Causal inequalities witness non-stabilizerness

Technical Summary: Causal Inequalities Witness Non-Stabilizerness

Problem Statement
The paper addresses a fundamental discrepancy within the resource theory of "magic" (non-stabilizerness), which is essential for achieving universal quantum computation beyond the classically simulable stabilizer fragment (Gottesman-Knill theorem). Specifically, it investigates the separation between two definitions of free operations:

1. Stabilizer Operations (SO): Operations constructible from Clifford gates, stabilizer state preparation, and adaptive Pauli measurements.
2. Completely Stabilizer Preserving Operations (CSPO): Channels that preserve the stabilizer polytope even in the presence of an ancilla.

While it is established that $SO \subsetneq CSPO$, the operational implications of this separation were not fully characterized. A key phenomenon, termed "Non-Stabilizerness Without Magic" (NSWM), was identified in Ref. [10]: certain ensembles of stabilizer states (specifically the SHIFT ensemble) cannot be perfectly discriminated using SO, despite being preparable via SO and perfectly discriminable via CSPO. The paper seeks to provide a principled understanding of NSWM, specifically determining necessary and sufficient conditions for its existence and exploring its relationship to causal structures.

Methodology
The authors analyze the problem of state discrimination within the stabilizer subtheory for $n$-qudit systems of prime dimension $d$. Their approach involves:

1. Formalizing Discrimination Protocols: They define adaptive stabilizer discrimination protocols as sequences of Pauli measurements and Clifford operations. They prove (Lemma 1) that the use of ancillary qudits provides no advantage for discrimination within this framework, allowing the analysis to focus on ancilla-free protocols.
2. Recursive Criterion Derivation: They introduce the stabilizer subgroup $S_B$ associated with a stabilizer basis $B$, defined as the intersection of the stabilizer groups of all states in the basis. They establish that $S_B$ contains all deterministic measurements. Using this, they derive a recursive condition (Theorem 1) to determine if a basis is perfectly discriminable.
3. Linking to Process Functions: For the specific case of Stabilizer Product Bases (SPBs), the authors leverage the known correspondence between unambiguous product bases and "process functions." A process function $\omega$ is a classical model describing how inputs are determined by outputs in a causal loop. They map the structure of an SPB to a unique process function, where the local bases of the qudits correspond to the functional dependence of the process.
4. Causal Analysis: They analyze the causal properties of these process functions. A process is "causal" if there exists a party in the global past of all others (a constant input component). If no such party exists, the process is "noncausal."

Key Contributions and Results

• Theorem 1 (General Criterion): The authors prove a necessary and sufficient condition for a stabilizer basis $B$ to be perfectly discriminable using SO. A basis is perfectly discriminable if and only if $S_B \neq \{1\}$ and, for every joint eigenvalue $\chi$ of the generators of $S_B$, the resulting sub-basis $B_\chi$ is also perfectly discriminable.

  • Corollary: If $S_B = \{1\}$, the basis exhibits NSWM. This explains the NSWM of the SHIFT ensemble, as its associated subgroup is trivial.

• Theorem 2 (Product Bases and Causality): Specializing to Stabilizer Product Bases (SPBs), the paper proves that an SPB is perfectly discriminable via SO if and only if its associated process function is causal.

  • Conversely, an SPB exhibits NSWM if and only if its associated process function is noncausal.

• Corollary 2 (Causal Inequalities): Since it is established that every noncausal process function violates a causal inequality, the authors conclude that an SPB exhibits NSWM if and only if its associated process function violates a causal inequality.

Significance and Claims
The paper claims to provide a new operational meaning to the violation of causal inequalities. Specifically, it establishes causal inequality violations as witnesses of non-stabilizerness, a form of computational nonclassicality.

The authors frame this as a trade-off between causal order and non-stabilizerness: the inability to perfectly discriminate a stabilizer product basis using only stabilizer operations (NSWM) is inextricably linked to the noncausal nature of the process function describing that basis. By allowing communication via noncausal process functions (as modeled in the process-matrix framework), one can implement the separable measurements required to perfectly discriminate these bases using local operations alone.

The work generalizes previous observations of NSWM (specifically the SHIFT ensemble) to arbitrary prime dimensions and provides a rigorous, necessary, and sufficient characterization of the phenomenon. The authors note that while the results are derived for product bases, the recursive criterion (Theorem 1) applies to general stabilizer bases, including those with entangled states. They leave open the question of whether NSWM entangled ensembles exist that are not Clifford-equivalent to product bases, noting that such a case would require relaxing assumptions of logical consistency in the causal interpretation.