Topology-Preserving Meshing of Implicit Scalar Fields via Monotonicity Constraints

Bei Wang; Hanqi Guo; Jixian Li; Joshua A. Levine; Tanner Finken

REVIEW 2 major objections 5 minor 35 references

Topology-Preserving Meshing of Implicit Scalar Fields via Monotonicity Constraints

T0 review · 2 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read If every edge of a triangulation is monotone along an implicit scalar field, the piecewise-linear approximation cannot invent or silently lose critical points.

desk verdict A genuinely useful meshing idea with a central theorem whose proof doesn't hold up as written—worth engaging, but the theoretical claim needs a corrected argument. read the letter →

arxiv 2608.12142 v1 pith:5XDYHA5Q submitted 2026-08-12 cs.CG

Tanner Finken , Jixian Li , Bei Wang , Hanqi Guo , Joshua A. Levine This is my paper · ORCID

classification cs.CG MSC 68U05

keywords Morse-SmalecompleximplicitscalarfieldsneuralrepresentationsmonotonicedgesDelaunayrefinementpiecewise-linearinterpolationcriticalpointsseparatrix

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

This paper establishes a sufficient condition under which a piecewise-linear (PL) triangulation of an implicit 2D scalar field faithfully reproduces the field's critical points: every mesh edge must be monotone along the field, meaning the field value changes monotonically along each edge. The paper proves that a PL mesh satisfying this condition cannot invent critical points, and any critical point it misses must be paired with another critical point inside the same triangle. It then presents a refinement algorithm that detects monotonicity violations by sampling the projected gradient along edges, splits offending edges with one-dimensional Newton steps, and further refines near separatrices, using only pointwise evaluations of the field and its derivatives. This matters because implicit neural representations and other implicit fields are increasingly used in visualization, yet standard Morse–Smale complex extraction assumes an explicit mesh; the paper gives a route to a topologically consistent explicit mesh from such fields.

What carries the argument

The load-bearing object is the monotone edge: an edge $v_i v_j$ is monotone with respect to $f$ when the restriction of $f$ to the segment between its endpoints is a monotone function of position along the segment, equivalently when the directional derivative $g'(t)$ along the edge does not change sign, so any isocontour crosses the edge at most once. Theorem 1 converts this one-dimensional property into a local counting argument inside the star of each vertex and inside single triangles: contour branches forced by a PL critical point, or by a critical point of $f$ inside a triangle, would have to cross some edge twice unless the critical points are genuine and appropriately paired. The algorithm's practical machinery is a Delaunay-refinement loop that samples the projected gradient on each edge at spacing less than $w$, uses one-dimensional Newton updates to locate sign changes of $g'(t)$, splits edges there, then refines near separatrices so PL steepest paths align with the true integral lines.

What would settle it

Take an implicit field that is monotone along an edge except for a tiny oscillation whose two sign changes of the directional derivative lie within a length shorter than $w$, run the pipeline with that $w$, and count critical points of the resulting PL mesh. If the mesh shows a spurious critical pair (or misses a true pair) with persistence above the numerical tolerance, the sampling-based monotonicity check has failed and the critical-point consistency claimed by the method does not hold for that input.

Watch

Extended reading notes

Core claim

The paper's central claim is Theorem 1: if $\hat f$ is the piecewise-linear interpolation of a Morse function $f$ on a triangulation whose every edge is monotone with respect to $f$, then every critical point of $\hat f$ is a critical point of $f$, and every critical point of $f$ either appears as a critical point of $\hat f$ or shares a triangle with at least one other critical point of $f$. The proof is a case analysis on isocontours: a monotone edge can be crossed by a given isocontour at most once, so the contour geometry forced by a PL extremum or saddle cannot be realized by a smooth Morse function unless the critical point is genuine, and an isolated critical point trapped inside a triangle would force an isocontour to cross some edge twice. The paper wraps this theoretical condition in a four-stage pipeline—density sampling, monotonicity-driven edge splitting, Newton-based critical-point repositioning, and separatrix refinement—that realizes the condition approximately with pointwise queries only, and demonstrates in experiments that the resulting meshes recover reference critical points with far fewer vertices than uniform sampling.

Load-bearing premise

The load-bearing premise is that sampling the projected gradient at spacing $w$ along each edge detects every change in the sign of the directional derivative; if a non-monotonic wiggle is narrower than $w$, the edge is declared monotone and the final mesh can violate the precondition of Theorem 1 without the algorithm knowing.

Editorial extensions

If this is right

On a monotone-edge mesh, every PL critical point is a genuine critical point of $f$, so spurious extrema and saddles do not need to be removed by persistence-based simplification.
If a true critical point is missing from the PL mesh, at least one other true critical point lies in the same triangle; topology can only be lost as a close pair, never as an isolated feature.
Topology-preserving meshing of an implicit field reduces to enforcing a local, checkable condition on edges, rather than solving a global optimization over vertex placement.
For implicit neural representations, the pipeline makes Morse–Smale complex extraction practical at modest mesh sizes: in the terrain INR experiment it matched all reference critical points with 6,359 vertices, far fewer than a 500×500 grid.

Reading between the lines

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

The theorem supplies a potential certificate: if a mesher could verify edge monotonicity with interval arithmetic or Lipschitz bounds on $g'$ instead of sampling, the critical-point guarantee would become rigorous; the paper itself relies on sampling at scale $w$.
The 'shares a triangle' bound suggests an adaptive multi-scale strategy: choose the sampling radius $R$ from a persistence or feature-scale estimate so that unresolved critical pairs are pushed below the desired scale, turning the refinement loop into a scale-space Morse–Smale complex extractor.
A natural 3D analogue would require a monotonicity condition on tetrahedral faces as well as edges, because isosurfaces can re-enter a face even when edges are monotone; the paper leaves 3D as future work.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Short version: the paper has a real new idea—use monotonic edges to guarantee critical-point consistency for PL approximations of implicit fields—and a refinement algorithm that clearly works in practice. But the proof of Theorem 1 has a genuine gap, and the experiments are too thin to verify the guarantee empirically.

What is actually new here: the monotonicity characterization itself, and the pipeline that samples gradients along edges, inserts Newton-refined split points, then refines around separatrices. The theorem is a plausible statement and the algorithm is clearly described. The code is public, which helps.

Now the soft spots, in order of importance. The proof of Theorem 1, Part 1, is not sound as written. The claim that an isocontour through an extremum vertex must be contained entirely within a single incident triangle is false: a smooth level curve can pass through the vertex from one triangle into a non-adjacent one without crossing any edge. The sharp-corner argument therefore doesn't go through. I agree with the stress-test that a correct proof needs a different argument—probably summing directional derivatives around the link to force the gradient to vanish. The theorem may still be true, but the manuscript doesn't establish it.

Second, the sampling-based monotonicity detector is heuristic. Sampling at spacing w cannot catch non-monotonic features narrower than w, so the algorithm never certifies the theorem's precondition. The paper repeatedly says "approximately monotonic," but there is no bound tying w to the feature scale of f. That's a real gap between the theory and the method.

Third, the experimental evaluation is qualitative: two datasets, no critical-point counts, no positional errors, no ablation of the monotonicity step. The INR comparison is muddied by persistence simplification on the reference grid. The code is available, but the paper itself doesn't give the reader enough to judge how often the method actually preserves topology.

These are specific issues, not a fundamental rejection. The direction is solid and the heuristic clearly helps in the shown examples. The paper deserves a serious referee, not a desk reject. The referee should ask for a corrected proof or a softened statement of Theorem 1, and a more quantitative evaluation. If the authors can close the proof gap, this becomes a solid contribution to topology-preserving meshing for implicit fields.

Referee Report

2 major / 5 minor

Summary. The paper proposes a pipeline for meshing implicitly defined 2D scalar fields so that the resulting piecewise-linear (PL) approximation is topologically consistent with the underlying field. The central theoretical claim is Theorem 1: if every mesh edge is monotonic with respect to a Morse function f, then (1) every critical point of the PL interpolant coincides with a critical point of f, and (2) any critical point of f either coincides with a PL critical point or shares a triangle with another critical point of f. The method uses Delaunay refinement, samples projected gradients along edges to detect monotonicity violations, inserts vertices via 1D Newton solves, refines critical point locations, and then adds separatrix refinement. Experiments are reported on a synthetic Griewank-plus-Gaussian function and on an implicit terrain INR, with qualitative images and vertex counts.

Significance. If Theorem 1 were correct, it would provide a simple and attractive criterion for PL/continuous critical-point consistency, and the paper's pipeline would be a useful engineering contribution for implicit fields such as INRs. The paper has clear strengths: it is well written, the code is publicly released, the method requires only pointwise evaluations of the function and its derivatives, and the intended application to implicit neural representations is timely. However, the central theorem is false as stated. Section 3.1 contains an invalid level-set containment inference, and there is an explicit counterexample showing that monotonic edges do not prevent spurious PL critical points. Consequently, the paper's main theoretical claim cannot support the proposed method, and the experimental evidence is not quantitative enough to compensate.

major comments (2)

[Section 3.1, Theorem 1(1)] The theorem is false as stated. Let f(x,y)=x, and let M be a triangulation of a small disk centered at the origin v=(0,0) whose link vertices all lie in the right half-plane x>0. On every edge of M, the restriction of f is affine, hence monotonic, so the mesh is monotonic with respect to f. The PL interpolant hat f has value 0 at v and positive values at every other vertex, so v is a strict local minimum of hat f. But ∇f(v)=(1,0)≠0, so v is not a critical point of f. This directly contradicts Part 1. The proof's assertion that the isocontour σ=f^{-1}(f(v)) 'must be contained entirely within a single triangle incident to x_m' is the exact false step: σ is the line x=0, which passes through v and continues through two (in general non-adjacent) incident triangles without crossing the interior of any edge. The differentiability argument then has nothing to apply to. If one insists on a Morse function with no critical points being vacuously Morse, this example suffices; alternatively, f(x,y)=x+y^2 with a sufficiently fine mesh gives the same local configuration with no critical point in the mesh neighborhood.
[Section 4, Monotonic Edges] The monotonicity detector does not certify the precondition of Theorem 1. The algorithm samples the projected gradient at spacing less than w and declares an edge monotonic when no sign change is observed. No bound is given relating w to the feature scale of f, such as a Lipschitz constant for the directional derivative along the edge. A non-monotonic feature narrower than w will therefore go undetected, and the final mesh may violate the monotonicity condition without the method knowing it. Since the algorithm can query f and its derivatives at arbitrary points, a rigorous certificate would require either a derivative-bounds argument or an adaptive subdivision rule with a proven stopping criterion; the paper provides neither.

minor comments (5)

[Section 5.2] The statement that the method 'successfully matches all reference critical points' is not backed by a quantitative comparison; please report critical-point counts, false positives/negatives, and positional errors with respect to the reference mesh.
[Figure 4 caption] The caption says '4567 fewer vertices (-42%)'; this should read '4,567 fewer vertices' and should specify the comparison baseline explicitly.
[Section 2] The definition of a monotonic edge should state whether constant-valued edges are allowed; the proof of Theorem 1 uses 'strictly above' for extrema, suggesting strict monotonicity may be intended, and the distinction matters for the theorem's validity.
[Section 3.1, Part 2 minimum case] The sentence 'As these contours expand, they must eventually intersect an edge' is terse; the proof should identify which edge is intersected and justify why continuing to increase the value forces a second intersection with the same edge.
[Section 4, Density Sampling] The claim that dense sampling ensures triangles 'do not contain pairs of critical points within a distance of 2R' is imprecise; it should be formulated as a geometric condition on the triangle circumradius and related explicitly to the scale of f.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the monotonic-edge theorem is an independent mathematical claim, and self-citations are background only.

full rationale

The paper's central derivation is Theorem 1, which asserts that a PL function on a mesh whose edges are monotonic with respect to a Morse function f has critical points consistent with those of f. This is not circular: the hypothesis concerns the behavior of f along mesh edges, while the conclusion concerns the critical points of the PL interpolant and of f. The two notions are not identified by definition, and the proof attempts to establish the implication geometrically. Whether that proof is fully sound is a correctness question, not a circularity question. The refinement algorithm detects non-monotonic edges by sampling the projected gradient and inserts vertices to split those edges; it does not fit parameters to the critical-point outputs and then claim those outputs as predictions. Experimental validation compares against ground-truth or uniform-mesh critical points, so it is an external check rather than a restatement of the method's inputs. The self-citations (Ma et al., Lu et al., Feng et al.) are background references or provide the INR terrain benchmark; none is load-bearing for Theorem 1, and no uniqueness theorem is imported from the authors' prior work. The sampling detector's inability to certify monotonicity is a stated limitation about meeting the theorem's precondition, not a circular reduction. Therefore no circular step is present.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The method introduces three user-set resolution parameters (R, w, r). It depends on the Morse assumption and on an ad hoc sampling-based certification of monotonicity. No new physical or mathematical entities are postulated.

free parameters (3)

R (density sampling radius) = 2.0, 1.0, 0.1, 0.05 in experiments
Controls initial sampling density; must be smaller than half the distance between critical points (stated in Section 4).
w (monotonicity sampling spacing) = 0.1, 0.01, 0.001 in experiments
Sets the sampling interval for detecting gradient orientation changes along edges; smaller w catches finer features at higher cost.
r (separatrix refinement radius) = 0.15 in Fig. 4b
Controls circumradius of triangles near PL separatrices during refinement.

assumptions (3)

domain assumption The underlying implicit field f is a Morse function (C^2 with non-degenerate critical points).
Invoked in Section 3.1 for differentiability and isolated critical points; implicit neural representations do not formally guarantee this.
standard math The piecewise-linear interpolation \hat f over the mesh has its critical points only at vertices.
Used in Theorem 1 proof; standard property of PL functions on triangulations.
ad hoc to paper Uniform sampling of projected gradients at spacing < w certifies edge monotonicity.
Assumed in Section 4 for the detection step; no rigorous link to the smallest feature scale of f.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Topology-Preserving Meshing of Implicit Scalar Fields via Monotonicity Constraints." pith.science (2026). https://pith.science/paper/5XDYHA5Q

@misc{pith2026260812142,
  author       = {Pith},
  title        = {Pith review of: Topology-Preserving Meshing of Implicit Scalar Fields via Monotonicity Constraints},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5XDYHA5Q}},
  note         = {Machine review of arXiv:2608.12142}
}

read the original abstract

Topological analysis of scalar fields yields structures such as the Morse-Smale complex (MSC) that summarize salient features across multiple scales. Existing MSC extraction algorithms typically assume an explicit representation of the input field, such as a discretely sampled mesh. However, recent advances in visualization have popularized implicit field representations, for which these assumptions no longer hold. In this work, we address the problem of extracting an MSC from an implicitly defined 2D scalar field. We present a method for constructing a triangulated piecewise-linear (PL) mesh that aims to preserve the critical points of an underlying implicit scalar field. Our central insight is that if all edges are monotonic with respect to the underlying field, then the resulting PL approximation is topologically consistent with respect to critical points. Based on this insight, we introduce a refinement procedure that mitigates monotonicity violations. Requiring only pointwise evaluations and modest mesh refinement, the approach produces PL meshes that are correct with regards to critical points in our experiments. Finally, we demonstrate that additional targeted refinement improves the geometric fidelity of MSC separatrices.

Figures

Figures reproduced from arXiv: 2608.12142 by the authors.

**Figure 3.** Pipeline overview. Our approach proceeds in four stages: [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

**Figure 4.** Griewank and Gaussian dataset (Eq. (2)) sampled via Poisson disk sampling (w = 0.1) at varying resolutions: (a) R = 2.0 (502 points), (b) R = 1.0 with separatrix refinement r = 0.15 (6,256 points), and (c) R = 0.1 (10,823 points). Extracted PL separatrices are shown in red (ascending) and blue (descending), with the assumed ground truth overlaid in white [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗

Figures from the paper (1 more)

**Figure 5.** Figure 5: Implicit Terrain INR dataset [11] reconstructed using our method. Parameters: Poisson disk sampling with R = 0.05 and w = 0.001. Critical points from our method are shown as spheres, while those from a uniform mesh (500 × 500 grid with consistent diagonals) are shown a…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 33 canonical work pages

[1]

A topological hierarchy for functions on triangulated sur- faces.IEEE Transactions on Visualization and Computer Graphics, 10(4):385–396, 2004

P-T Bremer, Bernd Hamann, Herbert Edelsbrunner, and Valerio Pas- cucci. A topological hierarchy for functions on triangulated sur- faces.IEEE Transactions on Visualization and Computer Graphics, 10(4):385–396, 2004. 1, 2, 4

work page 2004
[2]

Fast Poisson disk sampling in arbitrary dimensions

Robert Bridson. Fast Poisson disk sampling in arbitrary dimensions. SIGGRAPH sketches, 10(1):1, 2007. 3

work page 2007
[3]

Yi-Tang Chen, Neng Shi, Xihaier Luo, Wei Xu, and Han-Wei Shen. Explorable INR: an implicit neural representation for ensemble sim- ulation enabling efficient spatial and parameter exploration.IEEE Transactions on Visualization and Computer Graphics, 2025. 2

work page 2025
[4]

CRC Press Boca Raton, 2013

Siu-Wing Cheng, Tamal Krishna Dey, and Jonathan Shewchuk.De- launay mesh generation. CRC Press Boca Raton, 2013. 3

work page 2013
[5]

Guaranteed-quality triangular meshes

L Paul Chew. Guaranteed-quality triangular meshes. Technical Report CUCSDTR89983, Cornell University, 1989. 3

work page 1989
[6]

Stability of persistence diagrams

David Cohen-Steiner, Herbert Edelsbrunner, and John Harer. Stability of persistence diagrams. InProceedings of the Twenty-First Annual Symposium on Computational Geometry, SCG ’05, page 263–271, New York, NY , USA, 2005. Association for Computing Machinery. 2

work page 2005
[7]

Morse complexes for shape segmentation and homological analysis: discrete models and algorithms.Computer Graphics Forum, 34(2):761–785, 2015

Leila De Floriani, Ulderico Fugacci, Federico Iuricich, and Paola Magillo. Morse complexes for shape segmentation and homological analysis: discrete models and algorithms.Computer Graphics Forum, 34(2):761–785, 2015. 1

work page 2015
[8]

From data to functa: Your data point is a function and you can treat it like one.arXiv preprint arXiv:2201.12204, 2022

Emilien Dupont, Hyunjik Kim, SM Eslami, Danilo Rezende, and Dan Rosenbaum. From data to functa: Your data point is a function and you can treat it like one.arXiv preprint arXiv:2201.12204, 2022. 2

arXiv 2022

Show all 35 references

[9]

American Mathematical Soc., 2010

Herbert Edelsbrunner and John Harer.Computational topology: an introduction. American Mathematical Soc., 2010. 2

2010
[10]

Hierarchical Morse complexes for piecewise linear 2-manifolds

Herbert Edelsbrunner, John Harer, and Afra Zomorodian. Hierarchical Morse complexes for piecewise linear 2-manifolds. InProceedings of the seventeenth annual symposium on Computational geometry, pages 70–79, 2001. 1, 2

2001
[11]

ImplicitTerrain: a con- tinuous surface model for terrain data analysis

Haoan Feng, Xin Xu, and Leila De Floriani. ImplicitTerrain: a con- tinuous surface model for terrain data analysis. In2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 899–909, 2024. 1, 4

2024
[12]

A practical approach to Morse-Smale complex computation: Scalability and generality.IEEE Transactions on Visualization and Computer Graphics, 14(6):1619–1626, 2008

Attila Gyulassy, Peer-Timo Bremer, Bernd Hamann, and Valerio Pas- cucci. A practical approach to Morse-Smale complex computation: Scalability and generality.IEEE Transactions on Visualization and Computer Graphics, 14(6):1619–1626, 2008. 1

2008
[13]

Shared- memory parallel computation of Morse-Smale complexes with im- proved accuracy.IEEE Transactions on Visualization and Computer Graphics, 25(1), 2019

Attila Gyulassy, Peer-Timo Bremer, and Valerio Pascucci. Shared- memory parallel computation of Morse-Smale complexes with im- proved accuracy.IEEE Transactions on Visualization and Computer Graphics, 25(1), 2019. 1

2019
[14]

Jun Han and Chaoli Wang. CoordNet: Data generation and visual- ization generation for time-varying volumes via a coordinate-based neural network.IEEE Transactions on Visualization and Computer Graphics, 29(12):4951–4963, 2022. 2

2022
[15]

DCINR: A divide-and-conquer implicit neural representation for compressing time-varying volumetric data in hours

Jun Han and Fan Yang. DCINR: A divide-and-conquer implicit neural representation for compressing time-varying volumetric data in hours. IEEE Transactions on Visualization and Computer Graphics, 2025. 2

2025
[16]

Revisiting accu- rate geometry for Morse-Smale complexes

Son Le Thanh, Michael Ankele, and Tino Weinkauf. Revisiting accu- rate geometry for Morse-Smale complexes. In2024 IEEE Topologi- cal Data Analysis and Visualization (TopoInVis), pages 34–43. IEEE,
[17]

Haoyu Li and Han-Wei Shen. Improving efficiency of iso-surface ex- traction on implicit neural representations using uncertainty propaga- tion.IEEE Transactions on Visualization and Computer Graphics, 31(2):1513–1525, 2024. 2

2024
[18]

Y . Lu, K. Jiang, J. A. Levine, and M. Berger. Compressive Neural Representations of V olumetric Scalar Fields.Computer Graphics Fo- rum, 40(3):135–146, 2021. 1, 2

2021
[19]

Extracting complex topology from multivariate functional approxima- tion: Contours, Jacobi sets, and ridge-valley graphs

Guanqun Ma, David Lenz, Hanqi Guo, Tom Peterka, and Bei Wang. Extracting complex topology from multivariate functional approxima- tion: Contours, Jacobi sets, and ridge-valley graphs. In15th IEEE Symposium on Large Data Analysis and Visualization, LDAV 2025, Vienna, Austria, N...

2025
[20]

Critical Point Extraction from Multivariate Functional Approxima- tion

Guanqun Ma, David Lenz, Tom Peterka, Hanqi Guo, and Bei Wang. Critical Point Extraction from Multivariate Functional Approxima- tion . In2024 IEEE Topological Data Analysis and Visualization (TopoInVis), pages 12–22, Los Alamitos, CA, USA, October 2024. IEEE Computer Society. 1

2024
[21]

Parallel computation of piecewise linear Morse-Smale segmentations.IEEE Transactions on Visualiza- tion and Computer Graphics, 30(4):1942–1955, 2023

Robin GC Maack, Jonas Lukasczyk, Julien Tierny, Hans Hagen, Ross Maciejewski, and Christoph Garth. Parallel computation of piecewise linear Morse-Smale segmentations.IEEE Transactions on Visualiza- tion and Computer Graphics, 30(4):1942–1955, 2023. 1

1942
[22]

NeRF: Representing scenes as neural radiance fields for view synthesis.Communications of the ACM, 65(1):99–106, 2021

Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Bar- ron, Ravi Ramamoorthi, and Ren Ng. NeRF: Representing scenes as neural radiance fields for view synthesis.Communications of the ACM, 65(1):99–106, 2021. 2

2021
[23]

Implicit neural representation in medical imaging: A comparative survey

Amirali Molaei, Amirhossein Aminimehr, Armin Tavakoli, Amirhos- sein Kazerouni, Bobby Azad, Reza Azad, and Dorit Merhof. Implicit neural representation in medical imaging: A comparative survey. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 2...

2023
[24]

Foundations of multivariate functional approximation for scientific data

Tom Peterka, Youssef SG Nashed, Iulian Grindeanu, Vijay S Mahade- van, Raine Yeh, and Xavier Tricoche. Foundations of multivariate functional approximation for scientific data. In2018 IEEE 8th sympo- sium on large data analysis and visualization (LDAV), pages 61–71. IEEE, 2018. 1

2018
[25]

2D conforming triangulations and meshes

Laurent Rineau. 2D conforming triangulations and meshes. InCGAL User and Reference Manual. CGAL Editorial Board, 6.2 edition,
[26]

Vanessa Robins, Peter John Wood, and Adrian P Sheppard. The- ory and algorithms for constructing discrete Morse complexes from grayscale digital images.IEEE Transactions on pattern analysis and machine intelligence, 33(8):1646–1658, 2011. 1

2011
[27]

Parallel computation of 2D Morse-Smale complexes.IEEE Transactions on Visualization and Computer Graphics, 18(10):1757–1770, 2011

Nithin Shivashankar, M Senthilnathan, and Vijay Natarajan. Parallel computation of 2D Morse-Smale complexes.IEEE Transactions on Visualization and Computer Graphics, 18(10):1757–1770, 2011. 1

2011
[28]

Implicit neural representations with periodic activation functions.Advances in neural information processing sys- tems, 33:7462–7473, 2020

Vincent Sitzmann, Julien Martel, Alexander Bergman, David Lindell, and Gordon Wetzstein. Implicit neural representations with periodic activation functions.Advances in neural information processing sys- tems, 33:7462–7473, 2020. 1, 2

2020
[29]

ECNR: Efficient compressive neu- ral representation of time-varying volumetric datasets.arXiv preprint arXiv:2311.12831, 2023

Kaiyuan Tang and Chaoli Wang. ECNR: Efficient compressive neu- ral representation of time-varying volumetric datasets.arXiv preprint arXiv:2311.12831, 2023. 2

2023 arXiv
[30]

Fast neural representations for direct volume rendering.Computer Graph- ics Forum, 41(6):196–211, 2022

Sebastian Weiss, Philipp Herm ¨uller, and R ¨udiger Westermann. Fast neural representations for direct volume rendering.Computer Graph- ics Forum, 41(6):196–211, 2022. 2

2022
[31]

Distributed path compression for piecewise linear Morse-Smale seg- mentations and connected components

Michael Will, Jonas Lukasczyk, Julien Tierny, and Christoph Garth. Distributed path compression for piecewise linear Morse-Smale seg- mentations and connected components. In2024 IEEE 14th Sympo- sium on Large Data Analysis and Visualization (LDAV), pages 54–64. IEEE, 2024. 1

2024
[32]

Interactive volume visualization via multi-resolution hash encoding based neu- ral representation.IEEE Transactions on Visualization and Computer Graphics, 30(8):5404–5418, 2023

Qi Wu, David Bauer, Michael J Doyle, and Kwan-Liu Ma. Interactive volume visualization via multi-resolution hash encoding based neu- ral representation.IEEE Transactions on Visualization and Computer Graphics, 30(8):5404–5418, 2023. 2

2023
[33]

Distributed neural representation for re- active in situ visualization.IEEE Transactions on Visualization and Computer Graphics, 31(9):5199–5214, 2024

Qi Wu, Joseph A Insley, Victor A Mateevitsi, Silvio Rizzi, Michael E Papka, and Kwan-Liu Ma. Distributed neural representation for re- active in situ visualization.IEEE Transactions on Visualization and Computer Graphics, 31(9):5199–5214, 2024. 2

2024
[34]

Neural fields in visual computing and beyond

Yiheng Xie, Towaki Takikawa, Shunsuke Saito, Or Litany, Shiqin Yan, Numair Khan, Federico Tombari, James Tompkin, Vincent Sitzmann, and Srinath Sridhar. Neural fields in visual computing and beyond. Computer Graphics Forum, 41(2):641–676, 2022. 2

2022
[35]

Neural implicit represen- tations for multi-view surface reconstruction: A survey.IEEE Trans- actions on Visualization and Computer Graphics, 2025

Xinyun Zhang, Ruiqi Yu, and Shuang Ren. Neural implicit represen- tations for multi-view surface reconstruction: A survey.IEEE Trans- actions on Visualization and Computer Graphics, 2025. 2

2025

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

A topological hierarchy for functions on triangulated sur- faces.IEEE Transactions on Visualization and Computer Graphics, 10(4):385–396, 2004

P-T Bremer, Bernd Hamann, Herbert Edelsbrunner, and Valerio Pas- cucci. A topological hierarchy for functions on triangulated sur- faces.IEEE Transactions on Visualization and Computer Graphics, 10(4):385–396, 2004. 1, 2, 4

work page 2004

[2] [2]

Fast Poisson disk sampling in arbitrary dimensions

Robert Bridson. Fast Poisson disk sampling in arbitrary dimensions. SIGGRAPH sketches, 10(1):1, 2007. 3

work page 2007

[3] [3]

Yi-Tang Chen, Neng Shi, Xihaier Luo, Wei Xu, and Han-Wei Shen. Explorable INR: an implicit neural representation for ensemble sim- ulation enabling efficient spatial and parameter exploration.IEEE Transactions on Visualization and Computer Graphics, 2025. 2

work page 2025

[4] [4]

CRC Press Boca Raton, 2013

Siu-Wing Cheng, Tamal Krishna Dey, and Jonathan Shewchuk.De- launay mesh generation. CRC Press Boca Raton, 2013. 3

work page 2013

[5] [5]

Guaranteed-quality triangular meshes

L Paul Chew. Guaranteed-quality triangular meshes. Technical Report CUCSDTR89983, Cornell University, 1989. 3

work page 1989

[6] [6]

Stability of persistence diagrams

David Cohen-Steiner, Herbert Edelsbrunner, and John Harer. Stability of persistence diagrams. InProceedings of the Twenty-First Annual Symposium on Computational Geometry, SCG ’05, page 263–271, New York, NY , USA, 2005. Association for Computing Machinery. 2

work page 2005

[7] [7]

Morse complexes for shape segmentation and homological analysis: discrete models and algorithms.Computer Graphics Forum, 34(2):761–785, 2015

Leila De Floriani, Ulderico Fugacci, Federico Iuricich, and Paola Magillo. Morse complexes for shape segmentation and homological analysis: discrete models and algorithms.Computer Graphics Forum, 34(2):761–785, 2015. 1

work page 2015

[8] [8]

From data to functa: Your data point is a function and you can treat it like one.arXiv preprint arXiv:2201.12204, 2022

Emilien Dupont, Hyunjik Kim, SM Eslami, Danilo Rezende, and Dan Rosenbaum. From data to functa: Your data point is a function and you can treat it like one.arXiv preprint arXiv:2201.12204, 2022. 2

arXiv 2022