Sunflower Detectors: MDS Classification, Layered Optimization, and the Limits of Envelope Counting
Year: 2026
Type: preprint
Abstract: The interpolation method of Bucić, He, and Lee raises two connected questions: which prescribed spaces admit a normalized sunflower detector, and how efficiently can its layered monomial envelope be counted? We classify anchored detectors for arbitrary specified MDS spaces: sorted dimensions satisfy d_i ≥ i exactly when a detector exists, and the affine dimension is D[(D−1)⋯(D−m)−∏(D−d_i)]. The minimal profile is unique. For polynomial spaces over ordered fields, an even number of petals satisfies the corresponding all-direction rigidity criterion; odd petals and a four-element field illustrate its limits. Independently, we solve the cylinder-cover cost of a general complete profile envelope by equal binary relative entropies and a matching type-class witness. This gives explicit finite improvements, including F_4(N) < 2.312·1.988772^N, and the large-petal optimal saving (γ+o(1))/m^3 within the profile-counting and random-block framework, with 0.34938 < γ < 0.34940. These are structure and optimization results for a specified method; the inherited detector and block reduction retain their attribution.
Author: JIANG, ZHICHENG
Language: English
Subfield: Discrete Mathematics and Combinatorics
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7236831217
