On the existence of fully inseparable biseparable Gaussian states

This paper investigates fully inseparable biseparable Gaussian states and, through numerical analysis of archetypical families using finite-dimensional projections and entanglement witnesses, provides evidence supporting the conjecture that all fully inseparable Gaussian states are in fact genuinely multipartite entangled.

The Gist

The Big Picture: The "Entanglement" Puzzle

Imagine you have a group of three friends (let's call them Alice, Bob, and Charlie) who are playing a complex quantum game. In this game, entanglement is like a special, unbreakable bond where their actions are perfectly coordinated, no matter how far apart they are.

Physicists usually care about two types of this bond:

1. Genuine Multipartite Entanglement (GME): This is the "gold standard." It means Alice, Bob, and Charlie are all tangled up together in a single, inseparable knot. You can't split them into pairs without breaking the magic.
2. Fully Inseparable: This sounds similar, but it's a slightly looser definition. It means the group is so tangled that you can't separate any one person from the rest. However, mathematically, it might be possible that the group is just a "mixture" of different pairs being tangled in different ways, rather than one big three-way knot.

The Question: The authors ask: Is it possible to have a group that is "fully inseparable" (you can't split them) but is NOT "genuinely" tangled (it's just a mix of pairs)?

In the world of general quantum states, the answer is yes. You can have a "fake" three-way knot that is actually just a cocktail of two-way knots.

The Specific Focus: This paper looks at a specific, very common type of quantum state called Gaussian states. These are like the "smooth, round, predictable" states of the quantum world (think of them like a perfectly smooth hill, as opposed to a jagged, rocky mountain). The authors wanted to know: Do these "smooth" Gaussian states have this "fake knot" loophole, or are they always truly tangled?

The Investigation: Smoothing vs. Shaking

The researchers took several families of these "smooth" Gaussian states. They knew these states were "fully inseparable" (you couldn't split the group), but they also knew that, based on a standard test (looking only at the average position and speed of the particles), these states looked like they could be faked by mixing simpler pairs.

To find out if they were truly "Genuine" (a real three-way knot) or just "Faked" (a mix of pairs), the authors used a clever trick: Projection.

The Analogy: The 3D Sculpture and the Shadow
Imagine a complex 3D sculpture (the full quantum state). If you shine a light on it, you get a 2D shadow.

• The authors took their complex 3D quantum sculpture and projected it onto smaller, simpler 2D screens (finite-dimensional subspaces).
• They then checked these simpler 2D shadows for the "Genuine" knot.
• The Rule: If the simple shadow has a genuine knot, the original 3D sculpture must have had a genuine knot too. (You can't create a knot by squashing a shape down; you can only lose them).

They did this projection with increasing levels of detail:

1. Low Detail: Looking at the state as if it were made of simple "coins" (qubits).
2. Medium Detail: Looking at it as "dice" (qutrits).
3. High Detail: Looking at it as "four-sided dice" (ququarts).

The Findings: The Loophole Shrinks

Here is what they discovered as they increased the detail of their "shadows":

• At low detail: Some states looked like they might be fakes. The "Genuine" knot wasn't obvious.
• At medium detail: The "fake" area started to shrink. The states looked more and more like genuine knots.
• At high detail: The area where the state could be a fake almost disappeared. The more closely they looked, the more it became clear that the state was actually a genuine three-way knot.

The Metaphor: Imagine trying to identify a fake diamond.

• With a naked eye (low detail), it looks real.
• With a magnifying glass (medium detail), you see a tiny flaw that suggests it might be fake.
• With a high-powered microscope (high detail), you realize the "flaw" was just a trick of the light, and the stone is actually a perfect, genuine diamond.

In this paper, the "flaw" was the possibility that the state was a mixture of pairs. As they looked closer (increased the dimension of the projection), that possibility vanished.

The Conclusion: A Strong Guess

The authors did not find a single example of a "Gaussian state" that was fully inseparable but not genuinely entangled. In fact, every time they looked closer, the "fake" states turned out to be "real" ones.

They also noted a mathematical fact: If you mix different "smooth" (Gaussian) hills, you usually get a "bumpy" (non-Gaussian) shape. So, it's mathematically weird to think you could mix smooth states to get a smooth result that looks like a mix but isn't.

The Final Claim:
Based on all their tests, the authors propose a conjecture (a strong scientific guess):

All "fully inseparable" Gaussian states are actually "genuinely multipartite entangled."

In plain English: If a smooth quantum state is tangled enough that you can't split the group, it is definitely a real, three-way (or multi-way) knot. There are no "fake" knots in the smooth world of Gaussian states.

Why This Matters (According to the Paper)

If this guess is true, it makes life much easier for scientists.

• Before: To prove a state is truly entangled, you had to do very difficult, complex tests.
• After (if the guess is true): You only need to check if the state is "fully inseparable" (which is an easier test). If it passes that, you automatically know it's genuinely entangled.

The paper admits they haven't proved this 100% (mathematically, a counter-example could still exist), but their evidence is so strong that they are betting their reputation on it.

Technical Summary

Technical Summary: On the existence of fully inseparable biseparable Gaussian states

Problem Statement
The paper addresses the distinction between two quantum resources in continuous-variable (CV) systems: genuine multipartite entanglement (GME) and full inseparability. While all GME states are fully inseparable (entangled with respect to all bipartitions), the converse is not generally true for arbitrary quantum states. There exist "fully inseparable biseparable" (FIB) states, which are entangled across every bipartition but can be expressed as a convex mixture of states separable with respect to different partitions.

The specific problem investigated is whether such FIB states exist within the class of Gaussian states. Gaussian states are fully determined by their first and second moments (mean values and covariance matrix). A known limitation is that the "covariance matrix (CM) biseparability criterion" (Eq. 11) can detect GME only if a state's CM cannot be decomposed into a convex sum of CMs of partition-separable states. However, for any CM satisfying this decomposition, there exists a non-Gaussian biseparable state. Consequently, standard criteria based solely on second moments cannot distinguish between a Gaussian state that is genuinely multipartite entangled and one that is merely FIB, because the Gaussian state shares the same CM as a biseparable non-Gaussian state.

Methodology
The authors systematically investigate several archetypical families of three-mode Gaussian states that are fully inseparable (verified via the CV PPT criterion) yet satisfy the CM biseparability criterion (Eq. 11). To detect GME in these states, the authors employ a strategy of projecting the infinite-dimensional Gaussian states onto finite-dimensional subspaces (local projections to qubits, qutrits, and ququarts). Since local projections cannot create entanglement, if the projected finite-dimensional state is detected as GME, the original Gaussian state must also be GME.

The detection of GME in the projected states utilizes two primary methods:

1. Witness Inequality (10): A necessary condition for biseparability based on off-diagonal and diagonal elements of the density matrix, derived from Ref. [20] and extended in [31].
2. Fully Decomposable Witnesses: A semidefinite program (SDP) approach (Eq. 8) that distinguishes convex mixtures of PPT states for different bipartitions from GME states.

The density-matrix elements required for these tests are calculated from the Gaussian covariance matrices using quasiprobability distributions (Wigner, Husimi Q, and Glauber-Sudarshan P functions), specifically employing multidimensional Hermite polynomials for computational efficiency (Appendix A).

Key Contributions and Results
The study examines four families of Gaussian states:

1. Vacuum + TMSV: Gaussian states derived from mixing a two-mode squeezed vacuum (TMSV) with a vacuum state.
2. SMSV + TMSV: Mixing TMSV with a single-mode squeezed vacuum.
3. Thermal + TMSV: Mixing TMSV with a thermal state.
4. Coherent + TMSV: Mixing TMSV with a coherent state (non-zero first moments).
5. Noisy GHZ-like states: Mixing a pure GHZ-like Gaussian state with vacuum noise.

For all investigated families, the authors find that while the CMs satisfy the biseparability decomposition condition (11), the states are detected as genuinely multipartite entangled within specific non-trivial parameter ranges. Crucially, the region of parameters where GME is detected expands as the dimension of the local projection subspace increases (from qubits $d=2$ to qutrits $d=3$ to ququarts $d=4$). As the projection dimension grows, the "gap" between the region of full inseparability and the region where GME is detected shrinks, approaching the boundary of full inseparability.

Significance and Claims
The paper does not claim to provide a rigorous proof that FIB Gaussian states do not exist, nor does it present a counterexample. Instead, the authors formulate a conjecture based on their findings: All fully inseparable Gaussian states are genuinely multipartite entangled.

The significance of this conjecture, if true, is twofold:

1. Theoretical Simplification: It would imply that for Gaussian states, full inseparability and GME are equivalent. This would significantly simplify the detection of GME, as checking for full inseparability (which relies only on the CM) would be sufficient to confirm GME, bypassing the need for higher-moment analysis or complex witness constructions.
2. Implications for Multi-copy Activation: The paper notes that multi-copy GME activation (where multiple copies of a state reveal entanglement not present in a single copy) has been shown to exist in finite-dimensional and general infinite-dimensional systems. If the conjecture holds, this phenomenon would not exist for Gaussian states, as their single-copy full inseparability would already guarantee GME.

The authors conclude that while the question remains open due to the lack of a formal proof or a constructed counterexample, the observed behavior of detection regions with increasing projection dimensions and the absence of any known construction for FIB Gaussian states strongly support the conjecture that such states do not exist.