| name | quantum-tomography-retrodiction |
| description | Methodology connecting quantum state tomography and quantum retrodiction through the Petz recovery map. Shows Petz map is precisely the gradient update of log-likelihood in maximum-likelihood tomography. Includes noncommutative generalization for arbitrary quantum channels. Use for quantum state estimation, quantum error correction, quantum metrology, or measurement channel recovery. |
| metadata | {"arxiv_id":"2606.23777","published":"2026-06-22","authors":"Sebastian Murk, Ian Tan, Fabian Müller, Dominik Šafránek","tags":["quantum","tomography","retrodiction","statistics","estimation","petz-map"]} |
Quantum Tomography-Retrodiction Framework
Core Concept
Quantum state tomography and quantum retrodiction are manifestations of the same underlying principle: the Petz recovery map associated with a measurement channel is precisely the gradient update of the log-likelihood used in maximum-likelihood tomography.
Mathematical Framework
Petz Map as Gradient Update
Given a measurement channel E and measurement outcomes, the Petz recovery map P_E applied to the observed statistics equals the gradient ascent step on the log-likelihood function:
P_E(\rho) = argmax_\rho log L(\rho | data)
Key Properties
- Monotonic Likelihood: Repeated applications of the Petz map monotonically increase the likelihood
- Noncommutative Generalization: The Petz map generalizes from measurement channels to arbitrary quantum channels
- Unified View: Tomography (forward estimation) and retrodiction (backward inference) share the same mathematical structure
Usage Patterns
Pattern 1: Maximum-Likelihood Tomography via Petz Map
When performing quantum state tomography, use the Petz recovery map as an iterative optimization procedure:
- Define the measurement channel E from the POVM elements
- Initialize with a prior state (often maximally mixed)
- Apply Petz map iteratively: \rho_{n+1} = P_E(\rho_n)
- Convergence guaranteed by monotonic likelihood property
Pattern 2: Retrodiction for Error Correction
When a quantum error channel E has occurred:
- Model the error as a quantum channel E
- Construct the Petz recovery map P_E
- Apply P_E to the corrupted state for optimal recovery
- This generalizes beyond measurement channels to arbitrary noise
Pattern 3: Connection to Bayesian Inference
The Petz retrodiction map is the quantum analogue of Bayesian inference:
- Classical: P(A|B) = P(B|A)P(A)/P(B)
- Quantum: P_E(\rho) = \sqrt{\rho} E^*(\sigma^{-1}) \sqrt{\rho}
- Where \sigma = E(\rho) is the reference state
Methodology Steps
- Identify the channel: Determine the measurement or noise channel E
- Compute the Petz map: P_E(\cdot) = \sqrt{\rho} E^*(E(\rho)^{-1/2} (\cdot) E(\rho)^{-1/2}) \sqrt{\rho}
- Iterate for convergence: Repeated application monotonically improves likelihood
- Verify positivity: Petz map preserves complete positivity by construction
Pitfalls
Numerical Stability
- The Petz map requires matrix inversion (E(\rho)^{-1/2}) which can be numerically unstable
- Regularize with small identity term: (E(\rho) + \epsilon I)^{-1/2}
- Use SVD-based inversion for ill-conditioned channels
Channel Representation
- Ensure the channel E is completely positive and trace-preserving (CPTP)
- For measurement channels, verify the POVM elements sum to identity
- For general channels, verify Kraus representation validity
Activation
- quantum tomography, quantum retrodiction, Petz recovery map, maximum likelihood tomography, quantum state estimation, measurement channel recovery, quantum error correction, noncommutative Bayes, 量子态层析, 量子回溯推断