pith. sign in

arxiv: 2605.19244 · v1 · pith:K2ISMXKAnew · submitted 2026-05-19 · ⚛️ physics.flu-dyn · cs.NA· math.NA· physics.comp-ph

Multiresolution analysis on tessellation graphs for inertial particle dynamics

Keigo Matsuda , Thibault Maurel-Oujia , Kai Schneider This is my paper

Pith reviewed 2026-05-20 03:08 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn cs.NAmath.NAphysics.comp-ph
keywords multiresolution analysistessellation graphsinertial particleswavelet transformationparticle clusteringturbulent flowsLagrangian dynamicsDelaunay tessellation
0
0 comments X

The pith

A wavelet multiresolution technique on Delaunay tessellation graphs decomposes Lagrangian particle data into scale-dependent contributions to study clustering in turbulent flows.

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

The paper develops a method to analyze the dynamics of particles suspended in turbulent fluid at multiple spatial scales. It uses a graph defined by connecting nearby particles via Delaunay triangulation and applies a wavelet transform that respects the local volumes around each particle. This decomposition lets researchers isolate how particle velocities and clustering behave differently at small versus large scales. The approach is tested first on random particle placements and then on simulation data of inertial particles, where it successfully picks out the formation of caustics that drive strong local clustering.

Core claim

The multiresolution analysis on tessellation graphs splits spatial field data on millions of discrete particle positions into scale-dependent contributions by performing a wavelet transformation on the graph defined by the Delaunay tessellation, using Voronoi cell volumes to ensure volume conservation. This enables computation of scale-dependent statistics of particle dynamics from Lagrangian point particle data.

What carries the argument

The wavelet transformation applied to the graph constructed from Delaunay tessellation of particle positions, weighted by Voronoi cell volumes to maintain conservation.

If this is right

  • Computation of scale-dependent statistics of particle dynamics becomes feasible directly from Lagrangian data.
  • Characterization of particle clustering in turbulent flows improves by isolating contributions at different scales.
  • Verification against synthetic random particle distributions confirms the method's basic correctness.
  • Application to direct numerical simulation data of inertial particles in homogeneous isotropic turbulence yields scale-dependent velocity divergence matching Fourier results.
  • Wavelet-based filtering isolates the effect of caustics in inertial particle clustering.

Where Pith is reading between the lines

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

  • Similar graph-based wavelet methods could extend to other point-cloud datasets with irregular spacing, such as in molecular dynamics or astronomical surveys.
  • The technique might reveal how different scales interact to produce overall clustering patterns not visible in single-scale analyses.
  • Future work could test the method on experimental measurements of particle positions rather than simulations alone.

Load-bearing premise

That the Delaunay tessellation creates a graph suitable for wavelet decomposition on irregularly positioned particles and that Voronoi cell volumes provide accurate volume conservation in the transform.

What would settle it

If the energy spectrum of the particle velocity divergence extracted via this wavelet method deviates significantly from the spectrum obtained by a standard Fourier-based approach on the same turbulence simulation data.

Figures

Figures reproduced from arXiv: 2605.19244 by Kai Schneider, Keigo Matsuda, Thibault Maurel-Oujia.

Figure 1
Figure 1. Figure 1: Illustration of projection and prediction operators for 1D regular grids. [PITH_FULL_IMAGE:figures/full_fig_p010_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Point merging process on the Delaunay graph in 2D. (a) The graph before [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Multiresolution tessellation for a 2D uniform random particle distribution for [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) Probability density functions (PDFs) of volume, (b) first and (c) second [PITH_FULL_IMAGE:figures/full_fig_p018_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: PDFs of (a) volume and (b) projected particle velocity divergence for inertial [PITH_FULL_IMAGE:figures/full_fig_p020_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Wavelet and Fourier energy spectra of the particle velocity divergence [PITH_FULL_IMAGE:figures/full_fig_p022_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Joint PDF of the normalized wavenumber k ℓ i η and the logarithmically normalized energy τ 2 η Eℓ i of each wavelet coefficient for D in the HIT. The orange dash-dotted line is the conditionally averaged energy τ 2 η ⟨Eℓ i ⟩k for each wavenumber k ℓ i . wavenumbers larger than the peak wavenumber of the wavelet energy spec￾trum. The scale dependence of the skewness S ℓ is less significant compared with the… view at source ↗
Figure 8
Figure 8. Figure 8: (a) Scale-dependent flatness F ℓ and (b) skewness S ℓ of the band-pass filtered particle velocity divergence Dˇℓ for inertial particles in the HIT. ation of inertial particle trajectory from the fluid particle trajectory. In the caustic regions, particles can be adjacent to particles with significantly dif￾ferent path histories, and the divergence computation on the Delaunay graph can result in coexistence… view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of band-pass filtered particle velocity divergence [PITH_FULL_IMAGE:figures/full_fig_p025_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Spatial distributions of (a) original particle velocity divergence [PITH_FULL_IMAGE:figures/full_fig_p026_10.png] view at source ↗
read the original abstract

A multiresolution technique on tessellation graphs for particle dynamics is proposed. This allows to split spatial field data given on millions of discrete particle positions into scale-dependent contributions. The Delaunay tessellation is used to define the graph, and Voronoi cell volumes are used to satisfy volume conservation. Our approach enables computation of the scale-dependent statistics of particle dynamics by leveraging a wavelet transformation of Lagrangian point particle data and is useful for characterizing particle clustering in turbulent flows. The technique is systematically verified by using synthetic data of randomly distributed particles in a two-dimensional plane. Then the applicability of the technique is demonstrated by extracting the scale-dependent particle velocity divergence of inertial particles in homogeneous isotropic turbulence from direct numerical simulation data. The result is verified by comparing the energy spectrum of the divergence with that obtained by a Fourier-based approach. Finally, the wavelet-based filtering to the particle velocity divergence is demonstrated to extract the effect of caustics in inertial particle clustering.

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 paper proposes a multiresolution analysis technique for inertial particle dynamics that performs wavelet transformations on graphs constructed from the Delaunay tessellation of Lagrangian particle positions, using Voronoi cell volumes to enforce volume conservation in the discrete setting. The method is intended to decompose particle velocity fields into scale-dependent contributions directly from point data. Verification is performed on synthetic uniformly random particle distributions in 2D, followed by application to DNS data of inertial particles in homogeneous isotropic turbulence, where scale-dependent divergence statistics are extracted, cross-checked against Fourier spectra, and used to isolate caustic effects in clustering.

Significance. If the weighted inner product and filter construction remain conservative and orthogonal when Voronoi volumes vary by orders of magnitude, the approach would provide a useful grid-free tool for extracting scale-dependent Lagrangian statistics in turbulent flows, complementing existing Eulerian and Fourier methods for studying particle clustering.

major comments (2)
  1. [Verification on synthetic data] Verification section (synthetic random-particle data): the test employs uniformly random positions for which Voronoi volumes are nearly equal; this does not probe the strong inhomogeneity regime invoked for caustics, leaving open whether the weighted inner product ∑ f_i g_i V_i preserves exact constant reproduction and global conservation when volumes differ by orders of magnitude.
  2. [Method description] Method description (graph construction and wavelet filters): the manuscript must explicitly demonstrate that the graph-Laplacian and low-pass filters are defined with respect to the Voronoi-weighted inner product so that the decomposition remains orthogonal and energy is conserved across scales; without this, leakage cannot be excluded under the clustering conditions central to the application.
minor comments (2)
  1. [DNS application and Fourier comparison] The comparison of divergence energy spectra with the Fourier approach should include quantitative error metrics and a brief discussion of any shared interpolation artifacts.
  2. [Wavelet transform formulation] Clarify the precise definition of the discrete wavelet coefficients and the reconstruction formula used to obtain scale-dependent divergence fields.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address the major comments point by point below and will revise the manuscript accordingly to strengthen the presentation.

read point-by-point responses

  1. Referee: [Verification on synthetic data] Verification section (synthetic random-particle data): the test employs uniformly random positions for which Voronoi volumes are nearly equal; this does not probe the strong inhomogeneity regime invoked for caustics, leaving open whether the weighted inner product ∑ f_i g_i V_i preserves exact constant reproduction and global conservation when volumes differ by orders of magnitude.

    Authors: We acknowledge that the synthetic test case employs uniformly random particle positions, for which Voronoi volumes exhibit only small variations. This was intended as a baseline verification to confirm the basic functionality of the multiresolution analysis on tessellation graphs. The weighted inner product is constructed to enforce volume conservation by design for arbitrary positive volume weights, and constant reproduction holds exactly in the discrete setting regardless of volume variation. Nevertheless, to directly address the referee's concern regarding strong inhomogeneity, we will include an additional verification test in the revised manuscript using particle distributions with artificial clustering that induces volume variations spanning orders of magnitude. In this test, we will explicitly verify constant reproduction and global conservation properties. revision: yes

  2. Referee: [Method description] Method description (graph construction and wavelet filters): the manuscript must explicitly demonstrate that the graph-Laplacian and low-pass filters are defined with respect to the Voronoi-weighted inner product so that the decomposition remains orthogonal and energy is conserved across scales; without this, leakage cannot be excluded under the clustering conditions central to the application.

    Authors: We agree that an explicit demonstration is necessary to confirm orthogonality and energy conservation under the weighted inner product, particularly for the application to clustered particle data. In the revised version of the manuscript, we will expand the method section to clearly define the graph-Laplacian and low-pass filters using the Voronoi-weighted inner product. We will also add a proof or detailed derivation showing that the wavelet decomposition is orthogonal and conserves energy across scales. This will be supported by numerical checks in the verification section to rule out leakage in regimes with large volume variations. revision: yes

Circularity Check

0 steps flagged

No circularity: method and validations are independent of fitted inputs or self-citation chains

full rationale

The paper defines a graph wavelet transform from Delaunay tessellation plus Voronoi volumes, then applies it to synthetic uniform-random particles and to DNS inertial-particle data, cross-checking the divergence spectrum against a separate Fourier method. These checks use external data and an independent transform; no equation reduces a claimed prediction to a fitted parameter or to a prior result by the same authors. The construction is presented as a new tool whose correctness is tested rather than assumed by definition.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The approach rests on standard assumptions from computational geometry and wavelet theory applied to point-cloud data; no free parameters or new entities are explicitly introduced in the abstract.

axioms (2)
  • domain assumption Delaunay tessellation produces a valid connectivity graph for arbitrary particle positions that supports consistent multiresolution decomposition.
    Invoked when defining the graph for the wavelet transform.
  • domain assumption Voronoi cell volumes assigned to particles ensure exact volume conservation in the discrete field representation.
    Stated as the mechanism to satisfy volume conservation.

pith-pipeline@v0.9.0 · 5704 in / 1456 out tokens · 48709 ms · 2026-05-20T03:08:16.234013+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

47 extracted references · 47 canonical work pages · 1 internal anchor

  1. [1]

    Maxey, The gravitational settling of aerosol particles in homogeneous turbulence and random flow fields, J

    M. Maxey, The gravitational settling of aerosol particles in homogeneous turbulence and random flow fields, J. Fluid Mech. 174 (1987) 441–465

    work page 1987

  2. [2]

    K. D. Squires, J. K. Eaton, Measurements of particle dispersion obtained from direct numerical simulations of isotropic turbulence, J. Fluid Mech. 226 (1991) 1–35

    work page 1991

  3. [3]

    Gustavsson, B

    K. Gustavsson, B. Mehlig, Ergodic and non-ergodic clustering of inertial particles, Europhysics Letters 96 (6) (2011) 60012. 32

    work page 2011

  4. [4]

    A. D. Bragg, L. R. Collins, New insights from comparing statistical theories for inertial particles in turbulence: I. spatial distribution of particles, New J. Physics 16 (2014) 055013

    work page 2014

  5. [5]

    A. D. Bragg, P. J. Ireland, L. R. Collins, On the relationship between the non-local clustering mechanism and preferential concentration, J. Fluid Mech. 780 (2015) 327–343

    work page 2015

  6. [6]

    Brandt, F

    L. Brandt, F. Coletti, Particle-laden turbulence: progress and perspec- tives, Annual Review of Fluid Mechanics 54 (1) (2022) 159–189

    work page 2022

  7. [7]

    J. Bec, K. Gustavsson, B. Mehlig, Statistical models for the dynamics of heavy particles in turbulence, Annual Review of Fluid Mechanics 56 (1) (2024) 189–213

    work page 2024

  8. [8]

    Marchioli, M

    C. Marchioli, M. Bourgoin, F. Coletti, R. Fox, J. Magnaudet, M. Reeks, O. Simonin, M. Sommerfeld, F. Toschi, L.-P. Wang, et al., Particle-laden flows, International Journal of Multiphase Flow (2025) 105291

    work page 2025

  9. [9]

    Matsuda, K

    K. Matsuda, K. Schneider, K. Yoshimatsu, Scale-dependent statistics of inertial particle distribution in high Reynolds number turbulence, Phys. Rev. Fluids 6 (2021) 064304

    work page 2021

  10. [10]

    Matsuda, K

    K. Matsuda, K. Yoshimatsu, K. Schneider, Heavy particle clustering in inertial subrange of high–Reynolds number turbulence, Physical Review Letters 132 (23) (2024) 234001

    work page 2024

  11. [11]

    Monchaux, M

    R. Monchaux, M. Bourgoin, A. Cartellier, Preferential concentration of heavy particles: a Vorono¨ ı analysis, Physics of Fluids 22 (10) (2010)

    work page 2010

  12. [12]

    Ebeling, G

    H. Ebeling, G. Wiedenmann, Detecting structure in two dimensions combining Vorono¨ ı tessellation and percolation, Physical Review E 47 (1) (1993) 704

    work page 1993

  13. [13]

    Obligado, T

    M. Obligado, T. Teitelbaum, A. Cartellier, P. Mininni, M. Bourgoin, Preferential concentration of heavy particles in turbulence, Journal of Turbulence 15 (5) (2014) 293–310

    work page 2014

  14. [14]

    Monchaux, M

    R. Monchaux, M. Bourgoin, A. Cartellier, Analyzing preferential con- centration and clustering of inertial particles in turbulence, International Journal of Multiphase Flow 40 (2012) 1–18. 33

    work page 2012

  15. [15]

    Oujia, K

    T. Oujia, K. Matsuda, K. Schneider, Divergence and convergence of inertial particles in high-Reynolds-number turbulence, J. Fluid Mech. 905 (2020) A14

    work page 2020

  16. [16]

    D. I. Shuman, S. K. Narang, P. Frossard, A. Ortega, P. Vandergheynst, The emerging field of signal processing on graphs, IEEE Signal Proc. Mag. 30 (2013) 83–98

    work page 2013

  17. [17]

    Mallat, A Wavelet Tour of Signal Processing, Elsevier, 1999

    S. Mallat, A Wavelet Tour of Signal Processing, Elsevier, 1999

    work page 1999

  18. [18]

    Cohen, N

    A. Cohen, N. Dyn, S. M. Kaber, M. Postel, Multiresolution schemes on triangles for scalar conservation laws, J. Comput. Phys. 161 (2000) 264–286

    work page 2000

  19. [19]

    H. J. Yu, J. B. Ra, Fast triangular mesh approximation of surface data using wavelet coefficients, Vis. Comput. 15 (1999) 9–20

    work page 1999

  20. [20]

    R. R. Coifman, M. Maggioni, Diffusion wavelets, Applied and Compu- tational Harmonic Analysis 21 (1) (2006) 53–94

    work page 2006

  21. [21]

    D. K. Hammond, P. Vandergheynst, R. Gribonval, Wavelets on graphs via spectral graph theory, Applied and Computational Harmonic Anal- ysis 30 (2) (2011) 129–150

    work page 2011

  22. [22]

    Ortega, P

    A. Ortega, P. Frossard, J. Kovaˇ cevi´ c, J. M. Moura, P. Vandergheynst, Graph signal processing: Overview, challenges, and applications, Pro- ceedings of the IEEE 106 (5) (2018) 808–828

    work page 2018

  23. [23]

    S. Chen, A. Singh, J. Kovaˇ cevi´ c, Multiresolution representations for piecewise-smooth signals on graphs, arXiv preprint arXiv:1803.02944 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  24. [24]

    Avena, F

    L. Avena, F. Castell, A. Gaudilli` ere, C. M´ elot, Intertwining wavelets or multiresolution analysis on graphs through random forests, Appl. Comput. Harmon. Anal. 48 (2020) 949–992

    work page 2020

  25. [25]

    Matsuda, K

    K. Matsuda, K. Schneider, T. Oujia, J. West, S. Jain, K. Maeda, Mul- tiresolution analysis of inertial particle tessellations for clustering dy- namics, in: Proceedings of the Summer Program 2022, Center for Tur- bulence Research, Stanford University, 2022, pp. 143–152. 34

    work page 2022

  26. [26]

    Matsuda, R

    K. Matsuda, R. Onishi, M. Hirahara, R. Kurose, K. Takahashi, S. Ko- mori, Influence of microscale turbulent droplet clustering on radar cloud observations, J. Atmos. Sci. 71 (2014) 3569–3582

    work page 2014

  27. [27]

    Aurenhammer, Voronoi diagrams—a survey of a fundamental geo- metric data structure, ACM Computing Surveys (CSUR) 23 (3) (1991) 345–405

    F. Aurenhammer, Voronoi diagrams—a survey of a fundamental geo- metric data structure, ACM Computing Surveys (CSUR) 23 (3) (1991) 345–405

    work page 1991

  28. [28]

    Maurel-Oujia, K

    T. Maurel-Oujia, K. Matsuda, K. Schneider, Computing differential op- erators of the particle velocity in moving particle clouds using tessella- tions, J. Comput. Phys. 498 (2024) 112658

    work page 2024

  29. [29]

    S. V. Apte, T. Oujia, K. Matsuda, B. Kadoch, X. He, K. Schneider, Clustering of inertial particles in turbulent flow through a porous unit cell, J. Fluid Mech. 937 (2022) A9

    work page 2022

  30. [30]

    J. West, T. Oujia, K. Matsuda, K. Schneider, S. Jain, K. Maeda, Di- vergence and curl of the inertial particle velocity in a four-way coupled channel flow, in: Proceedings of the Summer Program 2022, Center for Turbulence Research, Stanford University, 2022, pp. 163–172

    work page 2022

  31. [31]

    J. West, T. Maurel-Oujia, K. Matsuda, K. Schneider, S. Jain, K. Maeda, Clustering, rotation, and swirl of inertial particles in turbulent channel flow, Int. J. Multiphase Flow (2024) 104764

    work page 2024

  32. [32]

    Harten, Discrete multi-resolution analysis and generalized wavelets, Applied Numerical Mathematics 12 (1-3) (1993) 153–192

    A. Harten, Discrete multi-resolution analysis and generalized wavelets, Applied Numerical Mathematics 12 (1-3) (1993) 153–192

    work page 1993

  33. [33]

    Harten, Multiresolution algorithms for the numerical solution of hy- perbolic conservation laws, Communications on Pure and Applied Math- ematics 48 (12) (1995) 1305–1342

    A. Harten, Multiresolution algorithms for the numerical solution of hy- perbolic conservation laws, Communications on Pure and Applied Math- ematics 48 (12) (1995) 1305–1342

    work page 1995

  34. [34]

    Harten, Multiresolution representation of data: A general framework, SIAM Journal on Numerical Analysis 33 (3) (1996) 1205–1256

    A. Harten, Multiresolution representation of data: A general framework, SIAM Journal on Numerical Analysis 33 (3) (1996) 1205–1256

    work page 1996

  35. [35]

    Kobbelt, S

    L. Kobbelt, S. Campagna, J. Vorsatz, H.-P. Seidel, Interactive multi- resolution modeling on arbitrary meshes, in: Proceedings of the 25th Annual Conference on Computer Graphics and Interactive Techniques, 1998, pp. 105–114. 35

    work page 1998

  36. [36]

    I. S. Dhillon, Y. Guan, B. Kulis, Weighted graph cuts without eigen- vectors: A multilevel approach, IEEE Transactions on Pattern Analysis and Machine Intelligence 29 (11) (2007) 1944–1957

    work page 2007

  37. [37]

    Cohen, I

    A. Cohen, I. Daubechies, J.-C. Feauveau, Biorthogonal bases of com- pactly supported wavelets, Communications on pure and applied math- ematics 45 (5) (1992) 485–560

    work page 1992

  38. [38]

    Farge, K

    M. Farge, K. Schneider, Wavelet transforms and their applications to mhd and plasma turbulence: a review, Journal of Plasma Physics 81 (6) (2015) 435810602

    work page 2015

  39. [39]

    Farge, Wavelet transforms and their applications to turbulence, An- nual Review of Fluid Mechanics 24 (1) (1992) 395–458

    M. Farge, Wavelet transforms and their applications to turbulence, An- nual Review of Fluid Mechanics 24 (1) (1992) 395–458

    work page 1992

  40. [40]

    Wilkinson, B

    M. Wilkinson, B. Mehlig, Caustics in turbulent aerosols, Europhysics Letters 71 (2) (2005) 186

    work page 2005

  41. [41]

    Meibohm, K

    J. Meibohm, K. Gustavsson, B. Mehlig, Caustics in turbulent aerosols form along the Vieillefosse line at weak particle inertia, Physical Review Fluids 8 (2023) 024305

    work page 2023

  42. [42]

    Meibohm, L

    J. Meibohm, L. Sundberg, B. Mehlig, K. Gustavsson, Caustic formation in a non-Gaussian model for turbulent aerosols, Physical Review Fluids 9 (2024) 024302

    work page 2024

  43. [43]

    Bassenne, J

    M. Bassenne, J. Urzay, K. Schneider, P. Moin, Extraction of coherent clusters and grid adaptation in particle-laden turbulence using wavelet filters, Phys. Rev. Fluids 2 (2017) 054301

    work page 2017

  44. [44]

    Maurel-Oujia, K

    T. Maurel-Oujia, K. Matsuda, K. Schneider, Multiscale dynamics of in- ertial particles in turbulence with and without the effect of gravitational settling, Preprint, submitted (2025)

    work page 2025

  45. [45]

    Sweldens, Wavelets and the lifting scheme: A 5 minute tour, ZAMM- Zeitschrift fur Angewandte Mathematik und Mechanik 76 (2) (1996) 41–44

    W. Sweldens, Wavelets and the lifting scheme: A 5 minute tour, ZAMM- Zeitschrift fur Angewandte Mathematik und Mechanik 76 (2) (1996) 41–44

    work page 1996

  46. [46]

    Sweldens, The lifting scheme: A construction of second generation wavelets, SIAM Journal on Mathematical Analysis 29 (2) (1998) 511– 546

    W. Sweldens, The lifting scheme: A construction of second generation wavelets, SIAM Journal on Mathematical Analysis 29 (2) (1998) 511– 546. 36

    work page 1998

  47. [47]

    Deiterding, M

    R. Deiterding, M. O. Domingues, K. Schneider, Multiresolution analysis as a criterion for effective dynamic mesh adaptation–a case study for Euler equations in the SAMR framework AMROC, Computers & Fluids 205 (2020) 104583. 37

    work page 2020