Stability of PPT in equilibrium states
Marco Merkli, Mitch Zagrodnik · 2025
Read Stability of PPT in equilibrium states by Marco Merkli, Mitch Zagrodnik (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-06882.txt.
Introduction
Marco Merkli and Mitch Zagrodnik show that a key quantum property survives small disturbances at equilibrium. The positive partial transpose property, known as PPT, is a practical test related to whether a quantum state is entangled. They ask whether states at thermal equilibrium keep it when the Hamiltonian is slightly perturbed, in systems of infinite dimension.
Using basic tools from the theory of spectra and perturbations, they prove the property is stable under bounded perturbations. The catch is a condition: the temperature must be high enough, or the change small enough. The result reassures researchers that conclusions based on PPT about hot equilibrium states are not knife-edge false alarms.
Catalog record
| CREATOR | Marco Merkli, Mitch Zagrodnik |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-06882
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Stability of PPT in equilibrium states Marco Merkli, Mitch Zagrodnik Abstract We use simple spectral perturbation theory to show that the positive partial transpose property is stable under bounded perturbations of the Hamiltonian, for equilibrium states in infinite dimensions. The result holds provided the temperature is high enough, or equivalently, provided the perturbation is small enough. 1 Introduction The positive partial transposition (PPT), or Peres-Horodecki criterion, gives a necessary condition for a density matrix \rho of a bipartite quantum system SB (‘system-bath’) to be separable. Namely, if \rho is separable, then the partial transpose (relative to either subsystem, say B), T_{\rm B}[\rho] is a positive operator [ 22 , 12 ] . Equivalently, if \rho is not PPT, then it must be inseparable, also called entangled. If the dimension of the subsystem Hilbert spaces are two or three, then the converse statement is also correct [ 12 ] . However, in dimension \geq 4 , there are states which are PPT and yet entangled. In this situation the entanglement is called bound. It is called so since one cannot extract from such a state, pure singlet (entangled) states, which would be needed as a resource for quantum information purposes. In contrast, if it is possible to extract (by LOCC protocols) from an entangled mixed state (or many copies thereof), pairs of particles in a pure singlet state, then the mixed state is said to have ‘distillable’ entanglement [ 3 ] . The corresponding state is said to be distillable. It is shown in [ 16 ] (for finite dimensional systems) that if a state is PPT then it is not distillable. The converse is not true in dimension 3 or higher, though. Any possible entanglement in a PPT state is bound. For qubits (dimension 2 ) it was shown in [ 13 ] that any entangled state is distillable. However, in higher dimensions there are entangled states with bound entanglement, as shown in [ 27 ] . A great feature of the PPT criterion is that in principle, it is easy to apply. To verify that a given density matrix \rho of a bipartite system is entangled, one ‘simply’ has to check that the partial transpose T_{B}[\rho] is not a positive operator. We are addressing the following question here: Suppose a density matrix \rho_{0} is PPT, and consider a modified density matrix \rho=\rho_{0}+\rho^{\prime} , where \rho^{\prime} is a perturbation operator. Under what conditions is \rho still PPT? To investigate the question we first note that since \rho is hermitian ( = selfadjoint) then so is T_{B}[\rho] (see also ( 3.10 ) below). Thus T_{B}[\rho]\geq 0 if and only if all the eigenvalues of T_{B}[\rho] are non-negative. The partial trace is a linear operation, T_{B}[\rho]=T_{B}[\rho_{0}]+T_{B}[\rho^{\prime}] . Basic perturbation theory [ 17 ] tells us that the eigenvalues of T_{B}[\rho] lie within a neighbourhood of the size \|T_{B}[\rho^{\prime}]\|_{\infty} (operator norm of the perturbation) of the eigenvalues of T_{B}[\rho_{0}] . Since \rho_{0} is PPT we know that T_{B}[\rho_{0}]\geq 0 . If T_{B}[\rho_{0}] has a lowest eigenvalue \lambda_{0}>0 then for \|T_{B}[\rho^{\prime}]\|_{\infty}<\lambda_{0} the spectrum of T_{B}[\rho] is guaranteed to be \geq 0 , which means that \rho is PPT. This straightforward approach breaks down as soon as the dimension is infinite. The reason is that \rho_{0} and hence T_{B}[\rho_{0}] are Hilbert-Schmidt operators ( c.f. ( 3.9 )). In particular, T_{B}[\rho_{0}] is a compact operator and therefore, in the infinite-dimensional case, its eigenvalues must accumulate at the origin. Consequently, no matter how small we take \|T_{B}[\rho^{\prime}]\|_{\infty} , the simple argument given above does not work. We show in this paper how one can modify the simple perturbation argument for equilibrium states, where the perturbation is a bounded interaction term V in the Hamiltonian, \rho_{0}=\frac{e^{-\beta H_{0}}}{\mbox{Tr}e^{-\beta H_{0}}},\qquad\rho=\frac{e^{-\beta(H_{0}+V)}}{\mbox{Tr}e^{-\beta(H_{0}+V)}} (1.1) with H_{0}=H_{A}\otimes\mathchoice{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.5mul}{\rm 1\mskip-5.0mul}_{B}+\mathchoice{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.5mul}{\rm 1\mskip-5.0mul}_{A}\otimes H_{B}. Thermal states form the cornerstone of equilibrium statistical mechanics. Examining their quantum properties, one of which is measured by entanglement, is an important question. That said, the equilibrium setup ( 1.1 ) is mathematically quite general. Indeed, any faithful density matrix 1 1 1 A density matrix is called faithful if zero is not one of its eigenvalues. \rho_{0} can be written in the form \rho_{0}=e^{-H_{0}}/{\mbox{Tr}e^{-H_{0}}} for some Hermitian (self-adjoint) H_{0} , and similarly for \rho . Our approach is applicable to any such \rho_{0},\rho . In this work, we express the perturbation \rho_{0}\mapsto\rho as an interaction term V in the Hamiltonian because this is physically intuitive. Our main idea is to use the Dyson expansion to write e^{-\beta(H_{0}+V)}=e^{-\beta H_{0}/2}\big[\mathchoice{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.5mul}{\rm 1\mskip-5.0mul}+\mathcal{O}(V)\big]e^{-\beta H_{0}/2}, (1.2) where \mathcal{O}(V) is an operator which vanishes for V=0 . PPT for \rho will then follow provided T_{B}[\mathcal{O}(V)] is small enough such that \mathchoice{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.5mul}{\rm 1\mskip-5.0mul}+T_{B}[\mathcal{O}(V)]\geq 0. By factoring out the operators e^{-\beta H_{0}/2} in ( 1.2 ) we remove the problem of eigenvalues accumulating at the origin, as the unperturbed density matrix is effectively replaced by the operator \mathchoice{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.0mul}{\rm 1\mskip-4.5mul}{\rm 1\mskip-5.0mul} now, whose spectrum \{1\} is separated from the origin. The detailed control of \mathcal{O}(V) leading to the wanted bounds involves the size of V as well as the inverse temperature \beta . We assume a bound on the Hilbert-Schmidt norm of the imaginary time evolved interaction operator. Namely, we assume that there are constants a,b , and s_{*}>0 such that for all 0\leq s\leq s_{*} , we have the bound \|e^{-sH_{0}}Ve^{sH_{0}}\|_{2}\leq ae^{bs}. We then show in Theorem 2.2 that for large enough temperature \beta\leq\max\{\beta_{*},s_{*}\} , the perturbed state \rho is PPT. The upper bound \beta_{*} on \beta depends on the constants a,b . If a is small (say V contains an overall small coupling constant) then \beta_{*}\sim\ln(1/a) . So the smaller a is the larger we can take \beta (the smaller we can take the actual temperature 1/\beta ) for the result to hold. Conversely, if the coupling V is not small ( a sizeable) then the temperature has to be higher for the validity of our derivation. Literature. The question of separability of thermal equilibrium states has a rich history. Entanglement is a measure for the degree of their “quantumness”. Intuitively it is expected that a quantum to classical transition happens at high temperature T , that is, entanglement disappears for large T . From a quantum information point of view, this means that only cool enough materials may be used as a quantum resource. Many works deal with spin (qubit) chains in equilibrium. They analyze how 2-qubit entanglement along the chain depends on various parameters [ 1 , 11 , 28 , 29 , 30 , 19 , 31 , 10 , 4 , 21 ] , in particular finding temperature bounds which guarantee entanglement or separability. In this situation, one can conveniently use concurrence to quanitfy entanglement. An analysis for two spins of arbitrary length (where concurrence cannot be used as an entanglement measure) was carried out in [ 24 ] . In that work, the validity of the PPT criterion was linked to properties of the spin correlators. More generally, it was shown in [ 20 , 23 ] that for finite dimensional systems, thermal states with respect to any Hamiltonian are separable at high enough temperature, T\geq T_{c} , for some