发表机构
Sungkyunkwan University (SKKU); Rutgers, The State University of New Jersey; University of California San Diego(成均馆大学; 新泽西州立罗格斯大学; 加利福尼亚大学圣地亚哥分校)
机构由 AI 辅助整理,请以论文原文为准。AI 中文总结
本文研究圆球 Vietoris--Rips 共指标阈值,构造连续奇映射并利用投影码与色数估计,证明在特定条件下 Gromov--Hausdorff 距离严格大于共指标的一半,给出无穷反例。
AI 中文摘要
对于 $0\le m\le n$,设 $c_{m,n}$ 为圆球 $\mathbb{S}^m$ 的 Vietoris--Rips 过滤在存在从 $\mathbb{S}^n$ 到其上的连续奇映射时的尺度下确界。一个定量的 Borsuk--Ulam 定理给出 $c_{m,n}/2\le d_{\mathrm{GH}}(\mathbb{S}^m,\mathbb{S}^n)$,并且有人问是否总是相等。利用球面连接的同步乘积测度提升,我们构造了 Vietoris--Rips 度量加厚之间的连续奇映射,其目标尺度等于最大输入尺度。迭代给出对于每个整数 $d\ge0$ 有 $c_{m+d,n+d}\le c_{m,n}$。更一般地,有限连接法则分别加上 $m_i+1$ 和 $n_i+1$,并将所得的 $c$ 值界定为 $\max_i c_{m_i,n_i}$。对于 $0\le r<\pi$,满足 $c_{k-1,\ell-1}\le r$ 的配对 $(k,\ell)$ 在加法下封闭,因此 Fekete 引理给出当 $k\to\infty$ 时最大可容许 $\ell/k$ 的极限。这一结构、精确值以及投影码估计给出有限和渐近界。若 $1\le m_j<n_j$,$m_j\to\infty$,且 $\log(n_j+1)=o(m_j)$,则 $c_{m_j,n_j}\to\pi/2$。在固定的 $r\in(\pi/2,\pi)$ 处,投影码和投影覆盖估计给出满足 $c_{m,n}\le r$ 的最大 $n$ 的指数下界和上界。结合精确值,这产生一个尖锐的转变:在尺度 $\pi/2$ 处,允许奇映射的最大 $n$ 为 $m$,而在每个固定的 $r\in(\pi/2,\pi)$ 处,它随 $m$ 指数增长。我们最终回到猜想中的等式。我们引入一个来自 Borsuk 图过滤的色数的 Gromov--Hausdorff 稳定不变量。将 $c_{m,n}$ 的投影码上界与来自球冠体积的色数下界相结合,我们证明如果 $m_j<n_j$,$m_j\to\infty$,且 $n_j/m_j\to\lambda\ge9$,则对于所有足够大的 $j$,$d_{\mathrm{GH}}(\mathbb{S}^{m_j},\mathbb{S}^{n_j})>c_{m_j,n_j}/2$,从而产生无穷多个反例。
英文摘要
For $0\le m\le n$, let $c_{m,n}$ be the infimum of scales at which the Vietoris--Rips filtration of the round sphere $\mathbb{S}^m$ admits a continuous odd map from $\mathbb{S}^n$. A quantitative Borsuk--Ulam theorem gives $c_{m,n}/2\le d_{\mathrm{GH}}(\mathbb{S}^m,\mathbb{S}^n)$, and it was asked whether equality always holds. Using a synchronized product-measure lift of the spherical join, we construct continuous odd maps between Vietoris--Rips metric thickenings with target scale equal to the maximum input scale. Iteration gives $c_{m+d,n+d}\le c_{m,n}$ for every integer $d\ge0$. More generally, the finite join law adds $m_i+1$ and $n_i+1$ separately and bounds the resulting $c$-value by $\max_i c_{m_i,n_i}$. For $0\le r<π$, the pairs $(k,\ell)$ with $c_{k-1,\ell-1}\le r$ are closed under addition, so Fekete's lemma gives a limit for the maximal admissible $\ell/k$ as $k\to\infty$. This structure, exact values, and projective-code estimates give finite and asymptotic bounds. If $1\le m_j<n_j$, $m_j\to\infty$, and $\log(n_j+1)=o(m_j)$, then $c_{m_j,n_j}\toπ/2$. At fixed $r\in(π/2,π)$, projective-code and projective-covering estimates give lower and upper exponential bounds for the largest $n$ with $c_{m,n}\le r$. Together with exact values, these yield a sharp transition: the largest $n$ admitting an odd map is $m$ at scale $π/2$ and grows exponentially with $m$ at each fixed $r\in(π/2,π)$. We finally return to the conjectural equality. We introduce a Gromov--Hausdorff-stable invariant from chromatic numbers of Borsuk-graph filtrations. Combining projective-code upper bounds for $c_{m,n}$ with chromatic lower bounds from spherical-cap volumes, we show that if $m_j<n_j$, $m_j\to\infty$, and $n_j/m_j\toλ\ge9$, then $d_{\mathrm{GH}}(\mathbb{S}^{m_j},\mathbb{S}^{n_j})>c_{m_j,n_j}/2$ for all sufficiently large $j$, producing infinitely many counterexamples.