finite group and $\pi$ is a projectively faithful irreducible projective representation on a Hilbert space $V$. This generalizes the concept of "nice error bases."
• Detectable Errors: An error $x \in G$ is "detectable" with respect to a code space $W$ if the projection $P_W \pi(x) P_W$ is a scalar multiple of $P_W$. The set of detectable errors is denoted $D(G, \pi)(W)$.
• Code Definitions:
  • Stabilizer Codes: Defined by a normal subgroup $N \trianglelefteq G$ and a character $f: N \to \mathbb{T}$, where the code space is the common eigenspace of $\pi(N)$.
  • Weak Stabilizer Codes: A generalization where the subgroup $H$ need not be normal.
  • Clifford Codes: Defined via Clifford theory (induction/restriction). A code $W$ is a Clifford code if it is the image of an irreducible representation $\rho$ of a subgroup $L \leq G$ such that the induced representation $\text{Ind}_L^G \rho$ is isomorphic to the global error model $\pi$.
• Group Cohomology: The paper utilizes the second cohomology group $H^2(G, \mathbb{T})$ to analyze the obstructions (2-cocycles) that determine whether a projective representation can be "linearized" or reduced to a stabilizer structure.

3. Key Contributions

A. Introduction of Weak Stabilizer Codes

The author introduces Weak Stabilizer Codes, defined by a subgroup $H$ (not necessarily normal) and a function $f$. This class strictly contains stabilizer codes and captures examples like XP-stabilizer codes that were previously ambiguous in the standard framework.

B. Characterization of Detectable Errors

The paper establishes a precise relationship between the detectable errors of a code and its algebraic structure:

• Logical Operators ($L$): The subgroup of $G$ that preserves the code space $W$.
• Stabilizers ($S$): The subgroup of $G$ that acts as scalars on $W$.
• Theorem 6.3: For a Clifford code $W$, the set of detectable errors is exactly $D = (G \setminus L) \cup S$. This proves that the structure of a Clifford code is completely determined by its group of logical operators.

C. Obstructions and Cohomology

The paper identifies the obstruction for a code to be non-trivial or to be a stabilizer code using group cohomology:

• For a stabilizer code to exist, the restriction of the 2-cocycle $\sigma$ to the stabilizer subgroup must be a coboundary (i.e., trivial in cohomology).
• The paper derives similar cohomological obstructions for weak stabilizer codes.

D. Characterization for Groups of Central Type

For "nice error bases" (where $|G| = (\dim V)^2$, i.e., groups of central type), the author provides a complete characterization:

• Theorem 7.3: A Clifford code $W$ is a weak stabilizer code if and only if $|G| = |L(W)| \cdot |S(W)|$.
• Corollary 7.5: A Clifford code is a stabilizer code if and only if $S(W)$ is a normal subgroup of $G$ and the size condition $|G| = |L(W)| \cdot |S(W)|$ holds.

E. Construction of Counter-Examples

The paper constructs infinite families of codes that demonstrate the strict inclusions between the classes:

1. Non-Stabilizer Clifford Codes: Using groups like $C_2 \times D_{2n}$ and semi-direct products involving odd $n$, the author constructs Clifford codes that are not stabilizer codes.
2. Non-Clifford Weak Stabilizer Codes: Using permutation-invariant codes (e.g., Dicke states), the author constructs weak stabilizer codes that are not Clifford codes.
3. Product Construction: A method is provided to combine non-stabilizer Clifford codes to generate new, larger families of non-stabilizer Clifford codes.

4. Key Results

• Hierarchy of Codes: The paper clarifies the relationship:
  $$\{ \text{Stabilizer Codes} \} \subsetneq \{ \text{Clifford Codes} \}$$
  $$\{ \text{Stabilizer Codes} \} \subsetneq \{ \text{Weak Stabilizer Codes} \}$$
  However, the intersection of Clifford and Weak Stabilizer codes is not necessarily the set of Stabilizer codes (though no counter-example was found, the question remains open).
• Dimension Formulas:
  • For a Clifford code: $\dim W = \frac{\dim V}{|G|} |L(W)|$.
  • For a weak stabilizer code (in central type models): $\dim W = \frac{\dim V}{|S(W)|}$.
• Detectability: The set of detectable errors for a Clifford code is explicitly given by the complement of the logical group union the stabilizer group. This allows for a direct calculation of error-correcting capabilities based on group sizes.

5. Significance

• Unification: The paper unifies the study of various quantum error-correcting codes under the umbrella of projective representation theory, moving beyond the limitations of the Pauli group and abelian stabilizers.
• Theoretical Clarity: It resolves ambiguities regarding "nice error bases" and "groups of central type" by providing rigorous cohomological criteria for when a code is a stabilizer code.
• New Code Families: By proving the existence of infinite families of non-stabilizer Clifford codes, the paper expands the landscape of known quantum codes. This suggests that there are potentially more efficient or robust codes outside the traditional stabilizer formalism.
• Practical Implications: The characterization of detectable errors in terms of logical and stabilizer groups provides a practical tool for engineers and physicists to design codes with specific error-correcting properties without needing to check the Knill-Laflamme conditions for every individual error.
• Future Directions: The paper opens new research avenues, specifically regarding the intersection of Clifford and weak stabilizer codes and the extension of these results to subsystem codes and code distance metrics.

In summary, Eidesen's work provides a robust algebraic framework that generalizes stabilizer codes, introduces a new class of weak stabilizer codes, and uses group cohomology to precisely map the boundaries and relationships between these fundamental classes of quantum error-correcting codes.