In simulations from the University of Ulm, entanglement seemingly disappears entirely under the action of decoherence in the GRW and CSL collapse models, yet it remains hidden and can be recovered with the help of local quantum error correction. Even when the original signal drops to zero, the ratio of gravitational interaction to decoherence can be made arbitrarily large.
Researchers Yi-Xuan Wang, Julen S. Pedernales and Martin B. Plenio from the Institute of Theoretical Physics at the University of Ulm (Germany) carried out the analysis in a preprint published on 4 November 2026 on arXiv. They considered Newtonian gravity in combination with the noise of the GRW, CSL and Diósi–Penrose collapse models. The team applied a strictly local quantum error correction protocol that suppresses decoherence while preserving the gravitational interaction.
The key mechanism: in the GRW and CSL models, localization saturates at large length scales, whereas the Newtonian interaction decays more slowly. This allows a code with sufficient distance to make the effective dynamics entangling. By contrast, the Diósi–Penrose model does not exhibit such an enhancement under the same protocol.
Entanglement here is like a letter written in invisible ink: decoherence does not erase the text but merely hides it beneath a layer of noise. Local error correction acts as a developer, making the entanglement dynamics observable without creating entanglement from scratch.
The result changes the notion of entanglement as a resource: the loss of a detectable signal does not mean the loss of the resource itself. This opens the way to protocols for recovering gravitationally induced entanglement even with extremely weak signals.
The work, according to arXiv preprint 2611.05044, emphasizes that recoverability depends on the specific collapse mechanism. For GRW and CSL, the ratio η can be made arbitrarily large as the code distance grows, whereas for the DP model no such enhancement occurs.
Thus, hidden entanglement in these collapse models remains an accessible resource that can be brought back to observation with suitable correction.