Correlation Extraction versus Causal Exchange

Final probe correlations can come from pre-existing field correlations, causal exchange, or common apparatus noise. To claim correlation extraction, a protocol must separate these mechanisms using the field commutator and anticommutator, spacetime controls, and intervention tests; the final two-probe density matrix alone does not identify the source.

Required background. Field communication defines the induced channel, and entanglement harvesting defines the spacelike extraction claim.

Helpful background. Signaling and causal composition supplies the intervention test for causal influence.

Statistical and causal kernels

For a Hermitian scalar field, decompose the two-point function as

$$W(x,x')=\frac12H(x,x')+\frac{i}{2}\Delta(x,x'),$$

where

$$H(x,x')=\langle\{\phi(x),\phi(x')\}\rangle, \qquad i\Delta(x,x')=\langle[\phi(x),\phi(x')]\rangle.$$

$H$ carries state-dependent symmetrized correlations. $\Delta$ controls linear causal response and vanishes at spacelike separation. Detector cross terms can contain both; their precise coefficients depend on time ordering and switching.

Correlation extraction and causal signaling can share the same detector hardware. Their distinction lies in support geometry, intervention dependence, and which field kernels contribute. The diagram is schematic.

Matched spacelike and timelike geometries

Use identical detector gaps, local switching profiles, smearing, and proper durations in two arrangements. In the spacelike arrangement, $\Delta(f_A,f_B)=0$ for compact supports, while $H(f_A,f_B)$ can remain nonzero. In the timelike arrangement, both can contribute. Compute the joint detector statistics and the signaling contrast under a deliberate change of $A$‘s coupling. Pozas-Kerstjens and Martín-Martínez 2015, §§ II–IV provide a representative perturbative harvesting calculation with finite switching and separation.

The comparison separates three statements:

• nonzero $H$ with zero intervention contrast supports pre-existing correlation transfer;
• nonzero $\Delta$ and receiver dependence on $A$ supports causal exchange;
• correlations present even when both detector–field couplings are off indicate common apparatus preparation or analysis bias.

An operator split into “vacuum fluctuations” and “radiation reaction” can depend on ordering conventions, but the spacelike commutator null and the intervention contrast are operational controls.

Controlled removal of initial correlations

In a regulated lattice or mode model, construct a comparison field state whose local covariance blocks match the target but whose $A$–$B$ cross block is removed subject to positivity. Retain the same retarded dynamics. If timelike probe correlation remains, it can be produced by causal propagation; if strictly spacelike entanglement vanishes, that supports an extraction interpretation for the target state.

This control is model-specific: a modified covariance may not define the same continuum state class, and higher-order exchange can re-enter when supports are not strictly spacelike. Tjoa and Martín-Martínez 2021, §§ II–IV emphasize that detector entanglement generated through a field must not automatically be identified with harvested pre-existing entanglement.

The three columns are distinguished by controls: remove initial cross correlations, move supports across the causal boundary, and vary a sender intervention. No single final-state correlation performs all three tests. The map is schematic.

Higher-order check

At each perturbative order, list paths that connect $A$ and $B$. A vanishing leading commutator term does not exclude a higher-order path through an overlapping switching tail or a shared ancillary mode. Bound those remainders relative to the observed witness. If the bound is comparable to the effect, describe the result as field-mediated probe correlation rather than demonstrated harvesting.

References

• Pozas-Kerstjens, A., and Martín-Martínez, E. (2015). “Harvesting Correlations from the Quantum Vacuum.” Physical Review D 92, 064042. DOI. Open PDF.
• Tjoa, E., and Martín-Martínez, E. (2021). “When Entanglement Harvesting Is Not Really Harvesting.” Physical Review D 104, 125005. DOI. Open PDF.