Let be a Riemannian manifold. As before, one associates to it the curvature tensor
In the previous post, we saw a quantitative expression of how the curvature is a measure of the deviation from the flatness of . Given
, one can try to choose local coordinates around a point
which make the metric look like the euclidean metric to order 2 at
, i.e. local coordinates such that the coefficients near
are given by
However, we saw that the quadratic terms involve precisely the values of the curvature tensor at . Even in the best coordinates, one can’t generally make the coefficients of a metric look euclidean to order 3: the obstruction is precisely the curvature at
. Today, I’d like to describe the interpretation of curvature in terms of geodesics. Once again, the material is standard and can be found in introductory textbooks on Riemannian geometry.
1. Curvature and geodesic deviation
There’s another way to think of curvature, which also leads to this: curvature measures how nearby geodesics spread. To think about this, suppose we have a one-parameter family of geodesics in
, where
is the starting point of the variation. One then has a vector field
along the curve , which measures the infinitesimal “spreading” of the one-parameter family
. Now, a computation shows that
satisfies the equation
in other words that is a Jacobi field. Here
is covariant differentiation along the curve
.
This is a linear second-order ordinary differential equation, which determines in terms of
and the covariant derivative
. Once again, the values of the curvature
(now not just at
, but at nearby points) determine
and the spreading of the one-parameter family of geodesics
.
As an example, we could choose tangent vectors and consider the one-parameter family of geodesics
this determines a Jacobi field along the geodesic
with initial values
The values are given by the derivatives of the exponential map. That is,
and the preceding analysis shows that this is determined in terms of .
2. The Cartan-Hadamard theorem
As a result of the Jacobi equation, we saw above that the curvature tensor controls the infinitesimal spreading of a one-parameter family of geodesics . More precisely, though, the sign of the curvature can be used to make a qualitative assertion:
Principle: If the (sectional) curvature is negative, nearby geodesics diverge. If the (sectional) curvature is positive, then nearby geodesics converge (after a little while).
An example of this principle is the Cartan-Hadamard theorem.
Theorem 4 (Cartan-Hadamard) Let
be a complete Riemannian manifold of nonpositive curvature. Then the universal cover of
is diffeomorphic to
, so
is a
.
The idea behind the Cartan-Hadamard theorem is that if (as we can assume without loss of generality) is simply connected, then for any
, the exponential map
is a diffeomorphism.
To see this, note that the Hopf-Rinow theorem implies that the exponential map is surjective: that is, any point in the manifold can be joined to by a geodesic. To say that it is a diffeomorphism thus amounts to saying that
is nonsingular everywhere, as this implies (since the map is proper) that it is a covering map.
But to say that is nonsingular everywhere is precisely to say that nearby geodesics “diverge”: that is, for any
,
for . In other words, given a Jacobi field
along the geodesic
(coming from the infinitesimal variation of geodesics along
) with
but
, we have that
for
.
But in fact,
and this is positively proportional, by the Jacobi equation, to minus the sectional curvature of the plane spanned by . In particular, it is always nonnegative, and that means that since
, this derivative is always nonnegative and
just keeps growing in
. In other words, the infinitesimal geodesic deviation just keeps growing in time.
3. Magnification of distances
But one actually has a bit more:
Corollary 5 If
is simply connected and complete of nonpositive curvature, then the map
magnifies distances: that is, given tangent vectors
, one has
Proof: The idea is that
where the notation refers to the differential of the exponential map in the direction , at
: in other words, infinitesimally,
increases lengths. Since
is a diffeomorphism, this easily implies the result.
In order to see this, we observe that the previous formula
is a Jacobi field along the geodesic , and that it satisfies
. We saw previously that it is never zero after
. We now want to bound from below the growth rate. Let
. Then,
and for all
. By integrating twice, this implies that
for all . Since
, this provides the desired bound on differential of the exponential map.
Let’s now drop the assumption that is simply connected and complete. We can then use the above analysis to give a complete characterization of when
has negative curvature.
Theorem 6 A Riemannian manifold
has nonpositive curvature if and only if for every
, there is a neighborhood
of zero in
such that the exponential map is defined on
and
increases distances.
Geometrically, this means that given a small triangle in , the length of the third side will be larger than expected from euclidean space (in terms of the other two sides and the angle between them).
Proof: We’ve already more or less proved that if has negative curvature, then the exponential map increases distances. Now assume that the exponential map increases distances.
Choose orthonormal, and consider the Jacobi field
. We’ll look at the behavior of
for
small, which requires a short computation.
Namely, we know that , while
and
and
Therefore:
and
because by the Jacobi equation. Finally,
This follows by as one sees by differentiating the Jacobi equation. Observe that the sectional curvature through the plane spanned by
pops up.
Anyway, it follows that
If magnifies distances, then this is true infinitesimally: the norm of a Jacobi field has to grow more quickly than
, which implies by the above asymptotic formula that the sectional curvatures
are nonpositive.
January 3, 2013 at 12:47 am
Dear Akhil,
Probably you know all this, but just in case anyone else reading this might find it helpful:
When thinking about curvature of a Riemannian manifold, it’s also useful to note that curvature is essentially a property of surfaces. That is, the curvature tensor is determined by the sectional curvatures attached to all pairs of vectors X and Y, and a pair of vectors X and Y (at some point p, and of unit length, say) determine an infinitesimal piece of surface through p, whose curvature is then the sectional curvature of X and Y. So to understand curvature on a general Riemannian manifold, it suffices to understand the case of surfaces.
(This fits nicely with the curvature coming from 2nd order information. Those second order monomials precisely correspond to pairs of vector fields.)
In the case of a surface, one can then imagine that this surface is embeddded in Euclidean 3-space, and use the Gauss map point-of-view on curvature, which (for example) gives the curvature at a point as the product of the principal curvatures through that point. Although this is a non-intrinsic view-point, it can help with visualizing what different curvatures look like. In particular, thinking about geodesics passing through a saddle can help with visualizing the negative curvature case (just as thinking of geodesics on a sphere or ellipsoid helps with the positive curvature case).
Best wishes,
Matt
January 3, 2013 at 9:11 am
Dear Matt,
Thank you for these helpful comments.
By the way, to anyone else is reading this: the result about sectional curvatures being determined in terms of the curvatures of surfaces doesn’t seem to be stated explicitly in most introductory textbooks (unless I’ve missed it!), but it follows from the Gauss equations.