English

A Marstrand projection theorem for lines

Classical Analysis and ODEs 2023-10-27 v1 Combinatorics Metric Geometry

Abstract

Fix integers 1<k<n1<k<n. For VG(k,n)V\in G(k,n), let PV:RnVP_V: \mathbb{R}^n\rightarrow V be the orthogonal projection. For VG(k,n)V\in G(k,n), define the map πV:A(1,n)A(1,V)V. \pi_V: A(1,n)\rightarrow A(1,V)\bigsqcup V. PV(). \ell\mapsto P_V(\ell). For any 0<a<dim(A(1,n))0<a<\text{dim}(A(1,n)), we find the optimal number s(a)s(a) such that the following is true. For any Borel set AA(1,n)\boldsymbol{A} \subset A(1,n) with dim(A)=a\text{dim}(\boldsymbol{A})=a, we have dim(πV(A))=s(a),for a.e. VG(k,n). \text{dim}(\pi_V(\boldsymbol{A}))=s(a), \text{for a.e. } V\in G(k,n). When A(1,n)A(1,n) is replaced by A(0,n)=RnA(0,n)=\mathbb{R}^n, it is the classical Marstrand projection theorem, for which s(a)=min{k,a}s(a)=\min\{k,a\}. A new ingredient of the paper is the Fourier transform on affine Grassmannian.

Cite

@article{arxiv.2310.17454,
  title  = {A Marstrand projection theorem for lines},
  author = {Shengwen Gan},
  journal= {arXiv preprint arXiv:2310.17454},
  year   = {2023}
}

Comments

25 pages; 2 figures

R2 v1 2026-06-28T13:02:51.066Z