Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

Ultrahigh-Q Torsional Nanomechanics through Bayesian Optimization

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper reports the first solid-state torsion oscillator with a Q factor above 100 million, reached at room temperature in silicon nitride nanoribbons whose clamp fillets are shaped by Bayesian optimization.

desk verdict A genuine room-temperature Q>1e8 torsion oscillator, but the Bayesian-optimization attribution leans on two devices and a loss model that misses flexural losses by ~100x. read the letter →

arxiv 2506.02325 v1 pith:KYWBH5KH submitted 2025-06-02 cond-mat.mes-hall physics.app-ph

Atkin D. Hyatt , Aman R. Agrawal , Christian M. Pluchar , Charles A. Condos , Dalziel J. Wilson This is my paper
classification cond-mat.mes-hallphysics.app-ph
keywords Bayesianoptimizationtorsionoscillatordissipationdilutionsiliconnitridenanoribbonqualityfactorsoftclampingtorquesensingnanomechanics
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Torsion modes of tensioned nanoribbons can have their mechanical loss dramatically reduced by a 'dissipation dilution' effect, but the bending of the mode at the clamps eventually limits this gain. This paper argues that the shape of the fillet where the ribbon meets its support controls that clamp loss, and that a Bayesian optimization loop over fillet geometries in a finite-element simulator can find shapes with far lower loss. Applied to centimeter-scale silicon nitride nanoribbons, the search yields torsion oscillators with quality factors above 100 million at room temperature—the first solid-state torsion oscillator to reach that mark—and with quality-factor–frequency products above $10^{13}$ Hz, the threshold for quantum coherence at 300 K. The resulting thermal torque noise is at the level of $10^{-20}$ N·m/√Hz and the zero-point angular displacement noise near $10^{-10}$ rad/√Hz, which the authors argue makes these devices attractive for weak-torque sensing and room-temperature quantum optomechanics.

What carries the argument

The central object is the clamp fillet, whose geometry sets the mode curvature near the support and therefore the bending-loss term in the dissipation-dilution formula. The load-bearing identity is Q/Q0 ≈ K_tot/U_s (Eq. 5), which turns the finite-element mode solution into a predicted quality factor by comparing total kinetic energy with stored strain energy; minimizing U_s relative to K_tot maximizes Q. The optimization machinery is Bayesian optimization with Gaussian-process regression and an expected-improvement acquisition function, used to search the fillet parameters (e.g., elliptical fillet radii rx and ry, or a circular fillet with diagonal boundary) at a computational cost low enough to be practical.

What would settle it

Measure the ringdown Q of torsion modes on a set of ribbons whose fillet shapes are chosen to span the optimizer's predicted response surface, including shapes the optimizer judged poor. If the measured Q does not follow the simulated K_tot/U_s curve—for instance, if all shapes give nearly the same Q, or if the best measured geometry is not the simulated optimum—then clamp-region bending is not the dominant loss and the central claim fails. A simpler check: if the torsion Q is found to depend strongly on surface preparation or on ambient conditions, an unmodeled surface loss channel is at play.

Watch

Extended reading notes

Core claim

The paper's central claim is that the fundamental torsion mode of a strained silicon nitride nanoribbon can be 'soft-clamped' far beyond the conventional hard-clamping limit by optimizing the fillet geometry at the clamps. The authors identify the relevant loss as clamp-region bending, proportional to the integral of the squared mode curvature, and compute the dissipation-diluted Q in a two-step finite-element simulation using the ratio of total kinetic energy to stored strain energy. They then run a Bayesian optimizer with an expected-improvement acquisition function over a one- or two-parameter fillet description, obtaining predicted Q factors near 2×$10^{8}$. Fabricated 400-µm-wide, 7-mm-long, 90-nm-thick ribbons reach Q = 1.5×$10^{8}$ at room temperature and 1.7×$10^{8}$ at 4 K, with Q·f > $10^{13}$ Hz in both the fundamental and second torsion modes; the authors report this as the first solid-state torsion oscillator with Q exceeding 100 million.

Load-bearing premise

The load-bearing premise is that the finite-element model, which computes Q from the total-kinetic-to-strain-energy ratio accounting only for material loss and clamp-bending loss, faithfully captures every significant damping channel of the torsion mode; the paper's own observation that the same model overestimates flexural-mode Q by about a factor of one hundred shows this premise is not guaranteed.

Editorial extensions

If this is right

  • The reported Q·f > 10^13 Hz means the fundamental and second torsion modes of the optimized ribbons meet the room-temperature quantum-coherence condition Q·f > k_BT/h, opening a route to feedback cooling and quantum optomechanics without dilution refrigeration.
  • Thermal torque sensitivity near 10^-20 N·m/√Hz and zero-point angular displacement spectral density near 10^-10 rad/√Hz put weak-force sensing tasks—magnetometry, gravimetry, tests of short-range gravity and dark matter—within reach of a simple, photolithographically defined device.
  • Because torsional soft-clamping is preserved under heavy mass loading, a central pad can be added to the optimized ribbon without degrading Q, enabling the design of micro- to milligram-scale torsion pendula.
  • Cooling the device to 4 K increases Q from 1.2×10^8 to 1.7×10^8 and raises the quantum-coherence number about a hundredfold, suggesting further gains at millikelvin temperatures.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper's own data show flexural modes of the same ribbons are roughly a hundred times lower than the finite-element prediction; if a similar unmodeled loss affects the torsion mode, the absolute Q ceiling would be lower than the simulated optimum, even though the optimizer's ranking of geometries could still be correct.
  • A direct test of the mechanism would be to measure torsion Q while sweeping the fillet shape through the optimizer's predicted landscape; if the measured Q fails to track the simulated K_tot/U_s curve, then clamp-bending loss is not the only important loss channel.
  • The single-parameter fillet parametrization leaves unexplored geometries such as non-elliptical tapers, multiple fillets, or three-dimensional clamp structures; a higher-dimensional Bayesian search could push Q further if the model is trusted.
  • Because the objective is a scalar Q at a fixed width, an obvious extension is to maximize Q at a target frequency or moment of inertia, which would directly address sensing applications where the resonance frequency matters.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This Letter reports the use of Bayesian optimization over fillet geometry to maximize the simulated dissipation-dilution quality factor of the fundamental torsion mode of strained Si3N4 nanoribbons, using Eq. (5) as the FEM-based objective. The authors fabricated centimeter-scale, 90-nm-thick ribbons with optimized diagonal fillets and report stroboscopic ringdown Q factors of 1.5e8 and 1.2e8 at room temperature for two 400-um-wide devices, with Q*f > 1e13 Hz, plus a 4-K measurement showing Q = 1.7e8. They frame these results as the first solid-state torsion oscillator with Q > 1e8 and as evidence that Bayesian-optimized clamp geometry extends torsional dissipation dilution beyond the hard-clamping limit.

Significance. If the reported values are reproducible and attributable to the optimized geometry, the result is significant for torque sensing and room-temperature quantum optomechanics: the Q*f product exceeds the 300-K thermal decoherence threshold, and the inferred torque sensitivity (1e-20 N m / sqrt(Hz)) and zero-point angular displacement spectral density are competitive. The paper has strengths that should be credited: the optimization objective is a forward FEM calculation with material parameters from prior literature, so the geometry optimization is not circularly fit to the measured torsion Q; the ringdown method is standard; and the authors explicitly flag the large discrepancy between simulated and measured flexural-mode Q. The main weaknesses are under-reported device statistics and the absence of uncertainty quantification, both of which are load-bearing for the headline attribution claim.

major comments (3)
  1. [Experimental results; Fig. 3] The manuscript reports measured Q for only two of the fabricated ribbons and does not state the Q values of the remaining devices; the fabrication count is also internally inconsistent, as the text first says 'five 400-um-wide and two 600-um-wide ribbons' (seven total) and then says 'six ribbons fabricated and characterized'. Without complete reporting of all devices or an explicit pre-registered selection rule, the claim that Bayesian optimization reproducibly realizes Q > 1e8 is not supported, because the two highlighted devices could be selected outliers.
  2. [Fillet optimization via finite element simulation; Experimental results] Equation (5) is both the optimization objective and the basis for attributing the measured Q to the designed geometry, yet the same experimental section reports that the fundamental and second-order flexural modes of the first optimized device are roughly 100 times lower than simulated, with the paper attributing this to 'sensitivity to other forms of loss'. No argument or measurement is given that the torsion mode is immune to this missing loss channel; the factor-of-two agreement for two torsion devices is therefore not, by itself, sufficient evidence that the unmodeled loss is negligible. The authors should provide a quantitative test, for example a torsion-Q measurement across a geometry variation predicted by Eq. (5), an estimate of the missing loss contribution from the flexural discrepancy, or a direct comparison with non-optimized control devices of identical material.
  3. [Experimental results] No uncertainty is reported for any Q value and no repeated ringdown measurements are shown. This matters for several comparisons in the paper: the 'within a factor of two' agreement with simulation, the 40% increase from 1.2e8 to 1.7e8 at 4 K, and the claimed distinction from the earlier Q = 1.0e8 device in Ref. [5]. Please provide standard errors or repeated-measurement statistics, and state the systematic uncertainties in the ringdown extraction method.
minor comments (5)
  1. [Fig. 1(d) and main text] The figure label gives Q = 147 million while the text and summary quote Q = 150 million; these numbers should be made consistent.
  2. [Fabrication paragraph] The phrase 'coated with on both sides' should read 'coated on both sides'.
  3. [Fig. 1 caption] The caption says 'V on Mises stress profile'; it should be 'von Mises stress profile'.
  4. [Eq. (2)] The subscripts in k_shear,ext_E and k_bend,cl_E are not defined in the text; a one-line definition would improve readability.
  5. [Eq. (3)] The denominator 'Q0/h / 60 nm^-1' is awkwardly parenthesized; rewriting as (Q0/h)/(60 nm^-1) would clarify the scaling.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the simulated-Q optimization objective and the measured ringdown Q are independent, and the self-citations to prior work are externally grounded measurements rather than fitted inputs.

full rationale

Walked the paper's claimed derivation chain. The Bayesian optimization objective is Eq. (5), Q/Q0 = Ktot/Us, evaluated by a two-step COMSOL model. Material parameters (stress, Young's modulus, Q0) are taken from literature values as stated in Eq. (3) and ref. [20]; none are fitted to the measured Q values of the optimized ribbons. The experimental quality factors (Q = 1.5e8 and 1.2e8) are obtained from independent stroboscopic ringdown measurements, so the optimized fillet geometry is a genuine out-of-sample prediction rather than a fit renamed as a prediction. The analytical framework in Eqs. (1)-(3) and the prior observation of torsional dissipation dilution in ref. [5] come partly from the same authors, but ref. [5] is a published, externally falsifiable experimental result and the present paper independently reproduces the effect in its own devices. The flexural-mode discrepancy noted in the text (measured flexural Q ~100 times lower than simulated) is a model-validity concern for the FEM objective, not a circularity: it does not make Eq. (5) equivalent to the measured torsion Q. The inconsistent device count ('five 400-um and two 600-um' vs 'six ribbons fabricated and characterized') is a reporting inconsistency, not a circular step. No self-definitional, fitted-input-as-prediction, uniqueness-imported-from-authors, or ansatz-smuggled-via-citation pattern was found.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the dissipation dilution theory from prior work (Eqs. 1-3, from [5] and [19]) and on the fidelity of the FEM-based objective function. The material parameters (E, sigma, Q0) are taken from literature, not fitted. The only adjustable hand-set value is the Bayesian optimizer's exploration ratio (0.3), which does not change the found optimum as shown by repeated runs. No new entities are introduced.

free parameters (1)
  • Exploration ratio = 0.3
    Hyperparameter of the Bayesian optimizer set by the authors; affects sampling strategy but repeated gray traces in Fig. 2(c) converge to similar Q, so it likely does not affect the central claim.
assumptions (4)
  • domain assumption Dissipation dilution theory for torsion modes (Eqs. 1-3) is valid for the ribbon regime w>h, L>w.
    The paper relies on the lumped model of Buckley (1914) and the continuum theory of [19], as applied to torsion in [5], without re-deriving them. This is an established result but is assumed as background.
  • domain assumption The two-step COMSOL model (stationary stress plus eigenfrequency) accurately computes the stress redistribution and mode shape, and Eq. (5) reliably yields Q/Q0 for the first torsion mode.
    The optimization objective is based on this FEM model. The paper does not validate the FEM against an analytic solution specifically for torsion, and the flexural mode results suggest the model may miss some loss channels.
  • domain assumption The measured Q is set only by material loss Q0 and bending loss at clamps; gas damping and photothermal damping are negligible.
    High vacuum (<1e-7 mbar) and low optical duty cycle (<10%) are used to suppress these losses, but their residual contributions are not quantified. The 40% Q increase at 4 K indicates temperature-dependent losses that are not in the model.
  • standard math The Bayesian optimization (MATLAB expected-improvement-plus, exploration ratio 0.3) finds the global optimum of the black-box objective.
    The algorithm's convergence to similar Q in multiple gray traces (Fig. 2c) supports this, but no formal guarantee is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Ultrahigh-Q Torsional Nanomechanics through Bayesian Optimization." pith.science (2026). https://pith.science/paper/KYWBH5KH

@misc{pith2026250602325,
  author       = {Pith},
  title        = {Pith review of: Ultrahigh-Q Torsional Nanomechanics through Bayesian Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KYWBH5KH}},
  note         = {Machine review of arXiv:2506.02325}
}
abstract

Recently it was discovered that torsion modes of strained nanoribbons exhibit dissipation dilution, giving a route to enhanced torque sensing and quantum optomechanics experiments. As with all strained nanomechanical resonators, an important limitation is bending loss due to mode curvature at the clamps. Here we use Bayesian optimization to design nanoribbons with optimal dissipation dilution of the fundamental torsion mode. Applied to centimeter-scale Si$_3$N$_4$ nanoribbons, we realize $Q$ factors exceeding 100 million and $Q$-frequency products exceeding $10^{13}$ Hz at room temperature. The thermal torque sensitivity of the reported devices is at the level of $10^{-20}\;\text{N}\,\text{m}/\sqrt{\text{Hz}}$ and the zero point angular displacement spectral density is at the level of $10^{-10}\;\text{rad}/\sqrt{\text{Hz}}$; they are moreover simple to fabricate, have high thermal conductivity, and can be heavily mass-loaded without diminishing their $Q$, making them attractive for diverse fundamental and applied weak force sensing tasks.

Figures

Figures reproduced from arXiv: 2506.02325 by the authors.

Figure 1
Figure 1. FIG. 1. (a) Image of an optimized 7-mm-long, 400- [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Optimizing torsional dissipation dilution by fillet design. (a) Two geometries are considered: elliptical fillet with perpendicular [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. (a) Compilation of [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Torsion mode ringdowns of an optimized nanoribbon (iden [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Precision optomechanical accelerometer via hybrid test mass integration

    physics.app-ph 2025-08 unverdicted novelty 6.0 of 10

    A pick-and-place bonded 95 mg platinum sphere on a silicon nitride trampoline yields a chip-integrated accelerometer with 5.5 ng/√Hz peak sensitivity at 117 Hz in air.

  2. Optomechanical Accelerometer Search for Ultralight Dark Matter

    hep-ex 2025-09 conditional novelty 5.0 of 10

    A 39 kHz optomechanical membrane accelerometer was operated at 4 K to search for ultralight vector dark matter, yielding no signal and upper limits weaker than existing equivalence-principle bounds.

Reference graph

Works this paper leans on

52 extracted references · 51 canonical work pages · cited by 2 Pith papers

  1. [5]

    Nanoscale torsional dis- sipation dilution for quantum experiments and precision mea- surement,

    J. R. Pratt, A. R. Agrawal, C. A. Condos, C. M. Pluchar, S. Schlamminger, and D. J. Wilson, “Nanoscale torsional dis- sipation dilution for quantum experiments and precision mea- surement,” Phys. Rev. X13, 011018 (2023)

    work page 2023

  2. [1]

    Torsion balances, torsion pen- dulums, and related devices,

    G. T. Gillies and R. C. Ritter, “Torsion balances, torsion pen- dulums, and related devices,” Rev. Sci. Instrum. 64, 283–309 (1993), https://doi.org/10.1063/1.1144248

    work page doi:10.1063/1.1144248 1993

  3. [2]

    Tests of the grav- itational inverse-square law,

    E. Adelberger, B. Heckel, and A. Nelson, “Tests of the grav- itational inverse-square law,” Annual Review of Nuclear and Particle Science 53, 77–121 (2003)

    work page 2003

  4. [3]

    Torsion-balance search for ultralow-mass bosonic dark matter,

    E. Shaw, M. Ross, C. Hagedorn, E. Adelberger, and J. Gund- 5 lach, “Torsion-balance search for ultralow-mass bosonic dark matter,” Physical Review D105, 042007 (2022)

    work page 2022

  5. [4]

    Nanoscale torsional optomechanics,

    P. H. Kim, C. Doolin, B. D. Hauer, A. J. MacDonald, M. R. Freeman, P. E. Barclay, and J. P. Davis, “Nanoscale torsional optomechanics,” Applied Physics Letters 102 (2013)

    work page 2013

  6. [6]

    Nanomechanical torsional resonator torque magnetometry,

    J. Davis, D. Vick, P. Li, S. Portillo, A. Fraser, J. Burgess, D. Fortin, W. Hiebert, and M. Freeman, “Nanomechanical torsional resonator torque magnetometry,” Journal of Applied Physics 109 (2011)

    work page 2011

  7. [7]

    On-chip torsion balances with femtonewton force resolution at room temperature enabled by carbon nanotube and graphene,

    L. Cong, Z. Yuan, Z. Bai, X. Wang, W. Zhao, X. Gao, X. Hu, P. Liu, W. Guo, Q. Li, et al., “On-chip torsion balances with femtonewton force resolution at room temperature enabled by carbon nanotube and graphene,” Science Advances7, eabd2358 (2021)

    work page 2021

  8. [8]

    Single-crystal silicon high-q torsional oscillators,

    R. Kleiman, G. Kaminsky, J. Reppy, R. Pindak, and D. Bishop, “Single-crystal silicon high-q torsional oscillators,” Review of scientific instruments 56, 2088–2091 (1985)

    work page 1985

Show all 52 references
  1. [9]

    Microscale torsion resonators for short-range gravity experiments,

    J. Manley, C. Condos, S. Schlamminger, J. Pratt, D. Wilson, and W. Terrano, “Microscale torsion resonators for short-range gravity experiments,” Physical Review D110, 122005 (2024)

    2024

  2. [10]

    Stronger limits on hypothetical yukawa inter- actions in the 30–8000 nm range,

    Y .-J. Chen, W. K. Tham, D. E. Krause, D. López, E. Fischbach, and R. S. Decca, “Stronger limits on hypothetical yukawa inter- actions in the 30–8000 nm range,” Phys. Rev. Lett.116, 221102 (2016)

    2016

  3. [11]

    Laser cooling a 1-milligram torsional pendulum to 240 mi- crokelvins,

    S. Agafonova, P. Rossello, M. Mekonnen, and O. Hosten, “Laser cooling a 1-milligram torsional pendulum to 240 mi- crokelvins,” arXiv preprint arXiv:2408.09445 (2024)

    2024

  4. [12]

    Massive quantum systems as interfaces of quantum mechanics and gravity,

    S. Bose, I. Fuentes, A. A. Geraci, S. M. Khan, S. Qvarfort, M. Rademacher, M. Rashid, M. Toroš, H. Ulbricht, and C. C. Wanjura, “Massive quantum systems as interfaces of quantum mechanics and gravity,” Rev. Mod. Phys.97, 015003 (2025)

    2025

  5. [13]

    Ultra- coherent nanomechanical resonators via soft clamping and dis- sipation dilution,

    Y . Tsaturyan, A. Barg, E. S. Polzik, and A. Schliesser, “Ultra- coherent nanomechanical resonators via soft clamping and dis- sipation dilution,” Nature Nanotechnology12, 776–783 (2017)

    2017

  6. [14]

    Ultrahigh- quality-factor micro-and nanomechanical resonators using dis- sipation dilution,

    N. J. Engelsen, A. Beccari, and T. J. Kippenberg, “Ultrahigh- quality-factor micro-and nanomechanical resonators using dis- sipation dilution,” Nature Nanotechnology , 1–13 (2024)

    2024

  7. [15]

    Elastic strain engineering for ultralow mechanical dissipation,

    A. H. Ghadimi, S. A. Fedorov, N. J. Engelsen, M. J. Bereyhi, R. Schilling, D. J. Wilson, and T. J. Kippenberg, “Elastic strain engineering for ultralow mechanical dissipation,” Science 360, 764–768 (2018)

    2018

  8. [16]

    Hierarchical tensile structures with ultralow mechanical dissipation,

    M. J. Bereyhi, A. Beccari, R. Groth, S. A. Fedorov, A. Arab- moheghi, T. J. Kippenberg, and N. J. Engelsen, “Hierarchical tensile structures with ultralow mechanical dissipation,” Nature Communications 13, 3097 (2022)

    2022

  9. [17]

    Lxxxiv. the bifilar property of twisted strips,

    J. Buckley, “Lxxxiv. the bifilar property of twisted strips,” The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 28, 778–787 (1914)

    1914

  10. [18]

    Mémoire sur la torsion des prismes,[essay on twisting prisms], mémoires des savants étrangers [essays of foreign scholars],

    A. Barré Saint-Venant, “Mémoire sur la torsion des prismes,[essay on twisting prisms], mémoires des savants étrangers [essays of foreign scholars],” CR Acad. Sci 14, 233–560 (1855)

  11. [19]

    Generalized dissipation dilution in strained mechanical resonators,

    S. A. Fedorov, N. J. Engelsen, A. H. Ghadimi, M. J. Bereyhi, R. Schilling, D. J. Wilson, and T. J. Kippenberg, “Generalized dissipation dilution in strained mechanical resonators,” Physical Review B 99, 054107 (2019)

    2019

  12. [20]

    Evidence of surface loss as ubiquitous limiting damping mechanism in sin micro-and nanomechanical resonators,

    L. G. Villanueva and S. Schmid, “Evidence of surface loss as ubiquitous limiting damping mechanism in sin micro-and nanomechanical resonators,” Physical Review Letters 113, 227201 (2014)

    2014

  13. [21]

    Clamp-tapering increases the quality factor of stressed nanobeams,

    M. J. Bereyhi, A. Beccari, S. A. Fedorov, A. H. Ghadimi, R. Schilling, D. J. Wilson, N. J. Engelsen, and T. J. Kippen- berg, “Clamp-tapering increases the quality factor of stressed nanobeams,” Nano letters 19, 2329–2333 (2019)

    2019

  14. [22]

    Influ- ence of clamp-widening on the quality factor of nanomechani- cal silicon nitride resonators,

    P. Sadeghi, M. Tanzer, S. L. Christensen, and S. Schmid, “Influ- ence of clamp-widening on the quality factor of nanomechani- cal silicon nitride resonators,” Journal of Applied Physics 126 (2019)

    2019

  15. [23]

    Mechanical res- onators for quantum optomechanics experiments at room tem- perature,

    R. A. Norte, J. P. Moura, and S. Gröblacher, “Mechanical res- onators for quantum optomechanics experiments at room tem- perature,” Physical Review Letters 116, 147202 (2016)

    2016

  16. [24]

    Ultralow-noise sin trampoline resonators for sensing and op- tomechanics,

    C. Reinhardt, T. Müller, A. Bourassa, and J. C. Sankey, “Ultralow-noise sin trampoline resonators for sensing and op- tomechanics,” Physical Review X 6, 021001 (2016)

    2016

  17. [25]

    Control of material damping in high-q membrane microresonators,

    P.-L. Yu, T. Purdy, and C. Regal, “Control of material damping in high-q membrane microresonators,” Physical Review Letters 108, 083603 (2012)

    2012

  18. [26]

    Ultra-coherent nano-mechanical resonators for quantum optomechanics at room temperature,

    A. H. Ghadimi, “Ultra-coherent nano-mechanical resonators for quantum optomechanics at room temperature,” (2018)

    2018

  19. [27]

    COMSOL v5.4 Structural Mechanics Module User’s Guide

  20. [28]

    Topology optimization: The- ory, methods, and applications,

    M. P. Bendsoe and O. Sigmund, “Topology optimization: The- ory, methods, and applications,” (2013)

    2013

  21. [29]

    Bayesian approach to global optimization and application to multiobjective and constrained problems,

    J. B. Mockus and L. J. Mockus, “Bayesian approach to global optimization and application to multiobjective and constrained problems,” Journal of optimization theory and applications 70, 157–172 (1991)

    1991

  22. [30]

    Practical bayesian optimization of machine learning algorithms,

    J. Snoek, H. Larochelle, and R. P. Adams, “Practical bayesian optimization of machine learning algorithms,” Advances in neural information processing systems 25 (2012)

    2012

  23. [31]

    Ultra-coherent nanomechanical resonators based on inverse design,

    D. Høj, F. Wang, W. Gao, U. B. Hoff, O. Sigmund, and U. L. Andersen, “Ultra-coherent nanomechanical resonators based on inverse design,” Nature communications 12, 5766 (2021)

    2021

  24. [32]

    Pentagonal pho- tonic crystal mirrors: Scalable lightsails with enhanced ac- celeration via neural topology optimization,

    L. Norder, S. Yin, M. de Jong, F. Stallone, H. Aydog- mus, P. Sberna, M. Bessa, and R. Norte, “Pentagonal pho- tonic crystal mirrors: Scalable lightsails with enhanced ac- celeration via neural topology optimization,” arXiv preprint arXiv:2407.07896 (2024)

    2024 arXiv

  25. [33]

    Spiderweb nanomechanical resonators via bayesian optimization: inspired by nature and guided by machine learning,

    D. Shin, A. Cupertino, M. H. de Jong, P. G. Steeneken, M. A. Bessa, and R. A. Norte, “Spiderweb nanomechanical resonators via bayesian optimization: inspired by nature and guided by machine learning,” Advanced Materials34, 2106248 (2022)

    2022

  26. [34]

    Centimeter-scale nanomechanical resonators with low dissipation,

    A. Cupertino, D. Shin, L. Guo, P. G. Steeneken, M. A. Bessa, and R. A. Norte, “Centimeter-scale nanomechanical resonators with low dissipation,” Nature Communications 15, 4255 (2024)

    2024

  27. [35]

    Black box optimization, machine learning, and no-free lunch theo- rems,

    P. M. Pardalos, V . Rasskazova, M. N. Vrahatis, et al., “Black box optimization, machine learning, and no-free lunch theo- rems,” 170 (2021)

    2021

  28. [36]

    An intuitive tutorial to gaussian processes regres- sion,

    J. Wang, “An intuitive tutorial to gaussian processes regres- sion,” Computing in Science & Engineering (2023)

    2023

  29. [37]

    Acquisition functions in bayesian optimization,

    W. Gan, Z. Ji, and Y . Liang, “Acquisition functions in bayesian optimization,” , 129–135 (2021)

    2021

  30. [38]

    Portfolio alloca- tion for bayesian optimization

    M. Hoffman, E. Brochu, N. De Freitas, et al., “Portfolio alloca- tion for bayesian optimization.” UAI, , 327–336 (2011)

    2011

  31. [39]

    Efficient global optimization of expensive blackbox functions,

    D. R. Jones, M. Schonlau, and W. J. Welch, “Efficient global optimization of expensive blackbox functions,” Journal of Global Optimization 13, 455–492 (1998)

    1998

  32. [40]

    Note that while our analysis is limited to one or two parameters, the Bayesian optimization algorithm can easily handle more, at the expense of increased runtime [35]

  33. [41]

    Ultra-high-q membrane optomechanics with a twist,

    A. R. Agrawal, “Ultra-high-q membrane optomechanics with a twist,” (2025)

    2025

  34. [42]

    Quantum- 6 limited optical lever measurement of a torsion oscillator,

    C. M. Pluchar, A. R. Agrawal, and D. J. Wilson, “Quantum- 6 limited optical lever measurement of a torsion oscillator,” Op- tica 12, 418–423 (2025)

    2025

  35. [43]

    De- pendence of the quality factor of micromachined silicon beam resonators on pressure and geometry,

    F. Blom, S. Bouwstra, M. Elwenspoek, and J. Fluitman, “De- pendence of the quality factor of micromachined silicon beam resonators on pressure and geometry,” Journal of Vacuum Sci- ence & Technology B: Microelectronics and Nanometer Struc- tures Processing, Measurement, and Ph...

    1992

  36. [44]

    Cavity optomechanics with stoichiometric sin films,

    D. J. Wilson, C. A. Regal, S. B. Papp, and H. Kimble, “Cavity optomechanics with stoichiometric sin films,” Physical Review Letters 103, 207204 (2009)

    2009

  37. [45]

    Soft-clamped silicon nitride string resonators at millikelvin temperatures,

    T. Gisler, M. Helal, D. Sabonis, U. Grob, M. Héritier, C. L. Degen, A. H. Ghadimi, and A. Eichler, “Soft-clamped silicon nitride string resonators at millikelvin temperatures,” Physical Review Letters 129, 104301 (2022)

    2022

  38. [46]

    Silicon nitride mem- brane resonators at millikelvin temperatures with quality factors exceeding 108,

    M. Yuan, M. A. Cohen, and G. A. Steele, “Silicon nitride mem- brane resonators at millikelvin temperatures with quality factors exceeding 108,” Applied Physics Letters 107 (2015)

    2015

  39. [47]

    Optical probing of mechanical loss of a si 3n4 membrane below 100 mk,

    R. Fischer, N. Kampel, G. Assumpção, P.-L. Yu, K. Cicak, R. Peterson, R. Simmonds, and C. Regal, “Optical probing of mechanical loss of a si 3n4 membrane below 100 mk,” arXiv preprint arXiv:1611.00878 (2016)

    2016 arXiv

  40. [48]

    Active laser cooling of a centimeter-scale torsional oscillator,

    D.-C. Shin, T. M. Hayward, D. Fife, R. Menon, and V . Sudhir, “Active laser cooling of a centimeter-scale torsional oscillator,” Optica 12, 473–478 (2025)

    2025

  41. [49]

    Back action evasion in optical lever detection,

    S. Hao and T. P. Purdy, “Back action evasion in optical lever detection,” Optica 11, 10–17 (2024)

    2024

  42. [50]

    Measurement-based control of a mechanical oscillator at its thermal decoherence rate,

    D. J. Wilson, V . Sudhir, N. Piro, R. Schilling, A. Ghadimi, and T. J. Kippenberg, “Measurement-based control of a mechanical oscillator at its thermal decoherence rate,” Nature524, 325–329 (2015)

    2015

  43. [51]

    Ultralow loss tor- sion micropendula for chipscale gravimetry,

    C. Condos, J. Pratt, J. Manley, A. Agrawal, S. Schlam- minger, C. Pluchar, and D. Wilson, “Ultralow loss tor- sion micropendula for chipscale gravimetry,” arXiv preprint arXiv:2411.04113 (2024)

    2024 arXiv

  44. [52]

    Distinguishable consequence of clas- sical gravity on quantum matter,

    S. Kryhin and V . Sudhir, “Distinguishable consequence of clas- sical gravity on quantum matter,” Phys. Rev. Lett. 134, 061501 (2025)

    2025

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.