Understanding distance function on Riemannian Manifold

365 Views Asked by At

We have the next definition. This is from the book "Classical and Modern Morse Theory with Applications" by Mercuri, Piccione and Tausk.

Let $f:M \to \mathbb{R}^{n+p}$ be an isometric immersion of an $n$-dimensional Riemannian manifold. Fix $q\in \mathbb{R}^{n+p}$. The $\textit{distance function from q}$ is the function $$L_q : M \to \mathbb{R}, \; L_q (x) = \langle q-x, q-x \rangle.$$

I am not sure if on the definition it should be understood that $L_q (x) = \langle q-f(x), q-f(x) \rangle$, since $x$ belongs to the manifold and $q$ belongs to an euclidean space, and it cannot be the riemannian metric since that is defined in the tangent space. Am I right?

And also, it continues on the book: (here, $\overline{\nabla}$ denotes the Levi-Civita connection of $\mathbb{R} ^{n+p}$, and $\nabla$ is the Levi-Civita connection of $M$.)

We study now the critical points of $L_q$. Set $\xi( x)=q-f(x)$. Identifying (locally) $M$ with $f(M)$, we have $\overline{\nabla} _X \xi = -X, X_1 T_xM$. Hence:

$\bullet dL_q (x) X = -2 \langle X , \xi \rangle. $ In particular, $x\in M$ is a critical point of $L_q$ if and only if $\xi (x) \in \nu _{n} M .$

$\bullet $ If $x$ is a critical point of $L_q$, we have: $d^2 L_q (x) (X, Y) =-2Y \langle X , \xi \rangle =2 \langle (Id-A_{\xi} ) X , Y \rangle. $

I have recently been introduced to the concept of the Levi-Civita connection, and I know that it is defined as the only symmetric connection that is compatible with the metric, but I am still trying to understand how to work with it. There are several things that I don't understand:

What does the expression $\overline{\nabla} _X \xi = -X, X_1 T_xM$ mean and why do we have that? (The text doesn't say who is $X_1$ and I am guessing it was written by mistake but I don't get what it means).

Also, on the first bullet I think there is a little mistake because it should appear $\nu _{x} M$ instead of $\nu _{n} M$, since on a previous section $\nu _{x} M$ denotes the normal space corresponding to $f$ at $x$, that is $df_x (T_x M ) ^{\perp} $. But I don't really know how was the expression $ dL_q (x) X = -2 \langle X , \xi \rangle $ obtained.

On the second bullet I also don't know how to obtain that expression. There, $A_{\xi} X = -(\overline{\nabla} _X \xi)^T $, where for $z\in T_{f(x)} \mathbb{R} ^{n+p}$, $z^T$ denotes the component of $z$ in $df_x (T_x M)$.

I would be really thankful for any comment and suggestion.

1

There are 1 best solutions below

1
On BEST ANSWER
  • Yes, $L_q$ should be understood as $L_q (x) = \langle q-f(x), q-f(x) \rangle$. In the following they/we do not distinguish between $M$ and $f(M)$, $TM$ and $f_*(TM)$.

  • Note that $\overline\nabla $ on $\mathbb R^{n+p}$ is nothing but coordinate-wise directional derivatives. Thus $$\overline\nabla_X (q-f(x))= - \overline \nabla_X f(x)$$ By definition $X$ is really in $TM$, and the above calculation should give $$-\overline\nabla _X f(x) = -f_*X.$$ But then they identify $X$ with $f_*X$, so $$\overline\nabla_X \xi = -X.$$ I don't know what's $X_1 T_xM$

  • Yes it should be $\nu_xM$.

  • To calculate $dL_q(x) X$, we use product rule:
    \begin{align} dL_q(x) X &= X \langle \xi, \xi\rangle \\ &= 2 \langle \overline\nabla_X \xi, \xi\rangle \\ &=-2 \langle X, \xi\rangle. \end{align}

  • $d^2 L_q(x)$ is the Hessian of $L_q$ at $x$ and should be calculated as (assume that $x$ is a critical point, $dL_q(x) = 0$) \begin{align} d^2 L_q(x) (X, Y) &= X(dL_q(Y)) - dL_q (\nabla_X Y) \\ &=-2 X \langle Y, \xi\rangle \ \ \ \ \ \ \ \ \ \qquad \qquad \ \ \ \ \ \ \text{ since } dL_q (x) = 0, \\ &=-2 \langle \overline\nabla _X Y, \xi\rangle -2 \langle Y, \overline \nabla_X\xi\rangle \\ &= -2\langle A(X, Y), \xi\rangle +2\langle X, Y\rangle \end{align} here $A$ is the second fundamental form. We do not get what you suggested in the post. I don't think what you write is correct. Indeed, $\overline\nabla _X\xi$ is already tangential, there's no need to write $(\cdot)^T$.