⚛️ quantum physics

Estimating the best separable approximation of non-pure spin-squeezed states

This paper presents a method to quantitatively estimate the distance of mixed collective spin states to the set of fully separable states by deriving lower bounds from spin-squeezing inequalities and refining iterative algorithms to find optimal separable approximations, thereby enabling the precise characterization of entanglement in thermal and non-equilibrium phases of fully-connected XXZ models.

Original authors: Julia Mathé, Ayaka Usui, Otfried Gühne, Giuseppe Vitagliano

Published 2026-04-15

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Julia Mathé, Ayaka Usui, Otfried Gühne, Giuseppe Vitagliano

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you have a giant, complex machine made of thousands of tiny, spinning tops (these are our "spins" or quantum particles). Sometimes, these tops spin in perfect, independent harmony. Other times, they get tangled up in a mysterious, invisible web where the movement of one instantly affects the others. This "tangling" is called quantum entanglement, and it's the superpower that makes quantum computers so powerful.

However, in the real world, things are messy. The machine gets hot, it vibrates, and the tops don't just sit in a perfect, frozen state; they are a chaotic mix of possibilities. This is what physicists call a mixed state.

The big question this paper asks is: "How tangled is this messy machine?"

Measuring this "tangle" is incredibly hard. It's like trying to figure out exactly how much of a bowl of fruit salad is made of real fruit versus how much is just air or fake plastic. You can't just look at it; you have to do some heavy math.

Here is how the authors of this paper solved the problem, using some clever tricks:

1. The "Detective" vs. The "Mapmaker"

To measure the tangle, the authors used two different strategies, like a detective and a mapmaker working together.

The Detective (Lower Bounds): The detective looks for clues that prove the tops are tangled. They use a set of rules called Spin-Squeezing Inequalities (SSIs).
- The Analogy: Imagine you have a group of people holding hands in a circle. If they are just standing randomly, their movements are loose. But if they are "squeezing" together so tightly that if one moves, everyone else must move, that's a sign of a secret connection.
- The authors found a way to look at the "squeezing" of the whole group and instantly calculate a minimum guarantee of how tangled they are. It's like saying, "Based on how tight this knot is, there is at least this much rope involved." This is fast, easy to calculate, and works even for huge groups of particles.
The Mapmaker (Upper Bounds): The mapmaker tries to find the exact "untangled" version of the machine to see how far the real machine is from being simple.
- The Analogy: Imagine you have a messy pile of laundry (the entangled state). The mapmaker tries to fold it into a perfectly neat stack (the separable state) to see how much effort it took to mess it up.
- The problem is, the pile is huge, and there are billions of ways to fold it. The authors improved an old algorithm to do this folding job much faster by realizing that the laundry has a pattern (symmetry). Instead of checking every single sock, they realized, "Hey, all the socks are the same!" and folded them in batches. This gave them a maximum limit on the tangle.

2. The "Thermal" Surprise

Usually, physicists only study these machines when they are ice-cold (at absolute zero), where things are predictable. But this paper looked at what happens when the machine is warm (at a non-zero temperature).

They discovered something surprising:

The "Ordered" Trap: In some parts of the machine's "phase diagram" (a map of its behavior), the ground state (the coldest, most stable state) is actually not tangled at all. It's just a bunch of independent tops.
The Heat Effect: But, as soon as you add a little bit of heat, entanglement suddenly appears!
- The Analogy: Think of a quiet library where everyone is sitting still (no tangle). If you start playing loud music (heat), people start dancing and bumping into each other, creating a chaotic, connected crowd. The authors found that in certain quantum systems, "heating it up" actually creates the quantum connections that were missing in the cold, quiet state.

3. Why This Matters

This paper is a big deal for three reasons:

It's Practical: The "Detective" method (the lower bound) is so simple that it can be calculated on a regular computer for systems with hundreds of particles. This is huge because usually, these calculations crash computers with just a few dozen particles.
It's Accurate: They found that their "Detective" clues are often perfect. When the machine is at the edge of becoming tangled or untangled, their math hits the bullseye.
It Opens New Doors: By showing that heat can create entanglement in places we didn't expect, they are helping us understand how quantum systems behave in the real world (where things are never perfectly cold). This could help us build better quantum sensors and understand how materials change their properties when they heat up.

The Bottom Line

The authors built a new toolkit to measure the "quantum messiness" of large groups of particles. They found a fast way to get a "good enough" answer (the lower bound) and a smarter way to get a "best possible" answer (the upper bound) by using the symmetry of the system. Most importantly, they showed that heat doesn't just destroy quantum magic; sometimes, it creates it.

1. Problem Statement

The paper addresses the challenge of quantitatively characterizing entanglement in mixed many-body quantum states, specifically collective spin systems at non-zero temperatures. While entanglement in pure ground states is well-studied, characterizing mixed states (e.g., thermal states) is computationally intractable for large systems ( $N$ particles) because:

Deciding separability is NP-hard.
Exact calculation of entanglement monotones (measures of entanglement) is generally impossible for $N > 3$ or $4$ qubits.
Existing methods often rely on ground-state paradigms or semi-classical approximations that miss quantum correlations arising from mixing or higher-order spin sectors.

The authors aim to bridge this gap by developing a framework to estimate the Best Separable Approximation (BSA)—a distance-based entanglement monotone representing the minimal distance from a state to the set of fully separable states—for collective spin models, leveraging symmetries and spin-squeezing inequalities.

2. Methodology

The authors propose a dual-bound approach to estimate the BSA, $E_{BSA}(\varrho)$ , which is defined as the minimum weight $t$ in the decomposition $\varrho = (1-t)\sigma + t\nu$ , where $\sigma$ is separable and $\nu$ is an arbitrary state.

A. Lower Bounds via Spin-Squeezing Inequalities (SSIs)

Instead of optimizing over a vast set of linear entanglement witnesses, the authors utilize a complete set of non-linear Spin-Squeezing Inequalities (SSIs) derived from collective spin variances.

Theoretical Basis: The set of separable states in the space of first and second collective spin moments forms a polytope. States outside this polytope are entangled.
Optimization: The authors derive a single, compact, rotation-invariant spin-squeezing parameter, $\xi_{SS}(\varrho)$ , which implicitly optimizes over the entire family of generalized SSIs.
Bounding: They prove that $\xi_{SS}(\varrho)$ can be rescaled to provide a rigorous lower bound on the BSA:
$E_{BSA}(\varrho) \geq -\frac{1}{N_K} \xi_{SS}(\varrho)$
where $N_K$ is a normalization factor dependent on the number of positive eigenvalues of a specific moment matrix $X_\varrho$ . This bypasses the need for numerical witness optimization.

B. Upper Bounds via Symmetry-Improved Iterative Ansatz

To find the upper bound (the distance to the closest separable state), the authors improve existing iterative algorithms:

Iterative Search: They employ a "see-saw" algorithm that alternates between finding the product state with the maximal overlap with the target state and updating the convex mixture of product states.
Symmetry Exploitation: Crucially, they enforce permutational invariance (PI) and rotational symmetry (about the $z$ $z$ -axis) on the ansatz.
- Instead of searching the full Hilbert space, they restrict the search to the symmetric subspace using twirling maps ( $T_{PI}$ and $P_z$ ).
- This reduces the effective dimensionality of the problem, allowing the algorithm to handle significantly larger $N$ (up to $\sim 10$ qubits for generic states, and much larger for symmetric thermal states) compared to standard methods limited to $N=3$ or $4$.
Initialization: They use thermal states at high temperatures (where separability is known) as initial seeds to accelerate convergence at lower temperatures.

C. Target System

The methods are applied to the fully-connected XXZ model (Lipkin-Meshkov-Glick type), which includes ferromagnetic (FM) and anti-ferromagnetic (AFM) regimes. The study focuses on thermal states $\varrho_T = e^{-H/T}/Z$ across different phases and phase transitions.

3. Key Contributions

Analytic Lower Bound: Derivation of a closed-form, optimal spin-squeezing parameter that provides a tight lower bound to the BSA using only second-order collective moments. This allows for analytic scaling analysis in the thermodynamic limit ( $N \to \infty$ ).
Symmetry-Adapted Algorithm: Development of an iterative algorithm that exploits permutational and rotational symmetries to find upper bounds for the BSA, significantly extending the computable system size for mixed-state entanglement quantification.
Thermal Entanglement Characterization: Application of these methods to the XXZ model, revealing that entanglement persists in thermal states even in phases where the ground state is separable (ordered phases), particularly near quantum critical points.
Analytic Scaling: Derivation of asymptotic expressions for the BSA lower bound in the thermodynamic limit for specific models (e.g., the Heisenberg model), showing that the entanglement threshold temperature scales as $T_c \propto g$ (coupling strength).

4. Key Results

Tightness of Bounds:
- For $N=3$ qubits, the lower and upper bounds coincide across all temperatures, validating the method's accuracy for small systems.
- For larger $N$ , the lower bound (SSI-based) is often tight at $T=0$ and at the critical temperature where entanglement vanishes.
- The upper bound is generally looser for weakly entangled states but effectively captures full separability and strong entanglement.
Entanglement in Ordered Phases: The study observes that entanglement can arise at non-zero temperatures even when the ground state is separable (e.g., in the ferromagnetic phase). This occurs when the system is near a quantum phase transition (QPT), suggesting that thermal fluctuations can activate quantum correlations in regimes where the ground state is classical.
Phase Diagrams: The SSI-based lower bound successfully maps out entanglement regions in the phase diagram, distinguishing between fully polarized, singlet-dominated, and Dicke-like phases. The entanglement is not uniform within a phase; it is strongest near critical points and weaker in fully polarized regions.
Scalability: The symmetry-reduced algorithm allows for the calculation of separable decompositions for $N \approx 10$ qubits (significantly higher than the $N=3,4$ limit for generic states), and the lower bound is computable for $N$ up to hundreds or analytically for $N \to \infty$ .

5. Significance and Outlook

Beyond Ground States: The work moves the field beyond the "ground-state paradigm," providing tools to quantify entanglement in physically relevant mixed states (thermal and non-equilibrium).
Experimental Relevance: Since the lower bounds rely only on collective spin moments (variances and covariances), they are directly measurable in experiments (e.g., cold atomic gases) without full state tomography.
Thermodynamic Connection: The results link entanglement measures to thermodynamic quantities (like energy and susceptibility) via fluctuation-dissipation relations, suggesting that entanglement can be inferred from standard thermodynamic measurements.
Future Directions: The authors highlight the potential to apply these methods to out-of-equilibrium dynamics (e.g., global quenches) and systems with reduced symmetries (e.g., translational but not permutational), which remains a challenge for current numerical methods.

In summary, this paper provides a robust, scalable, and experimentally accessible framework for quantifying entanglement in mixed collective spin systems, demonstrating that quantum correlations are a ubiquitous feature of thermal states, even in phases traditionally considered classical.