Non-Gaussian Entanglement Hierarchy Based on the Schmidt Number

This paper introduces a quantitative witness, $E_{\rm NG}$, that establishes a natural hierarchy of non-Gaussian entanglement in bipartite bosonic systems by providing a lower bound on the Schmidt number irreducible by Gaussian transformations, offering both a sharp theoretical framework for pure states and an experimentally economical measurement protocol for identifying these resources.

The Gist

The Big Picture: Sorting Quantum "Messiness"

Imagine you are trying to organize a massive library of books (quantum states). Some books are neatly written in a standard font (Gaussian states), while others are written in wild, chaotic, handwritten scribbles (non-Gaussian states).

In the world of quantum physics, "entanglement" is like a magical thread tying two books together so that what happens to one instantly affects the other. This thread is the fuel for future quantum computers and super-precise sensors.

However, not all magical threads are created equal. Some threads can be tied using simple, standard tools (Gaussian operations). Others require complex, custom-built machinery (non-Gaussian operations). The problem is: How do we tell the difference? And more importantly, how do we measure how "strong" or "complex" the complex threads are?

This paper introduces a new tool to answer those questions.

The Problem: The "Standard" Ruler Doesn't Work

For the neat, standard books (Gaussian states), scientists already have a perfect ruler to measure the magic threads. But for the chaotic, scribbled books (non-Gaussian states), the old ruler breaks. It can't see the complexity hidden in the higher-order scribbles.

Furthermore, there's a specific type of "super-thread" called non-Gaussian entanglement. This is the kind of thread you cannot make just by using standard tools on simple, unentangled books. You need special, non-standard tools. The paper notes that some famous quantum states (like the "NOON" state, used for ultra-precise measurements) are this special type, but we didn't have a good way to prove it or measure their "depth."

The Solution: A New "Complexity Witness" (ENG)

The authors invented a new measuring stick called ENG. Think of it as a "stress test" for quantum states.

Here is how the test works, using a Kitchen Analogy:

1. The Setup: Imagine you have a messy, complicated dish (a quantum state).
2. The Test: You are allowed to use a set of standard kitchen tools (Gaussian operations) to try and simplify the dish. You can chop, mix, and heat it, but only with standard tools.
3. The Goal: Can you turn this messy dish into a simple, plain sandwich (a separable state) using only those standard tools?
   • If YES: The dish was just a "Gaussian-entanglable" state. It looked complex, but it was actually just a fancy version of a simple sandwich. The test result is 1.
   • If NO: Even after trying every possible combination of standard tools, the dish remains a complex, unique stew that cannot be simplified. This means it has Non-Gaussian Entanglement. The test result is greater than 1.

The Hierarchy: Counting the Layers of Complexity

The paper doesn't just say "Yes, it's complex" or "No, it's simple." It creates a ladder of complexity.

• Level 1: The dish can be simplified to a plain sandwich. (No special non-Gaussian entanglement).
• Level 2: You can simplify it, but you are left with a "core" that requires at least 2 ingredients to describe.
• Level 3: The core requires 3 ingredients.
• And so on...

The number you get from the test (rounded up) tells you the minimum number of ingredients needed to describe the "core" of the dish after you've stripped away everything the standard tools could remove.

Why does this matter?
The paper connects this to learning. If you want to teach a computer to recognize this specific quantum dish, the higher the level on the ladder, the harder it is to learn.

• Level 1: Easy to learn (like learning to recognize a sandwich).
• Level 10: Very hard to learn (like learning to recognize a complex, multi-layered cake).

Real-World Examples Tested

The authors tested their new ruler on famous quantum "dishes":

• NOON States: These are like super-sensitive rulers used in quantum metrology. The paper confirms that for small versions (1 or 2 photons), they are actually just fancy sandwiches (Level 1). But once you get to 3 or more photons, they become true "complex stews" (Level 2 or higher) that standard tools can't simplify.
• Squeezed Kerr States: These are states created by a specific type of non-linear interaction (like a spring that gets stiffer the more you pull it). The paper shows that as you pull the spring harder, the complexity level rises, making the state harder to learn but potentially more powerful.

Robustness and Practicality

The paper also checks if this test breaks if the "dish" gets spoiled (noise or loss).

• Result: The test is surprisingly tough. Even if the dish loses some ingredients (due to noise), the test can still detect the complexity, though the score drops slightly.

Finally, the authors realized that doing the full "stress test" on every single dish is too slow and expensive (it requires full state tomography, which is like photographing every single atom in the dish).

• The Shortcut: For the specific "NOON" dishes, they created a quick-check version. Instead of analyzing the whole dish, you only need to check four specific spots (four measurements). If those four spots show a certain pattern, you know immediately that the dish is a "complex stew" and not a simple sandwich.

Summary

• The Goal: To find a way to measure how "truly complex" quantum entanglement is, specifically for the kind that standard tools can't create.
• The Tool: A new number (ENG) that acts like a stress test. If the number is 1, it's simple. If it's higher, it's complex.
• The Benefit: It creates a ladder of complexity. The higher you are, the harder the state is to learn, but the more powerful it might be for quantum tasks.
• The Application: It helps scientists identify which quantum resources are "premium" (non-Gaussian) and provides a practical, quick way to check for them in the lab without needing expensive, full-scale equipment.

Technical Summary

Technical Summary: Non-Gaussian Entanglement Hierarchy Based on the Schmidt Number

Problem Statement
Entanglement is a fundamental resource for quantum communication, computation, and sensing. In continuous-variable (CV) bosonic systems, states and operations are naturally divided into Gaussian and non-Gaussian classes. While Gaussian operations can generate entanglement from separable inputs, they are fundamentally limited; they cannot achieve entanglement distillation, loophole-free Bell inequality violations, or universal quantum computation. Consequently, non-Gaussian entanglement is essential for next-generation bosonic quantum technologies.

However, characterizing non-Gaussian entanglement remains a significant challenge. Unlike Gaussian states, where the covariance matrix and criteria like the Peres-Horodecki positive partial transpose (PPT) condition provide complete descriptions, non-Gaussian states require higher-order correlations. Existing tools, such as witnesses based on higher-order moments or full state tomography, are often incomplete or experimentally costly. Furthermore, while the Schmidt number provides a hierarchy of entanglement dimensionality in finite-dimensional systems, an analogous quantitative hierarchy for non-Gaussian bosonic entanglement—specifically one that distinguishes entanglement irreducible by Gaussian transformations—has been an open challenge.

Methodology
The authors introduce a quantitative witness, denoted as $E_{NG}$, designed to certify and hierarchically classify non-Gaussian entanglement in bipartite bosonic systems.

1. Definition of the Witness: The witness is defined as the minimum trace norm of the partial transpose of a state $\rho$ after optimization over all two-mode Gaussian unitaries $\hat{U}_G$:
   $$E_{NG} := \inf_{\hat{U}_G \in \mathcal{G}_2} \left\| (\hat{U}_G \rho \hat{U}_G^\dagger)^{T_A} \right\|_1$$
   Here, $\mathcal{G}_2$ represents the set of all two-mode Gaussian unitaries, and $T_A$ denotes the partial transpose with respect to mode A.

2. Classification of States:

   • Gaussian-Entanglable (GE) States: States that can be generated by applying a Gaussian unitary to a separable input (which may be non-Gaussian, e.g., Fock states). For all GE states, $E_{NG} = 1$.
   • Non-Gaussian Entangled (NGE) States: States where $E_{NG} > 1$. This certifies that the entanglement cannot be removed or generated by Gaussian operations acting on separable inputs.
   • Genuinely Non-Gaussian Entangled (GNGE) States: A stricter class defined as states outside the convex hull of GE states. For pure states, $E_{NG} > 1$ is a sharp condition for GNGE.

3. Hierarchy Construction: The authors define a hierarchy based on the ceiling of the witness, $d = \lceil E_{NG} \rceil$. This integer $d$ serves as a lower bound on the Schmidt number that remains after optimizing over Gaussian transformations.

   • $d=1$: The state is Gaussian-entanglable.
   • $d > 1$: The state possesses non-Gaussian entanglement of dimension at least $d$.

4. Connection to State Learning: For pure states, the hierarchy is linked to the complexity of state learning. If a Gaussian unitary $\hat{U}_G^{(opt)}$ achieves the infimum, the state can be mapped to a "core" state $|\psi_{core}\rangle$. The number of real parameters required to learn the target state is lower-bounded by $N = 2d(2D - d) - 2$, where $D$ is the local Hilbert space dimension. Thus, higher $d$ implies higher learning complexity.

5. Experimental Practicality: To avoid the cost of full state tomography required for calculating $E_{NG}$, the authors derive a specific, experimentally economical witness for NOON-type states. This witness, $f = \text{Tr}[F\rho]$, requires only four projective measurements. An analytical threshold $f_G$ is derived for Gaussian-entanglable states; any state with $f > f_G$ is certified as genuinely non-Gaussian entangled.

Key Results

• Benchmarking on Paradigmatic States:
  • NOON States: The framework correctly identifies NOON states $|N, 0\rangle + |0, N\rangle$ as Gaussian-entanglable for $N=1, 2$ ($E_{NG}=1$), but certifies them as genuinely non-Gaussian entangled for $N \ge 3$ ($E_{NG} > 1$).
  • Squeezed States: Superpositions of two-mode squeezed vacuum (TMSV) states and tri-squeezed states generated by three-photon spontaneous parametric down-conversion (SPDC) show increasing $E_{NG}$ and $d$ with nonlinear strength, indicating growing non-Gaussian entanglement dimensions.
  • Kerr Nonlinearity: In a two-mode squeezed Kerr system, the interplay between self-Kerr nonlinearity and two-mode squeezing generates a hierarchy of non-Gaussian entanglement, with $d$ increasing as a function of detuning and squeezing strength.
• Robustness: The witness $E_{NG}$ demonstrates robustness against photon loss. In lossy Kerr states, while both standard entanglement witnesses and $E_{NG}$ decrease with loss, a significant portion of the non-Gaussian character is retained (e.g., ~89% retention at 10% loss rate).
• Entanglement Distillation: The paper demonstrates that non-Gaussian entangled states (where $E_{NG} > 1$) serve as effective resources for entanglement distillation in Gaussian settings. Using a two-copy distillation protocol with Gaussian unitaries and homodyne measurements, the authors show that the distilled state exhibits enhanced entanglement compared to the input, a feat impossible with purely Gaussian resources.
• NOON-Type Witness: The derived practical witness for NOON states requires only four density-matrix elements. The analytical threshold $f_G$ decreases with $N$ (for $3 \le N \le 8$), implying that higher-$N$ NOON states are easier to certify as genuinely non-Gaussian entangled.

Significance and Claims
The paper establishes an operationally meaningful and experimentally accessible framework for identifying non-Gaussian entanglement resources in continuous-variable platforms. By extending the Schmidt-number paradigm to the CV non-Gaussian regime, the work provides:

1. A quantitative hierarchy of non-Gaussian entanglement dimensions that is irreducible by Gaussian transformations.
2. A direct link between this hierarchy and the complexity of learning quantum states.
3. A practical route to identifying resources for tasks such as entanglement distillation, which are unattainable with Gaussian resources alone.

The authors conclude that their results offer a new avenue for characterizing entanglement and identifying essential resources for quantum technologies, including quantum metrology, computing, and communication. The work also notes the recent independent development of fidelity-based certification methods for genuine non-Gaussian entanglement.