Entanglement of Formation: Theory & Applications

• Entanglement of formation is a measure defining the minimum entanglement cost required to construct a mixed quantum state from pure entangled resources.
• In two-qubit systems, closed formulas based on concurrence simplify computation, while higher dimensions require complex convex-roof minimization techniques.
• Applications extend to continuous-variable systems, noise-tolerant certification, and numerical approaches like tree tensor operators for scalable entanglement estimation.

Entanglement of formation is a central quantitative measure of bipartite quantum correlations, designed to capture the minimum entanglement cost required to synthesize a mixed quantum state from pure-state entangled resources. It is rooted in the convex-roof construction and forms a cornerstone for the operational theory of quantum entanglement across diverse physical systems.

1. Definition and Mathematical Framework

Given a bipartite quantum system described by the density matrix $\rho_{AB}$ on $\mathcal{H}_A \otimes \mathcal{H}_B$, the entanglement of formation (EoF) is defined as: $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$ where the infimum is over all pure-state ensembles $\{p_i, |\psi_i\rangle\}$ such that $$\rho_{AB} = \sum_i p_i |\psi_i\rangle\langle\psi_i|$$, and $S(\sigma) = -\mathrm{Tr}[\sigma \log \sigma]$ is the von Neumann entropy. For pure bipartite states, EoF reduces to the entropy of entanglement—$$E_F(|\psi\rangle) = S(\mathrm{Tr}_B\,|\psi\rangle\langle\psi|)$$ (Kim, 2021).

EoF is a fully entanglement monotone: it is non-increasing under LOCC, convex, and vanishes on separable states. However, explicit calculation of EoF for mixed states is computationally hard due to the convex-roof minimization.

2. Exact Results in Finite Dimensions

For two-qubit ($2 \otimes 2$) mixed states, EoF admits Wootters' closed formula: $$E_F(\rho_{AB}) = h\left(\frac{1+\sqrt{1-C^2(\rho_{AB})}}{2}\right)$$ with $h(x) = -x \log_2 x - (1-x)\log_2(1-x)$ and $\mathcal{H}_A \otimes \mathcal{H}_B$0 the concurrence, itself computable as the largest singular value of the spin-flipped operator minus the sum of the others (Lastra et al., 2011, Oliveira et al., 2013, Bai et al., 2014).

For higher-dimensional cases, general closed forms do not exist. However, analytic results cover several broad yet structured cases:

• For $\mathcal{H}_A \otimes \mathcal{H}_B$1 systems, when the reduced $\mathcal{H}_A \otimes \mathcal{H}_B$2 marginal of a tripartite pure state is of X-form, EoF can be exactly evaluated via Koashi–Winter relations, reducing the task to analytic conditional-entropy minimizations (notably using results of Luo and Ali et al.) (Lastra et al., 2011).
• Special parameterizations, such as in the Tavis–Cummings model, yield closed formulae for EoF in $\mathcal{H}_A \otimes \mathcal{H}_B$3, $\mathcal{H}_A \otimes \mathcal{H}_B$4 (Lastra et al., 2011).

3. EoF in Continuous Variables and Gaussian States

For two-mode Gaussian states, the EoF is fully characterized via symplectic invariants of the covariance matrix. The optimal decomposition is constructed from single two-mode squeezed vacuum states displaced according to classical Gaussian weights: $\mathcal{H}_A \otimes \mathcal{H}_B$5 where $\mathcal{H}_A \otimes \mathcal{H}_B$6 is the smallest symplectic eigenvalue of the partially transposed covariance matrix, and $\mathcal{H}_A \otimes \mathcal{H}_B$7 as above (0809.0321, Akbari-Kourbolagh et al., 2014). In this sector, the optimal decomposition is Gaussian, and EoF is proven to be additive: $\mathcal{H}_A \otimes \mathcal{H}_B$8

General two-mode Gaussian states without simple symmetry can be addressed via single-parameter minimizations using bounds and compact expressions for upper and lower limits, reducing the minimization to a tractable univariate optimization (Tserkis et al., 2019). Explicit forms are available for symmetric and squeezed thermal cases (Akbari-Kourbolagh et al., 2014).

For superpositions of coherent states on circles (circular states, RICS), the Schmidt decomposition becomes analytic, enabling closed-form evaluation of EoF and detailed dependence on physical parameters such as the circle's radius and the number of components in the superposition (Horoshko et al., 2016).

4. Monogamy, Additivity, and Multipartite Distribution

EoF's behavior under monogamy is subtle. For three qubits, the squared EoF satisfies a monogamy relation analogous to Coffman-Kundu-Wootters (CKW) for squared concurrence: $\mathcal{H}_A \otimes \mathcal{H}_B$9 and similarly for $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$0-qubit mixed states (Oliveira et al., 2013, Bai et al., 2014). In contrast, the linear EoF (unsquared) does not universally obey CKW-type inequalities; explicit violations are seen in e.g., the three-qubit W state (Fanchini et al., 2011, Kim, 2021). However, rigorous upper bounds—such as $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$1 for three-qubit pure states—demonstrate limited shareability, and the sum is strongly constrained (Oliveira et al., 2013).

Recently, monogamy has been reframed via no-inequality conditions: strict concavity of von Neumann entropy ensures that EoF is monogamous in the sense that $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$2, for both pure and mixed states (Guo et al., 2018).

In multipartite systems, sufficient conditions for monogamy in terms of EoF have been constructed utilizing classical–classical–quantum (ccq) extensions: if certain mutual information additivity conditions hold for ccq states built from the reduced bipartite states, then the monogamy inequality for EoF is satisfied. This holds for generic $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$3-party systems, generalizing the tripartite case (Kim, 2021).

5. Additivity, Entanglement-Breaking Subspaces, and Entanglement Cost

EoF is not generally additive, but important classes of states satisfy additivity criteria. If the support of $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$4 lies in an entanglement-breaking (EB) subspace, then for any other state $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$5: $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$6 and the entanglement cost $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$7 equals $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$8 (Zhao et al., 2019). In practice, this principle allows construction of high-rank, high-dimensional states with computable entanglement cost and additive EoF via systematic combination of EB spaces.

Product states of two-mode Gaussian states and certain two-qubit, rank-2 states (when their support is on EB subspaces) also manifest additive EoF, directly linking operational cost to convex-roof quantification.

6. Operational and Numerical Approaches

The convex-roof construction of EoF is computationally inefficient in general, scaling poorly with system size. However, numerical frameworks, in particular the tree tensor operator (TTO) ansatz, provide scalable approaches to bipartite entanglement estimation in many-body lattice systems. TTOs enable efficient parameterization of pure-state decompositions and optimization directly on tensor components, making EoF estimation feasible for up to $$E_F(\rho_{AB}) = \inf_{\{p_i,|\psi_i\rangle\}} \sum_i p_i\,S(\mathrm{Tr}_B\,|\psi_i\rangle\langle\psi_i|)$$9 spins (Arceci et al., 2020).

Within this framework, universal scaling laws for EoF at finite temperature in 1D critical systems have been identified, extending the conformal field theory scaling of the entanglement entropy to mixed states.

For pure-state scenarios in arbitrary dimension, direct experimental protocols are available to measure EoF by reduction to two-qubit observables and local measurements, following appropriate operator basis decompositions (Li et al., 2012).

7. Applications, Noise-Tolerant Certification, and Multipartite Indicators

EoF directly quantifies the rate of entangled pure states required to assemble a target state under asymptotically many copies (the entanglement cost),

$\{p_i, |\psi_i\rangle\}$0

and is used to certify high-dimensional entanglement in experimental devices. Threshold non-local games leveraging the gap between quantum and classical values can be constructed such that passing the test guarantees the presence of $\{p_i, |\psi_i\rangle\}$1 bits of entanglement of formation, tolerant to constant levels of noise and obviating the need for error correction (Arnon et al., 2017).

In multipartite contexts, EoF-based indicators ($\{p_i, |\psi_i\rangle\}$2, $\{p_i, |\psi_i\rangle\}$3) have been developed to detect genuine $\{p_i, |\psi_i\rangle\}$4-partite entanglement even in the absence of pairwise concurrence or $\{p_i, |\psi_i\rangle\}$5-tangles. These indicators leverage the squared EoF monogamy relation and quantum discord computations, rendering them practical in both analytical and numerical studies (Bai et al., 2014).

---

References:

• (Kim, 2021, 0809.0321, Horoshko et al., 2016, Lastra et al., 2011, Arceci et al., 2020, Tserkis et al., 2019, Oliveira et al., 2013, Fanchini et al., 2011, Arnon et al., 2017, Bai et al., 2014, Akbari-Kourbolagh et al., 2014, Zhao et al., 2019, Guo et al., 2018, Li et al., 2012)