Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:quant-ph/9811018 · Physical Review Letters

Separability of Very Noisy Mixed States and Implications for NMR Quantum Computing

Samuel L. Braunstein, Carlton M. Caves, Richard Jozsa, Noah Linden, Sandu Popescu, Rüdiger Schack

Published Aug 2, 1999• 566 citations
arXivHTMLPDF

Abstract

We give a constructive proof that all mixed states of $N$ qubits in a sufficiently small neighborhood of the maximally mixed state are separable (unentangled). The construction provides an explicit representation of any such state as a mixture of product states. We give upper and lower bounds on the size of the neighborhood, which show that its extent decreases exponentially with the number of qubits. The bounds show that no entanglement appears in the physical states at any stage of present NMR experiments. Though this result raises questions about NMR quantum computation, further analysis would be necessary to assess the power of the general unitary transformations, which are indeed implemented in these experiments, in their action on separable states.

DOI
10.1103/physrevlett.83.1054

Related papers

New considerations on the separability of very noisy mixed states and implications for NMR quantum computing

J.D. Bulnes, Roberto S. Sarthour, Eduardo Ribeiro de Azevêdo +5 more

We revise the problem first addressed by Braunstein and co-workers (Phys. Rev. Lett. {\bf 83} (5) (1999) 1054) concerning the separability of very noisy mixed states represented by general density matrices with the form $ρ_ε= (1-ε)M_d+ερ_1$. From a detailed numerical analysis, it is shown that: (1) there exist infinite values in the interval taken for the density matrix expansion coefficients, $-1\le c_{α_1,...,α_N}\le 1$, which give rise to {\em non-physical density matrices}, with trace equal to 1, but at least one {\em negative} eigenvalue; (2) there exist entangled matrices outside the predicted entanglement region, and (3) there exist separable matrices inside the same region. It is also shown that the lower and upper bounds of $ε$ depend on the coefficients of the expansion of $ρ_1$ in the Pauli basis. If $ρ_{1}$ is hermitian with trace equal to 1, but is allowed to have {\em negative} eigenvalues, it is shown that $ρ_ε$ can be entangled, even for two qubits.

arXiv:quant-ph/0404020•Apr 3, 2004•1 citation

Separable balls around the maximally mixed multipartite quantum states

Leonid Gurvits, Howard Barnum

We show that for an m-partite quantum system, there is a ball of radius ${2}^{\ensuremath{-}(m/2\ensuremath{-}1)}$ in Frobenius norm, centered at the identity matrix, of separable (unentangled) positive semidefinite matrices. This can be used to derive an \ensuremath{\epsilon} below which mixtures of \ensuremath{\epsilon} of any density matrix with $1\ensuremath{-}\ensuremath{\epsilon}$ of the maximally mixed state will be separable. The \ensuremath{\epsilon} thus obtained is exponentially better (in the number of systems) than existing results. This gives a number of qubits below which nuclear magnetic resonance with standard pseudopure-state preparation techniques can access only unentangled states; with parameters realistic for current experiments, this is 23 qubits (compared to 13 qubits via earlier results). A ball of radius 1 is obtained for multipartite states separable over the reals.

arXiv:quant-ph/0302102•Physical Review A•Oct 14, 2003•98 citations

Entangled states close to the maximally mixed state

Roland Hildebrand

This paper deals with the radius of the largest ball of separable mixed states around the maximally mixed state for multiqubit systems. This radius determines how close entangled states can be to the maximally mixed state. In Aubrun and Szarek (e-print arXiv:quant-ph/0503221) an upper bound on the radius was given, while in Gurvits and Barnum (e-print arXiv:quant-ph/0409095) a lower bound was provided. In this paper we improve both the upper and the lower bound, bringing the ratio of these bounds down to a constant $c=\sqrt{34∕27}\ensuremath{\approx}1.122$, as opposed to a term of order $\sqrt{m\phantom{\rule{0.2em}{0ex}}\mathrm{log}\phantom{\rule{0.2em}{0ex}}m}$ for the best bounds known previously, where $m$ is the number of qubits in the system. We construct concrete examples of separable states on the boundary to entanglement which realize the upper bounds. As a by-product, we compute the radii of the largest balls that fit into the projective tensor product of three and four unit balls in ${\mathbb{R}}^{3}$ and in the projective tensor product of an arbitrary number of unit balls in ${\mathbb{R}}^{n}$ for $n=2$, 4, and 8.

arXiv:quant-ph/0503194•Physical Review A•Jun 26, 2007•22 citations

Upper bound on the region of separable states near the maximally mixed state

Piotr Deuar, W J Munro, Kae Nemoto

A lower bound on the amount of noise that must be added to a GHZ-like entangled state to make it separable (also called the random robustness) is found using the transposition condition. The bound is applicable to arbitrary numbers of subsystems, and dimensions of Hilbert space, and is shown to be exact for qubits. The new bound is compared with previous such bounds on this quantity, and found to be stronger in all cases. It implies that increasing the number of subsystems, rather than increasing their Hilbert space dimension, is a more effective way of increasing entanglement. An explicit decomposition into an ensemble of separable states, when the state is not entangled, is given for the case of qubits.

arXiv:quant-ph/0002002•Journal of Optics B Quantum and Semiclassical Optics•May 11, 2000•7 citations

Upper Bound on the region of Separable States near the Maximally Mixed State

Piotr Deuar, William J. Munro, Kae Nemoto

A lower bound on the amount of noise that must be added to a GHZ-like entangled state to make it separable (also called the random robustness) is found using the transposition condition. The bound is applicable to arbitrary numbers of subsystems, and dimensions of Hilbert space, and is shown to be exact for qubits. The new bound is compared to previous such bounds on this quantity, and found to be stronger in all cases. It implies that increasing the number of subsystems, rather than increasing their Hilbert space dimension is a more effective way of increasing entanglement. An explicit decomposition into an ensemble of separable states, when the state is not entangled,is given for the case of qubits.

•Jan 1, 2000•7 citations

Bounds on the entanglement attainable from unitary transformed thermal states in liquid-state nuclear magnetic resonance

Terri M. Yu, Kenneth R. Brown, Isaac L. Chuang

The role of mixed-state entanglement in liquid-state nuclear magnetic resonance (NMR) quantum computation is not yet well understood. In particular, despite the success of quantum-information processing with NMR, recent work has shown that quantum states used in most of those experiments were not entangled. This is because these states, derived by unitary transforms from the thermal equilibrium state, were too close to the maximally mixed state. We are thus motivated to determine whether a given NMR state is entanglable---that is, does there exist a unitary transform that entangles the state? The boundary between entanglable and nonentanglable thermal states is a function of the spin system size $N$ and its temperature $T$. We provide bounds on the location of this boundary using analytical and numerical methods; our tightest bound scales as $N\ensuremath{\sim}T$, giving a lower bound requiring at least $N\ensuremath{\sim}22\phantom{\rule{0.2em}{0ex}}000$ proton spins to realize an entanglable thermal state at typical laboratory NMR magnetic fields. These bounds are tighter than known bounds on the entanglability of effective pure states.

arXiv:quant-ph/0409170•Physical Review A•Mar 24, 2005•11 citations
We give a constructive proof that all mixed states of NNN qubits in a sufficiently small neighborhood of the maximally mixed state are separable (unentangled). The construction provides an explicit representation of any such state as a mixture of product states. We give upper and lower bounds on the size of the neighborhood, which show that its extent decreases exponentially with the number of qubits. The bounds show that no entanglement appears in the physical states at any stage of present NMR experiments. Though this result raises questions about NMR quantum computation, further analysis would be necessary to assess the power of the general unitary transformations, which are indeed implemented in these experiments, in their action on separable states.
We revise the problem first addressed by Braunstein and co-workers (Phys. Rev. Lett. {\bf 83} (5) (1999) 1054) concerning the separability of very noisy mixed states represented by general density matrices with the form ρε=(1−ε)Md+ερ1ρ_ε= (1-ε)M_d+ερ_1ρε​=(1−ε)Md​+ερ1​. From a detailed numerical analysis, it is shown that: (1) there exist infinite values in the interval taken for the density matrix expansion coefficients, −1≤cα1,...,αN≤1-1\le c_{α_1,...,α_N}\le 1−1≤cα1​,...,αN​​≤1, which give rise to {\em non-physical density matrices}, with trace equal to 1, but at least one {\em negative} eigenvalue; (2) there exist entangled matrices outside the predicted entanglement region, and (3) there exist separable matrices inside the same region. It is also shown that the lower and upper bounds of εεε depend on the coefficients of the expansion of ρ1ρ_1ρ1​ in the Pauli basis. If ρ1ρ_{1}ρ1​ is hermitian with trace equal to 1, but is allowed to have {\em negative} eigenvalues, it is shown that ρερ_ερε​ can be entangled, even for two qubits.
We show that for an m-partite quantum system, there is a ball of radius ${2}^{\ensuremath{-}(m/2\ensuremath{-}1)}$ in Frobenius norm, centered at the identity matrix, of separable (unentangled) positive semidefinite matrices. This can be used to derive an \ensuremath{\epsilon} below which mixtures of \ensuremath{\epsilon} of any density matrix with $1\ensuremath{-}\ensuremath{\epsilon}$ of the maximally mixed state will be separable. The \ensuremath{\epsilon} thus obtained is exponentially better (in the number of systems) than existing results. This gives a number of qubits below which nuclear magnetic resonance with standard pseudopure-state preparation techniques can access only unentangled states; with parameters realistic for current experiments, this is 23 qubits (compared to 13 qubits via earlier results). A ball of radius 1 is obtained for multipartite states separable over the reals.
This paper deals with the radius of the largest ball of separable mixed states around the maximally mixed state for multiqubit systems. This radius determines how close entangled states can be to the maximally mixed state. In Aubrun and Szarek (e-print arXiv:quant-ph/0503221) an upper bound on the radius was given, while in Gurvits and Barnum (e-print arXiv:quant-ph/0409095) a lower bound was provided. In this paper we improve both the upper and the lower bound, bringing the ratio of these bounds down to a constant $c=\sqrt{34∕27}\ensuremath{\approx}1.122$, as opposed to a term of order mlogm\sqrt{m\phantom{\rule{0.2em}{0ex}}\mathrm{log}\phantom{\rule{0.2em}{0ex}}m}mlogm​ for the best bounds known previously, where mmm is the number of qubits in the system. We construct concrete examples of separable states on the boundary to entanglement which realize the upper bounds. As a by-product, we compute the radii of the largest balls that fit into the projective tensor product of three and four unit balls in R3{\mathbb{R}}^{3}R3 and in the projective tensor product of an arbitrary number of unit balls in Rn{\mathbb{R}}^{n}Rn for n=2n=2n=2, 4, and 8.
A lower bound on the amount of noise that must be added to a GHZ-like entangled state to make it separable (also called the random robustness) is found using the transposition condition. The bound is applicable to arbitrary numbers of subsystems, and dimensions of Hilbert space, and is shown to be exact for qubits. The new bound is compared with previous such bounds on this quantity, and found to be stronger in all cases. It implies that increasing the number of subsystems, rather than increasing their Hilbert space dimension, is a more effective way of increasing entanglement. An explicit decomposition into an ensemble of separable states, when the state is not entangled, is given for the case of qubits.
A lower bound on the amount of noise that must be added to a GHZ-like entangled state to make it separable (also called the random robustness) is found using the transposition condition. The bound is applicable to arbitrary numbers of subsystems, and dimensions of Hilbert space, and is shown to be exact for qubits. The new bound is compared to previous such bounds on this quantity, and found to be stronger in all cases. It implies that increasing the number of subsystems, rather than increasing their Hilbert space dimension is a more effective way of increasing entanglement. An explicit decomposition into an ensemble of separable states, when the state is not entangled,is given for the case of qubits.
The role of mixed-state entanglement in liquid-state nuclear magnetic resonance (NMR) quantum computation is not yet well understood. In particular, despite the success of quantum-information processing with NMR, recent work has shown that quantum states used in most of those experiments were not entangled. This is because these states, derived by unitary transforms from the thermal equilibrium state, were too close to the maximally mixed state. We are thus motivated to determine whether a given NMR state is entanglable---that is, does there exist a unitary transform that entangles the state? The boundary between entanglable and nonentanglable thermal states is a function of the spin system size NNN and its temperature TTT. We provide bounds on the location of this boundary using analytical and numerical methods; our tightest bound scales as $N\ensuremath{\sim}T$, giving a lower bound requiring at least $N\ensuremath{\sim}22\phantom{\rule{0.2em}{0ex}}000$ proton spins to realize an entanglable thermal state at typical laboratory NMR magnetic fields. These bounds are tighter than known bounds on the entanglability of effective pure states.