It was shown by Hamilton in the 1990s that the isoperimetric ratio $C_H$ on the $2$-sphere improves along the Ricci flow.

A way to prove this is to use the fact that if $(M^2, g(t))$ is a solution of the Ricci flow on a topological $2$-sphere, then the isoperimetric ratios $C_H (\rho , t)$ of parallel loops $\gamma_{\rho}$ measured with respect to the metric $g(t)$ satisfy a heat-type equation

$\frac{\partial }{\partial t} (\log C_H) = \frac{\partial^2}{\partial {\rho}^2 } (\log C_H) + \frac{\Gamma}{L} \frac{\partial}{\partial \rho} (\log C_H)+ \frac{4 \pi - C_H}{A} \bigg( \frac{A+}{A_-} + \frac{A_-}{A_+} \bigg) $,

where $\Gamma$ is an integral of the signed curvature and more definitions can be found in the textbook of Chow and Knopf.

The isoperimetric ratio $C_H$ is similar to, but distinct from the Cheeger isoperimetric constant.

1.) Is it known if the Cheeger constant $h$ also satisfies a heat-type equation for a solution of the Ricci flow on a topological $2$-sphere?

2.) For a topological $2$-sphere, is it the case that

$C_H (M) \leq h (M)?$


1 Answer 1


Answer to (2) is negative. For a counterexample, consider a $2$-sphere and blow up the radius. The isoperimetric ratio $C_H$ remains the same, but the Cheeger constant $h$ shrinks, essentially because $C_H$ is a dimensionless quantity which is independent of the scaling, whereas $h$ is not.

Actually in Lemma 5.85 of the textbook of Chow and Knopf, it is shown that if $(M^2, g)$ is a closed orientable Riemannian surface, then

$C_H (M) \leq 4 \pi,$

which is obviously not true for the Cheeger constant.

I did a few calculations and looks as if answer to (1) is positive.

  • 1
    $\begingroup$ About one year ago, I feel the $h(M)$ maybe monotone along Ricci flow before singularity. But I don't know how to prove it, even though on special manifold. Since I just be a beginner of Ricci flow, I give up it. But I still be interested in it. If possible, could you tell me the calculation about (1). Thanks very much. $\endgroup$
    – Enhao Lan
    Feb 15, 2023 at 12:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.