Skip to main content
1 of 2
John Harvey
  • 599
  • 2
  • 11

The answer is no. In fact, it is not even true for Riemannian manifolds.

Consider a flat Riemannian manifold $(M,g)$ of dimension $m$ with diameter $d$ realized by the points $x$ and $y$. Let the volume be $Cd^m$. Let $M_n$ denote the Riemannian manifold $(M,n^2 g)$, where $n$ is a positive integer.

Attach a cylinder of radius $\delta$, where $\delta \ll d$ at the point $y \in M_n$. (More formally, remove $y$ from $M_n$, and change the metric so that it is a product outside a compact subset.) Do this so that the curvature remains $\geq -1$. The "bending" required to attach the cylinder can be the same for every $n$, because the metric is flat.

Now in $M_n$ we have $\mathrm{vol}(B_x(md)) = Cn^md^m + \epsilon_1$ and $\mathrm{vol}(B_x(2md)) = Cn^md^m + nd\delta^{m-1} + \epsilon_2$, where $\epsilon_1$ and $\epsilon_2$ are determined by the bending, and so are independent of $n$.

Then the ratio of the volumes tends to 1 as $n \to \infty$. You can attach copies of $M_n$ for every $n$ to the Euclidean plane to get a Riemannian manifold of curvature $\geq -1$ which does not satisfy the condition.

John Harvey
  • 599
  • 2
  • 11