Skip to content

Bounded Markov Order Nonunital Extension Candidate

Claim/Theorem

Current proof candidate for the minimal Conjecture-1 target: restrict to a depth-independent process tensor of bounded Markov order relative to the intervention class generated by the local random circuit ensemble. Represent the noise by a bounded-memory process tensor on a sliding time window, or equivalently by a finite memory register coupled to the system over consecutive layers. Then attempt to adapt the PRX Quantum 2024 Markovian nonunital moment argument to the enlarged system-plus-memory evolution. If the nonunital contraction survives as a depth-uniform bias on this enlarged transfer operator, one expects a second-moment lower bound of the form

\[ \mathbb E_{U}\!\left[q^n \sum_x p_{U,\Upsilon}^{(D)}(x)^2\right] \ge 1 + c_{\mathrm{ac}} \]

with a constant depending on the memory parameters but not on \(D\).

Dependencies

  • [[markovian-nonunital-anticoncentration-obstruction.md]]
  • [[process-tensor-framework.md]]
  • [[instrument-specific-finite-quantum-markov-order.md]]
  • [[unital-noisy-random-circuit-uniformization.md]]

Conflicts/Gaps

  • The missing step is a rigorous transfer from an instrument-specific bounded-memory description to the uniform-over-histories formulation used in Conjecture 1.
  • It is not yet proved that the enlarged memory register can be chosen with bounded effective dimension under the exact assumptions of the conjecture, rather than under a more model-specific collision or hidden-Markov construction.
  • The current cycle does not yet bridge spatial locality of the process tensor to the moment calculation; it only identifies the most promising temporal-memory reduction.

Sources

  • 10.1103/PRXQuantum.5.030317
  • 10.1103/PhysRevA.97.012127
  • 10.1103/PhysRevLett.122.140401
  • 10.1103/PhysRevA.99.042108
  • 10.1103/PRXQuantum.3.040329