Tabletop Quantum Gravity

Tabletop quantum-gravity research uses controlled quantum systems to test specified properties of gravitational interactions. Measuring a gravitational phase, constraining a semiclassical model, and witnessing nonclassical mediation are different goals. Reviewed 11 October 2026, this assessment includes a 5 October 2026 QGEM white paper; that document sets proposed requirements and does not report a completed gravity-mediated entanglement experiment.

Helpful background. The quantum-foundations/gravity interface explains the stable scope. Gravity-mediated entanglement examines one particular inference.

Gravitational phase and channel discrimination

A quantum probe in a classical gravitational potential can acquire a phase. For two prescribed paths,

$$\Delta\phi=-\frac{1}{\hbar}\int [V_1(t)-V_2(t)]\,dt.$$

Its observation tests a quantum system’s response to gravity under that model. It does not alone show that the gravitational source or mediator possesses quantum degrees of freedom.

A stronger experiment might distinguish an interaction channel from local operations and classical communication, or test a nonlinear mean-field prediction. The source state, preparation independence, permitted nongravitational channels, and measured witness then become part of the conclusion.

Established within the stated models: Bose and colleagues derived a spin-entanglement witness in 2017. Lami, Pedernales, and Plenio instead bound how faithfully local operations and classical communication can reproduce a target evolution across a specified input ensemble. Their oscillator proposal uses coherent states and need not generate entanglement. Its discriminator is an ensemble-dependent channel-fidelity inequality, not an entanglement witness. Both routes exclude a defined class of channels; neither rules out every conceivable classical–quantum dynamics by terminology alone.

The current experimental bottleneck is integrated control

The October 2026 white paper identifies the 1–10 Hz regime and low-frequency acceleration and gravity-gradient noise as important for its QGEM parameter choices. The requirements combine massive objects, spatially separated branches, close separation between objects, and long coherence. Treat these as coupled engineering targets; separate demonstrations of the components do not establish their simultaneous operation.

Three contributions are especially useful: an independently reproducible noise model; a control protocol that suppresses a nuisance without suppressing the signal; or a witness robust to realistic preparation and readout. A proposal should show how electrostatic, magnetic, Casimir, and shared control interactions are bounded where relevant.

Bose et al.’s 2025 review compares massive-system platforms and the physical resources behind their proposed tests. Use it to choose a credible preparation and control setting before importing the noise targets of a particular protocol.

Noise amplitude spectral densities cannot be compared directly with phase uncertainty without the interferometer’s response function. With a stated spectral convention,

$$\operatorname{Var}\phi= \int_0^\infty S_a(f)\,|F_a(f)|^2\,df,$$

where $F_a$ includes the physical conversion from acceleration to phase.

A numerical acceleration-to-phase check

For prescribed paths separated by $\Delta x(t)$ along a spatially uniform acceleration fluctuation, the phase is $\delta\phi=(m/\hbar)\int\Delta x(t)\delta a(t)\,dt$, up to the chosen path sign. Thus

$$F_a(f)=\frac{m}{\hbar}\int_0^T\Delta x(t)e^{2\pi ift}\,dt.$$

This expression has units $\mathrm{s^2/m}$; a one-sided acceleration power spectral density has units $\mathrm{m^2s^{-4}/Hz}$. Their integral gives dimensionless phase variance. An amplitude spectral density must be squared before insertion.

As an illustrative rectangular hold, take $m=10^{-14}\,\mathrm{kg}$, $\Delta x=10\,\mu\mathrm m$, and $T=1\,\mathrm s$. Then $F_a(0)=9.48\times10^{14}\,\mathrm{s^2/m}$. A constant differential acceleration of $10^{-15}\,\mathrm{m/s^2}$ produces about $0.95$ rad, not a negligible phase. For ideal white one-sided noise of amplitude $10^{-15}\,\mathrm{m/s^2/\sqrt{Hz}}$, integrating the squared sinc response gives $\operatorname{Var}\phi\simeq0.45$ and Gaussian phase averaging reduces visibility by $e^{-\operatorname{Var}\phi/2}\simeq0.80$.

This is a units and integration check, not an achieved sensitivity or a full QGEM noise forecast. Real splitting and recombination change the filter; correlations between interferometers change the differential spectrum; drift may violate stationarity. A common acceleration can cancel only to the degree that the two response functions match. The active question is whether a trajectory with an adequate interaction phase retains an acceptable witness margin under those measured, correlated disturbances.

An entry project: a complete response-function budget

Baseline. Reproduce the phase-noise prediction for one proposed interferometer trajectory from the 2026 white paper or its technical references. Derive its acceleration response rather than inserting a free constant.

Resources. A laptop, Fourier analysis, classical trajectories, and access to an experimentalist who can judge achievable spectra. Public or synthetic spectra suffice for a methodology project.

Deliverable. A frequency-resolved error budget showing which bands dominate and how the answer changes with duration, separation, and filtering. Recover the rectangular-hold result above and compare the spectral integral with direct sampling of a stationary Gaussian acceleration process. Include units, one-sided versus two-sided conventions, preparation/readout duty cycle, and a convergence plot for sample rate and record length. A limited frequency record cannot justify an extrapolated low-frequency noise floor.

Failure criterion. Stop if noise values are borrowed from incompatible geometries or frequency ranges, or if a cancellation assumes independently controlled degrees of freedom that the apparatus cannot address.

Research extension. Optimize trajectories against a measured colored spectrum while preserving interaction phase and experimental control constraints. A credible negative feasibility bound can redirect effort as usefully as an optimistic optimum.

What would change the assessment?

A meaningful milestone combines verified coherence, controlled interactions, and a witness in the same apparatus. A stronger inference also needs exclusion of alternative channels and an explicit mediator model. Neither a larger test mass nor a lower isolated noise figure establishes the full claim.

Students should select the bottleneck their group can measure or calculate now. A complete end-to-end demonstration is normally a collaborative program; a well-bounded noise or inference problem can be an excellent graduate project.

References

• S. Bose et al., “Spin entanglement witness for quantum gravity,” Physical Review Letters 119, 240401 (2017). arXiv:1707.06050.
• S. Bose et al., “White paper: 1-10 Hz matter-wave interferometer to test the spin entanglement witness for quantum gravity,” arXiv:2610.07422v1 (5 October 2026), preprint. Requirements paper.
• S. Bose et al., “Massive quantum systems as interfaces of quantum mechanics and gravity,” Reviews of Modern Physics 97, 015003 (2025). DOI — platform and control overview.
• L. Lami, J. S. Pedernales, and M. B. Plenio, “Testing the quantum nature of gravity without entanglement,” Physical Review X 14, 021022 (2024). DOI — LOCC simulation-fidelity bounds and the coherent-state oscillator proposal.

Entry points

Read one protocol’s observable and noise model in full. Build its resource ledger before comparing it with another proposal’s headline sensitivity.