WorksMathematics
Wills' (1981) conjecture holds if and only if the dimension is at most 4
Abstract
Let be a lattice polytope in that is symmetric about the origin and has the origin as its only interior lattice point. In 1981 Wills conjectured that each coefficient of the Ehrhart polynomial of is at most the corresponding coefficient of the cube , that is, . Wills proved this for , and Henze showed in 2014 that it fails for every ; the dimensions , and remained open. We settle them. In dimension the conjecture holds; the new bounds are and , and, up to unimodular equivalence, only for the cube, one explicit -cell and one explicit cross-polytope. In dimension we give a polytope with vertices in whose linear Ehrhart coefficient is ; products with segments then give counterexamples in every dimension . Hence Wills' conjecture holds in if and only if . The proof in dimension combines Ehrhart reciprocity, the lower bound theorem for triangulated -spheres and an exhaustive computer enumeration of about three million sets of lattice points, listed up to symmetry.
Introduction
A lattice polytope in is the convex hull of finitely many points of . Ehrhart [6] proved that for a lattice polytope of dimension the number of lattice points in the dilate is a polynomial in the integer ,
the Ehrhart polynomial of . Here , the leading coefficient is the volume of , and is half the lattice surface area of (§ 2); the remaining coefficients have no such simple description and may be negative [3].
Minkowski’s first theorem [14] states that a convex body that is symmetric about the origin and has no lattice point other than 0 in its interior has volume at most , the volume of the cube . Wills [20] looked for a discrete analog of this theorem, one that bounds every Ehrhart coefficient of a lattice polytope. Let be the set
An expanded version of this paper, with further results, is [7], at https://doi.org/10.5281/zenodo.239555454 of lattice polytopes with whose only interior lattice point is the origin. The cube belongs to , and , so . Wills called these numbers the expected values of the coefficients [20 Remark (b)], and the following statement is known as Wills’ conjecture. We state it as in [11 Conjecture 1.1]; see also [8].
Conjecture 1.1 (Wills’ conjecture). For every , every and every ,
For there is equality, and for the inequality is Minkowski’s theorem. Wills proved the case in every dimension [20 Theorem (i)]; hence the conjecture holds for . Wills also proved it for [20 Theorem (i), (ii)], and, in dimension 4, the bounds , and [20 Theorem (iii)]. A discrete version of Minkowski’s first theorem by Betke, Henk and Wills [4 Theorem 2.1] implies that for every and every integer , which is the inequality of the conjecture for the values of the two polynomials instead of their coefficients. For , a bound in terms of the successive minima that implies Wills’ bound is due to Henk, Schürmann and Wills [10]. Henze [11] constructed a family of polytopes and showed that has ; taking products with segments, he obtained counterexamples in every dimension . On the positive side, Henze proved the conjecture for every whose Ehrhart roots all have real part [11 Corollary 3.3]; by a theorem of Bey, Henk and Wills [5], every such polytope is reflexive. About the remaining dimensions, Henze wrote [11 Remark 2.2]: “In dimensions 4, 5 and 6 we tested Wills’ conjecture on lattice polytopes that have a similar structure to . [. . . ] None of these examples turned out to be a counterexample.” We settle these three dimensions.
Results. Our main results are Theorems 1.2 and 1.3 and Corollary 1.4 below, and Theorem 7.1 in § 7, which determines the polytopes with . Table 1 compares them with the best earlier results, and Figure 1 shows the status of the conjecture in each dimension. Two of the three polytopes with have Ehrhart roots off the line , so Henze’s theorem does not cover them (Remark 7.3).
| Best result before this paper | New result |
| [20 Theorem (iii)] | for every , with equality exactly for , a 24-cell and a cross-polytope , up to unimodular equivalence (Theorems 1.2 and 7.1) |
| and [20 Theorem (iii)] | for every , with equality exactly for , up to unimodular equivalence (Theorem 1.2 and Corollary 7.2) |
| no counterexample known for ; the tested families in dimensions 4, 5, 6 contain none [11 Remark 2.2] | the polytope of Table 2 is in and has ; counterexamples in every dimension (Theorem 1.3) |
| true for [20]; false for [11] | Conjecture 1.1 holds in if and only if (Corollary 1.4) |
Table 1. The best results before this paper and the results of this paper.

Figure 1. The status of Wills’ conjecture in each dimension . Before this paper the conjecture was known to hold for (Wills [20]) and to fail for (Henze [11]); the dimensions 4, 5 and 6 (hatched) were open. This paper proves it for (Theorem 1.2 and Corollary 1.4) and disproves it for every (Theorem 1.3); the dimensions settled here are drawn in full colour, inside the thick frame.
Theorem 1.2. Every satisfies and .
Theorem 1.3. The polytope of Table 2 belongs to , and
In particular . For every the polytope belongs to and has .
Table 2. The polytope is the convex hull of the 50 points , where runs over the 25 points below ( stands for ). All 50 points are vertices of , and has 92 facets.
Corollary 1.4. Wills’ conjecture holds in if and only if .
Proof. For this is Wills’ theorem [20]. For , by (E1) of § 2, by [20 Theorem (i)], by Minkowski’s theorem, and , are bounded by Theorem 1.2. For , Theorem 1.3 gives a counterexample. □
Idea of the proofs. Theorem 1.3 is a computation. The polytope is given by its 50 vertices, and since has only one interior lattice point, its linear coefficient is determined by four numbers: the numbers of lattice points in , in and in the interior of , and the volume of . Products with segments carry the example to every higher dimension.
For Theorem 1.2 we use Ehrhart reciprocity. Since has exactly one interior lattice point, ; comparing this with shows that is equivalent to the inequality
between the number of lattice points of and its lattice surface area , and that this inequality also gives (Proposition 4.2). A triangulation of the boundary of that uses every boundary lattice point, together with Walkup’s lower bound theorem for triangulated 3-spheres [19], proves the inequality when has at most 29 lattice points (Lemma 4.3).
For with at least 31 lattice points (their number is odd, since they come in pairs together with 0) we look at the integer linear functionals that take values in on . Either they contain a basis of , and then is unimodularly equivalent to a polytope in the cube ; or they span without containing a basis, and then a linear map carries , together with its lattice, onto a polytope in with vertices in a sublattice of index 2; or they span a proper subspace. In the last case we fix four lattice points of whose determinant is as large as possible; in their coordinates the lattice points of lie in , in one of 39 lattices, up to signed permutations of the coordinates (§ 5). This reduction, by four lattice points of largest determinant, follows Averkov and Schymura [2], who used it to compute the largest number of columns of a -modular integer matrix for small parameters; their enumeration builds on that of Averkov, Borger and Soprunov [1]. In each case only finitely many sets of lattice points remain, and they are enumerated by computer (§ 6).
Section 2 collects the facts we use from Ehrhart theory. Section 3 proves Theorem 1.3. Sections 4–6 prove Theorem 1.2, and Section 7 identifies the equality cases. The expanded version [7] of this paper also gives a second proof of Case III of Section 5, determines the second largest value of in dimension 4, and describes the computations in more detail.
Preliminaries
Ehrhart polynomials
Let , let be a lattice (a discrete subgroup of rank ), and let be a polytope of dimension whose vertices lie in . Then is a polynomial in of degree [6]; its coefficients depend on , which will be clear from the context, and we write when . We call two pairs and isomorphic if a linear isomorphism of satisfies and ; then for every , so . When , the map lies in , and the two lattice polytopes are called unimodularly equivalent. We use four standard facts [3 Chapters 3–5].
(E1) and .
(E2) Reciprocity (Ehrhart, Macdonald): for every integer .
(E3) For a facet of let be the primitive vector of the dual lattice that is an outer normal of , and let be the integer with . Let be the -dimensional volume of , measured so that a fundamental domain of the lattice has volume 1. Then , where is the lattice surface area. If , then , by cutting into the pyramids with apex 0 over its facets.
(E4) for lattice polytopes and , with the lattice , since .
In (E3), is an integer because contains points of ; when it is positive, and it is the lattice distance of from the origin.
The class
Let be the set of polytopes with vertices in , and ; so . A polytope in has dimension , because 0 is an interior point. We write . Since the lattice points of other than 0 come in pairs , is odd.
Lemma 2.1. Let be a set with whose convex hull has dimension . Then .
Proof. The polytope is a lattice polytope with . It has an interior point , so is interior too, and so is . Finally , whose only lattice point is .
The same proof applies to a sublattice in place of : if is symmetric and spans , then . We use this in § 5 and § 6.
Lemma 2.2 (cf. [11 Proposition 2.3]). If and , then and .
Proof. The product is a symmetric lattice polytope, and contains only the lattice point . By (E4) and (E1), the coefficient of in is .
A counterexample in dimension five
We compute four numbers for the polytope of Table 2, three lattice-point counts and the volume; by Proposition 3.1 they determine .
Proposition 3.1. Let . Then
Proof. Write and . By (E2) with , and . Hence
Eliminating and using (E1) gives .
Proof of Theorem 1.3. The 50 points of Table 2 lie in , form a symmetric set and span ; by Lemma 2.1, . Direct computation gives
so and, by Proposition 3.1, . The whole polynomial is determined by and the values for , namely 117, 1815, 10765, 39927 and 112639. All numbers in this proof can be computed with Normaliz [16] or LattE [12]; the input files are in [17].
For , Lemma 2.2 gives and , since . □
Dimension four: the key inequality and polytopes with few lattice points
From now on . For we write , (with respect to ) and .
Lemma 4.1. For , .
Proof. We have , and by (E2) . Subtracting the second equation from the first, and adding the two, gives the two identities. □
Proposition 4.2. Let . Then and . Moreover
and if then .
Proof. The bound is Minkowski’s theorem, applied in the coordinates of a basis of . Since every , (E3) gives ; this is Wills’ argument for [20 Theorem (i)]. By Lemma 4.1, , so if and only if . If this holds, then , and Lemma 4.1 gives . □
So Theorem 1.2 is equivalent to the key inequality on . Let be the normalized volume of a facet, so that a lattice tetrahedron in has normalized volume a positive integer. The inequality reads
Lemma 4.3. If and , then (1) holds with strict inequality; hence .
Proof. Take a triangulation of the boundary of whose vertex set is the set of all boundary lattice points and in which every tetrahedron lies in a facet. To construct one, take a triangulation of whose vertices are exactly the points of ; one exists, for example by strong pulling ([9 Lemma 2.1(2)], applied in the coordinates of a basis of ). Its faces that lie in the boundary of form . Then is a triangulated -sphere with vertices. Each of its tetrahedra is a lattice tetrahedron in the affine hull of a facet, so it has normalized volume at least , and . Walkup’s lower bound theorem [19 Theorem 1] gives for every triangulated -sphere, and the Dehn–Sommerville relations , give . Hence , and because . □
:::
The proof gives a bound on for every : by Lemma 4.1 and (E3), , and , so (the solid line in Figure 3).
Figure 3. The largest values of in Case I, and the three equality cases of Theorem 7.1, against the number of lattice points. The small grey dots show, for each odd , the largest value of in Case I (Computation 1). The three larger green dots mark the polytopes , and of Theorem 7.1: belongs to Case I, and and belong to Case III. For Cases II and III only the largest value of over all is recorded: at most 6 in Case II (Computation 3), and at most 8 in Case III, attained only by and (Computation 5). The dashed line is . In the shaded region , the solid line is the bound stated after the proof of Lemma 4.3.
Remark 4.4. With Euler’s formula in place of Walkup’s theorem, a similar triangulation argument gives Wills’ theorem in dimension 3 (see [7]). In dimension 4 the lower bound theorem proves (1) only when ; this is why the computations of § 6 are needed.
Dimension four: three cases
By Lemma 4.3 it remains to prove (1) for with . For such let
The set is finite (since contains a ball around 0) and symmetric, and for and . We distinguish three cases: contains a basis of (Case I); but contains no basis (Case II); (Case III). Figure 2 shows how each case is settled.

Figure 2. The proof of Theorem 1.2 splits into four parts. Polytopes with at most 29 lattice points are handled by Lemma 4.3 (the blue box, on the left); for the others (the three orange boxes), the integer functionals bounded by 1 in absolute value on decide which finite enumeration of Section 6 contains , up to isomorphism (Section 2). The frame lattices are defined after Lemma 5.1.
Case I
For a lattice , a set is called -saturated if ; when the lattice is clear, we say saturated. If contains a basis of , let have rows . Then , so for the set , which is symmetric and saturated.
Case II
Let , and let be linearly independent, with the matrix of rows and . Then , since otherwise would contain a basis. The map carries onto the sublattice of index and carries into ; so is isomorphic to , , , and for the -saturated symmetric set . In particular
Moreover : the set spans , so it contains four linearly independent points of ; their determinant is a nonzero multiple of , and it is at most 16 in absolute value by Hadamard’s inequality.
The sublattices of index 2 are for nonempty . A point of lies in exactly when an even number of its coordinates in are nonzero, so
Computation 2 shows that a sublattice of index contains at most 27 points of whenever these points span . Since , in Case II we have and with .
Case III: the frame
Now let . Choose lattice points of for which is as large as possible, and let be the matrix with columns . We call a frame of . Since spans , . The map carries onto the lattice , of index over , and carries onto a set .
Lemma 5.1. In this situation:
(i) , , and contains the unit vectors ;
(ii) for all ;
(iii) and ;
(iv) contains no point with ;
(v) .
Proof. For we have by the choice of , which gives (ii). For (i), write a lattice point of as ; by Cramer’s rule . Property (iii) holds because carries and to the corresponding statements for and . For (iv), the open cross-polytope lies in the interior of , so by (iii) its only point of is .
For (v), contains the cross-polytope of volume , and , so . If , then and . The translates , , have disjoint interiors (an overlap would put a nonzero lattice point in ) and total density , so they tile : the union of the translates is closed, and its complement is open, invariant under and of density ; a nonempty such set would contain a ball and all its -translates, hence have positive density. A convex polytope that tiles space by lattice translations has centrally symmetric facets (Minkowski [15]; see Venkov [18] and McMullen [13]), but the facets of a cross-polytope are tetrahedra. So . □
Let be the set of lattices of index at most over that satisfy (iv); a lattice of index over lies in , so is finite. We call the lattices in the frame lattices. For let be the family of sets satisfying (i)–(iii). The signed permutation matrices act on , and Computation 4 shows that consists of 39 orbits; we fix one representative in each. Integer functionals transfer as well. For a set let and . Since and , a vector lies in if and only if lies in ; so . In Case III we therefore have a lattice among the 39 representatives and a set with and such that is isomorphic to (by , followed by a signed permutation of the coordinates); in particular . Lemma 5.1 does not use , so this reduction applies to every . We use it only in Case III, where the condition shortens the search of Computation 5; the polytopes with are listed directly, in Cases I and II.
The computations
All enumerations below are exhaustive and use exact integer and rational arithmetic; the repository [17] contains the programs and, for each computation, its output. Computations 1, 3 and 5 list sets of lattice points up to the symmetries named. For each set the Ehrhart polynomial of is computed exactly: in Computations 1 and 3 from $G = #S and the volume and the lattice surface area of (taken from a triangulation of its boundary), by Lemma 4.1, (E1) and (E3); in Computation 5 from the numbers of lattice points of for and of the interior of for , which give its values at by (E1) and (E2) and so determine it. Figure 3 shows, for each , the largest value of in Case I (Computation 1) and the three equality cases of Theorem 7.1, and, for , the bound on stated after the proof of Lemma 4.3.
Computations 1 and 3 walk downward from the full set, or , deleting one pair of opposite vertices at a time. By the next lemma, such a walk reaches every saturated symmetric subset of the full set. Sets that a symmetry carries onto each other form a class, and the walk keeps one set of each class; classes are recognized by a canonical form, the least image of the set under the symmetry group in a fixed total order. This loses no class: a symmetry maps saturated symmetric sets to saturated symmetric sets and the vertices of to those of , so the sets obtained from by deleting a pair of opposite vertices are the images under of those obtained from . Computation 5 uses a different method, a sandwich search, described after its statement.
Lemma 6.1. Let be a lattice and a finite symmetric set with .
(i) If is a vertex of , then is symmetric and satisfies .
(ii) If is symmetric with , then some vertex of is not in , and .
Proof. (i) Since is a vertex, , and by symmetry ; as , it equals . (ii) If every vertex of were in , then . A vertex has by symmetry. □
So every saturated symmetric is reached from through saturated symmetric sets that contain ; in particular a search that stops at 31 points misses no set with at least 31 points.
Computation 1. Case I. We list all symmetric sets with and , up to the 384 signed permutations of the coordinates. Each such spans (a hyperplane through contains at most 27 points of , since on it one coordinate is determined by the other three), so by Lemma 2.1. There are 2,547,694 classes; for all of them the linear coefficient of is at most 8, with equality only for .
Computation 2. Index bound. We list every sublattice of index with once, as with in Hermite normal form under column operations (); there are 40,406 of them. For each whose points in span we count these points. The maximum is 45 for , and it is at most 27 for every .
Computation 3. Case II. A permutation of coordinates carries onto when , so it suffices to take for . For each of these three lattices we list all -saturated symmetric sets that span and have , up to the signed permutations that preserve ; each such lies in (see the paragraph after Lemma 2.1). There are 531, 58 and 27 classes for , and for all of them the linear coefficient of with respect to is at most 6.
Computation 4. The frame lattices. We list all lattices in . For the frame matrices and give the same lattice , so it suffices to take in Hermite normal form under row operations (), with determinant , and to keep the lattices that satisfy (iv). There are 1,040 frame lattices, in 39 orbits under signed permutations, with indices .
Computation 5. Case III. For each of the 39 representatives we list all with and , up to the signed permutations that preserve . There are 588,592 classes; for all of them the linear coefficient of with respect to is at most 8, with equality for exactly two of them, one with and one with .
The search follows [2 Algorithm 1]. It keeps a lower set and an upper set , and it starts from and . Every member of lies between these two sets: it lies in by (i), and it contains by (i) and (iii). The set is itself a member: its points have , so (ii) holds by Hadamard’s inequality and (iii) by (iv).
For a pair in the search branches: either is excluded and is replaced by , or is included and is replaced by ; every saturated set between and lies in exactly one branch. An include branch is abandoned when the new is not contained in , violates (ii), or has a nonzero point of in the interior of its convex hull, since then no set in the branch satisfies (i)–(iii). A pair in for which violates (ii) is removed from , since no member contains it together with . A branch is abandoned when or ; the second rule uses that reverses inclusion, that is, whenever , so that every subset of a set with also has . The search stops at the pairs with . Each include step keeps in , so these sets are members, and they are the sets found. A pair that a signed permutation preserving carries onto a pair met earlier in the search is discarded, since the permutation carries the sets between the two pairs onto each other; the sets found are then reduced to one of each class by the canonical form described before Lemma 6.1.
Proof of Theorem 1.2. Let . If , Lemma 4.3 gives . If , then by Section 5 the pair is isomorphic to for a set listed in Computation 1 (Case I), in Computation 3 (Case II, using Computation 2), or in Computation 5 (Case III, using Computation 4), where is the lattice of that computation (, or a frame lattice); so . In each list . Hence , and Proposition 4.2 gives . □
The equality cases
Let be the convex hull of the 24 points , , where for and . This is the 24-cell written in the basis of the lattice ; in other words, is isomorphic to the pair formed by this 24-cell and this lattice. Let
Theorem 7.1. A polytope has if and only if it is unimodularly equivalent to , or .
We first check that and lie in , that , and have , and that .
The polytope is the set of points with for all and , so its lattice points in other than 0 are the 24 vertices, the 8 points and the 16 points ; each has a coordinate or all four coordinates , so it lies on one of the facets or . Hence and . The dual lattice of is ; its primitive vectors normal to these facets are and , and they take the value 2 on them, so every facet of has lattice distance 2 from the origin. Moreover , since has volume 8 and the lattice has determinant .
For , a direct computation (see [17]) shows that 0 is its only interior lattice point, that , and that each of its 16 facets has lattice distance 3 from the origin; and , since the vectors , , , have determinant 18.
If every facet of has lattice distance , then (E3) gives , and Lemma 4.1 gives and . For we have and , and its facets have lattice distance 1. With and this gives
For , let and . Then for , so at most one is nonzero, and it is . Here for and . If , then and is even, so . Otherwise for one and , so , which is impossible. So .
For , the conditions for the first three vertices, namely , and , leave nine triples , one for each . The fourth condition, , requires or . Of the nine triples only satisfies this, and then . So .
Proof of Theorem 7.1. The three polytopes have , and they are pairwise inequivalent because they have 81, 49 and 33 lattice points. Conversely, let . By Lemma 4.3, , so, as in the proof of Theorem 1.2, is isomorphic to for a set with in one of Computations 1, 3, 5. In Computation 1 this is the cube, and the isomorphism lies in ; Computation 3 has none; in Computation 5 it is one of the two sets with and . Now and have , so they fall under Case III, and and are isomorphic to pairs with in Computation 5 and ; comparing , these are the two sets found there. If and are both isomorphic to the same pair (after a signed permutation fixing its lattice, if needed), then composing the two isomorphisms gives a linear map that carries onto and onto , that is, a map in , and in the same way for . Hence, in this case, is unimodularly equivalent to or to . □
Corollary 7.2. A polytope has if and only if it is unimodularly equivalent to .
Proof. The cube has . Conversely, let . Since by Theorem 1.2, Proposition 4.2 gives , so , and then by Lemma 4.1. The same lemma gives , and , force . By Theorem 7.1, is unimodularly equivalent to , or , and of these only has 81 lattice points. □
Remark 7.3. Henze’s theorem [11 Corollary 3.3] covers but neither nor . All roots of have real part , but those of and do not: if all four roots of , for , have real part , they sum to , so by Vieta’s formulas; but has and has . Alternatively, by [5] such a would be reflexive, that is, all its facets would have lattice distance 1 from the origin, while the facets of and have lattice distance 2 and 3.
Acknowledgments
This work was done overwhelmingly by AI models, steered by the author in a continuous critical dialogue, alternating between Anthropic’s Claude and OpenAI’s ChatGPT. The author takes full responsibility for the paper.
References
Email address: [email protected]
References
- [1]G. Averkov, C. Borger and I. Soprunov, Classification of triples of lattice polytopes with a given mixed volume, Discrete Comput. Geom. 66 (2021), no. 1, 165–202.DOI
- [2]G. Averkov and M. Schymura, On the maximal number of columns of a Δ-modular integer matrix: bounds and computations, Math. Program. 206 (2024), 61–89.arxiv.org/abs/2111.06294
- [3]M. Beck and S. Robins, Computing the Continuous Discretely, 2nd ed., Springer, New York, 2015.
- [4]U. Betke, M. Henk and J. M. Wills, Successive-minima-type inequalities, Discrete Comput. Geom. 9 (1993), 165–175.
- [5]C. Bey, M. Henk and J. M. Wills, Notes on the roots of Ehrhart polynomials, Discrete Comput. Geom. 38 (2007), 81–98.DOI
- [6]E. Ehrhart, Sur les polyèdres rationnels homothétiques à n dimensions, C. R. Acad. Sci. Paris 254 (1962), 616–618.
- [7]D. Goel, Wills’ (1981) conjecture holds if and only if the dimension is at most 4, and other miscellanea, Zenodo, 2026, https://doi.org/10.5281/zenodo.23955454
- [8]P. Gritzmann and J. M. Wills, Lattice points, in: Handbook of Convex Geometry, Vol. B, North-Holland, Amsterdam, 1993, 765–797.DOI
- [9]C. Haase, A. Paffenholz, L. C. Piechnik and F. Santos, Existence of unimodular triangulations—positive results, Mem. Amer. Math. Soc. 270 (2021), no. 1321; arXiv:1405.1687.
- [10]M. Henk, A. Schürmann and J. M. Wills, Ehrhart polynomials and successive minima, Mathematika 52 (2005), 1–16.DOI
- [11]M. Henze, On counterexamples to a conjecture of Wills and Ehrhart polynomials whose roots have equal real parts, Electron. J. Combin. 21(1) (2014), #P1.28.arxiv.org/abs/1309.0725
- [12]J. A. De Loera, R. Hemmecke, J. Tauzer and R. Yoshida, Effective lattice point counting in rational convex polytopes, J. Symbolic Comput. 38 (2004), 1273–1302.
- [13]P. McMullen, Convex bodies which tile space by translation, Mathematika 27 (1980), 113–121.
- [14]H. Minkowski, Geometrie der Zahlen, Teubner, Leipzig, 1896.
- [15]H. Minkowski, Allgemeine Lehrsätze über die convexen Polyeder, Nachr. Ges. Wiss. Göttingen (1897), 198–219.
- [16]W. Bruns, B. Ichim, C. Söger and U. von der Ohe, Normaliz. Algorithms for rational cones and affine monoids, available at https://www.normaliz.uni-osnabrueck.de
- [17]D. Goel, Supporting materials for “Wills’ (1981) conjecture holds if and only if the dimension is at most 4”, GitHub repository, https://github.com/david44hello/paper_wills-1981-conjecture
- [18]B. A. Ven’kov, On a class of Euclidean polyhedra, Vestnik Leningrad. Univ. Ser. Mat. Fiz. Him. 9 (1954), 11–31.
- [19]D. W. Walkup, The lower bound conjecture for 3- and 4-manifolds, Acta Math. 125 (1970), 75–107.DOI
- [20]J. M. Wills, On an analog to Minkowski’s lattice point theorem, in: The Geometric Vein (C. Davis, B. Grünbaum, F. A. Sherk, eds.), Springer, New York, 1981, 285–288.DOI
