Is entanglement a reversible resource?

The question in perspective

Two laboratories can perform local quantum operations and exchange classical messages, abbreviated LOCC. They cannot send quantum systems for free. Under this restriction, shared entanglement enables tasks that local preparation cannot supply. Can all forms of entanglement be converted into one currency and back without loss?

A case with an exact currency

A bipartite pure state has a Schmidt decomposition

\[\ket\psi=\sum_j\sqrt{\lambda_j}\ket j_A\ket j_B,\qquad \lambda_j\geq0,\quad\sum_j\lambda_j=1.\]

Tracing out B gives \(\rho_A=\sum_j\lambda_j\ket j\bra j\). Its entropy \(H(\lambda)=-\sum_j\lambda_j\log_2\lambda_j\) measures pure-state entanglement in bits. One maximally entangled qubit pair, \(\ket{\Phi^+}=(\ket{00}+\ket{11})/\sqrt2\), supplies one ebit.

For many independent copies of \(\ket\psi\), typical Schmidt strings have probability approximately \(2^{-nH(\lambda)}\), and about \(2^{nH(\lambda)}\) strings contain nearly all probability. This suggests an effective maximally entangled dimension \(2^{nH(\lambda)}\). Concentration and dilution theorems make the intuition precise: with unrestricted LOCC and vanishing error, the asymptotic Bell-pair rates in both directions equal \(H(\lambda)\). This is a theorem about many copies, not deterministic conversion of one copy.

For \(\ket\psi=\sqrt{0.8}\ket{00}+\sqrt{0.2}\ket{11}\), the rate is \(h_2(0.2)\approx0.722\) ebits per copy. Single-copy postselection instead lets Alice apply \(K=\operatorname{diag}(1/2,1)\). The unnormalized output is \(\sqrt{0.2}(\ket{00}+\ket{11})\), a Bell pair with success probability \(0.4\). Different success and copy regimes give different conversion statements.

What is established

For a mixed state \(\rho_{AB}\), define \(E_D\) as the best asymptotic number of Bell pairs extracted per input copy, and \(E_C\) as the Bell-pair cost of preparing a copy, both under LOCC with vanishing error. In general,

\[E_D(\rho)\leq E_C(\rho),\]

and equality need not hold. The entanglement review describes bound entanglement and irreversible transformations. Distillation and error-correction protocols connect shared-state purification to communication tasks. Private key from bound entanglement also shows why “useful resource” depends on the task.

What remains unresolved

For many mixed-state families, achievable distillation rates and matching converse bounds remain difficult to obtain. Finite-copy error, restricted classical communication, computationally efficient local operations, and multipartite sharing lead to different resource theories. One must state the allowed operations before asking whether a transformation is reversible.

A concrete next investigation

Reproduce: verify the local filter above, including \(K^\dagger K\leq I\), its success probability, and the normalized conditional state.

Check: distinguish the 0.4 single-copy success probability from the approximately 0.722 asymptotic rate. Neither quantity contradicts the other.

Extend: add a specified noise channel to this pure-state family and compare an explicit distillation protocol with a computable upper bound. Narrowing that gap for a defined noise range is a plausible contribution; merely detecting entanglement does not determine its distillable rate.

Selected references

1. Quantum entanglement

   Ryszard Horodecki et al.. 2007. Review

   A starting point for entanglement detection, quantification, and its use as a resource.

2. Mixed State Entanglement and Quantum Error Correction

   Charles H. Bennett, David P. DiVincenzo, John A. Smolin, William K. Wootters. 1996. Research paper

   Connects purification of shared entanglement with quantum error correction.

3. Secure key from bound entanglement

   Karol Horodecki, Michał Horodecki, Paweł Horodecki, Jonathan Oppenheim. 2003. Research paper

   Shows that secret-key utility and distillable Bell-pair entanglement are different resources.

Years refer to first arXiv submission. Publication details are available in the linked records.