Arithmetic in the Boij-Söderberg Cone

Adam Boocher, Noah Huang, Harrison Wolf · 2025

Read Arithmetic in the Boij-Söderberg Cone by Adam Boocher, Noah Huang, Harrison Wolf (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-24320.txt.

Introduction

Adam Boocher, Noah Huang, and Harrison Wolf attack two old conjectures in algebra. The conjectures give lower bounds for the Betti numbers of graded modules. Those numbers encode the shape of a module's resolution. The authors recast the bounds as arithmetic. The new home is the Boij-Söderberg cone.

Inside that cone, a counterexample would leave a numeric trace. It would match an explicit Diophantine obstruction from a pure resolution. Number theory sorts those obstructions fully in codimension three. Delicate links emerge between Betti tables and classic Diophantine equations. New cases fall in codimensions five and six. Those concern Gorenstein algebras. There, studying the variety cut out by the equations settles the conjecture.

Catalog record

CREATORAdam Boocher, Noah Huang, Harrison Wolf
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE48 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-24320

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Arithmetic in the Boij-Söderberg Cone Adam Boocher, Noah Huang, Harrison Wolf Abstract . We study two long-standing conjectures concerning lower bounds for the betti numbers of a graded module over a polynomial ring. We prove new cases of these conjectures in codimensions five and six by reframing the conjectures as arithmetic problems in the Boij-Söderberg cone. In this setting, potential counterexamples correspond to explicit Diophantine obstructions arising from the numerics of pure resolutions. Using number-theoretic methods, we completely classify these obstructions in the codimension three case revealing some delicate connections between betti tables, commutative algebra and classical Diophantine equations. The new results in codimensions five and six concern Gorenstein algebras where a study of the variety determined by these Diophantine equations is sufficient to resolve the conjecture in this case. 1. Introduction Let R=k[x_{1},\ldots,x_{n}] be a polynomial ring over a field k and let M be a finitely-generated graded R -module. Two long-standing conjectures concern the betti numbers of M and their relationship to c , the codimension of M (the height of the annihilator of M ). Main Conjectures . Let R and M be as above and let c=\codim M . Then (1) \beta_{i}(M)\geq{c\choose i} , (2) If M is not (isomorphic to) a complete intersection then \sum\beta_{i}(M)\geq 2^{c}+2^{c-1} . We give a brief history of these conjectures in Section 2, but mention now that they are open when c\geq 5 . The first main result of this paper establishes new cases of these conjectures with the assumption that M=R/I defines a Gorenstein ring. Theorem 1 . Suppose that M=R/I is a Gorenstein ring of codimension c , not isomorphic to a complete intersection. • If c=6 then Conjecture (1) holds for i=0,1,2,4,5,6 without any further assumption on M . • With the added assumption that M is pure, Conjecture (2) is true if c\in\{5,6\} and Conjecture (1) is true if c=5 . A module M is called pure if it is Cohen-Macaulay and for all i\geq 0 its i th syzygy module is generated in a single degree d_{i} . To such a module we associate its degree sequence \{0,d_{1},\ldots,d_{c}\}. 1 1 1 Without loss of generality, we take d_{0}=0 By a result of [ 16 ] , there is (almost) a formula for the betti numbers of such modules (the so-called Herzog-Kühl Equations 2.1 ). Indeed, for each D there is a list of integers B(D) such that if M is pure with degree sequence D then its list of betti numbers, \beta(M) satisfies \beta(M)=LB(D) for some integer L . Example 2 . Suppose M is a pure module with c=3 and degree sequence D . (1) If D=\{0,2,3,8\} then B(D)=\{5,20,16,1\} and thus \beta(M)=\{5L,20L,16L,L\} . (2) If D=\{0,1,3,4\} then B(D)=\{1,2,2,1\} and thus \beta(M)=\{L,2L,2L,L\} . Notice that the numerics of the two examples are fundamentally different. For the first, regardless of what L is, B(D) is “large” enough to ensure that M satisfies both conjectures. On the other hand, in the second example, B(D) itself violates the bounds of both conjectures. When this happens we say that D is an arithmetic obstruction to the conjecture. In other words, if L=1 then \beta(M)=B(D) would violate the conjectures and so to prove that such a module M does indeed satisfy the conjecture, one must show that it is impossible for L to be 1 . This is simple enough in this case, since such a module would have the minimal free that begins R^{2}\to R^{1}\to M and then evidently M\cong R/I where I is an ideal generated by 2 minimal generators, which is impossible for an ideal of height c=3 by Krull’s altitude theorem. In principle, one approach to resolving the Main Conjectures would be to classify all arithmetic obstructions and then rule out the corresponding small values of L . In fact, one can get away without classifying all obstructions (such an analysis in codimensions 5 and 6, is precisely how we prove Theorem 1 ). At the same time it is possible to completely classify all obstructions in codimension 3: Theorem 3 . If c=3 then • There are infinitely many arithmetic obstructions to Main Conjecture (1). These are described explicitly in Theorem 15 and all have either B_{1}=2 or B_{2}=2 . • Regarding Main Conjecture (2), there are exactly 4 degree sequences D with \sum B(D)<12 . These are described explicitly in Theorem 21 . If M is a pure module that is not a complete intersection and its degree sequence D is one of the offending sequences above, then in the equation \beta(M)=LB(D) , L must be at least 2, recovering that the Main Conjectures hold in this case. Taken together, the above Theorems give a proof of concept that this framework can shed light on questions concerning the ranks of syzygies. From the point of view of Boij-Söderberg Theory (see e.g. [ 3 , 10 , 11 ] ), the set of all betti tables of Cohen-Macaulay modules of codimension c lies in a simplicial cone whose extremal rays correspond to pure modules. If \rho_{D} is the extremal ray corresponding to degree sequence D then, B(D) gives the coordinates of the smallest integer point on that ray, and the theorem of Herzog-Kühl asserts that every betti table of any pure module of type D will lie on \rho_{D} as an integer multiple of B(D) . Our approach is similar in spirit to the one initiated in [ 12 ] , because although calculating B(D) is immediate it can be quite difficult to tell whether or not the numbers B(D) are indeed the betti numbers of a module. Because the numerics of the Boij Söderberg cone are rather subtle, we find pure modules to be an excellent testing ground for the Main Conjectures. When 1\cdot B(D) obeys the bounds, that provides interesting positive evidence for the conjectures and when not, it directs our attempt to understand for which L , L\cdot B(D) can be the betti sequence of a module. In an effort to systematically study the general case, we propose the following. Program 4 . Let c be a positive integer. (1) Identify those degree sequences D=\{0,d_{1},\ldots,d_{c}\} with the property that B(D) violates Main Conjecture (1) or (2). We will call these arithmetic obstructions . (2) Recalling that if M is a pure module of type D then \beta(M)=LB(D) , for some integer L\geq 1 , prove that if M is a pure module with betti sequence L\beta(D) , then L must be sufficiently large so that \beta(M) satisfies the conjecture violated by B(D) . When c\in\{1,2\} this program is trivial and Theorem 3 settles the case when c=3 . For c\geq 4 this program seems quite difficult (e.g. see the discussion in Section 3.1 ) and we hope that techniques from neighboring fields might help. Why is this interesting? If one views Theorem 3 as a new way of verifying a special case of the Main Conjectures when c=3 , then this is perhaps not so interesting. After all, both conjectures have been long known to be true for any module in codimension c\leq 4 . However, there are many features of Program 4 that are remarkable and and worth more careful study.

Read the complete text