6620HW1
6620HW1
6620HW1
Important info on HW: Homework is always due on Mondays, on paper at the beginning of class.
There are two types of problems. Type 1 are essential checks that you understand the material and no-
tation and you should complete all of them. If you are totally following the lectures and the reading and
understanding everything, they should not take much time. But, doing the reading and thinking about the
lectures does take much time, so don’t expect these to be totally effortless!
Type 2 questions are meant to make you curious. Each week, you should explain something you did on at
least one of these problems. Maybe you have a proof, a good starting point, or you explore a special case.
If you get really interested in a problem and do some outside reading, that’s great too. However, anything
you submit must be understandable by a student in our class. That means: use the notation from Lee, only
use concepts that we have defined so far or that are covered in undergraduate calculus or the core graduate
courses, etc. If you collaborate extensively, please indicate the other members of your group.
Reading this week: Lee, ch 2. You can skip the sections on pseudo-Riemannian and other generalizaions
Type 1 problems:
1 2 n 2 2
1. Consider the space {(x1 , x2 , . . . xn , y) ∈ Rn+1 : y > 0}, with metric ds2 = (dx ) +...+(dx
y2
) +dy
. We write
1 2 n
(x, y) for (x , x , . . . x , y) to be short.
(a) For any given x, prove that the shortest length path between (x, a) and (x, b) is a vertical line. Your
answer should involve integrals somewhere
(b) Prove that the distance from any point to the boundary plane y = 0 is infinite.
2. Let N and S denote the north and south poles of S 2 . Define a Riemannian metric on S 2 such that
d(N, S) > d(x, y) for all points x, y ∈ S 2 not equal to N, S. (It’s ok to give a proof without intensive
computation, as long as your argument is actually a proof). Now, modify your metric so that d(N, S) = 27.
3. Let S n (r) be the sphere of radius r in Rn+1 . After removing the one point from S n (r) where the last coor-
dinate is r, we have a stereographic projection map φ to Rn defined by φ(y1 , . . . yn+1 ) = r−y1n+1 (y1 , . . . yn ).
The round metric on the sphere is the induced metric from the Euclidean space Rn+1 . Compute the pull-
back of the round metric to Rn via φ.
Type 2 problems:
1. Suppose Γ1 and Γ2 are subgroups of translations of the Euclidean space Rn , each isomorphic to Zn . When
is Rn /Γ1 isometric to Rn /Γ2 ?
2. Given a Riemannian metric on M , we defined a distance d(p, q) as an infininum of lengths of paths. Show
that d(p, q) = 0 implies p = q. (Note: if M is not Hausdorff this fails – can you give a counterexample?
– so you should use Hausdorff somehow!)
3. Let S1 and S2 be closed surfaces in R3 . Let gi be the metric on Si induced from the Euclidean metric on
R3 . Suppose (S1 , g1 ) and (S2 , g2 ) are isometric. Is there necessarily an isometry of R3 taking S1 to S2 ?
Note: this is not true for closed curves in R3 , or even for closed curves in R2 (explain why?)
Some ideas: is it easier if you assume S1 is a round sphere? If you make other assumptions? Is “closed”
really needed here? ...
4. Let (M, g) be a Riemannian n-manifold. Show that, near any point p ∈ M , there is a local orthonormal
frame field (a set of n smooth vector fields, orthonormal at each point). (Lee gives a not very detailed
proof if you want hints)
Naively, you might try to solve this by making a coordinate chart such that the coordinate vector fields
are orthonormal. As part of your solution, show that this naive approach is doomed: there is no such
chart at any point of the round sphere. If a manifold does have a point p with such a chart, what can
you say about a neighborhood of p?