4
$\begingroup$

It's well known that a sequence of two dimensional Riemannian manifolds with uniform sectional curvature lower bound can Gromov-Hausdorff converge to a cone.

Let $y=|x|$, by rotating around the y-axis, we get a cone $X$. Now I want to construct functions $y=f_i(x)$ such that their rotations around the y-axis (denoted by $M_i$) are smooth Riemannian manifolds and Gromov-Hausdorff converge to the cone $X$. I also want to compute the Gaussian curvature of $M_i$.

I considered $f_i(x)=|x|^{a_i}$ such that $a_i>1$ decrease to 1, but $f_i''(0)=\infty$, so it's not smooth enough, and the curvature at $x=0$ can't be defined. I also considered mollification of $|x|$ by convolution of $\eta(x)=c\exp(\frac{1}{|x|^2-1})$, but we can't write down the explicit form of the integration.

Can any one give good approximations such that $f_i(x)$ can explicit be written down?

Thanks to Thomas Richard's answer, below we compute the Gaussian curvatures. Let $y=\sqrt{t^2+a^2}$, then the length of the curve $$ l(t)=\int_0^t \sqrt{1+[y'(t)]^2}=\int_0^t (2t^2+a^2)^{\frac12}(t^2+a^2)^{-\frac12}dt $$ The metric can be written as $dl^2+t^2 d\theta^2$, by the curvature formula of the warped products, and note that $t''(l)=-\frac{l''(t)}{(l'(t))^3}$, the Gaussian curvature $$ K=-\frac{t''(l)}{t}=\frac{l''(t)}{(l'(t))^3 t}=\frac{2a^2}{(2t^2+a^2)^2}. $$ At $t=0$, $K=\frac{2}{a^2}\to \infty$ as $a\to 0$.

$\endgroup$
1
  • 1
    $\begingroup$ Instead of closing the question would you edit it to include the curvature formula for Thomas Richard's function. $\endgroup$ Commented Nov 27, 2017 at 12:24

1 Answer 1

2
$\begingroup$

What about $f(x)=\sqrt{x^2+\varepsilon_i^2}$ with $\varepsilon_i$ going to zero ? If you want to vary the angle of the limit cone, consider $f(x)=C\sqrt{x^2+\varepsilon_i^2}$.

$\endgroup$
4
  • 2
    $\begingroup$ Oops! I think I should close the question. $\endgroup$ Commented Nov 27, 2017 at 11:01
  • $\begingroup$ Maybe I miscalculated but when I computed the curvature of the graph of $f(x,y)=\sqrt{x^2+y^2+\epsilon}$ I got $\epsilon (2x^2+2y^2+\epsilon)^{-2}$ which at the origin gives $\epsilon^{-1}$, i.e., not uniformly bounded. What am I missing? $\endgroup$ Commented Nov 27, 2017 at 12:57
  • 1
    $\begingroup$ The curvature is positive. The question only asked for a bound from below. $\endgroup$
    – Tim Carson
    Commented Nov 27, 2017 at 13:07
  • $\begingroup$ @TimCarson: thank you, I missed the requirement. I confused the question with a related example of a sequence of metrics whose limit is $C^{1,\alpha}$ but not $C^{1,1}$ as in [Peters, Convergence of Riemannian manifolds, Compositio Mathematica, 62, no 1 (1987), p. 3-16]. $\endgroup$ Commented Nov 27, 2017 at 13:12

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .