⚛️ general relativity

Torsion balances as operational probes of semiclassical gravity: Matched-filter bounds, torque-diffusion constraints, and quantum-noise benchmarks

This paper establishes a comprehensive framework for using calibrated torsion-balance spectra to constrain both deterministic and stochastic deviations from Newtonian gravity, deriving matched-filter bounds and torque-diffusion limits that link experimental data to semiclassical gravity searches without directly testing the full relativistic semiclassical Einstein equation.

Original authors: Jyotirmaya Mohanta, Yutaka Shikano

Published 2026-08-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jyotirmaya Mohanta, Yutaka Shikano

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

In the quiet corners of physics laboratories, scientists have long used delicate instruments to weigh the invisible pull of gravity. These devices, known as torsion balances, consist of a bar suspended by a thin fiber that twists when a nearby mass exerts a gravitational force. For centuries, they have been the gold standard for measuring the strength of gravity and testing whether all objects fall at the same rate, regardless of their composition. However, a new frontier has opened where the rules of the very large, which govern planets and stars, meet the rules of the very small, which govern atoms and light. In this strange middle ground, known as semiclassical gravity, physicists wonder if the gravitational field itself might behave differently when it interacts with quantum systems. Does gravity simply react to the average position of a quantum object, or does it react to the specific, random outcomes of quantum events? Answering this question could reshape our understanding of the universe, but it requires instruments sensitive enough to detect the faintest whispers of force that might appear if these theories are true.

A team of researchers has now provided a clear, practical map for how to listen for these whispers using torsion balances. Rather than proposing a new machine or claiming to have found a new force, the authors have developed a standardized language and a set of tools to translate raw experimental data into meaningful limits on these theories. Their work acts as a universal translator, converting the complex, noisy signals recorded by sensitive detectors into a format that can be directly compared against the predictions of various gravitational models. By doing so, they allow scientists to distinguish between the expected background noise of the laboratory and any potential new signal that might hint at a deeper reality.

The core of their achievement lies in separating two different types of questions that physicists ask about gravity. The first type involves looking for a specific, predictable pattern of force, such as a twist that repeats in a regular rhythm. The second type involves looking for a random, jittery noise that might be added to the system by the very nature of gravity itself. The researchers showed that while these sound like different problems, they can both be addressed using the same statistical methods. For the predictable patterns, they refined a technique called matched filtering, which is like tuning a radio to a specific station to hear a clear signal amidst static. For the random noise, they developed a way to subtract all the known sources of disturbance—such as the heat of the room, the vibration of the ground, and the limitations of the measuring device—to see what, if anything, remains.

What remains after this subtraction is the key to the mystery. If the theories of semiclassical gravity are correct, there should be a tiny, extra amount of random torque, or twisting force, that cannot be explained by ordinary physics. The authors demonstrated how to calculate the maximum size of this extra noise based on the data. They found that in many cases, this extra noise cannot be described by a single number, but must be understood as a spectrum that changes depending on the frequency of the measurement. Only in a very specific, simplified case where the noise is completely random and uniform can it be compressed into a single value, which they call a diffusion coefficient. This distinction is crucial because many modern theories predict noise that is not uniform, and treating them as if they were would lead to incorrect conclusions.

To make these abstract concepts concrete, the team applied their methods to real-world examples. They analyzed data from a recent experiment that used an ultra-sensitive torsion balance to search for these effects. By taking the reported sensitivity of that experiment, they calculated a conservative upper limit on the amount of random gravitational noise that could be present. Their calculation showed that if such noise exists, it must be incredibly small, far below the level of the thermal noise generated by the heat of the room. They also provided a benchmark for a standard laboratory setup, showing that the natural heat-induced shaking of the balance is currently millions of times larger than the quantum limits that future experiments hope to reach. This highlights that while the instruments are incredibly precise, the biggest obstacle right now is not the quantum nature of the detector, but the classical noise of the environment.

The paper also clarified how to test specific, famous ideas about gravity, such as the Page-Geilker experiment, which tried to see if gravity reacts to the different possible outcomes of a quantum event. The researchers showed that these tests, which often seem to involve deep philosophical questions about the nature of reality, can be reduced to a straightforward statistical problem: does the data show a twist that matches the prediction of one theory better than the prediction of another? They emphasized that while these tests address different physical questions, the mathematical tool used to evaluate them is the same. This unification means that future experiments can be designed with greater confidence, knowing exactly how to interpret their results against a wide range of theoretical possibilities.

Ultimately, this work does not claim to have solved the mystery of how gravity and quantum mechanics fit together. Instead, it provides the essential toolkit for the next generation of scientists to do so. By establishing a clear, calibrated way to report data and set limits, the authors have ensured that when a future experiment finally detects a deviation from standard physics, the scientific community will know exactly what it means. They have turned a complex theoretical challenge into a manageable engineering task, defining the boundaries of what is currently possible and pointing the way toward the measurements that could one day reveal the true nature of the gravitational force.

Technical Summary: Torsion Balances as Operational Probes of Semiclassical Gravity

Problem Statement
The paper addresses the challenge of translating calibrated torsion-balance data into rigorous constraints on semiclassical and post-quantum gravity models. While torsion balances are established tools for measuring Newton's constant ( $G$ ), testing the Weak Equivalence Principle (WEP), and searching for inverse-square-law violations, their application to semiclassical gravity requires a distinct operational interface. The core difficulty lies in distinguishing between standard experimental noise (thermal, environmental, Newtonian, and quantum backaction) and potential new physics, such as deterministic branch-dependent torques or stochastic torque noise arising from semiclassical sourcing rules. The authors note that existing literature often conflates linear quantum measurement theory, torsion-balance metrology, and semiclassical gravity models; this work seeks to separate these domains to provide a standardized framework for comparison.

Methodology
The authors develop a linear, calibrated framework that maps observed angle spectra to bounds on torque templates and stochastic noise. The methodology proceeds through the following steps:

Spectral Conventions and Calibration: The paper establishes a consistent one-sided power spectral density (PSD) convention. It defines the torsional susceptibility $\chi(\omega)$ and converts the measured angle spectrum $S_{yy}(\omega)$ into an equivalent torque spectrum using the relation $S_{\tau\tau}^{eq}(\omega) = S_{yy}(\omega) / |\chi(\omega)|^2$ . This allows all torque-like sources to be compared in a common unit ( $N^2 m^2/Hz$ ).
Deterministic Analysis (Matched Filtering): For deterministic signals (e.g., WEP violations, Yukawa potentials, or Page–Geilker branch-dependent torques), the authors apply the finite-time matched-filter formalism. They derive the Cramér–Rao bound for an unknown amplitude parameter $g_x$ associated with a known torque template $\Lambda_x(t)$ . This bound explicitly accounts for finite integration times, windowing functions, and leakage factors.
Stochastic Analysis (Residual Spectra): For stochastic models, the authors propose subtracting a "standard noise budget" (thermal, Newtonian, environmental, imprecision, and backaction) from the observed spectrum to define a residual spectrum. This residual is converted to an equivalent residual torque spectrum.
- In the Markovian white-noise limit, this frequency-resolved bound is compressed into a single scalar parameter: the torque-diffusion coefficient $D_\tau$ , defined by $S_{\tau\tau} = 4D_\tau$ .
- For colored or non-Markovian models (e.g., $1/f$ , Lorentzian), the authors argue that the full frequency-dependent residual spectrum must be used for comparison, rather than a single diffusion coefficient.
Noise Budgeting: The paper details the construction of the noise budget, distinguishing between additive Newtonian noise (gravity gradients) and multiplicative stiffness fluctuations (resonance drifts). It emphasizes the necessity of auxiliary channels (tilt, barometric, hydrological) to subtract correlated environmental noise.

Key Contributions
The paper makes four specific, modest contributions to the field:

Standardization: It standardizes the measurement interface by explicitly maintaining one-sided PSD conventions, transfer-function factors, and angle-to-torque conversions from the raw output spectrum to the final constraint, preventing ambiguities when comparing different gravitational sourcing models.
Finite-Time Formalism: It applies the standard matched-filter/Cramér–Rao formalism specifically to finite-time torsion-balance torque templates, providing explicit expressions for amplitude bounds that account for windowing and leakage.
Spectral vs. Scalar Bounds: It formally separates the general residual equivalent-torque spectrum from the special case of Markovian white-noise compression ( $D_\tau$ ). The paper asserts that for colored noise models, the spectrum itself is the correct comparison object, not a single diffusion coefficient.
Unified Interpretation: It demonstrates that the Page–Geilker branch discrimination test and the Fedida–Kent mixture-equivalence principle (MEP) test, while addressing different physical sourcing questions, reduce to the same statistical problem of matched-filter template discrimination once their competing predictions are projected onto torsion-balance torque templates.

Results and Benchmarks
The authors provide quantitative benchmarks to illustrate the scale of these constraints:

Representative Cavendish Benchmark: For a room-temperature torsion balance with $I = 5.0 \times 10^{-5} \text{ kg m}^2$ and $Q = 5.0 \times 10^3$ , the resonant thermal angle ASD is calculated to be $1.36 \times 10^{-4} \text{ rad}/\sqrt{\text{Hz}}$ . This exceeds the measurement-added Standard Quantum Limit (SQL) of $3.25 \times 10^{-12} \text{ rad}/\sqrt{\text{Hz}}$ by approximately seven orders of magnitude in amplitude. The corresponding viscous thermal white-diffusion scale is $D_\tau \approx 1.85 \times 10^{-30} \text{ N}^2 \text{ m}^2 \text{ s}$ .
Yan et al. (2025) Benchmark: Using the public sensitivity floor of $0.3 \text{ \mu rad}/\sqrt{\text{Hz}}$ at $2.5 \text{ mHz}$ from a recent optomechanical torsion-balance search, the authors derive a conservative observed-floor bound for white torque noise of $D_\tau \lesssim 2.4 \times 10^{-23} \text{ N}^2 \text{ m}^2 \text{ s}$ . They clarify that this is an observed-floor bound, not a fully subtracted residual constraint, and note that the platform's inferred thermal noise floor is significantly lower ( $\sim 10^{-29} \text{ N}^2 \text{ m}^2 \text{ s}$ ), indicating that thermal noise is not the dominant limitation in that specific frequency band.

Significance and Claims
The paper explicitly claims not to provide a direct test of the full relativistic semiclassical Einstein equation. Instead, it offers a "Newtonian-limit measurement interface." Its significance lies in providing a common language and mathematical framework to:

Translate calibrated torsion-balance data into bounds on deterministic torque templates.
Translate data into bounds on stochastic semiclassical gravity searches via residual torque spectra.
Clarify that the comparison between theory and experiment must occur at the level of the projected torque waveform or torque-noise spectrum, which depends on the specific geometry and mass distribution of the apparatus.

The authors emphasize that their framework is a tool for data analysis and constraint setting. They do not propose new experimental setups but rather provide the necessary "calibrated language" to interpret existing and future data in the context of semiclassical and post-quantum gravity models. The work underscores that for non-Markovian models, the full spectral shape is the critical observable, and that distinguishing between different noise sources (thermal, Newtonian, quantum) is a prerequisite for claiming any detection of new physics.

Technical Summary: Torsion Balances as Operational Probes of Semiclassical Gravity

More like this