pith. sign in

arxiv: 2606.24824 · v1 · pith:B237TSMGnew · submitted 2026-06-23 · 💻 cs.AI

Solving Inverse Problems of Chaotic Systems with Bidirectional Conditional Flow Matching

Peiyan Hu , Jian Zhang , Jiashu Pan , Ruiqi Feng , Tao Zhang , Zhi-Ming Ma , Yuan-Sen Ting , Gongjie Li
show 1 more author
Tailin Wu
This is my paper

Pith reviewed 2026-06-25 23:15 UTC · model grok-4.3

classification 💻 cs.AI
keywords chaotic systemsinverse problemsflow matchingbidirectional mappingconservation lawsLorenz systemplanetary dynamicsglobular clusters
0
0 comments X

The pith

Bidirectional conditional flow matching learns mappings between initial and final state distributions to solve inverse problems in chaotic systems.

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

The paper proposes Bidirectional Conditional Flow Matching (Bi-CFM) as a way to infer initial conditions from observed final states in chaotic dynamics, where the problem is ill-posed due to instability and non-uniqueness. It trains conditional flow models in both directions to capture the stochastic spread of possible trajectories and reduce error blow-up when running dynamics backward. An extension called CBi-CFM adds explicit constraints to preserve conservation laws. Tests on Lorenz, circuit, and high-dimensional Lorenz 96 systems show gains on five distribution metrics plus more than 100-fold speedup over baselines. On three-body scattering and real globular-cluster data, the constrained version keeps conservation errors at levels matching direct simulation.

Core claim

Bi-CFM learns bidirectional mappings between distributions of initial and final states to capture the stochasticity of chaotic evolution and mitigate exponential error accumulation over time; CBi-CFM further respects conservation laws with errors comparable to ground truth.

What carries the argument

Bidirectional Conditional Flow Matching (Bi-CFM), which trains conditional flow models to transport probability mass in both forward and reverse directions between state distributions.

If this is right

  • Bi-CFM improves five distribution-level metrics over baselines on Lorenz, circuit, and Lorenz 96 systems.
  • It delivers more than two orders of magnitude speedup in solving the inverse mapping.
  • CBi-CFM yields conservation-law errors on three-body problems that match those of ground-truth trajectories.
  • The same constrained model improves accuracy on real globular-cluster observations evolved over 10 Gyr.

Where Pith is reading between the lines

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

  • If the bidirectional training stabilizes reverse-time trajectories, the same architecture could be tested on other stiff or chaotic inverse problems such as weather reanalysis or molecular dynamics.
  • The reported scalability to million-body collisional systems suggests the method might be combined with reduced-order models for even longer astrophysical timescales.
  • Because the approach works directly on state distributions rather than single trajectories, it may reduce sensitivity to the precise choice of integrator step size in the original simulator.

Load-bearing premise

That bidirectional conditional flow matching can reliably capture the stochasticity of chaotic evolution and mitigate exponential error accumulation in time-reverse dynamics without introducing new instabilities or biases.

What would settle it

Forward integration of states sampled from the learned reverse map produces final distributions whose distance to the target final distribution exceeds that of baselines or ground-truth trajectories by a statistically significant margin on the Lorenz or three-body test cases.

read the original abstract

Modeling chaotic systems is crucial yet challenging. Inverse problems in chaotic dynamics, namely inferring initial conditions from final states, remain largely unsolved because of ill-posedness, non-uniqueness, instability, and potentially chaotic time-reverse dynamics. We address this open problem with Bidirectional Conditional Flow Matching (Bi-CFM), which learns bidirectional mappings between distributions of initial and final states to capture the stochasticity of chaotic evolution and mitigate exponential error accumulation over time. Furthermore, for systems with conservation laws, we extend it to Conservation-constrained Bi-CFM (CBi-CFM). Across the classic Lorenz, Circuit, and high-dimensional Lorenz 96 systems, Bi-CFM improves five distribution-level metrics over baselines while achieving a speedup of more than two orders of magnitude. In the three-body planet-planet scattering problem in planetary dynamics, CBi-CFM better respects conservation laws, with conservation errors comparable to those of the ground truth. Finally, on real observations of globular clusters, collisional million-body systems shaped by $\sim 10^{10}$ years (10 Gyr) of evolution, our method represents an advance in accuracy, establishing a scalable route to solving inverse problems of long-timescale real-world chaotic dynamics.

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

3 major / 2 minor

Summary. The manuscript proposes Bidirectional Conditional Flow Matching (Bi-CFM) to solve inverse problems for chaotic systems by learning bidirectional mappings between distributions of initial and final states, thereby capturing stochasticity and mitigating exponential error growth in time-reversed dynamics. An extension, Conservation-constrained Bi-CFM (CBi-CFM), is introduced for systems obeying conservation laws. Experiments on Lorenz, Circuit, and high-dimensional Lorenz-96 systems report gains on five distribution-level metrics plus >100x speedup versus baselines; CBi-CFM is shown to match ground-truth conservation errors on three-body scattering and real globular-cluster observations.

Significance. If the central claims hold after verification, the work would constitute a meaningful advance on an open problem in chaotic inverse modeling. The bidirectional flow-matching formulation offers a learned, scalable alternative to direct integration or optimization-based inversion, with demonstrated applicability to both synthetic chaotic attractors and long-timescale astrophysical data. Explicit credit is due for the reproducible experimental protocol implied by the multi-system evaluation and the conservation-error comparison against ground truth.

major comments (3)
  1. [Abstract, §4] Abstract and §4 (results): the claim that Bi-CFM 'mitigates exponential error accumulation' and 'captures the stochasticity of chaotic evolution' is load-bearing for the central contribution, yet the reported distribution metrics do not include explicit quantification of reverse-time divergence rates, Lyapunov-exponent estimates, or bias checks against direct forward integration on the same trajectories.
  2. [§3.1–3.2] §3.1–3.2 (method): the bidirectional conditional flow matching is presented as addressing ill-posedness without new instabilities, but no derivation or ablation demonstrates that the learned reverse mapping remains stable under the positive Lyapunov exponents characteristic of the target systems; the abstract's metric gains alone do not establish this property.
  3. [Table 1, §4.2] Table 1 / §4.2 (Lorenz-96): the five distribution-level metrics and speedup are reported without error bars, data-split details, or baseline hyper-parameter counts, preventing assessment of whether the reported improvements are robust or sensitive to implementation choices.
minor comments (2)
  1. [§3] Notation for the conditional vector field and the bidirectional loss should be unified between §3 and the supplementary material to avoid ambiguity when reproducing the training objective.
  2. [Figure 3] Figure 3 (three-body conservation plots): axis labels and units for the conservation-error curves are missing, complicating direct comparison with the ground-truth reference.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the positive assessment of the work's potential significance and for the constructive major comments. We respond point-by-point below and will incorporate revisions to address the identified gaps in explicit quantification, stability analysis, and reporting details.

read point-by-point responses

  1. Referee: [Abstract, §4] Abstract and §4 (results): the claim that Bi-CFM 'mitigates exponential error accumulation' and 'captures the stochasticity of chaotic evolution' is load-bearing for the central contribution, yet the reported distribution metrics do not include explicit quantification of reverse-time divergence rates, Lyapunov-exponent estimates, or bias checks against direct forward integration on the same trajectories.

    Authors: The distribution-level metrics (MMD, Wasserstein, etc.) are intended to demonstrate that the learned bidirectional mappings recover the target distributions accurately; mismatched distributions would result from unmitigated exponential divergence. We nevertheless agree that direct evidence via reverse-time divergence rates, Lyapunov estimates on the learned reverse map, and bias checks versus forward integration would strengthen the central claim. We will add these analyses (including numerical estimates and comparisons) to a new subsection in §4. revision: yes

  2. Referee: [§3.1–3.2] §3.1–3.2 (method): the bidirectional conditional flow matching is presented as addressing ill-posedness without new instabilities, but no derivation or ablation demonstrates that the learned reverse mapping remains stable under the positive Lyapunov exponents characteristic of the target systems; the abstract's metric gains alone do not establish this property.

    Authors: The bidirectional conditional flow-matching objective is constructed to enforce consistency between forward and reverse conditionals, which in principle avoids introducing additional instabilities. We acknowledge, however, that the manuscript lacks an explicit derivation of stability under positive Lyapunov exponents and a supporting ablation. We will add a short theoretical argument in §3.2 together with a targeted ablation study in the experiments that varies system Lyapunov exponents and measures reverse-mapping stability. revision: yes

  3. Referee: [Table 1, §4.2] Table 1 / §4.2 (Lorenz-96): the five distribution-level metrics and speedup are reported without error bars, data-split details, or baseline hyper-parameter counts, preventing assessment of whether the reported improvements are robust or sensitive to implementation choices.

    Authors: We agree that error bars, explicit data-split information, and baseline hyper-parameter counts are necessary for evaluating robustness. These details were omitted for brevity but will be supplied in an expanded Table 1, added to §4.2, and included in full in the supplementary material. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical ML method with independent validation

full rationale

The paper introduces Bi-CFM as a learned bidirectional conditional flow matching model for inverse problems in chaotic systems, with extensions to CBi-CFM for conservation constraints. All reported results consist of empirical metric improvements, speedups, and conservation error comparisons on Lorenz, Circuit, Lorenz 96, three-body, and globular cluster data. No derivation chain reduces any prediction or central claim to a fitted parameter by construction, a self-citation load-bearing premise, or an ansatz smuggled via prior work; the method is presented as a trainable mapping whose performance is assessed against external benchmarks rather than internal redefinitions.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard assumptions from flow matching and chaotic dynamics literature plus the domain assumption that bidirectional distribution learning suffices to handle instability and non-uniqueness.

axioms (1)
  • domain assumption Bidirectional mappings between initial and final state distributions can capture the stochasticity of chaotic evolution and mitigate exponential error accumulation.
    Invoked as the core mechanism for addressing ill-posedness and instability in the abstract.

pith-pipeline@v0.9.1-grok · 5768 in / 1187 out tokens · 23658 ms · 2026-06-25T23:15:35.425139+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

121 extracted references · 36 canonical work pages · 11 internal anchors

  1. [1]

    Proceedings of the IEEE 75(8), 1009–1021 (2005)

    Hasler, M.J.: Electrical circuits with chaotic behavior. Proceedings of the IEEE 75(8), 1009–1021 (2005)

    2005

  2. [2]

    American Journal of Physics 68(8), 758–763 (2000)

    Sprott, J.C.: Simple chaotic systems and circuits. American Journal of Physics 68(8), 758–763 (2000)

    2000

  3. [3]

    Bulletin of the American Meteorological Society70(1), 14–23 (1989)

    Tsonis, A., Elsner, J.: Chaos, strange attractors, and weather. Bulletin of the American Meteorological Society70(1), 14–23 (1989)

    1989

  4. [4]

    Scientific American273(5), 62–68 (1995)

    Broecker, W.S.: Chaotic climate. Scientific American273(5), 62–68 (1995)

    1995

  5. [5]

    Nature361(6413), 608–612 (1993)

    Laskar, J., Robutel, P.: The chaotic obliquity of the planets. Nature361(6413), 608–612 (1993)

    1993

  6. [6]

    Contopoulos, G.: Order and Chaos in Dynamical Astronomy vol. 21. Springer, Heidelberg (2002)

    2002

  7. [7]

    Nature267(5607), 117–120 (1977)

    Barrow, J.D.: A chaotic cosmology. Nature267(5607), 117–120 (1977)

    1977

  8. [8]

    Nature267(5607), 177–178 (1977)

    Olsen, L.F., Degn, H.: Chaos in an enzyme reaction. Nature267(5607), 177–178 (1977)

    1977

  9. [9]

    Nature343(6259), 653–657 (1990) 40

    Chialvo, D.R., Gilmour Jr, R.F., Jalife, J.: Low dimensional chaos in cardiac tissue. Nature343(6259), 653–657 (1990) 40

    1990

  10. [10]

    Physical review letters64 8, 821–824 (1990)

    Pecora, L.M., Carroll, T.L.: Synchronization in chaotic systems. Physical review letters64 8, 821–824 (1990)

    1990

  11. [11]

    SIAM/ASA J

    Kantas, N., Beskos, A., Jasra, A.: Sequential monte carlo methods for high- dimensional inverse problems: A case study for the navier-stokes equations. SIAM/ASA J. Uncertain. Quantification2, 464–489 (2013)

    2013

  12. [12]

    2016 28th International Conference on Microelectronics (ICM), 193–196 (2016)

    Ouannas, A., Azar, A.T., Radwan, A.G.: On inverse problem of generalized synchronization between different dimensional integer-order and fractional-order chaotic systems. 2016 28th International Conference on Microelectronics (ICM), 193–196 (2016)

    2016

  13. [13]

    Meteorology and Atmospheric Physics60, 165–178 (1996)

    Bennett, A.F., Chua, B.S., Leslie, L.M.: Generalized inversion of a global numeri- cal weather prediction model. Meteorology and Atmospheric Physics60, 165–178 (1996)

    1996

  14. [14]

    Proceedings

    Gallet, A., Rigby, S., Tallman, T., Kong, X., Hajirasouliha, I., Liew, A., Liu, D., Chen, L., Hauptmann, A., Smyl, D.J.: Structural engineering from an inverse problems perspective. Proceedings. Mathematical, Physical, and Engineering Sciences478(2021)

    2021

  15. [15]

    Annals of Physics275(2), 151–189 (1999)

    Jung, C., Lipp, C., Seligman, T.: The inverse scattering problem for chaotic hamiltonian systems. Annals of Physics275(2), 151–189 (1999)

    1999

  16. [16]

    Springer, New York (2005)

    Kaipio, J.P., Somersalo, E.: Statistical and Computational Inverse Problems. Springer, New York (2005)

    2005

  17. [17]

    A., & Ford, E

    Rasio, F.A., Ford, E.B.: Dynamical instabilities and the for- mation of extrasolar planetary systems. Science274(5289), 954–956 (1996) https://doi.org/10.1126/science.274.5289.954 https://www.science.org/doi/pdf/10.1126/science.274.5289.954

    work page doi:10.1126/science.274.5289.954 1996

  18. [18]

    The Astrophysical Journal614(1), 497 (2004)

    Goldreich, P., Lithwick, Y., Sari, R.: Final stages of planet formation. The Astrophysical Journal614(1), 497 (2004)

    2004

  19. [19]

    B., Matsumura, S., & Rasio, F

    Chatterjee, S., Ford, E.B., Matsumura, S., Rasio, F.A.: Dynamical outcomes of planet-planet scattering. The Astrophysical Journal686(1), 580–602 (2008) https://doi.org/10.1086/590227

    work page doi:10.1086/590227 2008

  20. [20]

    Draft version for use in AMath585(6), 112 (1998)

    LeVeque, R.J.: Finite difference methods for differential equations. Draft version for use in AMath585(6), 112 (1998)

    1998

  21. [21]

    Monthly Notices of the Royal Astronomical Society533(3), 3708–3723 (2024)

    Lu, T., Hernandez, D.M., Rein, H.: Trace: a code for time-reversible astrophysi- cal close encounters. Monthly Notices of the Royal Astronomical Society533(3), 3708–3723 (2024)

    2024

  22. [22]

    Numerische Mathematik97(2), 219–268 (2004)

    Binev, P., Dahmen, W., DeVore, R.: Adaptive finite element methods with 41 convergence rates. Numerische Mathematik97(2), 219–268 (2004)

    2004

  23. [23]

    Nature Communications15(1), 10416 (2024)

    Du, P., Parikh, M.H., Fan, X., Liu, X.-Y., Wang, J.-X.: Conditional neural field latent diffusion model for generating spatiotemporal turbulence. Nature Communications15(1), 10416 (2024)

    2024

  24. [24]

    In: The Thirteenth International Conference on Learning Representations (2024)

    Hu, P., Wang, R., Zheng, X., Zhang, T., Feng, H., Feng, R., Wei, L., Wang, Y., Ma, Z.-M., Wu, T.: Wavelet diffusion neural operator. In: The Thirteenth International Conference on Learning Representations (2024)

    2024

  25. [25]

    Astronomy & Astrophysics701, 149 (2025)

    Alibert, Y., Davoult, J., Marques, S.: A transformer-based generative model for planetary systems. Astronomy & Astrophysics701, 149 (2025)

    2025

  26. [26]

    Scientific Reports14(1), 3143 (2024)

    Wang, M., Li, J.: Interpretable predictions of chaotic dynamical systems using dynamical system deep learning. Scientific Reports14(1), 3143 (2024)

    2024

  27. [27]

    Journal of Computational physics378, 686–707 (2019)

    Raissi, M., Perdikaris, P., Karniadakis, G.E.: Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational physics378, 686–707 (2019)

    2019

  28. [28]

    Journal of Computational Physics425, 109913 (2021)

    Yang, L., Meng, X., Karniadakis, G.E.: B-pinns: Bayesian physics-informed neu- ral networks for forward and inverse pde problems with noisy data. Journal of Computational Physics425, 109913 (2021)

    2021

  29. [29]

    Nature Machine Intelligence5(12), 1466–1475 (2023)

    Bastek, J.-H., Kochmann, D.M.: Inverse design of nonlinear mechanical metama- terials via video denoising diffusion models. Nature Machine Intelligence5(12), 1466–1475 (2023)

    2023

  30. [30]

    In: International Conference on Machine Learning, pp

    Molinaro, R., Yang, Y., Engquist, B., Mishra, S.: Neural inverse operators for solving pde inverse problems. In: International Conference on Machine Learning, pp. 25105–25139 (2023). PMLR

    2023

  31. [31]

    Advances in Neural Information Processing Systems 37, 130291–130323 (2024)

    Huang, J., Yang, G., Wang, Z., Park, J.J.: Diffusionpde: Generative pde-solving under partial observation. Advances in Neural Information Processing Systems 37, 130291–130323 (2024)

    2024

  32. [32]

    In: The Eleventh International Conference on Learning Representations (2023)

    Lipman, Y., Chen, R.T., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. In: The Eleventh International Conference on Learning Representations (2023)

    2023

  33. [33]

    In: The Eleventh International Conference on Learning Representations (2023)

    Liu, X., Gong, C.,et al.: Flow straight and fast: Learning to generate and transfer data with rectified flow. In: The Eleventh International Conference on Learning Representations (2023)

    2023

  34. [34]

    Statistical Science, 69–90 (1992) 42

    Berliner, L.M.: Statistics, probability and chaos. Statistical Science, 69–90 (1992) 42

    1992

  35. [35]

    Physica D: Nonlinear Phe- nomena98(2), 481–491 (1996) https://doi.org/10.1016/0167-2789(96)00107-8

    Leith, C.E.: Stochastic models of chaotic systems. Physica D: Nonlinear Phe- nomena98(2), 481–491 (1996) https://doi.org/10.1016/0167-2789(96)00107-8 . Nonlinear Phenomena in Ocean Dynamics

    work page doi:10.1016/0167-2789(96)00107-8 1996

  36. [36]

    Nature576(7787), 406–410 (2019)

    Stone, N.C., Leigh, N.W.: A statistical solution to the chaotic, non-hierarchical three-body problem. Nature576(7787), 406–410 (2019)

    2019

  37. [37]

    In: Universality in Chaos, 2nd Edition, pp

    Lorenz, E.N.: Deterministic nonperiodic flow 1. In: Universality in Chaos, 2nd Edition, pp. 367–378. Routledge, Cambridge (2017)

    2017

  38. [38]

    Physics Letters A373(40), 3637–3642 (2009)

    Dadras, S., Momeni, H.R.: A novel three-dimensional autonomous chaotic sys- tem generating two, three and four-scroll attractors. Physics Letters A373(40), 3637–3642 (2009)

    2009

  39. [39]

    In: Proc

    Lorenz, E.N.: Predictability: A problem partly solved. In: Proc. Seminar on Predictability, vol. 1, pp. 1–18 (1996). Reading

    1996

  40. [40]

    Journal of Machine Learning Research9, 2579–2605 (2008)

    Maaten, L., Hinton, G.E.: Visualizing data using t-sne. Journal of Machine Learning Research9, 2579–2605 (2008)

    2008

  41. [41]

    Chaos: An interdisciplinary journal of nonlinear science20(4) (2010)

    Karimi, A., Paul, M.R.: Extensive chaos in the lorenz-96 model. Chaos: An interdisciplinary journal of nonlinear science20(4) (2010)

    2010

  42. [42]

    Journal of machine learning research9(11) (2008)

    Maaten, L., Hinton, G.: Visualizing data using t-sne. Journal of machine learning research9(11) (2008)

    2008

  43. [43]

    Distill (2016) https://doi.org/10.23915/distill.00002

    Wattenberg, M., Vi´ egas, F., Johnson, I.: How to use t-sne effectively. Distill (2016) https://doi.org/10.23915/distill.00002

    work page doi:10.23915/distill.00002 2016

  44. [44]

    The Astrophysical Journal581(1), 666 (2002)

    Kokubo, E., Ida, S.: Formation of protoplanet systems and diversity of planetary systems. The Astrophysical Journal581(1), 666 (2002)

    2002

  45. [45]

    Ida, S., Lin, D.N.: Toward a deterministic model of planetary formation. i. a desert in the mass and semimajor axis distributions of extrasolar planets. The Astrophysical Journal604(1), 388 (2004)

    2004

  46. [46]

    , year = 1996, month = feb, volume =

    Chambers, J.E., Wetherill, G.W., Boss, A.P.: The stability of multi-planet systems. Icarus119(2), 261–268 (1996) https://doi.org/10.1006/icar.1996.0019

    work page doi:10.1006/icar.1996.0019 1996

  47. [47]

    The Astrophysical Journal477(2), 781 (1997)

    Lin, D., Ida, S.: On the origin of massive eccentric planets. The Astrophysical Journal477(2), 781 (1997)

    1997

  48. [48]

    Icarus163(2), 290–306 (2003) https://doi.org/10.1016/ s0019-1035(03)00081-2

    Adams, F.C., Laughlin, G.: Migration and dynamical relaxation in crowded systems of giant planets. Icarus163(2), 290–306 (2003) https://doi.org/10.1016/ s0019-1035(03)00081-2

    2003

  49. [49]

    The Astrophysical Journal693(2), 113 (2009) 43

    Scharf, C., Menou, K.: Long-period exoplanets from dynamical relaxation. The Astrophysical Journal693(2), 113 (2009) 43

    2009

  50. [50]

    The Astrophysical Journal637(2), 137 (2006) https://doi.org/10.1086/500613

    Boss, A.P.: On the formation of gas giant planets on wide orbits. The Astrophysical Journal637(2), 137 (2006) https://doi.org/10.1086/500613

    work page doi:10.1086/500613 2006

  51. [51]

    J., Davies, M

    Mustill, A.J., Davies, M.B., Johansen, A.: The effects of exter- nal planets on inner systems: multiplicities, inclinations and pathways to eccentric warm jupiters. Monthly Notices of the Royal Astronomical Society468(3), 3000–3023 (2017) https://doi. org/10.1093/mnras/stx693 https://academic.oup.com/mnras/article- pdf/468/3/3000/13628841/stx693.pdf

    work page doi:10.1093/mnras/stx693 2017

  52. [52]

    Monthly Notices of the Royal Astronomical Society452(1), 376–388 (2015)

    Rein, H., Tamayo, D.: Whfast: a fast and unbiased implementation of a symplec- tic wisdom–holman integrator for long-term gravitational simulations. Monthly Notices of the Royal Astronomical Society452(1), 376–388 (2015)

    2015

  53. [53]

    Celestial Mechanics and Dynamical Astronomy74(4), 287–295 (1999)

    Mikkola, S., Tanikawa, K.: Explicit symplectic algorithms for time-transformed hamiltonians. Celestial Mechanics and Dynamical Astronomy74(4), 287–295 (1999)

    1999

  54. [54]

    The Astronomical Journal112, 1487 (1996) https://doi.org/10.1086/118116

    Harris, W.E.: A Catalog of Parameters for Globular Clusters in the Milky Way. The Astronomical Journal112, 1487 (1996) https://doi.org/10.1086/118116

    work page doi:10.1086/118116 1996

  55. [55]

    The Gravitational Million-Body Problem

    Heggie, D.C.: The Gravitational Million-Body Problem. In: Makino, J., Hut, P. (eds.) Astrophysical Supercomputing Using Particle Simulations. IAU Sympo- sium, vol. 208, p. 81 (2003). https://doi.org/10.48550/arXiv.astro-ph/0111045

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.astro-ph/0111045 2003

  56. [56]

    The Astrophysical Journal775(2), 134 (2013) https://doi.org/10.1088/ 0004-637X/775/2/134 arXiv:1308.2257 [astro-ph.GA]

    VandenBerg, D.A., Brogaard, K., Leaman, R., Casagrande, L.: The Ages of 55 Globular Clusters as Determined Using an Improved ∆V HB TO Method along with Color-Magnitude Diagram Constraints, and Their Implications for Broader Issues. The Astrophysical Journal775(2), 134 (2013) https://doi.org/10.1088/ 0004-637X/775/2/134 arXiv:1308.2257 [astro-ph.GA]

    Pith/arXiv arXiv 2013

  57. [57]

    Astronomy & Astrophysics630, 4 (2019) https: //doi.org/10.1051/0004-6361/201936135 arXiv:1906.08271 [astro-ph.GA]

    Massari, D., Koppelman, H.H., Helmi, A.: Origin of the system of globular clusters in the Milky Way. Astronomy & Astrophysics630, 4 (2019) https: //doi.org/10.1051/0004-6361/201936135 arXiv:1906.08271 [astro-ph.GA]

    work page doi:10.1051/0004-6361/201936135 2019

  58. [58]

    On the effects of dynamical evolution on the initial mass function of globular clusters

    Vesperini, E., Heggie, D.C.: On the effects of dynamical evolution on the initial mass function of globular clusters. Monthly Notices of the Royal Astronom- ical Society289(4), 898–920 (1997) https://doi.org/10.1093/mnras/289.4.898 arXiv:astro-ph/9705073 [astro-ph]

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/289.4.898 1997

  59. [59]

    doi:10.1046/j.1365-8711.2003.06897.x , eprint =

    Baumgardt, H., Makino, J.: Dynamical evolution of star clusters in tidal fields. Monthly Notices of the Royal Astronomical Society340(1), 227–246 (2003) https://doi.org/10.1046/j.1365-8711.2003.06286.x arXiv:astro- ph/0211471 [astro-ph]

    work page doi:10.1046/j.1365-8711.2003.06286.x 2003

  60. [60]

    doi:10.1111/j.1365-2966.2010.17535.x , eprint =

    Giersz, M., Heggie, D.C.: Monte Carlo simulations of star clusters - VII. The 44 globular cluster 47 Tuc. Monthly Notices of the Royal Astronomical Soci- ety410(4), 2698–2713 (2011) https://doi.org/10.1111/j.1365-2966.2010.17648.x arXiv:1008.3048 [astro-ph.GA]

    work page doi:10.1111/j.1365-2966.2010.17648.x 2011

  61. [61]

    The initial conditions of observed star clusters - I. Method description and validation

    Pijloo, J.T., Portegies Zwart, S.F., Alexander, P.E.R., Gieles, M., Larsen, S.S., Groot, P.J., Devecchi, B.: The initial conditions of observed star clusters - I. Method description and validation. Monthly Notices of the Royal Astro- nomical Society453(1), 605–637 (2015) https://doi.org/10.1093/mnras/stv1546 arXiv:1507.04372 [astro-ph.SR]

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stv1546 2015

  62. [62]

    10, held in Cambridge, England, August 12-15, 1970.)

    H´ enon, M.: Monte Carlo Models of Star Clusters (Part of the Proceedings of the IAU Colloquium No. 10, held in Cambridge, England, August 12-15, 1970.). Astrophysics and Space Science13(2), 284–299 (1971) https://doi.org/10.1007/ BF00649159

    1970

  63. [63]

    Comparison with N-body simulations

    Giersz, M., Heggie, D.C., Hurley, J.R., Hypki, A.: MOCCA code for star cluster simulations - II. Comparison with N-body simulations. Monthly Notices of the Royal Astronomical Society431(3), 2184–2199 (2013) https://doi.org/10.1093/ mnras/stt307 arXiv:1112.6246 [astro-ph.GA]

    Pith/arXiv arXiv 2013

  64. [64]

    Million-Body Star Cluster Simulations: Comparisons between Monte Carlo and Direct $N$-body

    Rodriguez, C.L., Morscher, M., Wang, L., Chatterjee, S., Rasio, F.A., Spurzem, R.: Million-body star cluster simulations: comparisons between Monte Carlo and direct N-body. Monthly Notices of the Royal Astronomical Society463(2), 2109– 2118 (2016) https://doi.org/10.1093/mnras/stw2121 arXiv:1601.04227 [astro- ph.IM]

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stw2121 2016

  65. [65]

    ApJS247(2), 61 (2020) https://doi.org/10.3847/1538-4365/ ab7ccd arXiv:1912.01621 [astro-ph.GA]

    Kremer, K., Ye, C.S., Rui, N.Z., Weatherford, N.C., Chatterjee, S., Fragione, G., Rodriguez, C.L., Spera, M., Rasio, F.A.: Modeling Dense Star Clusters in the Milky Way and Beyond with the CMC Cluster Catalog. The Astrophysical Jour- nal Supplement Series247(2), 48 (2020) https://doi.org/10.3847/1538-4365/ ab7919 arXiv:1911.00018 [astro-ph.HE]

    work page doi:10.3847/1538-4365/ 2020

  66. [66]

    The Astrophysical Journal912(2), 102 (2021) https://doi.org/10.3847/ 1538-4357/abed49 arXiv:2103.05033 [astro-ph.GA]

    Rui, N.Z., Kremer, K., Weatherford, N.C., Chatterjee, S., Rasio, F.A., Rodriguez, C.L., Ye, C.S.: Matching Globular Cluster Models to Observa- tions. The Astrophysical Journal912(2), 102 (2021) https://doi.org/10.3847/ 1538-4357/abed49 arXiv:2103.05033 [astro-ph.GA]

    arXiv 2021

  67. [67]

    the globular cluster m4

    Heggie, D.C., Giersz, M.: Monte carlo simulations of star clus- ters – v. the globular cluster m4. Monthly Notices of the Royal Astronomical Society389(4), 1858–1870 (2008) https://doi.org/10. 1111/j.1365-2966.2008.13702.x https://academic.oup.com/mnras/article- pdf/389/4/1858/3870208/mnras0389-1858.pdf

    arXiv 2008

  68. [68]

    stellar- mass black holes in the globular cluster m22

    Heggie, D.C., Giersz, M.: Mocca code for star cluster simulations–iii. stellar- mass black holes in the globular cluster m22. Monthly Notices of the Royal Astronomical Society439(3), 2459–2467 (2014) 45

    2014

  69. [69]

    science153(3731), 34–37 (1966)

    Bellman, R.: Dynamic programming. science153(3731), 34–37 (1966)

    1966

  70. [70]

    In: The Thirteenth International Conference on Learning Representations (2025)

    Frans, K., Hafner, D., Levine, S., Abbeel, P.: One step diffusion via shortcut mod- els. In: The Thirteenth International Conference on Learning Representations (2025)

    2025

  71. [71]

    Zico, K., Kaiming, H.: Mean flows for one-step generative modeling

    Zhengyang, G., Mingyang, D., Xingjian, B., J. Zico, K., Kaiming, H.: Mean flows for one-step generative modeling. In: The Thirty-Ninth Annual Conference on Neural Information Processing Systems (2025)

    2025

  72. [72]

    In: International Conference on Machine Learning, pp

    Song, Y., Dhariwal, P., Chen, M., Sutskever, I.: Consistency models. In: International Conference on Machine Learning, pp. 32211–32252 (2023). PMLR

    2023

  73. [73]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Meng, C., Rombach, R., Gao, R., Kingma, D., Ermon, S., Ho, J., Salimans, T.: On distillation of guided diffusion models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 14297–14306 (2023)

    2023

  74. [74]

    In: International Conference on Machine Learning, pp

    Rezende, D., Mohamed, S.: Variational inference with normalizing flows. In: International Conference on Machine Learning, pp. 1530–1538 (2015). PMLR

    2015

  75. [75]

    Dormand and P.J

    Dormand, J.R., Prince, P.J.: A family of embedded runge-kutta formulae. Journal of Computational and Applied Mathematics6(1), 19–26 (1980) https: //doi.org/10.1016/0771-050X(80)90013-3

    work page doi:10.1016/0771-050x(80)90013-3 1980

  76. [76]

    In: Gesammelte Abhandlungen- Collected Papers, pp

    Noether, E.: Invariante variationsprobleme. In: Gesammelte Abhandlungen- Collected Papers, pp. 231–239. Springer, ??? (1983)

    1983

  77. [77]

    Advances in neural information processing systems26(2013)

    Cuturi, M.: Sinkhorn distances: Lightspeed computation of optimal transport. Advances in neural information processing systems26(2013)

    2013

  78. [78]

    In: The 22nd International Conference on Artificial Intelligence and Statistics, pp

    Feydy, J., S´ ejourn´ e, T., Vialard, F.-X., Amari, S.-i., Trouv´ e, A., Peyr´ e, G.: Inter- polating between optimal transport and mmd using sinkhorn divergences. In: The 22nd International Conference on Artificial Intelligence and Statistics, pp. 2681–2690 (2019). PMLR

    2019

  79. [79]

    The annals of mathematical statistics35(2), 876–879 (1964)

    Sinkhorn, R.: A relationship between arbitrary positive matrices and doubly stochastic matrices. The annals of mathematical statistics35(2), 876–879 (1964)

    1964

  80. [80]

    IEEE Transactions on Information Theory55(5), 2392–2405 (2009)

    Wang, Q., Kulkarni, S.R., Verd´ u, S.: Divergence estimation for multidimensional densities viak-nearest-neighbor distances. IEEE Transactions on Information Theory55(5), 2392–2405 (2009)

    2009

Showing first 80 references.