⚛️ quantum physics

Bosonic quantum communication beyond the thermal threshold

This paper demonstrates that non-Gaussian input states can achieve positive quantum capacity for bosonic thermal attenuators in high-noise regimes where the standard Gaussian-based lower bound vanishes, thereby proving that quantum communication is possible beyond previously established thermal thresholds.

Original authors: Francesco Anna Mele, Giuseppe Catalano, Marco Fanizza, Vittorio Giovannetti, Ludovico Lami

Published 2026-07-31

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Francesco Anna Mele, Giuseppe Catalano, Marco Fanizza, Vittorio Giovannetti, Ludovico Lami

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 are trying to send a secret message through a very noisy, crowded room. In the world of quantum physics, this "room" is a channel that carries light particles (photons) to transmit information. The challenge is that the room is filled with "thermal noise"—like a chaotic crowd of people bumping into your message, scrambling it. Scientists have long been trying to figure out the absolute maximum speed at which you can send quantum secrets through this noisy room without them getting lost. This speed limit is called the "quantum capacity."

For decades, physicists have relied on a specific type of message carrier called a "Gaussian state." Think of these as perfectly round, smooth balloons. They are the standard, easy-to-make tools for sending quantum information, and for a long time, everyone assumed that if you wanted to get the best possible speed, you just needed to find the perfect size and shape for these balloons. The big question was: Could a weirdly shaped, lumpy, or jagged balloon (a "non-Gaussian" state) squeeze through the noise better than the smooth ones? Until now, no one knew for sure if the smooth balloons were truly the best, or if there was a secret, weird shape that could outperform them in the noisiest conditions.

This paper, titled "Bosonic quantum communication beyond the thermal threshold," answers that question with a resounding "yes." The researchers, a team of physicists from Italy and France, discovered that the old rulebook was incomplete. They proved that the best "smooth balloon" strategy actually hits a hard wall: in certain very noisy conditions, the best smooth balloon can't send any information at all. However, by using a strange, lumpy, non-Gaussian shape, they found a way to send a tiny but real amount of information through that same wall.

The team focused on a specific scenario where the noise is high (one thermal photon in the environment) and the signal gets weakened to 80% of its original strength (a transmissivity of $\eta=0.8$ ). Under these conditions, the old "smooth balloon" method predicts that the quantum capacity is zero—meaning no information can get through. But the authors constructed a specific, weird quantum state (a "rank-two non-Gaussian state" supported on only six energy levels) that defies this. They mathematically proved that this strange state can achieve a coherent information of at least $4.7 \times 10^{-4}$ qubits per channel use. While that number sounds tiny, it is strictly positive, meaning communication is possible where the old theory said it was impossible.

To be absolutely sure their math wasn't just a lucky computer glitch, the authors didn't just run a simulation; they used a rigorous "computer-assisted certification" method. This is like checking a math proof with a calculator that never makes a rounding error, ensuring the result is mathematically bulletproof. They found that while their specific weird shape works, there are even better shapes hidden in the math. When they let a computer search for the best possible weird shapes, it found a strategy that could push the information rate up to at least $8.4 \times 10^{-3}$ qubits per channel use at that same point.

The paper also maps out exactly where this new trick works. They identified a "danger zone" between two thresholds: a lower limit where the channel is so noisy it's impossible to send anything (the "antidegradability threshold" at $\eta \le 0.75$ ) and an upper limit where the old smooth balloons start working again (the "thermal threshold" at $\eta = 0.8$ ). In the gap between 0.75 and 0.8, the old theory said the door was locked. The new research proves the door is actually unlocked, provided you use the right weird key. They even found a specific point at $\eta = 0.7841$ where they can rigorously prove the capacity is positive, expanding the known safe zone for quantum communication.

Interestingly, the authors were honest about how they found this solution. They admitted that they had tried to find this "weird shape" for years using traditional methods and even with earlier versions of AI, but failed. It was only when they used a newer, more advanced AI model (ChatGPT 5.6 Sol) that the AI suggested the specific families of non-Gaussian states that worked. The authors emphasize that while the AI was the crucial tool that handed them the key, the heavy lifting of proving it worked and understanding the physics was done entirely by the human researchers.

In short, this paper shatters a long-held assumption in quantum physics. It shows that in the noisiest, most difficult environments, sticking to the standard, smooth "Gaussian" tools isn't enough. To push the boundaries of how much quantum information we can send, we need to embrace the weird, the lumpy, and the non-Gaussian. This discovery opens up new high-noise regimes where reliable quantum communication is now proven to be possible, changing the map of what is achievable in the quantum world.

Technical Summary: Bosonic Quantum Communication Beyond the Thermal Threshold

Problem Statement
The quantum capacity $Q(\mathcal{N})$ of a noisy quantum channel $\mathcal{N}$ , defined as the regularized coherent information, remains unknown for the bosonic thermal attenuator $\Phi_{\eta, \nu}$ when the environment contains thermal noise ( $\nu > 0$ ). While the capacity of the pure-loss channel ( $\nu=0$ ) is exactly known due to degradability, the thermal attenuator is generally non-degradable.

Historically, the best-known lower bound on $Q(\Phi_{\eta, \nu})$ was established by Holevo and Werner (1999) using thermal input states. This bound vanishes when the transmissivity $\eta$ falls below the "thermal threshold" $\eta_G(\nu)$ . Conversely, the channel is known to be antidegradable (and thus has zero capacity) only when $\eta \leq \eta_{AD}(\nu)$ , where $\eta_{AD}(\nu) < \eta_G(\nu)$ . This leaves an intermediate region $\eta_{AD}(\nu) < \eta \leq \eta_G(\nu)$ where the thermal-state lower bound is zero, yet the channel is not antidegradable. The central open questions addressed are:

Do single-mode Gaussian states maximize the coherent information of the thermal attenuator?
Is the quantum capacity strictly positive in the intermediate region where the thermal-state lower bound vanishes?

Methodology
The authors employ a combination of analytical proofs and computer-assisted numerical certification:

Optimality of Thermal States within Gaussian Class: The authors first prove that among all single-mode Gaussian states (including squeezed states), the supremum of the coherent information is achieved by thermal states in the infinite-energy limit. This confirms that the Holevo–Werner bound is the exact optimum for the Gaussian subclass.
Construction of Non-Gaussian Witnesses: To disprove Gaussian optimality, the authors construct specific rank-two non-Gaussian states supported on a finite number of Fock levels. These states are mixtures of orthogonal superpositions of Fock states with specific parity properties (e.g., modulo-4 and modulo-5 structures).
Finite-Dimensional Approximation and Certification: Since the thermal attenuator acts on an infinite-dimensional Hilbert space, the authors introduce a "cutoff-erasure" channel $T_M$ that truncates the output to the first $M$ Fock levels and flags higher levels as erasures. By the data processing inequality, the coherent information of the truncated channel provides a rigorous lower bound for the original channel.
Rigorous Interval Arithmetic: To transform numerical estimates into mathematical proofs, the authors utilize interval arithmetic. They represent matrix entries as certified intervals and use Gershgorin's theorem on a transformed basis to bound the eigenvalues of the resulting density matrices. This ensures that rounding errors do not invalidate the positivity of the coherent information.

Key Contributions and Results

Disproof of Gaussian Optimality: The paper proves that the supremum of coherent information over single-mode Gaussian states is strictly less than the supremum over all states. Specifically, at the thermal threshold point $(\eta=0.8, \nu=1)$ , the coherent information for any single-mode Gaussian input is zero, whereas a specific non-Gaussian state yields a strictly positive value.
Explicit Non-Gaussian Witness: The authors define a rank-two state $\rho_\star$ supported on six Fock levels (specifically levels 0, 4, 8, 1, 5, 9). They rigorously certify that for $\Phi_{0.8, 1}$ , the coherent information satisfies $I_c(\rho_\star, \Phi_{0.8, 1}) \geq 4.7 \times 10^{-4}$ qubits per channel use.
Extension of the Positive Capacity Region: By optimizing over families of non-Gaussian states, the authors identify a region where the quantum capacity is strictly positive even below the thermal threshold.
- At $\nu=1$ , they certify $Q(\Phi_{\eta, 1}) > 0$ down to $\eta = 0.7841$ .
- This is significant because the antidegradability threshold is $\eta_{AD}(1) = 0.75$ . Thus, the paper establishes reliable quantum communication in the interval $0.75 < \eta \leq 0.7841$ , a regime where previous bounds suggested zero capacity.
Numerical vs. Certified Bounds: While the certified lower bound is conservative ( $4.7 \times 10^{-4}$ ), numerical optimization over similar non-Gaussian families suggests values up to $8.4 \times 10^{-3}$ at the same point, indicating the certified witness is far from optimal but sufficient to prove the existence of a positive gap.

Significance
The paper fundamentally alters the understanding of bosonic quantum communication in high-noise regimes. It demonstrates that restricting optimization to Gaussian states is insufficient for determining the quantum capacity of thermal attenuators. The results establish that reliable quantum communication is possible in parameter regions previously thought to be inaccessible (below the thermal threshold), provided non-Gaussian encoding strategies are employed.

The work implies that a complete theory of noisy bosonic quantum communication must account for non-Gaussian inputs, moving beyond the "Gaussian optimizer" conjecture which held for other bosonic channels. The authors note that AI (specifically ChatGPT 5.6 Sol) played a crucial role in identifying the specific non-Gaussian state families that served as counterexamples, though the mathematical proofs and certification remain the authors' responsibility.

Technical Summary: Bosonic Quantum Communication Beyond the Thermal Threshold

More like this