Second-level collapse of the exact PPT entanglement-cost hierarchy
- Field
- Topics
Problem
Does the semidefinite hierarchy for exact PPT entanglement cost collapse at its second level for every finite-dimensional bipartite density operator \(\rho_{AB}\)? Let \(\Gamma=\operatorname{id}_A\otimes T_B\) denote partial transpose. For an integer \(p\geq0\), define
where the variables in Eq. (1) are Hermitian operators on \(A\otimes B\) and the inequalities are in the positive-semidefinite order. Prove or disprove
A counterexample to Eq. (2) must establish a strict gap.
Source
Conjecture S33 in the Supplement subsection “Open problem: hierarchy collapse” of Lami, Mele, and Regula explicitly poses Eq. (2); their Definition S5 gives Eq. (1) [LMR25]. This is a precise conjecture within the supplied note’s broader single-letter PPT-cost question.
Progress
Reports do not certify correctness or automatically change the problem's status. Progress policy.
The hierarchy converges to the exact asymptotic PPT entanglement cost. Equality of consecutive levels forces all later levels to coincide, so Eq. (2) would identify that cost with \(E_{\chi,2}\); see the Supplement subsection “Open problem: hierarchy collapse” [LMR25]. Here exact cost means the asymptotic number of maximally entangled qubit pairs per target copy under completely PPT-preserving channels, with zero preparation error at each block length.
The 2025 peer-reviewed paper reports numerical evidence for Eq. (2), without proving finite collapse [LMR25]. This conjecture was subsequently disproved by the preprint below.
Brinster et al. disprove Eq. (2). Their Appendix C, Theorem 8 constructs states \(\rho_m\) on \(\mathbb C^{2^m}\otimes\mathbb C^{2^m}\) with \(E_{\chi,m-2}(\rho_m)<E_{\chi,m-1}(\rho_m)\) for \(m\geq3\). The hierarchy in their Eq. (11) agrees with Eq. (1). Taking \(m=4\) yields a \(16\times16\) counterexample: Eqs. (154)–(156) give \(\chi_3(\rho_4)=16/6\) and \(\chi_2(\rho_4)\leq15/6\). These are analytic bounds, not numerical evidence [BKK+26].
- Reported progress: PR #77.
Comment
The answer is negative, based on an arXiv preprint submitted 17 September 2026. The construction also rules out a universal finite collapse level for this hierarchy; it does not settle the distinct exact PPT-distillation problem.