Capacity of multimode quantum Gaussian channels

This paper derives explicit formulas for the capacity of multimode quantum Gaussian channels, demonstrating that increasing the number of modes is always optimal under fixed power constraints and providing analytical results for ensemble-averaged Holevo capacities under random passive transformations, homodyne, and heterodyne detection.

The Gist

The Big Picture: Sending More Messages Through the Same Pipe

Imagine you are trying to send a massive amount of data (like a movie or a huge file) from one place to another using light. In the old days, we thought of this like sending a single stream of water through a single hose. But modern technology allows us to use Multiple-Input Multiple-Output (MIMO) systems. Think of this not as one hose, but as a whole garden sprinkler system with dozens of nozzles sending water (light) at the same time.

This paper asks a fundamental question: If we have a limited amount of energy (power) to send our light, how many "nozzles" (modes) should we use to send the most information?

The authors, Maria Popławska and Marcin Jarzyna, use the laws of quantum mechanics (the rules that govern how tiny particles like photons behave) to answer this. They found that using more modes is almost always better, even if the total power stays the same.

The Core Concepts

1. The Quantum "Noise" Problem

In the real world, light doesn't travel perfectly. It hits dust, air, or fibers, which creates "noise."

• Classical View: Imagine a radio signal getting static. You can just turn up the volume to overcome it.
• Quantum View: The paper explains that at the quantum level, there is a "floor" of noise that you cannot eliminate. It's like trying to hear a whisper in a room where the air itself is constantly buzzing with a faint hum. You can't turn the volume up forever because the quantum laws say there's a limit to how clearly you can distinguish the signal from that hum.

2. The "Water-Filling" Strategy

The paper describes a clever way to distribute your limited energy. Imagine you have a bumpy floor (representing the different paths or "modes" your light can take). Some paths are smooth and clear (high quality), while others are full of holes and rocks (high noise).

If you pour a bucket of water (your power) onto this floor, the water naturally fills the deepest holes first.

• The Paper's Finding: To get the best result, you shouldn't pour water evenly everywhere. You should pour it into the "deepest" (best) paths first. This is called the water-filling algorithm.
• The Surprise: Even with this smart strategy, the paper shows that if you keep adding more paths (modes) to your system, the total amount of information you can send keeps growing. It's like having a giant field of pipes; even if some are clogged, having more pipes gives you more total capacity than just a few perfect pipes.

3. Random Scattering (The "Whirling Dervish" Effect)

Sometimes, the path your light takes isn't fixed. Imagine throwing a ball through a room full of spinning fans (random scatterers). The ball might bounce off a fan here, a wall there, and end up in a different spot than you aimed.

The paper models this as a random transformation. They asked: "If the path of the light is completely random and chaotic, can we still predict how much information gets through?"

• The Result: Yes. They derived a formula (a mathematical recipe) to calculate the average capacity.
• The Analogy: It's like guessing how much rain will hit a field if the wind is blowing in a totally random direction. You can't predict the exact drop, but you can calculate the average amount that will land on the crops. They found that even with this chaos, having more modes (more "crops" to catch the rain) increases the total harvest.

4. The "Passive" vs. "Active" Distinction

The paper distinguishes between two types of changes the light might undergo:

• Passive: The light just gets shuffled around or dimmed (like water flowing through a maze of pipes). This is the main focus of the paper.
• Active: The light gets amplified or squeezed (like a pump adding extra pressure). The paper briefly looked at what happens if we add a little bit of this "active" help. They found that sometimes it helps, and sometimes it hurts, depending on how many pipes you have.

The Main Takeaways

1. More is Better: If you have a fixed budget for energy, splitting that energy across many different "modes" (channels) of light allows you to send more information than focusing it all on just one or two channels.
2. Smart Distribution: You shouldn't treat all channels equally. You should focus your energy on the channels that are clearest and avoid the ones that are too noisy.
3. Randomness is Manageable: Even if the environment is chaotic and scatters your light randomly, you can still calculate exactly how much information you can send on average.
4. Quantum Limits: The paper confirms that quantum mechanics sets a hard "ceiling" on how much information can be sent, but by using many modes and smart strategies, we can get very close to that ceiling.

What They Did Not Claim

• They did not build a new physical device or a new internet cable.
• They did not claim this will immediately fix your home Wi-Fi.
• They did not discuss medical applications or clinical uses.
• They focused strictly on the mathematical theory of how much information can be sent under specific quantum rules, not on how to build the hardware to do it tomorrow.

In short, this paper is a theoretical map. It tells us that if we want to build the ultimate high-speed optical communication system, we should use many channels, distribute our energy smartly, and we can handle random chaos with the right math.

Technical Summary

Technical Summary: Capacity of Multimode Quantum Gaussian Channels

Problem Statement
Modern optical communication systems, ranging from free-space links to fiber networks, increasingly utilize multiple-input multiple-output (MIMO) architectures to enhance data rates. While classical MIMO theory is well-established, the fundamental limits of optical MIMO channels require a quantum mechanical description due to the inherent quantum nature of light and unavoidable quantum noise. Existing models often treat optical signals as classical waveforms or focus on single-mode scenarios. This paper addresses the gap in characterizing the fundamental information capacity of multimode optical channels modeled as bosonic Gaussian quantum channels. Specifically, it investigates how the number of modes, random scattering (passive transformations), and weak active transformations affect the channel capacity under fixed power constraints.

Methodology
The authors model optical communication links using the framework of bosonic Gaussian quantum channels. The system consists of $N$ input signal modes, $K$ output signal modes, and $M$ environmental modes. The channel dynamics are described by symplectic transformations acting on the quadrature operators of the signal and environment.

1. Channel Decomposition: For passive and phase-insensitive transformations, the authors demonstrate that the multimode channel can be decomposed into parallel single-mode subchannels via Singular Value Decomposition (SVD). Crucially, they show that for matrices representing these specific transformations, the unitary matrices in the SVD are also symplectic, ensuring the decomposition corresponds to valid physical optical transformations.
2. Capacity Calculation:
   • Holevo Capacity: The fundamental capacity is derived using the Holevo bound, which represents the maximum accessible classical information via optimal quantum measurements. The authors derive an explicit formula for the Holevo information of the ensemble-averaged output state, assuming Gaussian modulation of coherent states.
   • Detection Strategies: The paper also derives capacities achievable with specific, suboptimal detection strategies: homodyne and heterodyne detection. These are expressed as sums of classical mutual informations for the decomposed single-mode channels.
   • Optimization: Power allocation across modes is optimized using a quantum analogue of the "water-filling" algorithm, subject to a total power constraint.
3. Random Scattering Model: To model random signal scattering (analogous to Rayleigh fading in radio), the authors treat the channel transformation matrix as a random unitary matrix sampled from the Haar measure. They utilize the theory of the Jacobi ensemble to derive the spectral density of the channel's eigenvalues. This allows for the analytical calculation of the expected channel capacity by integrating the single-mode capacity function over the eigenvalue distribution.
4. Active Transformations: The model is extended to include weak active transformations (squeezing) by sampling squeezing parameters from a Gaussian distribution, evaluated via Monte Carlo simulations.

Key Contributions and Results

• Explicit Capacity Formulas: The paper provides closed-form analytical expressions for the Holevo capacity of multimode Gaussian channels under passive and phase-insensitive operations. The capacity is shown to be a sum of individual mode contributions, each dependent on the singular values of the channel transformation and the effective noise.
• Optimality of Mode Expansion: A central finding is that under a fixed total power constraint, increasing the number of transmission modes ($N$) always increases the channel capacity. In noiseless scenarios, the capacity grows indefinitely with $N$, whereas in noisy scenarios, it saturates at a value determined by the noise strength.
• Random Passive Channels: For channels governed by random passive transformations, the authors derive an analytical formula for the expected capacity (Eq. 29) based on the ensemble-averaged spectral density of the Jacobi ensemble. Results indicate that while capacity increases with the number of signal modes, a larger environment size ($M$) generally reduces capacity due to signal leakage.
• Detection Strategy Comparison: The study compares Holevo capacity with homodyne and heterodyne detection limits. It finds that for small numbers of modes, heterodyne detection performs better, but for large numbers of modes, homodyne detection approaches the quantum bound more closely, though neither reaches the full Holevo limit.
• Active Transformations: The inclusion of weak active transformations shows a nuanced impact: depending on the configuration of transmitter ($N$), receiver ($K$), and environment ($M$) modes, active random transformations can either increase or decrease the Holevo information. However, they consistently increase the capacities achievable via homodyne and heterodyne detection.

Significance
The paper establishes a rigorous quantum mechanical foundation for understanding MIMO optical communication. By extending the classical concepts of water-filling and random scattering to the quantum regime, it demonstrates that the quantum description imposes specific constraints and offers distinct scaling behaviors compared to classical models. The derivation of analytical formulas for random passive channels provides a practical tool for estimating the performance of optical links subject to complex, random scattering environments without resorting solely to numerical simulations. The work confirms that the fundamental limits of optical information transmission are governed by the interplay between the number of available modes, the channel's transmission properties, and the quantum noise floor, independent of specific receiver architectures.