⚛️ quantum physics

Geometric bounds on multiparameter Heisenberg scaling in optical metrology with limited squeezed resources

This paper establishes a geometric upper bound of $n_{\rm HS}\le \min\{p,k(k+3)/2\}$ on the number of independent parameter combinations in a passive linear optical network that can achieve Heisenberg scaling when probed by $k$ squeezed states, demonstrating that this limit arises from distinct contributions of squeezing-enhanced fluctuations and first-moment shifts.

Original authors: Atmadev Rai, Paolo Facchi, Vincenzo Tamma

Published 2026-07-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Atmadev Rai, Paolo Facchi, Vincenzo Tamma

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 take the sharpest possible photograph of a mysterious object, but instead of a camera, you are using a beam of light. In the world of quantum physics, there is a special "super-power" called Heisenberg scaling. Think of it as a magical rule that says: if you double the amount of light (photons) you use, you don't just get a slightly clearer picture; you get a picture that is four times more precise. It's the difference between seeing a blurry smudge and reading the tiny text on a grain of sand. This is the holy grail of quantum metrology, the science of measuring things with extreme accuracy.

However, creating this super-precise light is hard. You can't just use a regular flashlight; you need "squeezed" light. Imagine squeezing a balloon: you can't make it smaller in every direction at once, but you can squash it flat in one direction while it puffs out in another. In quantum terms, this "squeezing" reduces the fuzziness (noise) in one part of the measurement while increasing it in another. The big question scientists have been asking is: If you have a limited supply of this special squeezed light, how many different things can you measure at once with that super-precise power? Can you measure the color, the temperature, and the shape of an object all at the same time, or does the magic only work for one thing?

This paper tackles that exact puzzle. The authors, working with complex networks of mirrors and beam splitters (called interferometers), discovered a strict "budget" for how many things you can measure with Heisenberg scaling. They found that the number of independent measurements you can make isn't determined by how big your machine is or how many knobs you turn, but by a simple geometric rule based on how many "squeezed" beams you inject into the system.

Here is the surprising math they uncovered: If you have k squeezed beams entering your machine, the maximum number of independent things you can measure with super-precision is k(k + 3)/2.

To understand why, imagine your squeezed beams are like a team of specialized workers. The authors found that these workers have two different jobs, and each job has a limit on how many tasks it can handle:

The "Fluctuation" Team (Covariance): These workers look at how the light jiggles and wobbles. Because of the way squeezing works, a team of k workers can only handle k(k + 1)/2 different types of wobbles. It's like having k people trying to juggle; they can only keep a certain number of balls in the air at once before they drop them. This part of the measurement relies purely on the "noise" or fluctuations of the light.
The "Push" Team (First-Moment): These workers look at how the center of the light beam moves or gets pushed. This team is smaller and simpler; k workers can only push in k different directions at once. However, to get this super-precision, you need to mix your squeezed light with a regular, strong "coherent" beam (like a standard laser pointer) to act as a reference for the push.

When you add these two teams together, you get the total limit: k(k + 1)/2 (from the wobbles) plus k (from the pushes) equals k(k + 3)/2.

The paper proves that this isn't just a guess; it is a hard, geometric limit. No matter how cleverly you arrange your mirrors or how many extra channels you add to your machine, you cannot break this rule. If you want to measure p different parameters (like 100 different phases in a giant optical network) with this super-precision, you need to calculate how many squeezed beams (k) you must start with. The authors show that you need at least k squeezed beams where k is roughly the square root of 8p. For example, if you want to measure 100 parameters, you don't need 100 squeezed beams; you only need about 10 or 11, because the math allows each squeezed beam to help measure multiple things at once.

Crucially, the authors didn't just say "this is the limit." They also built a theoretical machine—a specific arrangement of mirrors and beam splitters—that actually reaches this limit. They showed that if you place your regular laser beam in a specific spot (an unsqueezed channel) and arrange your squeezed beams just right, you can hit this maximum number of measurements. This means the limit is real and achievable, not just a theoretical ceiling that no one can reach.

So, the takeaway is a mix of good news and a reality check. The good news is that you don't need a million squeezed beams to measure a million things; the math is efficient, and a small number of squeezed resources can unlock a surprisingly large number of super-precise measurements. The reality check is that there is a hard cap. You can't just keep adding more parameters and expect the precision to stay super-high forever; eventually, you run out of the "geometric room" in your squeezed beams to hold more information. This paper gives scientists a clear map: it tells them exactly how many resources they need to build the ultimate quantum sensors, and it warns them when they are trying to measure too many things with too little light.

Technical Summary: Geometric Bounds on Multiparameter Heisenberg Scaling in Optical Metrology

Problem Statement
In quantum metrology, particularly within the context of large photonic interferometers and distributed sensing, a fundamental limitation arises when attempting to simultaneously estimate multiple unknown parameters. While nonclassical resources, such as squeezed states of light, can theoretically enable Heisenberg-scaling precision ( $O(1/N^2)$ ) where $N$ is the total average input photon number, practical constraints limit the number of available nonclassical probes. The central question addressed by this work is: given a finite number of single-mode squeezed states (SMSS) injected into a passive linear optical network, what is the maximum number of independent parameter combinations that can inherit Heisenberg scaling? Previous results established the potential for enhanced precision in multiparameter settings but did not determine the fundamental geometric limits on the number of estimable parameters in a generic passive multimode network with limited nonclassical resources.

Methodology
The authors analyze arbitrary passive linear optical networks ( $U(\phi) \in U(M)$ ) encoding $p$ unknown parameters, probed by a multimode Gaussian state. The input state consists of $k$ single-mode squeezed vacua and $M-k$ coherent or vacuum states. The analysis relies on the continuous-variable formalism where the quantum Fisher information matrix (QFIM) is expressed in terms of the evolution of the first moments (mean vector) and second moments (covariance matrix).

The QFIM is decomposed into two distinct contributions:

Covariance Contribution ( $F_\Gamma$ ): Associated with parameter-dependent fluctuations and the orientation of the noise ellipsoid in phase space.
First-Moment Contribution ( $F_\mu$ ): Associated with the parameter-dependent displacement of the Gaussian center.

The authors perform an asymptotic expansion of these contributions with respect to the squeezing parameter $s = e^{2r}$ (where $N_s \sim s$ ). They identify the leading-order terms that scale as $N^2$ (Heisenberg scaling) and analyze the rank of the resulting coefficient matrices. The problem is reduced to a finite-dimensional rank problem on the active squeezed input subspace, specifically examining the dimensionality of the space spanned by the generators of the symplectic algebra projected onto the squeezed modes.

Key Contributions and Results

Geometric Bounds on Heisenberg Scaling:
The paper establishes that the rank of the Heisenberg-scaling coefficient of the QFIM, denoted as $n_{HS}$ , is bounded by the number of squeezed modes $k$ and the total number of parameters $p$ . The bound is given by:
$n_{HS} \leq \min\left\{p, \frac{k(k+3)}{2}\right\}$
This result indicates that the number of independent parameter combinations achievable with Heisenberg scaling is determined by the geometry of the squeezed input subspace, not merely by the number of modes or encoded parameters.
Separation of Contributions:
The authors demonstrate that the total bound separates into two geometrically distinct contributions:
- Covariance Contribution ( $n_\Gamma^{HS}$ ): Driven by squeezing-enhanced fluctuations, this term provides at most $k(k+1)/2$ independent parameter combinations. This corresponds to the dimension of the space of real symmetric $k \times k$ matrices, $\text{Sym}(k, \mathbb{R})$ .
- First-Moment Contribution ( $n_\mu^{HS}$ ): Driven by the amplified displacement of the mean field (requiring both squeezed and coherent resources), this term provides at most $k$ additional independent parameter combinations.
Saturation Conditions and Constructive Proof:
The paper identifies the conditions required to saturate these bounds. Saturation of the covariance bound requires the parameter-induced generators to span the full space of symmetric matrices within the squeezed subspace. Saturation of the full bound requires the first-moment contribution to provide $k$ independent directions distinct from those in the covariance contribution. The authors construct a specific family of passive optical networks (using a specific set of generators involving beam splitters and phase shifters) that simultaneously saturates both the covariance and first-moment bounds, proving that the derived limits are tight.
Resource Requirements for Full Estimation:
A direct consequence of the bound is a lower bound on the number of SMSS ( $k$ ) required to estimate all $p$ parameters in a network with Heisenberg scaling. The minimum number of squeezed states required is:
$k \geq k_{\min}(p) = \left\lceil \frac{\sqrt{8p + 9} - 3}{2} \right\rceil$
For large $p$ , this implies that the number of required squeezed inputs grows only as $\sqrt{p}$ .

Significance and Claims
The authors claim that their work provides a fundamental, parameterization-independent bound on the dimension of the leading Heisenberg-scaling subspace of the Fisher matrix. Unlike previous works that focused on optimizing specific states or measurements, this result defines the intrinsic geometric capacity of passive Gaussian networks to support Heisenberg scaling.

The significance lies in establishing a "resource-counting principle" for large photonic sensors. The paper asserts that simply increasing the number of optical modes, phases, or tunable elements does not inherently increase the number of directions capable of achieving Heisenberg scaling; rather, this number is strictly limited by how the available squeezed inputs couple to the encoded parameters through the covariance and first moments. By providing a concrete benchmark and demonstrating that these bounds are achievable with finite passive interferometers, the work offers a theoretical limit for the design and calibration of multiparameter quantum sensors.

Technical Summary: Geometric Bounds on Multiparameter Heisenberg Scaling in Optical Metrology

More like this