The universe of quantum information has a stubborn asymmetry. You can use pure, maximally entangled states to create noisier, less useful entangled states — but you cannot go backwards. The transformation comes with an entropy-like cost that cannot be recovered. For more than two decades, physicists have known that this irreversibility is not just a practical nuisance; it is woven into the mathematical structure of entanglement theory. Yet, how deep that irreversibility runs remained an open question, formally captured as a conjecture by two leading theorists in 2023. A preprint posted to arXiv on July 29, 2026 — whose author list is not yet public — now settles the matter with finality. [arXiv:2607.27195]
The Core Finding
The new work constructs specific families of mixed quantum states for which any attempt to restore reversibility between entanglement distillation and entanglement cost incurs an error that balloons exponentially with the number of state copies. The paper proves a strict separation between two exponential strong-converse quantities: the strong-converse distillable entanglement and the strong-converse entanglement cost. In other words, even if you allow the error tolerance to grow modestly — polynomially in the number of copies, rather than keeping it fixed — the gap persists, ruling out any scheme that could convert distilled entanglement back into its original pure form with reasonable resources.
“Some mixed entangled states require pure entanglement for their preparation, although no pure entanglement can be recovered from them by local operations and classical communication.”
Think of it like a currency-exchange booth that offers a perfect rate when buying foreign notes, but charges an exponentially worsening fee if you try to convert back. The preprint’s analysis, which relies on semidefinite programming lower bounds and an analytical family of antisymmetric states under a specific set of operations (completely PPT-preserving maps), confirms that the fee is not a bug: it is a feature of entanglement itself.
The State of the Field
Since the early 2000s, researchers including Fernando Brandão, Martin Plenio, and the Horodecki family established that entanglement transformations are irreversible even under the most general operations that cannot create entanglement — a result that set entanglement theory apart from classical thermodynamics, where reversibility is achievable in macroscopic limits. In 2023, Ludovico Lami (now at the University of Amsterdam) and Bartosz Regula (RIKEN) sharpened the picture in a Nature Physics paper, conjecturing that an even stronger form of irreversibility might exist: an exponential separation between distillation and cost that kicks in the moment one deviates from the single-copy regime. The new preprint resolves that conjecture affirmatively and goes further, demonstrating that the separation survives if you permit errors growing with the square of the copy number, a qualitatively stronger claim.
The work arrives at a time when the quantum computing industry is racing to build the first truly fault-tolerant machines. In Google’s 2024 Willow chip and IBM’s 2025 Heron revision, logical qubit count is climbing, and error correction codes such as the surface code depend intimately on prescribed amounts of entanglement. Understanding the fundamental limits of entanglement manipulation could reshape resource estimates and error thresholds for those platforms.
From Lab to Reality
For theorists, the preprint opens a new toolkit. The semidefinite-programming lower bound on the exponential strong-converse cost under non-entangling operations provides a concrete numerical handle that can be applied to other resource theories, including the theory of magic states, which powers non-Clifford gates in quantum error correction. For engineers, the result is a warning bell: any protocol that relies on finely balanced entanglement recycling — an idea sometimes floated to reduce overhead — must now factor in an unavoidable exponential penalty when attempting to reverse processes.
From an industrial viewpoint, the quantum error correction market, projected to reach $2 billion by 2035 according to BCG, hinges on reducing the physical-to-logical qubit ratio. If entanglement distillation incurs an irreducible and exponential overhead in practical settings, the hardware requirements for fault-tolerant quantum computing could be larger than current roadmaps anticipate. But in the near term, the finding primarily guides algorithm-co-design choices, especially in noisy intermediate-scale quantum (NISQ) devices, where entanglement distribution remains a bottleneck.
What Still Needs to Happen
Two concrete challenges stand out. First, the strongest separation results are proved under completely PPT-preserving operations, a broader class than the local operations and classical communication (LOCC) that real-world quantum networks can implement. The paper notes that, remarkably, no analogous exponential strong-converse gap is known under LOCC. Bridging this gap will require new mathematical insights into the structure of LOCC protocols, a notoriously difficult problem that groups at the University of Bristol and the Technion are tackling using tensor-network methods.
Second, the analytical families of states constructed here are extreme and may be fragile in a physical setting. Demonstrating that realistic noisy channels — such as those in superconducting or trapped-ion architectures — exhibit a measurable exponential irreversibility would require experimental proposals that can produce and manipulate the necessary antisymmetric states. Teams at Delft University of Technology and the Max Planck Institute of Quantum Optics have the capabilities to test small-scale entanglement reversibility, but a full validation is likely five to ten years away.
Conclusion
In short: entanglement irreversibility is exponentially robust, deepening the divergence between entanglement theory and thermodynamics and imposing new fundamental limits on quantum error correction protocols. The paper transforms a conjecture into a theorem, equipping the field with rigorous bounds that will guide resource estimation for years to come.