The Weighted Gaussian Curvature Derivative of a Space-Filling Diagram

Arseniy Akopyan; Herbert Edelsbrunner

REVIEW 2 major objections 3 minor 15 references

The Weighted Gaussian Curvature Derivative of a Space-Filling Diagram

T0 review · 2 major / 3 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper derives an explicit formula for the derivative of the weighted Gaussian curvature of a space-filling diagram, completing the gradient of the morphometric solvation free energy.

desk verdict The paper's central gradient formula is invalid: a sign error in Eq. (21) propagates into Theorem 5, so the main result as stated does not hold. read the letter →

arxiv 1908.06777 v2 pith:3PNVEOAS submitted 2019-08-19 cs.CG physics.bio-ph

Arseniy Akopyan , Herbert Edelsbrunner This is my paper · ORCID

classification cs.CGphysics.bio-ph MSC 52A3868U05

keywords moleculardynamicsproteinsspace-fillingdiagramsintrinsicvolumealphashapesinclusion-exclusionderivativesdiscontinuities

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 completes the derivative formulas for the four weighted intrinsic volumes of a space-filling diagram, adding the weighted Gaussian curvature to the already known volume, area, and mean curvature derivatives. The morphometric approach writes solvation free energy as a linear combination of these four weighted volumes, so an explicit Gaussian-curvature gradient is the last missing piece for computing solvation forces directly from atom geometry in molecular dynamics simulations. The main result, Theorem 5, expresses the derivative at a state $x$ with momentum $t$ as an inner product $\langle g, t\rangle$, with each atom receiving a gradient component built from four contributions coming from sphere patches, circular arcs, and corners. The formula is explicit enough for computer implementation and is accompanied by an analysis of where the gradient fails to be continuous.

What carries the argument

The machinery is the decomposition of the weighted Gaussian curvature into three sums, over sphere patches, circular arcs, and corners, with weights $w_i$ assigned to atoms, plus the corner-splitting rule that assigns each corner's curvature to its three atoms by the areas of spherical quadrangles. The derivative computation reduces every term to three ingredients: the derivative of $\lambda_{ij}$, the combined projected normal length, in Lemma 2; the derivative of $\varphi_{i,jk}$, the spherical-quadrangle area, in Lemma 4; and previously derived derivatives of the patch and arc fractions from [1]. The spherical-quadrangle derivative is made explicit by the substitution $a=\cos^2(\varphi_{ij}/2)$, $b=\cos^2(\varphi_{jk}/2)$, $c=\cos^2(\varphi_{ki}/2)$, under which the spherical triangle area becomes $S(a,b,c)=2\arcsin\sqrt{(4abc-(a+b+c-1)^2)/(4abc)}$ and the radius of its circumcap has a closed form; these identities carry the geometric complexity and turn the final gradient into finite sums over the $\alpha$-complex boundary.

What would settle it

On a generic configuration of three unequal overlapping spheres, move one center by a small step, evaluate Eq. (5) before and after, and compare the central difference with the inner product of the Theorem 5 gradient against the step vector; many random trials would confirm the formula, and a sustained mismatch would refute it.

Watch

Extended reading notes

Core claim

The paper's main result is Theorem 5: for a space-filling diagram of $n$ balls at state $x$ moving with momentum $t$, the derivative of the weighted Gaussian curvature is $D\mathrm{gauss}_x(t)=\langle g,t\rangle$, where the per-atom gradient component is $g_i=d_i+e_i+f_i+h_i$, with the four pieces given in Eqs. (47), (58), (61), and (66). These four pieces come from the sphere-patch term, the arc-length term, the projected-normal-length term, and the corner term of the curvature function. The corner contribution is split among the three atoms meeting there by coefficients that are areas of spherical quadrangles, a split consistent with the equal-halves rule used for arc curvature in the companion mean-curvature paper. The paper further establishes that the gradient is continuous on the complement of a $(3n-1)$-dimensional set of non-generic states and is undefined or discontinuous exactly at the surface-topology events in that set.

Load-bearing premise

The corner curvature split is chosen geometrically, not forced by physics; if the physical model assigns each corner differently, the gradient formula would not match the solvation free energy.

Editorial extensions

If this is right

The four weighted-intrinsic-volume derivatives now form a complete set, so the gradient of any morphometric solvation free energy built from volume, area, mean curvature, and Gaussian curvature is explicitly computable.
The gradient remains continuous through Delaunay flips; only surface topological events in the alpha complex can make it discontinuous, so molecular dynamics integrators face a known, finite set of singular events.
For equal weights, the weighted gradient specializes to the classical unweighted behavior: derivative zero almost everywhere and undefined at topology changes.
The per-atom formula is a sum over incident boundary edges and triangles of the alpha shape, so it can be evaluated with standard alpha-complex data structures.

Reading between the lines

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

The analytic gradient could be verified numerically on random non-degenerate ball configurations by comparing $\langle g,t\rangle$ with central finite differences of Eq. (5); the formulas in the paper make this check straightforward.
Because the corner-split coefficients are the only part of the construction not forced by a physical principle, a morphometric model that assigns corner curvature differently would change per-atom forces while leaving total curvature unchanged; the formula's physical applicability inherits the split's validity.
The same spherical-trigonometry substitution used for $\varphi_{i,jk}$ is self-contained enough to be reused for derivatives of curvature terms in unions of other quadric surfaces or for higher-order intersection features.

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

The paper's central result is wrong as stated. Equation (21) misstates dS/da. I worked through the chain rule in (18)–(20) and the stress-test is correct: the product simplifies to (b+c−a−1)/(a√(4abc−(a+b+c−1)^2)), not −(a+b+c−1)/(...). A numeric check with the equilateral triangle confirms: finite differences of (13) give about −0.943, while (21) gives about −4.714. This error flows into the chain-rule expressions (30)–(32), into Lemma 4's coefficients, and from there into h′ in (42) and the gradient formula in Theorem 5. The main theorem as written does not follow from the paper's own derivation.

What the paper does well: the framing is clear — the weighted Gaussian curvature is the last missing piece for the morphometric solvation gradient — and the decomposition into patch, arc, and corner contributions is sensible. The corner-splitting via spherical quadrangles is explicit and geometrically motivated, and the authors are upfront that it is a convention. The overall four-term structure (d, e, f, h) follows the pattern of the earlier companion papers, and if (21) is fixed, the same proof strategy should go through.

Soft spots beyond the algebraic error: the proof of Formula 3 in the appendix has a hand-wavy 'canceling the terms' step, though the identity checks out. Theorem 6 defers to the companion paper [1], which is acceptable but leaves the paper not fully self-contained for the continuity part. The corner-splitting convention is a modeling choice, not forced by the physics; the authors do not justify it from the morphometric model. That is a legitimate concern but secondary.

I disagree with the reader's soundness score of 7. This is a load-bearing flaw, not a typo. Still, it is the kind of error a careful referee can catch, and the fix is likely local. I think the paper deserves peer review — not because it is nearly acceptable, but because the approach is valuable and the error is identifiable and correctable. A serious referee should verify the corrected formula and recompute the affected coefficients. Unless you specifically want to discuss spherical triangle derivative geometry, I'd wait for a corrected version rather than read the current one.

Referee Report

2 major / 3 minor

Summary. The paper defines the weighted Gaussian curvature of a space-filling diagram and derives an explicit gradient formula, Theorem 5, expressed as the sum of four vector fields corresponding to patch, arc, and corner contributions. The derivation computes the derivative of the projected normal length λ_ij and of the spherical-quadrangle areas φ_i,jk, then assembles the gradient using redistribution over the alpha complex. The stated aim is to complete the set of derivative formulas needed for the morphometric approach to solvation free energy.

Significance. If the formulas were correct, the paper would provide a useful, explicit computational component for molecular dynamics and would complete the derivative program begun in the companion papers [1,4,7]. The geometric setup is clear, the decomposition is principled, and the reliance on companion papers is declared. However, the central derivative computation contains a concrete algebraic error in Eq. (21), and the area formula in Eq. (13) has a branch ambiguity that is not addressed. The final gradient theorem is therefore not established as written, although the overall approach remains plausible and correctable.

major comments (2)

[Section 4.2, Eq. (21)] The printed derivative dS/da is algebraically incorrect. Multiplying the factors in (18)–(20) gives dS/da = sign(s)(b+c-a-1)/(a sqrt(4abc-s^2)) with s=a+b+c-1, not -s/(a sqrt(4abc-s^2)); in addition, Eq. (18) should contain |s| in the denominator. For a=b=c=3/4, the correct value is -0.943, while Eq. (21) predicts -4.714, and finite differences of Eq. (13) confirm the former. Because Eqs. (30)–(32), Lemma 4, and Theorem 5 all build on dS/da, the final gradient formula is not established as stated.
[Section 4.2, Eq. (13)] The function S=2 arcsin(...) only produces values in [0,π] and is therefore not the area of an arbitrary spherical triangle. For a valid corner configuration with three unit outward normals at mutual angle about 118°, realized by unit balls centered at -n_i with P=0 on the boundary of their union, the spherical triangle has area about 276°, while Eq. (13) gives about 83°. The paper must either prove that such cases cannot occur for corners of space-filling diagrams or replace Eq. (13) with a branch-correct area formula; the derivative and the gradient depend on this choice.

minor comments (3)

[Section 3, corner splitting] The weighted Gaussian curvature in Eq. (5) depends on the corner-splitting rule α_i defined in Section 3; if the morphometric free energy requires a different assignment of corner curvature to atoms, the gradient of the solvation free energy would differ. The theorem should be understood as the derivative of the function defined here, contingent on this modeling choice.
[Appendix A, proof of Formula 3] The final sentence of the proof says 'canceling the terms of degree 4, 5, 6 and more' without displaying the cancellation; please expand this computation for the reader or provide it as supplementary material.
[Eq. (60)] Equation (60) appears to have a misbalanced parenthesis or bracket in the displayed formula; please fix the typesetting.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the gradient is derived by direct differentiation of an explicitly defined function, with self-citations used only as auxiliary technical lemmas.

full rationale

Score 0. No circular step is exhibited. The paper defines a weighted Gaussian curvature function in Eq. (5) with an explicit corner-splitting rule, then differentiates that function directly using the chain rule. The final gradient in Theorem 5 is the rearrangement of those derivatives, so it is not an input to its own derivation. Citations to the authors' prior papers [1,4,7] supply technical formulas for derivatives of simpler fractional measures (sigma_i', sigma_ij', dphi_ij/d||xi-xj||), which are parameter-free identities independent of the target Gaussian-curvature derivative; they are auxiliary lemmas, not the conclusion. The corner-splitting convention from [1] is a stated modeling definition rather than a hidden ansatz, and the derivative is explicitly the derivative of the function so defined. The possible algebraic error in Eq. (21) flagged in review would be a correctness defect, not a circular reduction: a wrong chain-rule simplification does not make the derivation equivalent to its inputs. The stray heading 'Solvation free-energy estimated from macroscopic continuum theory:' before Appendix A has no bearing on circularity. Thus no claim reduces by construction to a fit, a definition, or a self-citation.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The central claim (the gradient formula) rests on the definition of weighted intrinsic volumes from the morphometric approach, the corner-splitting convention, general position, and the prior derivatives of sigma_i and sigma_ij from [1]. No free parameters are fitted, and no new physical entities are introduced.

assumptions (6)

domain assumption The weighted intrinsic volume formulas in Proposition 1 (Eqs. 2-5) define vol, area, mean, gauss; these are taken from the morphometric approach literature.
The paper's derivative is of these functions; the formulas are stated without derivation here.
domain assumption The corner-splitting convention in Section 3 (alpha_i as spherical quadrangle areas, with alpha_i + alpha_j + alpha_k = 1) determines the weighted Gaussian curvature function.
The h term of the gradient depends directly on this convention; it is a modeling choice inherited from [1].
domain assumption The state x of the space-filling diagram is in general position (Conditions I and II in Section 6) when the derivative formula is applied.
The gradient formula holds for generic states; non-generic states are analyzed separately in Section 6.
domain assumption The derivatives of sigma_i and sigma_ij, and the angle parametrization alpha_P, from the companion paper [1] are correct.
The d and e terms of the gradient rely directly on these cited results; they are not rederived here.
standard math Standard spherical trigonometry identities, including the area formula (11) and radius formula (22) from Chauvenet [5], hold.
Used to define S and R in Section 4.2.
standard math Gauss-Bonnet theorem (Eq. 1) as background to the unweighted case.
Motivates why the weighted case is different; not directly used in the derivative algebra.

how reviews work

0 comments

Cite this review

Pith. "Pith review of The Weighted Gaussian Curvature Derivative of a Space-Filling Diagram." pith.science (2026). https://pith.science/paper/3PNVEOAS

@misc{pith2026190806777,
  author       = {Pith},
  title        = {Pith review of: The Weighted Gaussian Curvature Derivative of a Space-Filling Diagram},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3PNVEOAS}},
  note         = {Machine review of arXiv:1908.06777}
}

read the original abstract

The morphometric approach [HRC13,RHK06] writes the solvation free energy as a linear combination of weighted versions of the volume, area, mean curvature, and Gaussian curvature of the space-filling diagram. We give a formula for the derivative of the weighted Gaussian curvature. Together with the derivatives of the weighted volume in [EdKo03], the weighted area in [BEKL04], and the weighted mean curvature in [AkEd19], this yields the derivative of the morphometric expression of solvation free energy.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 4 canonical work pages

[1]

The weighted mean curvature derivative of a space-filling diagram

A.\ Akopyan and H.\ Edelsbrunner. The weighted mean curvature derivative of a space-filling diagram. Manuscript, IST Austria, Klosterneuburg, Austria, 2019

work page 2019
[2]

The Morse theory of Cech and Delaunay complexes

U.\ Bauer and H.\ Edelsbrunner. The Morse theory of Cech and Delaunay complexes. Trans.\ Amer.\ Math.\ Soc. 369 (2017), 3741--3762

2017
[3]

M\' e moire sur la th\' e orie g\' e n\' e rale des surfaces

P.O.\ Bonnet. M\' e moire sur la th\' e orie g\' e n\' e rale des surfaces. J.\ de l'\' E cole Polytechnique 32 (1848), 1--46

work page
[4]

The area derivative of a space-filling diagram

R.\ Bryant, H.\ Edelsbrunner, P.\ Koehl and M.\ Levitt. The area derivative of a space-filling diagram. Discrete Comput.\ Geom. 32 (2004), 293--308

2004
[5]

Treatise on Plane and Spherical Trigonometry

W.\ Chauvenet. Treatise on Plane and Spherical Trigonometry. Ninth edition, Lippincott Company, Philadelphia, Pennsylvania, 1887

work page
[6]

Geometry and Topology for Mesh Generation

H.\ Edelsbrunner. Geometry and Topology for Mesh Generation. Cambridge Univ.\ Press, Cambridge, England, 2001

2001
[7]

The weighted-volume derivative of a space-filling diagram

H.\ Edelsbrunner and P.\ Koehl. The weighted-volume derivative of a space-filling diagram. Proc.\ Natl.\ Acad.\ Sci. 100 (2003), 2203--2208

2003
[8]

Three-dimensional alpha shapes

H.\ Edelsbrunner and E.P.\ M\" u cke. Three-dimensional alpha shapes. ACM Trans.\ Graphics 13 (1994), 43--72

1994

Show all 15 references

[9]

u r konvexe K\

H.\ Hadwiger. Beweis eines Funktionalsatzes f\" u r konvexe K\" o rper. Abh.\ Math.\ Sem.\ Univ.\ Hamburg 17 (1951), 11--23

1951
[10]

Solvation of proteins: linking thermodynamics to geometry

H.\ Hansen-Goos, R.\ Roth, K.R.\ Mecke and S.\ Dietrich. Solvation of proteins: linking thermodynamics to geometry. Phys.\ Rev.\ Lett. 99 (2007), 128101:1--4

2007
[11]

A morphometric approach for the accurate solvation thermodynamics of proteins and ligands

Y.\ Harano, R.\ Roth and S.\ Chiba. A morphometric approach for the accurate solvation thermodynamics of proteins and ligands. J.\ Comput.\ Chem. 34 (2013), 1969--1974

2013
[12]

Morphological thermodynamics of fluids: shape dependence of free energies

P.-M.\ K\" o nig, R.\ Roth and K.R.\ Mecke. Morphological thermodynamics of fluids: shape dependence of free energies. Phys.\ Rev.\ Lett. 93 (2004), 160601:1--4

2004
[13]

A morphological model for complex fluids

K.R.\ Mecke. A morphological model for complex fluids. J.\ Phys.: Condens.\ Matter 8 (1996), 9663--9667

1996
[14]

Morphometric approach to the solvation free energy of complex molecules

R.\ Roth, Y.\ Harano and M.\ Kinoshita. Morphometric approach to the solvation free energy of complex molecules. Phys.\ Rev.\ Lett. 97 (2006), 078101

2006
[15]

Solvation free-energy estimated from macroscopic continuum theory:

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

The weighted mean curvature derivative of a space-filling diagram

A.\ Akopyan and H.\ Edelsbrunner. The weighted mean curvature derivative of a space-filling diagram. Manuscript, IST Austria, Klosterneuburg, Austria, 2019

work page 2019

[2] [2]

The Morse theory of Cech and Delaunay complexes

U.\ Bauer and H.\ Edelsbrunner. The Morse theory of Cech and Delaunay complexes. Trans.\ Amer.\ Math.\ Soc. 369 (2017), 3741--3762

2017

[3] [3]

M\' e moire sur la th\' e orie g\' e n\' e rale des surfaces

P.O.\ Bonnet. M\' e moire sur la th\' e orie g\' e n\' e rale des surfaces. J.\ de l'\' E cole Polytechnique 32 (1848), 1--46

work page

[4] [4]

The area derivative of a space-filling diagram

R.\ Bryant, H.\ Edelsbrunner, P.\ Koehl and M.\ Levitt. The area derivative of a space-filling diagram. Discrete Comput.\ Geom. 32 (2004), 293--308

2004

[5] [5]

Treatise on Plane and Spherical Trigonometry

W.\ Chauvenet. Treatise on Plane and Spherical Trigonometry. Ninth edition, Lippincott Company, Philadelphia, Pennsylvania, 1887

work page

[6] [6]

Geometry and Topology for Mesh Generation

H.\ Edelsbrunner. Geometry and Topology for Mesh Generation. Cambridge Univ.\ Press, Cambridge, England, 2001

2001

[7] [7]

The weighted-volume derivative of a space-filling diagram

H.\ Edelsbrunner and P.\ Koehl. The weighted-volume derivative of a space-filling diagram. Proc.\ Natl.\ Acad.\ Sci. 100 (2003), 2203--2208

2003

[8] [8]

Three-dimensional alpha shapes

H.\ Edelsbrunner and E.P.\ M\" u cke. Three-dimensional alpha shapes. ACM Trans.\ Graphics 13 (1994), 43--72

1994