Skip to content

Recovery Maps and Approximate Markovianity

Small conditional mutual information guarantees that a tripartite state can be recovered approximately from its ABAB marginal by acting on BB. The theorem is quantitative and state dependent; it does not guarantee that the recovering channel is unique, geometrically local, causal, energy feasible, or easy to implement.

Required background. Use Conditional Mutual Information and Quantum Markov Structure. Helpful background. Markov generators and semigroups distinguish dynamical memory loss, while data processing supplies the information-loss perspective.

For a finite-dimensional tripartite state, Fawzi and Renner proved that there exists a channel RBBC\mathcal R_{B\to BC} satisfying

I(A:CB)ρ2logF ⁣(ρABC,(idAR)(ρAB)),I(A{:}C\mid B)_\rho \geq -2\log F\!\left( \rho_{ABC}, ({\rm id}_A\otimes\mathcal R)(\rho_{AB}) \right),

where FF is the unsquared fidelity. Thus IδI\leq\delta guarantees a recovered state with Feδ/2F\geq e^{-\delta/2} for natural logarithms. See Fawzi and Renner 2015, pp. 575–611.

The structural figure shows the one-way logic: correlation loss bounds attainable recovery, while choosing a concrete map is a further step.

Small conditional mutual information points from correlation structure to an approximate recovery channel, subject to fidelity, domain, and resource choices.

Recovery translates small conditional mutual information into closeness of states. The bound asserts existence on a specified algebra; it does not supply spacetime locality, energy control, uniqueness, or an implementation cost. Schematic.

Take three neighboring blocks in a finite harmonic chain. Compute I(A:CB)I(A{:}C\mid B) from one covariance matrix and construct a Gaussian channel from BB to BCBC. Compare the recovered and exact covariance matrices and then their fidelity. This tests a restricted Gaussian recovery family; the theorem’s optimal channel may be non-Gaussian.

To claim a QFT limit, control the continuum embeddings and the energy carried by the recovered modes. A channel that reproduces ultraviolet correlations using unbounded squeezing can satisfy a formal fidelity target at each cutoff yet have no finite-energy limiting implementation.

Causal support is separate. An abstract operation on the algebra BB need not be generated inside the spacetime domain one intends to call local. The admissible operation algebra and time window must therefore be stated in an operational claim.

The validity map separates the theorem’s hypotheses from common overinterpretations.

Recovery theorems fix the tripartite algebra, support, channel, fidelity, and resource set; causal support or energy feasibility does not follow automatically.

Changing the recovery domain, using postselection, ignoring rank support, or removing the energy bound changes the guarantee. Abstract recoverability is weaker than a local laboratory protocol. Schematic.

Report the recovered marginal, channel domain and codomain, fidelity convention, error, regulator, energy behavior, and any spacetime locality requirement. Without those data, “approximately Markov” is incomplete.

  • Fawzi, Omar, and Renato Renner. “Quantum Conditional Mutual Information and Approximate Markov Chains.” Communications in Mathematical Physics 340 (2015): 575–611. DOI. Open preprint.
  • Sutter, David, Omar Fawzi, and Renato Renner. “Universal Recovery Map for Approximate Markov Chains.” Proceedings of the Royal Society A 472 (2016): 20150623. DOI. Open preprint.