Namespaces
Variants
Actions

Difference between revisions of "Hopf-Rinow theorem"

From Encyclopedia of Mathematics
Jump to: navigation, search
m (moved Hopf–Rinow theorem to Hopf-Rinow theorem: ascii title)
(TeX)
Line 1: Line 1:
If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480101.png" /> is a connected Riemannian space with distance function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480102.png" /> and a [[Levi-Civita connection|Levi-Civita connection]], then the following assertions are equivalent:
+
{{TEX|done}}
 +
If $M$ is a connected Riemannian space with distance function $\rho$ and a [[Levi-Civita connection|Levi-Civita connection]], then the following assertions are equivalent:
  
1) <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480103.png" /> is complete;
+
1) $M$ is complete;
  
2) for every point <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480104.png" /> the [[Exponential mapping|exponential mapping]] <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480105.png" /> is defined on the whole tangent space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480106.png" />;
+
2) for every point $p\in M$ the [[Exponential mapping|exponential mapping]] $\exp_p$ is defined on the whole tangent space $M_p$;
  
3) every closed set <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480107.png" /> that is bounded with respect to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480108.png" /> is compact.
+
3) every closed set $A\subset M$ that is bounded with respect to $\rho$ is compact.
  
 
===Corollary:===
 
===Corollary:===
Any two points <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h0480109.png" /> can be joined in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801010.png" /> by a geodesic of length <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801011.png" />. This was established by H. Hopf and W. Rinow [[#References|[1]]].
+
Any two points $p,q\in M$ can be joined in $M$ by a geodesic of length $\rho(p,q)$. This was established by H. Hopf and W. Rinow [[#References|[1]]].
  
A generalization of the Hopf–Rinow theorem (see [[#References|[4]]]) is: If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801012.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801013.png" /> are two points in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801014.png" />, then either there exists a curve joining them in a shortest way or there exists a geodesic <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801015.png" /> emanating from <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801016.png" /> with the following properties: 1) <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801017.png" /> is homeomorphic to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801018.png" />; 2) if a sequence of points on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801019.png" /> does not have limit points on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801020.png" />, then it does not have limit points in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801021.png" />, that is, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801022.png" /> is closed in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801023.png" />; 3) <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801024.png" /> contains the shortest connection between any two points on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801025.png" />; 4) <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801026.png" /> for every point <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801027.png" />; and 5) the length of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801028.png" /> is finite and does not exceed <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801029.png" />. Here the function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801030.png" /> is not necessarily symmetric, and every point can be joined in a shortest possible (not necessarily unique) way to any point in a certain neighbourhood <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801031.png" />.
+
A generalization of the Hopf–Rinow theorem (see [[#References|[4]]]) is: If $p$ and $q$ are two points in $M$, then either there exists a curve joining them in a shortest way or there exists a geodesic $L$ emanating from $p$ with the following properties: 1) $L$ is homeomorphic to $0\leq t<1$; 2) if a sequence of points on $L$ does not have limit points on $L$, then it does not have limit points in $M$, that is, $L$ is closed in $M$; 3) $L$ contains the shortest connection between any two points on $L$; 4) $\rho(p,x)+\rho(x,q)=\rho(p,q)$ for every point $x\in L$; and 5) the length of $L$ is finite and does not exceed $\rho(p,q)$. Here the function $\rho(p,q)$ is not necessarily symmetric, and every point can be joined in a shortest possible (not necessarily unique) way to any point in a certain neighbourhood $U_p$.
  
 
===Corollary:===
 
===Corollary:===
If there are no bounded rays in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801032.png" />, then every bounded set in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801033.png" /> is compact.
+
If there are no bounded rays in $M$, then every bounded set in $M$ is compact.
  
 
====References====
 
====References====
Line 21: Line 22:
  
 
====Comments====
 
====Comments====
Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801034.png" />. The manifold <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801035.png" /> is called geodesically complete at <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801036.png" /> if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801037.png" /> is defined on all of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801038.png" />. The manifold <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801039.png" /> is geodesically complete if this holds for all <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801040.png" />. The Hopf–Rinow theorem also includes the statement that geodesic completeness is equivalent to geodesic completeness at one <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801041.png" />.
+
Let $p\in M$. The manifold $M$ is called geodesically complete at $p$ if $\exp_p$ is defined on all of $T_pM$. The manifold $M$ is geodesically complete if this holds for all $p$. The Hopf–Rinow theorem also includes the statement that geodesic completeness is equivalent to geodesic completeness at one $p\in M$.
  
A geodesic joining <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801042.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h048/h048010/h04801043.png" /> and of minimal length is called a minimizing geodesic.
+
A geodesic joining $p$ and $q$ and of minimal length is called a minimizing geodesic.
  
 
====References====
 
====References====
 
<table><TR><TD valign="top">[a1]</TD> <TD valign="top">  W. Klingenberg,  "Riemannian geometry" , de Gruyter  (1982)  (Translated from German)</TD></TR></table>
 
<table><TR><TD valign="top">[a1]</TD> <TD valign="top">  W. Klingenberg,  "Riemannian geometry" , de Gruyter  (1982)  (Translated from German)</TD></TR></table>

Revision as of 06:33, 15 August 2014

If $M$ is a connected Riemannian space with distance function $\rho$ and a Levi-Civita connection, then the following assertions are equivalent:

1) $M$ is complete;

2) for every point $p\in M$ the exponential mapping $\exp_p$ is defined on the whole tangent space $M_p$;

3) every closed set $A\subset M$ that is bounded with respect to $\rho$ is compact.

Corollary:

Any two points $p,q\in M$ can be joined in $M$ by a geodesic of length $\rho(p,q)$. This was established by H. Hopf and W. Rinow [1].

A generalization of the Hopf–Rinow theorem (see [4]) is: If $p$ and $q$ are two points in $M$, then either there exists a curve joining them in a shortest way or there exists a geodesic $L$ emanating from $p$ with the following properties: 1) $L$ is homeomorphic to $0\leq t<1$; 2) if a sequence of points on $L$ does not have limit points on $L$, then it does not have limit points in $M$, that is, $L$ is closed in $M$; 3) $L$ contains the shortest connection between any two points on $L$; 4) $\rho(p,x)+\rho(x,q)=\rho(p,q)$ for every point $x\in L$; and 5) the length of $L$ is finite and does not exceed $\rho(p,q)$. Here the function $\rho(p,q)$ is not necessarily symmetric, and every point can be joined in a shortest possible (not necessarily unique) way to any point in a certain neighbourhood $U_p$.

Corollary:

If there are no bounded rays in $M$, then every bounded set in $M$ is compact.

References

[1] H. Hopf, W. Rinow, "Ueber den Begriff der vollständigen differentialgeometrischen Flächen" Comm. Math. Helv. , 3 (1931) pp. 209–225
[2] G. de Rham, "Sur la réducibilité d'un espace de Riemann" Comm. Math. Helv. , 26 (1952) pp. 328–344
[3] D. Gromoll, W. Klingenberg, W. Meyer, "Riemannsche Geometrie im Grossen" , Springer (1968)
[4] S.E. Cohn-Vossen, "Some problems of differential geometry in the large" , Moscow (1959) (In Russian)


Comments

Let $p\in M$. The manifold $M$ is called geodesically complete at $p$ if $\exp_p$ is defined on all of $T_pM$. The manifold $M$ is geodesically complete if this holds for all $p$. The Hopf–Rinow theorem also includes the statement that geodesic completeness is equivalent to geodesic completeness at one $p\in M$.

A geodesic joining $p$ and $q$ and of minimal length is called a minimizing geodesic.

References

[a1] W. Klingenberg, "Riemannian geometry" , de Gruyter (1982) (Translated from German)
How to Cite This Entry:
Hopf-Rinow theorem. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Hopf-Rinow_theorem&oldid=22591
This article was adapted from an original article by M.I. Voitsekhovskii (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article