Quantum entanglement can be distilled: given many copies of a weakly entangled state, you can produce fewer copies of a strongly entangled state using only local operations and one-way classical communication. The rate at which this is possible — the one-way distillable entanglement — is a fundamental measure of quantum correlation. But computing it requires optimizing over an infinite number of copies (a “regularized” formula), which is generally intractable.
Note (2026-11-14; read against the paper's full text): the essay's account is close to the paper but leaves out two conditions. The flagged-mixture result holds when one component has orthogonal support on Alice's system and zero one-way distillable entanglement, which the paper presents as a stability result (degradability is typically lost under such mixing). The spin-alignment entropy-minimization rule is a proposed principle that the paper says it establishes in several key cases, including a complete Renyi-2 result; it is not proved in general, and the essay says it 'helps minimize the conditional entropy'. The weakened conditions (regularized less noisy and informationally degradable) are proved to guarantee additivity and so a single-letter formula, as the essay says, and the paper gives explicit examples such as flagged mixtures of amplitude damping channels. The closing claim that redundancy, not simplicity, makes entanglement computable is the essay's own.
For degradable states — states where the environment's information is a degraded version of what the receiver gets — a single-letter formula exists: you only need to optimize over one copy. But most states aren't degradable. Ahmed, Smith, and Wu (arXiv: 2603.23417) find three families of non-degradable states that nonetheless have single-letter formulas.
The trick: weaken the degradability condition. “Regularized less-noisy” and “informationally degradable” are new conditions, strictly weaker than degradable but strong enough to guarantee that the multi-copy optimization collapses to a single-copy one. The entanglement formula doesn't simplify because the state is simple — it simplifies because the state has just enough structure to prevent the multi-copy correlations from helping.
Additional results: orthogonally flagged mixtures (states built from orthogonal components) are stable under these conditions, and a generalized spin-alignment principle helps minimize the conditional entropy.
The through-claim: computability is a property of structure, not of simplicity. Non-degradable states can have simple formulas. The single-letter condition doesn't require the state to be nice — it requires the state's structure to prevent multi-copy exploitation. The formula collapses not because one copy is enough information, but because additional copies are redundant. Redundancy, not simplicity, is what makes entanglement computable.
Ahmed, Smith & Wu, 2603.23417. Quantum information / entanglement distillation / single-letter formulas / degradability / one-way communication.