pith. sign in

arxiv: 2303.16204 · v3 · submitted 2023-03-28 · 🌌 astro-ph.HE · astro-ph.GA· gr-qc

The Decoupling of Binaries from Their Circumbinary Disks

Alexander J. Dittmann , Geoffrey Ryan , M. Coleman Miller This is my paper

Pith reviewed 2026-05-24 09:43 UTC · model grok-4.3

classification 🌌 astro-ph.HE astro-ph.GAgr-qc
keywords supermassive black hole binariescircumbinary disksdecouplinggravitational wavesaccretionLISAinspiral
0
0 comments X

The pith

Timescale-based predictions overestimate the separations at which black hole binaries decouple from their disks by a factor of about three.

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

The paper studies accreting supermassive black hole binaries inspiraling under gravitational radiation and how they lose contact with their circumbinary gas disks. Analytic arguments and high-resolution simulations both show that the usual method of comparing evolutionary timescales places the decoupling point at separations roughly three times too large. A criterion based on comparing velocities instead marks the transition more accurately. This shift changes expectations for when gas stops affecting the orbit and for the strength and timing of any electromagnetic signals that accompany the gravitational waves. The simulations reveal that lower-viscosity disks decouple earlier, producing a sharp drop in accretion that could pinpoint the host galaxy inside a LISA localization volume.

Core claim

We demonstrate analytically and numerically that timescale-based predictions overestimate the binary separations at which decoupling occurs by factors of ∼3, and illustrate the utility of a velocity-based decoupling criterion. High-viscosity (ν ≳ 0.03 GM/c²) circumbinary systems decouple late (a_b ≲ 15 GM/c²) and have qualitatively similar morphologies near merger to circumbinary systems with constant binary separations. Lower-viscosity circumbinary disks decouple earlier and exhibit qualitatively different accretion flows, which lead to precipitously decreasing accretion onto the binary. Even when dynamically negligible, gas may leave a detectable imprint on the phase of gravitational waves

What carries the argument

velocity-based decoupling criterion that compares the binary's orbital evolution speed to the local gas flow speed

If this is right

  • High-viscosity disks decouple at separations ≲15 GM/c² and retain similar accretion morphologies to fixed-separation cases.
  • Lower-viscosity disks decouple earlier and produce a sharp, observable drop in accretion rate onto the binary.
  • A detected drop in accretion can uniquely identify the host galaxy of a LISA event within its error volume.
  • Accretion amplitude and variability change gradually as the binary decouples over the course of its inspiral.
  • Gas can still imprint a measurable phase shift on the gravitational-wave signal even after it becomes dynamically unimportant.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The revised decoupling separations imply that electromagnetic follow-up strategies for LISA events should focus on smaller binary separations than previously assumed.
  • The velocity criterion could be tested against analytic disk models with varying radial profiles to see whether the factor-of-three offset persists.
  • If future observations detect the predicted accretion drop, it would strengthen the case for using velocity rather than timescale comparisons in population synthesis models.

Load-bearing premise

The numerical hydrodynamics accurately captures the gas flow and decoupling point down to scales of ∼0.04 GM/c² without significant resolution or artificial viscosity artifacts affecting the late-inspiral accretion morphology.

What would settle it

A simulation at higher resolution or with different artificial viscosity that instead finds decoupling at the larger separations given by timescale comparisons would falsify the central result.

Figures

Figures reproduced from arXiv: 2303.16204 by Alexander J. Dittmann, Geoffrey Ryan, M. Coleman Miller.

Figure 1
Figure 1. Figure 1: Surface density profiles from our ν0 = 0.03 and ν0 = 0.003 simulations over the course of their inspirals, during which decoupling occurs around ab ∼ 14 Rg and ab ∼ 56 Rg respectively. Each panel is 800 Rg long in each dimension. The flow of gas around the binary near merger is qualitatively different when the system decouples at earlier times. which removes gas at a rate 600(0.05a0/rs) 2ν0Ω0(1 − (ri/rs) 4… view at source ↗
Figure 2
Figure 2. Figure 2: The top panel displays the measurements of the cavity semi-major axis as a function of binary semi-major axis from each of our simulations. The bottom panel dis￾plays our inferences of the characteristic binary semi-major axis at decoupling compared with timescale-based predic￾tions (Equation (5), a solid green line) and velocity-based predictions (Equation (6), a dashed orange line). Although decoupling i… view at source ↗
Figure 3
Figure 3. Figure 3: The accretion rates, normalized by M˙ 0 = 3πνΣ0, for our ν0 = 0.03 and ν0 = 0.003 simulations. The leftmost panels display the accretion rates over the course of the entire simulations, after the initial burn-in stage. The other panels in each row focus on narrower windows, displaying shorter-term variation in the binary accretion rate. Vertical dashed lines indicate the time at which each binary reaches a… view at source ↗
Figure 4
Figure 4. Figure 4: Lomb-Scargle periodograms of the accretion rate time series from our ν0 = 0.01 and ν0 = 0.001 simulations during three epochs: blue lines plot the periodograms over the first 50πΩ −1 0 after burn-in, orange lines plot the periodograms over a period of 50πΩ −1 0 centered on the time of decoupling, and red lines plot the periodogram over the final 50πΩ −1 0 of each simulation before merger. The left column p… view at source ↗
Figure 5
Figure 5. Figure 5: The evolution of black hole binaries in the LISA band at z = 2 over the final three years prior to merger. Pur￾ple, green, and blue dots mark the points where each binary reaches ad, given by Equation 7), for ν0 = {0.1, 0.01, 0.001} respectively. The dashed markings along each line mark one year, one month, one day, and one hour before merger. De￾pending on the system mass and disk viscosity, many binaries… view at source ↗
read the original abstract

We have investigated, both analytically and numerically, accreting supermassive black hole binaries as they inspiral due to gravitational radiation to elucidate the decoupling of binaries from their disks and inform future multi-messenger observations of these systems. Our numerical studies evolve equal-mass binaries from initial separations of $100 GM/c^2$ until merger, resolving scales as small as $\sim0.04 GM/c^2$, where $M$ is the total binary mass. Our simulations accurately capture the point at which the orbital evolution of each binary decouples from that of their circumbinary disk, and precisely resolve the flow of gas throughout the inspiral. We demonstrate analytically and numerically that timescale-based predictions overestimate the binary separations at which decoupling occurs by factors of $\sim3$, and illustrate the utility of a velocity-based decoupling criterion. High-viscosity ($\nu\gtrsim0.03 GM/c$) circumbinary systems decouple late ($a_b\lesssim 15 GM/c^2$) and have qualitatively similar morphologies near merger to circumbinary systems with constant binary separations. Lower-viscosity circumbinary disks decouple earlier and exhibit qualitatively different accretion flows, which lead to precipitously decreasing accretion onto the binary. If detected, such a decrease may unambiguously identify the host galaxy of an ongoing event within a LISA error volume. We illustrate how accretion amplitude and variability evolve as binaries gradually decouple from their circumbinary disks, and where decoupling occurs over the course of binary inspirals in the LISA band. We show that, even when dynamically negligible, gas may leave a detectable imprint on the phase of gravitational waves.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript investigates accreting equal-mass supermassive black hole binaries inspiraling under gravitational-wave emission from initial separations of 100 GM/c² to merger. Through analytic arguments and numerical hydrodynamical simulations that resolve down to ~0.04 GM/c², it claims that conventional timescale-based decoupling criteria overestimate the binary separation at which the binary decouples from its circumbinary disk by a factor of ~3; a velocity-based decoupling criterion is shown to be more accurate. High-viscosity (ν ≳ 0.03 GM/c) disks are found to decouple late (a_b ≲ 15 GM/c²) with morphologies similar to fixed-separation cases, while lower-viscosity disks decouple earlier and exhibit a sharp drop in accretion that may serve as an electromagnetic identifier within LISA error volumes. The work also examines the imprint of gas on gravitational-wave phase even when dynamically negligible.

Significance. If the reported factor-of-3 discrepancy and the velocity-based criterion hold, the results would revise expectations for the timing and morphology of electromagnetic counterparts to LISA sources and provide a concrete observational diagnostic via accretion-rate drops. The analytic demonstration combined with simulations that track the full inspiral is a clear strength, as is the explicit mapping of accretion amplitude and variability across the LISA band. The absence of reported convergence tests on the numerical decoupling measurement, however, limits the immediate weight that can be assigned to the quantitative claims.

major comments (2)
  1. [Abstract and Numerical Methods] Abstract and Numerical Methods section: the statement that the simulations 'accurately capture' the decoupling point at scales ~0.04 GM/c² is not accompanied by any convergence tests, resolution studies, or sensitivity analysis with respect to grid spacing or the viscosity parameter ν. Because the headline result (timescale-based predictions overestimate decoupling separation by ~3) is extracted directly from the measured decoupling location in these runs, the lack of such verification is load-bearing for the central claim.
  2. [Results] Results section (discussion of high- vs. low-viscosity cases): the reported decoupling separations (a_b ≲ 15 GM/c² for ν ≳ 0.03 GM/c) and the qualitative change in accretion morphology are presented without error bars or explicit checks that the measured transition radius remains stable when ν or numerical diffusivity is varied; this directly affects the robustness of the claimed factor-of-3 offset from timescale-based estimates.
minor comments (2)
  1. Notation for binary separation a_b and total mass M is used without an explicit reminder of the units (GM/c²) in every figure caption or equation block, which could confuse readers comparing to other literature.
  2. [Analytic section] The analytic derivation of the velocity-based criterion would benefit from an explicit equation number and a short appendix showing the algebraic steps that lead to the factor-of-3 improvement over the timescale criterion.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading of the manuscript and for highlighting the need for additional numerical verification. We address the major comments point by point below.

read point-by-point responses

  1. Referee: [Abstract and Numerical Methods] Abstract and Numerical Methods section: the statement that the simulations 'accurately capture' the decoupling point at scales ~0.04 GM/c² is not accompanied by any convergence tests, resolution studies, or sensitivity analysis with respect to grid spacing or the viscosity parameter ν. Because the headline result (timescale-based predictions overestimate decoupling separation by ~3) is extracted directly from the measured decoupling location in these runs, the lack of such verification is load-bearing for the central claim.

    Authors: We agree that explicit convergence tests are required to substantiate the quantitative decoupling measurements. The resolution ~0.04 GM/c² was selected on the basis of earlier fixed-separation studies, but we did not report dedicated resolution or ν-sensitivity runs for the full-inspiral decoupling diagnostic. In the revised manuscript we will add a new subsection to Numerical Methods that presents resolution studies at two additional grid spacings and quantifies the resulting variation in the measured decoupling separation; we will also report the outcome of limited ν-variation tests around the fiducial values. revision: yes

  2. Referee: [Results] Results section (discussion of high- vs. low-viscosity cases): the reported decoupling separations (a_b ≲ 15 GM/c² for ν ≳ 0.03 GM/c) and the qualitative change in accretion morphology are presented without error bars or explicit checks that the measured transition radius remains stable when ν or numerical diffusivity is varied; this directly affects the robustness of the claimed factor-of-3 offset from timescale-based estimates.

    Authors: We accept that the absence of error bars and stability checks weakens the presentation of the factor-of-3 result. In revision we will attach uncertainty estimates to the quoted decoupling separations (derived from the spread across the simulation suite) and will add a short paragraph in Results that demonstrates the transition radius remains stable under modest changes in ν and numerical diffusivity. revision: yes

Circularity Check

0 steps flagged

No circularity: decoupling separation measured directly from simulation outputs compared to external timescale predictions

full rationale

The paper's headline result compares numerical measurements of decoupling (identified when binary orbital evolution separates from disk evolution) against pre-existing timescale-based analytic predictions, showing a factor-of-~3 overestimate. The decoupling criterion itself is extracted from the time evolution of orbital separation and disk quantities in the hydrodynamical runs; it is not defined in terms of the reported factor or any fitted parameter. No self-citation chain, ansatz smuggling, or renaming of known results appears in the derivation. The claim that simulations 'accurately capture' the decoupling point is an assertion of numerical fidelity rather than a tautology that reduces the result to its own inputs by construction. The derivation chain therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Abstract provides limited detail; viscosity appears as an explored parameter rather than a fitted constant, and standard GR inspiral is assumed without new entities introduced.

free parameters (1)
  • viscosity ν
    Explored in high (≳0.03 GM/c) and lower regimes; value controls decoupling time and accretion morphology.
axioms (1)
  • domain assumption Binary inspiral is driven solely by gravitational wave emission on the timescales considered
    Stated in the setup of evolving binaries from 100 GM/c² to merger.

pith-pipeline@v0.9.0 · 5840 in / 1317 out tokens · 24788 ms · 2026-05-24T09:43:07.861286+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

63 extracted references · 63 canonical work pages

  1. [1]

    2007, Classical and Quantum Gravity, 24, S689

    Ajith, P., Babak, S., Chen, Y., et al. 2007, Classical and Quantum Gravity, 24, S689. https://dx.doi.org/10.1088/0264-9381/24/19/S31

    work page doi:10.1088/0264-9381/24/19/s31 2007

  2. [2]

    J., & Natarajan, P

    Armitage, P. J., & Natarajan, P. 2002, ApJL, 567, L9

    work page 2002

  3. [3]

    Artymowicz, P., & Lubow, S. H. 1994, ApJ, 421, 651

    work page 1994

  4. [4]

    E., & Hernquist, L

    Barnes, J. E., & Hernquist, L. 1992, ARA&A, 30, 705

    work page 1992

  5. [5]

    C., Blandford, R

    Begelman, M. C., Blandford, R. D., & Rees, M. J. 1980, Nature, 287, 307

    work page 1980

  6. [6]

    2019, MNRAS, 486, 4044

    Bonetti, M., Sesana, A., Haardt, F., Barausse, E., & Colpi, M. 2019, MNRAS, 486, 4044

    work page 2019

  7. [7]

    B., Mewes, V., Noble, S

    Bowen, D. B., Mewes, V., Noble, S. C., et al. 2019, ApJ, 879, 76

    work page 2019

  8. [8]

    2016, MNRAS, 463, 2145

    Charisi, M., Bartos, I., Haiman, Z., et al. 2016, MNRAS, 463, 2145

    work page 2016

  9. [9]

    2020, MNRAS, 499, 2245

    Chen, Y.-C., Liu, X., Liao, W.-T., et al. 2020, MNRAS, 499, 2245

    work page 2020

  10. [10]

    G., Campanelli, M., et al

    Combi, L., Lopez Armengol, F. G., Campanelli, M., et al. 2022, ApJ, 928, 187

    work page 2022

  11. [11]

    2019, eXtremeGravityInstitute/LISA Sensitivity: Version 1, v1, Zenodo, doi:10.5281/zenodo.2636514

    Cornish, N. 2019, eXtremeGravityInstitute/LISA Sensitivity: Version 1, v1, Zenodo, doi:10.5281/zenodo.2636514. https://doi.org/10.5281/zenodo.2636514

    work page doi:10.5281/zenodo.2636514 2019

  12. [12]

    R., Haiman, Z., & MacFadyen, A

    Corrales, L. R., Haiman, Z., & MacFadyen, A. 2010, MNRAS, 404, 947

    work page 2010

  13. [13]

    2022, PhRvD, 105, 103010

    Davelaar, J., & Haiman, Z. 2022, PhRvD, 105, 103010

    work page 2022

  14. [14]

    2021, MNRAS, 501, 3540

    MacFadyen, A. 2021, MNRAS, 501, 3540

    work page 2021

  15. [15]

    J., & Ryan, G

    Dittmann, A. J., & Ryan, G. 2021, ApJ, 921, 71 —. 2022, MNRAS, 513, 6158 —. 2023, in preparation D’Orazio, D. J., & Haiman, Z. 2017, MNRAS, 470, 1198

    work page 2021

  16. [16]

    Eggleton, P. P. 1983, ApJ, 268, 368

    work page 1983

  17. [17]

    D., Duffell, P., MacFadyen, A

    Farris, B. D., Duffell, P., MacFadyen, A. I., & Haiman, Z. 2014, ApJ, 783, 134 —. 2015a, MNRAS, 447, L80 —. 2015b, MNRAS, 446, L36

    work page 2014

  18. [18]

    Shapiro, S. L. 2012, PhRvL, 109, 221102

    work page 2012

  19. [19]

    G., & Stone, J

    Felker, K. G., & Stone, J. M. 2018, Journal of Computational Physics, 375, 1365

    work page 2018

  20. [20]

    2022, MNRAS, 517, 1339

    Mayer, L. 2022, MNRAS, 517, 1339

    work page 2022

  21. [21]

    B., Shapiro, S

    Gold, R., Paschalidis, V., Etienne, Z. B., Shapiro, S. L., & Pfeiffer, H. P. 2014, PhRvD, 89, 064060

    work page 2014

  22. [22]

    J., Djorgovski, S

    Graham, M. J., Djorgovski, S. G., Stern, D., et al. 2015, MNRAS, 453, 1562 Guti´ errez, E. M., Combi, L., Noble, S. C., et al. 2022, ApJ, 928, 137

    work page 2015

  23. [23]

    2009, ApJ, 700, 1952

    Haiman, Z., Kocsis, B., & Menou, K. 2009, ApJ, 700, 1952

    work page 2009

  24. [24]

    Hunter, J. D. 2007, Computing in Science Engineering, 9, 90

    work page 2007

  25. [25]

    Khan, A., Paschalidis, V., Ruiz, M., & Shapiro, S. L. 2018, PhRvD, 97, 044036

    work page 2018

  26. [26]

    2011, PhRvD, 84, 024032 10 Dittmann, Ryan, & Miller

    Kocsis, B., Yunes, N., & Loeb, A. 2011, PhRvD, 84, 024032 10 Dittmann, Ryan, & Miller

    work page 2011

  27. [27]

    Kormendy, J., & Ho, L. C. 2013, ARA&A, 51, 511

    work page 2013

  28. [28]

    M., Davelaar, J., Haiman, Z., et al

    Krauth, L. M., Davelaar, J., Haiman, Z., et al. 2023, arXiv e-prints, arXiv:2304.02575

    work page arXiv 2023

  29. [29]

    Lai, D., & Mu˜ noz, D. J. 2022, arXiv e-prints, arXiv:2211.00028

    work page arXiv 2022

  30. [30]

    2021, MNRAS, 500, 4025

    Liao, W.-T., Chen, Y.-C., Liu, X., et al. 2021, MNRAS, 500, 4025

    work page 2021

  31. [31]

    K., Wu, X.-B., & Cao, S

    Liu, F. K., Wu, X.-B., & Cao, S. L. 2003, MNRAS, 340, 411

    work page 2003

  32. [32]

    Liu, T., Gezari, S., & Miller, M. C. 2018, ApJL, 859, L12

    work page 2018

  33. [33]

    2019, ApJ, 884, 36

    Liu, T., Gezari, S., Ayers, M., et al. 2019, ApJ, 884, 36

    work page 2019

  34. [34]

    Lomb, N. R. 1976, Ap&SS, 39, 447

    work page 1976

  35. [35]

    Lubow, S. H. 1991, ApJ, 381, 259

    work page 1991

  36. [36]

    Lynden-Bell, D., & Pringle, J. E. 1974, MNRAS, 168, 603

    work page 1974

  37. [37]

    2013, Classical and Quantum Gravity, 30, 244008 Milosavljevi´ c, M., & Merritt, D

    Mayer, L. 2013, Classical and Quantum Gravity, 30, 244008 Milosavljevi´ c, M., & Merritt, D. 2003a, ApJ, 596, 860 Milosavljevi´ c, M., & Merritt, D. 2003b, in American Institute of Physics Conference Series, Vol. 686, The Astrophysics of Gravitational Wave Sources, ed. J. M. Centrella, 201–210 Milosavljevi´ c, M., & Phinney, E. S. 2005, ApJL, 622, L93

    work page 2013

  38. [38]

    2015, MNRAS, 452, 2396

    Miranda, R., & Lai, D. 2015, MNRAS, 452, 2396

    work page 2015

  39. [39]

    J., & Lai, D

    Miranda, R., Mu˜ noz, D. J., & Lai, D. 2017, MNRAS, 466, 1170

    work page 2017

  40. [40]

    Moody, M. S. L., Shi, J.-M., & Stone, J. M. 2019, ApJ, 875, 66

    work page 2019

  41. [41]

    C., Krolik, J

    Noble, S. C., Krolik, J. H., Campanelli, M., et al. 2021, ApJ, 922, 175

    work page 2021

  42. [42]

    C., Mundim, B

    Noble, S. C., Mundim, B. C., Nakano, H., et al. 2012, ApJ, 755, 51 O’Neill, S. M., Miller, M. C., Bogdanovi´ c, T., Reynolds, C. S., & Schnittman, J. D. 2009, ApJ, 700, 859

    work page 2012

  43. [43]

    Peters, P. C. 1964, Physical Review, 136, 1224

    work page 1964

  44. [44]

    Ragusa, E., Alexander, R., Calcino, J., Hirsh, K., & Price, D. J. 2020, MNRAS, 499, 3362

    work page 2020

  45. [45]

    Ragusa, E., Lodato, G., & Price, D. J. 2016, MNRAS, 460, 1243

    work page 2016

  46. [46]

    J., & Liu, C

    Robson, T., Cornish, N. J., & Liu, C. 2019, Classical and Quantum Gravity, 36, 105011

    work page 2019

  47. [47]

    Roe, P. L. 1981, Journal of Computational Physics, 43, 357

    work page 1981

  48. [48]

    Ruiz, M., Tsokaros, A., & Shapiro, S. L. 2023, arXiv e-prints, arXiv:2302.09083

    work page arXiv 2023

  49. [49]

    Scargle, J. D. 1982, ApJ, 263, 835

    work page 1982

  50. [50]

    D., & Krolik, J

    Schnittman, J. D., & Krolik, J. H. 2008, ApJ, 684, 835

    work page 2008

  51. [51]

    I., & Sunyaev, R

    Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 500, 33

    work page 1973

  52. [52]

    Shapiro, S. L. 2010, PhRvD, 81, 024019

    work page 2010

  53. [53]

    2005, ApJL, 620, L79

    Springel, V., Di Matteo, T., & Hernquist, L. 2005, ApJL, 620, L79

    work page 2005

  54. [54]

    M., & Gardiner, T

    Stone, J. M., & Gardiner, T. 2009, NewA, 14, 139

    work page 2009

  55. [55]

    M., Tomida, K., White, C

    Stone, J. M., Tomida, K., White, C. J., & Felker, K. G. 2020, ApJS, 249, 4

    work page 2020

  56. [56]

    2018, MNRAS, 476, 2249

    Tang, Y., Haiman, Z., & MacFadyen, A. 2018, MNRAS, 476, 2249

    work page 2018

  57. [57]

    M., Greene, C

    Thyng, K. M., Greene, C. A., Hetland, R. D., Zimmerle, H. M., & DiMarco, S. F. 2016, Oceanography. https://doi.org/10.5670/oceanog.2016.66

    work page doi:10.5670/oceanog.2016.66 2016

  58. [58]

    2020, ApJ, 900, 43

    Tiede, C., Zrake, J., MacFadyen, A., & Haiman, Z. 2020, ApJ, 900, 43

    work page 2020

  59. [59]

    Townsend, R. H. D. 2010, ApJS, 191, 247

    work page 2010

  60. [60]

    J., Smith, B

    Turk, M. J., Smith, B. D., Oishi, J. S., et al. 2011, ApJS, 192, 9 van der Walt, S., Colbert, S. C., & Varoquaux, G. 2011, Computing in Science Engineering, 13, 22

    work page 2011

  61. [61]

    G., et al

    Vaughan, S., Uttley, P., Markowitz, A. G., et al. 2016, MNRAS, 461, 3145

    work page 2016

  62. [62]

    2021, MNRAS, 506, 2408

    Xin, C., & Haiman, Z. 2021, MNRAS, 506, 2408

    work page 2021

  63. [63]

    R., & Mayer, L

    Zwick, L., Capelo, P. R., & Mayer, L. 2023, MNRAS, arXiv:2209.04060

    work page arXiv 2023