Certified coherent, informative, and non-entanglement-breaking fixed points of future-referential quantum feedback

Technical Summary: Certified Coherent, Informative, and Non-Entanglement-Breaking Fixed Points of Future-Referential Quantum Feedback

Problem Statement
The paper investigates quantum processes where information extracted from a forward simulation is returned as input to an earlier internal time of the simulated dynamics. While externally the protocol operates as an ordinary, causally ordered quantum circuit, internally it is "future-referential." The central challenge is to determine which multi-time processes admit a fixed point that is simultaneously:

1. Stable: Attractive under repeated application.
2. Informative: Carrying positive quantum mutual information about a designated future variable.
3. Feedable: Generated by a trace-preserving instrument and reinsertable compositionally.
4. Genuinely Quantum: Preserving input entanglement (non-entanglement-breaking), rather than merely preparing a coherent state via a measure-and-prepare (entanglement-breaking) channel.

The work distinguishes this framework from closed timelike curve (CTC) models (e.g., Deutsch's nonlinear formulation) by maintaining linearity and avoiding hidden postselection, focusing instead on the classification of fixed-point properties within standard quantum channel theory.

Methodology
The author employs a process tensor formalism to model the interaction between a simulated world ($W$), a message register ($M$), a future-event register ($F$), and a leakage probe ($L$). A single feedback round induces a completely positive trace-preserving (CPTP) map $\Phi$ on the message register. The study proceeds through a sequence of increasingly complex models:

1. Classical Baseline: A solvable classical anti-predictor model is analyzed to demonstrate that deterministic anti-compliance can lead to instability (two-cycles) or randomization, motivating the need for quantum coherence.
2. Unitary-Dilation Models: The author constructs a family of qubit channels using three qubits ($M, F, L$) and a unitary evolution involving a world unitary ($U_W$), a weak probe coupling ($U_{weak}$), a partial swap ($U_{fb}$), and a controller rotation.
3. Analytical and Numerical Certification:
   • Exact Solutions: A two-parameter family (full swap, $\theta=0$) is solved exactly to demonstrate coherent stationary outputs.
   • Perturbation Analysis: A four-parameter family (partial swap, $\theta > 0$) is analyzed using continuity arguments and rigorous interval arithmetic.
   • Machine-Verified Certificates: The author uses directed rounding and ball arithmetic to compute "outward-rounded working enclosures" for channel parameters, fixed points, and spectral properties. This ensures that all inequalities (stability margins, coherence, mutual information, and Choi negativity) hold strictly within certified bounds.

Key Contributions and Results

• Separation of Concepts (Propositions 1 & Lemma 1): The paper proves that a coherent stationary output does not imply the preservation of entanglement. In a model with perfect future records (orthogonal states in $F$), the induced channel is necessarily entanglement-breaking, even if the output state possesses coherence. This resolves a conflation often found in informal discussions of "information from the future."

• Existence of Non-Entanglement-Breaking Phases (Proposition 2): By relaxing the orthogonality of future records (introducing partial swaps, $\theta > 0$), the author establishes the existence of a non-empty open set of parameters where the channel is not entanglement-breaking, while remaining stable, coherent, and informative.

• Explicit Analytic Neighborhood (Proposition 3): An explicit neighborhood of the reference point in the parameter space is identified where the Choi state of the channel has a negative partial transpose (NPT), serving as a witness for non-entanglement-breaking behavior.

• Certified Coexistence (Proposition 4): The primary result is a rigorous certification of a specific parameter square (half-width $0.0013\pi$) where the feedback channel simultaneously satisfies all four operational criteria:

  • Strict Contraction: Stability margin $\ge 0.237$.
  • Coherence: Stationary coherence $C_{\ell_1} \ge 0.416$.
  • Informativeness: Quantum mutual information $I(F:L) \ge 0.172$ bits (evaluated at the stationary message).
  • Non-Entanglement-Breaking: NPT margin $\ge 0.188$.

  The author notes that direct numerical evaluation suggests these properties persist over a region an order of magnitude larger, indicating that the certified square is a conservative "proof of principle" rather than a physical phase boundary.

Significance and Claims
The paper claims to provide an operational framework for analyzing future-referential feedback without invoking retrocausality in external laboratory time. Its significance lies in:

1. Operational Classification: It rigorously separates the notions of coherent output, entanglement preservation, stability, and informativeness, showing they are distinct properties that require specific conditions to coexist.
2. Verification: It offers a machine-verified, end-to-end reproducible certificate for the coexistence of these quantum properties in a specific feedback loop, moving beyond theoretical existence proofs to explicit, certified parameter regions.
3. Limitations: The author explicitly states that the work does not establish retrocausality, nor does it claim the certified region is the maximal possible phase. The certified square is intentionally conservative, relying on generic perturbation bounds that overestimate local variations by factors of 2.5 to 5. The work serves as a "proof of principle" that such a regime exists and can be rigorously certified, rather than a complete map of the phase space.

The paper concludes by framing future-referential feedback as a concrete program for channel classification and fixed-point analysis, distinct from but related to CTC models, quantum causal models, and coherent quantum feedback control.