Zero two-way quantum capacity of endpoint Holevo–Werner channels

Problem

Is the two-way assisted quantum capacity of the Holevo–Werner channel \(\mathcal W_d\) zero for every integer \(d\geq3\)? For \(d\geq3\), define \(\mathcal W_d:\mathcal L(\mathbb C^d)\to\mathcal L(\mathbb C^d)\) by

\begin{equation} \mathcal W_d(X):=\frac{\operatorname{Tr}(X)I_d-\frac12X^{\mathsf T}}{d-\frac12}, \tag{1} \end{equation}

where \(\mathsf T\) denotes the transpose in the computational basis. Let \(S_d\) be the swap operator on \(\mathbb C^d\otimes\mathbb C^d\), and let \(\phi_d:=\lvert\phi_d\rangle\!\langle\phi_d\rvert\) with \(\lvert\phi_d\rangle:=d^{-1/2}\sum_{j=0}^{d-1}\lvert jj\rangle\). The channel in Eq. (1) has the normalized Choi state

\begin{equation} \rho_d:=(\operatorname{id}\otimes\mathcal W_d)(\phi_d) =\frac{I_{d^2}-\frac12S_d}{d^2-\frac d2}. \tag{2} \end{equation}

Let \(Q_{\leftrightarrow}(\mathcal W_d)\) be the supremum of the rates, in qubits per channel use, that can be transmitted with vanishing error by protocols that interleave uses of \(\mathcal W_d\) with adaptive local operations and unlimited two-way classical communication, starting without shared entanglement. The question is whether

\begin{equation} Q_{\leftrightarrow}(\mathcal W_d)=0 \qquad\text{for every integer }d\geq3 . \tag{3} \end{equation}

Source

The Choi states in Eq. (2) belong to the Werner family in which DiVincenzo, Shor, Smolin, Terhal, and Thapliyal conjecture the existence of undistillable states with non-positive partial transpose (Section II) [DSS+00]. Bharti, Gajjala, and Haug state undistillability of this endpoint for all copy numbers as the remaining unrestricted many-copy conjecture (Eq. (4)) [BGH26]. The channel formulation uses the teleportation covariance of Holevo–Werner channels [CGP18].

Progress

In the Werner-state convention \(\rho_{d,\alpha}:=(I_{d^2}+\alpha S_d)/(d^2+\alpha d)\) with \(-1\leq\alpha\leq1\), Eq. (2) is \(\rho_d=\rho_{d,-1/2}\). Partial transposition on the second factor gives

\begin{equation} \rho_d^{T_B}=\frac{I_{d^2}-\frac d2\,\phi_d}{d^2-\frac d2}, \tag{4} \end{equation}

whose eigenvalue on \(\lvert\phi_d\rangle\) is negative for every \(d\geq3\), so zero capacity cannot follow from a PPT Choi state. Up to a positive factor, Eq. (4) is the operator \(\lambda I-(\lambda+1)\lvert\Phi_0\rangle\!\langle\Phi_0\rvert\) with \(\lambda=2/(d-2)\) of DiVincenzo, Shor, Smolin, Terhal, and Thapliyal, the transition point between one-copy distillable Werner states and Werner states with nonnegative expectation on all Schmidt-rank-two vectors (Section III A); following Horodecki et al., they conjecture that some NPT Werner states are undistillable [DSS+00]. The partial transpose and this identification were checked directly for this entry.
Cope, Goodenough, and Pirandola define Holevo–Werner channels as the channels whose Choi states are Werner states and show that they are teleportation covariant; in the present parameterization, their Eq. (8) reads

\begin{equation} \mathcal W_d(UXU^\dagger)=\overline U\,\mathcal W_d(X)\,U^{\mathsf T} \qquad(U\in\mathrm U(d)). \tag{5} \end{equation}

The channel in Eq. (1) is their Eq. (6) with expectation parameter \(\operatorname{Tr}(\rho_dS_d)=(2-d)/(2d-1)\), and their \(\alpha\)-representation uses the opposite sign of \(\alpha\) [CGP18]. By Eq. (5), the channel is Choi-stretchable: each use can be simulated by an LOCC operation acting on one copy of \(\rho_d\) (Proposition 2) [PLOB17].
Teleportation stretching reduces an adaptive protocol with \(n\) uses of \(\mathcal W_d\) to an LOCC operation on \(\rho_d^{\otimes n}\), while sending halves of \(\phi_d\) through the channel gives the reverse inequality; hence

\begin{equation} Q_{\leftrightarrow}(\mathcal W_d)=D_{\leftrightarrow}(\rho_d), \tag{6} \end{equation}

where \(D_{\leftrightarrow}\) is the two-way LOCC distillable entanglement. Cope, Goodenough, and Pirandola state this identity for finite-dimensional teleportation-covariant channels through a reduction to entanglement distillation (Appendix A) [CGP18]; see also Supplementary Note 10 of [PLOB17]. By the finite-copy distillability criterion (Lemmas 1 and 2) [DSS+00], Eq. (3) is equivalent to requiring, for every \(d\geq3\),

\begin{equation} \langle\psi\rvert\bigl(I_{d^2}-\tfrac d2\,\phi_d\bigr)^{\otimes k} \lvert\psi\rangle\geq0 \quad\text{for all }k\geq1\text{ and all }\lvert\psi\rangle\text{ with } \operatorname{SR}_{R^k:B^k}(\lvert\psi\rangle)\leq2, \tag{7} \end{equation}

where \(\operatorname{SR}_{R^k:B^k}\) is the Schmidt rank between the \(k\) reference systems and the \(k\) output systems.
Four preprints posted in July 2026 prove the two-copy threshold: for every \(d\geq2\) and \(-1\leq\alpha\leq1\), the state \(\rho_{d,\alpha}\) is two-copy undistillable if and only if \(\alpha\geq-1/2\) (Theorem 1.1 of [FGP26], Theorem 5 of [SC26], Corollary C of [FHPV26], and Corollary 1.2 of [BGH26]). This establishes Eq. (7) for \(k\leq2\) in every dimension, but not for \(k\geq3\).
Bharti, Gajjala, and Haug give equivalent forms of the all-copy endpoint condition, including 2-positivity of \(\Phi_d^{\otimes k}\) for every \(k\), where \(\Phi_d(X):=\operatorname{Tr}(X)I_d-\frac12X\). They explain why the two-copy proof does not induct: a partial trace can increase rank, and tensor products need not preserve 2-positivity. They prove the endpoint inequality for tensor-factorized witnesses whose rank-two factor is supported on at most two copies, and they construct constants \(\gamma_k>0\), with \(\gamma_3=1/6\), such that \(\rho_{d,\alpha}\) is \(k\)-copy undistillable for \(\alpha\geq-\gamma_k\) in every dimension. For \(k\geq3\), neither result reaches \(\alpha=-1/2\) [BGH26].
For three copies, write a test vector as \(\lvert\psi\rangle=\operatorname{vec}C\), so that its Schmidt rank is \(\operatorname{rank}C\). Wu and Zou prove the three-copy endpoint inequality for every positive semidefinite and every normal \(C\) of rank at most two (Theorem 1.3 and Corollary 6.1), and for nonnormal \(C\) under local support conditions (Theorems 7.2 and 7.3). General nonnormal rank-two witnesses remain open, so three-copy undistillability of \(\rho_d\) is not established [WZ26].

Comment

Literature checked through 15 September 2026. No proof or disproof of Eq. (7) for any \(k\geq3\) and \(d\geq3\) was located; the remaining gap is undistillability of \(\rho_{d,-1/2}\) at three or more copies. The two-copy results and the three-copy sector results are preprints. A proof of Eq. (3) for one fixed \(d\geq3\) would already exhibit NPT bound entanglement, answering the existence question of the NPT bound entanglement problem affirmatively; the present record asks for more, namely undistillability of this specific endpoint state in every dimension. Restricting Schmidt-rank-two test vectors to \(\mathbb C^{d'}\otimes\mathbb C^{d'}\) in each copy shows that \(k\)-copy distillability of \(\rho_{d',-1/2}\) implies that of \(\rho_{d,-1/2}\) for every \(d\geq d'\); hence Eq. (3) holds for every \(d\geq3\) as soon as it holds for infinitely many \(d\). Mixing with separable Werner states likewise transfers undistillability from \(\alpha=-1/2\) to every \(-1/2\leq\alpha<-1/d\). These two reductions are deductions made for this entry.

References

[DSS+00]: D. P. DiVincenzo, P. W. Shor, J. A. Smolin, B. M. Terhal, and A. V. Thapliyal, “Evidence for Bound Entangled States with Negative Partial Transpose,” Physical Review A 61, 062312 (2000).DOI arXiv
[BGH26]: K. Bharti, R. Gajjala, and T. Haug, “Two-Copy Nondistillability of Werner States: Sharp Partial-Trace Inequalities and Finite-Copy Extensions,” arXiv preprint (2026).arXiv
[CGP18]: T. P. W. Cope, K. Goodenough, and S. Pirandola, “Converse Bounds for Quantum and Private Communication over Holevo–Werner Channels,” Journal of Physics A: Mathematical and Theoretical 51, 494001 (2018).DOI arXiv
[PLOB17]: S. Pirandola, R. Laurenza, C. Ottaviani, and L. Banchi, “Fundamental Limits of Repeaterless Quantum Communications,” Nature Communications 8, 15043 (2017).DOI arXiv
[FGP26]: J. Fu, L. Gao, and S.-J. Park, “A Solution to 2-Copy Distillability of Werner States,” arXiv preprint (2026), version 2.arXiv
[SC26]: Z. Song and L. Chen, “A Partial-Trace Matrix Inequality and Werner-State Distillability,” arXiv preprint (2026).arXiv
[FHPV26]: T. C. Fraser, F. Huber, B. Pozsgay, and I. Vona, “On the Two-Copy Distillability of Werner States and a New Partial Trace Inequality,” arXiv preprint (2026).arXiv
[WZ26]: T. Wu and Q. Zou, “Sharp Plücker Geometry for Three-Copy Werner Distillation,” arXiv preprint (2026).arXiv

Zero two-way quantum capacity of endpoint Holevo–Werner channels

Problem

Source

Progress

Comment

References

Cite this problem

BibTeX

Plain text

Problem

Source

Progress

Comment

References

Related problems