A theoretical arXiv preprint reaches a result with two sides. It shows that the full Bayer-Stillman regularity-preservation result fails in the multigraded setting for some multihomogeneous ideals, whatever monomial ordering is chosen. It then develops weaker lower bounds for ordinary regularity and applies them to the fractional Helly problem for d-Leray simplicial complexes. The paper concludes that two stated colorful conjectures hold. In this setting, colorful refers to faces assembled across designated blocks of variables or vertices, a structure that lets the algebraic argument address a combinatorial counting problem.
The algebraic break
The mathematical objects are multigraded polynomial rings, with variables partitioned into blocks, and multihomogeneous ideals. The paper applies a generic block-diagonal change of coordinates and then takes the initial ideal. That output is the generic block-diagonal initial ideal. The central algebraic question is whether this passage preserves ordinary graded regularity, the numerical measure the paper uses for the quotient's graded structure.
The answer is no in full generality. The paper shows that the full Bayer-Stillman preservation statement fails for some multihomogeneous ideals regardless of monomial ordering. In practical terms, one cannot assume that moving to a multigraded generic initial ideal leaves ordinary regularity unchanged. The result marks a boundary on the method, rather than claiming that every multihomogeneous ideal behaves this way.
That failure motivates a different target: a lower bound rather than equality. Under the paper's stated hypotheses, the ordinary regularity of the quotient R/I is at least the sum of the blockwise degree parameters q1 through qk. Here R denotes the multigraded polynomial ring, I the multihomogeneous ideal, and the q terms are the theorem's selected blockwise parameters. The bound is obtained with almost regular sequences and reverse-lexicographic orders in which the variables follow a prescribed order.
The proof also uses blockwise Koszul cycles. It combines cycles from separate blocks to build a cycle in the Koszul complex for R/I, giving the lower-bound argument a construction that reflects the multigrading. The paper's algebraic strategy therefore moves from a failed preservation claim to a controlled statement about how large regularity must be under its hypotheses.
From algebra to colored faces
For the combinatorial part, the paper turns to colored algebraic shifting. The operation preserves the flag f-vector, meaning the relevant counts of faces with fixed color patterns, and it has a color-shifted replacement property. With a reverse-lexicographic order that places specified early-block variables above later-block variables, the resulting colored shifted complex excludes the theorem's set F. That forbidden-set statement lets the regularity lower bound become a restriction on which colored faces can occur.
The first counting consequence concerns mixed-color faces, those containing one vertex from each listed block. In a d-Leray complex with d+1 blocks, the number of such faces is bounded above by a product-minus-product expression: multiply the block sizes, then subtract the product of the corresponding differences between each block size and its within-class maximum face bound.
The headline combinatorial consequence is the colorful fractional Helly bound. Suppose a fraction α greater than zero and no greater than one of all possible mixed-color choices form faces. Then, under the d-Leray and block-size assumptions, at least one color class contains a face whose size reaches the theorem's threshold. In words, that threshold is the size of the selected class multiplied by one minus the (d+1)th root of one minus α. The result turns a density condition on mixed-color faces into a guarantee of a large face in at least one class.
That implication is the paper's route to the two conjectures named in its introduction. It concludes that Conjectures 1.5 and 1.6 hold. The authors present the lower-bound machinery and colored shifting as sufficient to prove the colorful fractional Helly theorem and derive broader face and Betti-number bounds.
Bounds beyond the main result
The framework reaches beyond the mixed-color count. A broader corollary gives an upper bound for a general flag face number by counting t-faces that contain none of the forbidden sets in the relevant family. The paper says this bound is tight: its defining simplicial complex is d-collapsible, so the upper limit can be attained. Another theorem lower-bounds a topological Betti number of the complex by summing colored-shifted faces that satisfy a per-block extension condition, over tuples whose entries add up to i.
The conditions matter
The claims come with specific conditions. The regularity result is a lower bound under theorem-specific hypotheses, not a general equality between an ideal and its multigraded generic initial ideal. The colored-face exclusion depends on the prescribed reverse-lexicographic variable order, and the colored-shifting discussion assumes characteristic 0 for its strong color-stability property. These are conditional mathematical conclusions, with no statistical uncertainty attached.
The supplied document is an arXiv version 4 preprint dated 3 September 2026. Its acknowledgments list support from IBS-R029-C1 and partial support from National Science Foundation award 2402145.
Paper data and sources
Original title: Multi-graded generic initial ideals, regularity, and the optimal colorful fractional Helly theorem for $d$-Leray complexes
Authors: Daniel McGinnis
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text