Removing the logarithmic loss in a lower bound for sum-difference exponents
For a finite rational slope family containing $0$, $1$ and $\infty$, the sum-difference exponent measures how much entropy an output projection can have relative to the largest input projection entropy. Tao bounded this exponent from below by $2-O_k\!\left(\log(2+D)/D\right)$, where $D$ is the rational complexity of the output relative to the inputs and $k$ is the number of additional input slopes. For $k\ge1$, we prove the lower bound $2-C_k/D$, with an explicit constant depending only on $k$. The proof retains Tao's evaluated-polynomial construction and adds a covering estimate between two comparable degree-and-coefficient scales. This estimate holds for sets of evaluated values, including when polynomial evaluations coincide. A separate appendix evaluates the corrected folded-entropy functional associated with the three-slope continuous limit. Its infimum is $\log(2/\sqrt{3})$ for every folding parameter, by the classical sharp entropy Brascamp--Lieb inequality and concentration into a single strip.
Authors
- Sam Vaseghi (ORCID: https://orcid.org/0009-0006-4105-1949)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24221525
- Primary Topic
- Limits and Structures in Graph Theory
- Type
- preprint