Exercise Sheet 7 - Solutions
Exercise Sheet 7 - Solutions
Exercise 1:
(b) Let M be a smooth manifold. Show that there exists a smooth map f : M → [0, +∞)
that is proper.
[Hint: Use a function of the form f = +∞ +∞
P
i=1 ci ψi , where (ψi )i=1 is a partition of unity
and the ci ’s are real numbers.]
Solution:
(a) The fact that G is smooth follows immediately from part (b) of Exercise 4, Sheet 3,
and the fact that the differential of G at p ∈ N has the above form follows readily from
part (b) of Exercise 1, Sheet 4 and Exercise 3, Sheet 4.
(b) Let U = (Ui )i∈N be a countable basis for the topology of M such that U i is compact for
each i ∈ N, and let (ψi ) be a smooth partition of unity subordinate to U. Consider now
a sequence (ci )i∈N of non-negative real numbers satisfying limi→∞ ci = +∞ (for instance,
take ci = i) and define the smooth function
X
f : M → R, x 7→ ci ψi (x).
i∈N
1
We may view f (x) as a weighted average ofP the numbers ci , using the coefficients ψi (x)
as weights, which satisfy 0 ≤ ψi (x) ≤ 1 and i ψi (x) = 1 for every x ∈ M . In particular,
if Ix ⊆ N is the set of indices i such that Ui contains the point x ∈ M , then any upper or
lower bound for the numbers ci with i ∈ Ix is also an upper or lower bound for f (x). Thus,
if f (x) < c for some c > 0, then for any i ∈ N such that ψi (x) ̸= 0 (there are only finitely
many such indices by construction of a partition of unity) it holds that ci < c, so x is
contained in the union of the corresponding (first few) Ui ’s, since then x ∈ supp ψi ⊆ Ui .
We will now show that f is proper. Let K ⊆ R be a compact set. Take any number
c > 0 such that K ⊆ (−c, c) and pick an index ic ∈ N such that ci ≥ c for every i ≥ ic .
The preimage f −1 (K) consists S of points x ∈ M satisfying f (x) < c, and is therefore
−1
contained in the compact set i<ic U i . Since the set f (K) is closed, we conclude that
it is compact, as desired.
(c) By part (b) there exists a smooth proper function f : M → R. Consider now the map
G : M → N × R, x 7→ F (x), f (x) ,
which is smooth and whose differential has the form dG = dF, df by part (a). Since F
is injective, one immediately sees that G is also injective. Moreover, since F is a smooth
immersion, and thus its differential
dFp is injective at every point p ∈ M , it follows
readily that dGp = dFp , dfp is also injective at every point p ∈ M . Consequently, G is
an injective smooth immersion.
Next, we claim that G is a proper map. Given a compact subset K ⊆ N × R,
we will show that G−1 (K) is a compact subset of M . To this end, since N × R is a
Hausdorff space, K is in particular a closed subset of N × R, and since G is continuous,
the preimage G−1 (K) is a closed subset of M . Now, since the projection to the second
factor pr2 : N × R → R is continuous, the image pr2 (K) is a compact subset of R, and
since f is proper by assumption, the preimage f −1 pr2 (K) is a compact subset of M ,
which contains the closed set G−1 (K). Hence, G−1 (K) is a compact subset of M , as
claimed.
In conclusion, G is a smooth embedding by the above and by Proposition 4.6 (b), as
asserted.
Exercise 2:
(a) Show that the inclusion map ι : Sn ,→ Rn+1 is a smooth embedding.
(b) Consider the map
F : R → R2 , t 7→ (2 + tanh t) · (cos t, sin t).
(i) Show that F is an injective smooth immersion.
(ii) Show that F is a smooth embedding.
Solution:
(a) Consider the graph coordinates Ui± ∩ Sn , φ±
i for Sn ; see Example 1.8 (2). We have
shown in Example 2.7 that the inclusion map ι : Sn ,→ Rn+1 is smooth, because its
coordinate representation with respect to any of the graph coordinates is
p
bι(u1 , . . . , un ) = u1 , . . . , ui−1 , ± 1 − ∥u∥2 , ui , . . . , un ,
2
which is smooth on its domain, the unit ball Bn = {u = (u1 , . . . , un ) ∈ Rn | ∥u∥ < 1}.
ι = ι ◦ (φ±
The Jacobian matrix of the coordinate representation b i )
−1
of ι with respect to
the graph coordinates has the form
1 0 ... 0 0 0 ... 0 0
0 1 ... 0 0 0 ... 0 0
.. .. .. .. .. .. ..
. . . . . . .
0 0 ... 1 0 0 ... 0 0
∓u1 2 i−1 i i+1 n−1 n
√ √ ∓u ... √∓u √ ∓u √∓u ... √∓u √ ∓u 2 .
1−∥u∥2 1−∥u∥2 1−∥u∥2 1−∥u∥2 1−∥u∥2 1−∥u∥2 1−∥u∥
0 0 ... 0 0 1 ... 0 0
.. .. .. .. .. .. ..
. . . . . . .
0 0 ... 0 0 0 ... 1 0
0 0 ... 0 0 0 ... 0 1
∥F (t)∥ = 2 + tanh t, t ∈ R
is strictly increasing, which implies that F is injective. Finally, to show that F is a smooth
immersion, it suffices to show that F ′ (t) ̸= 0 for every t ∈ R. To this end, recall that
d 1
tanh t = , t ∈ R,
dt cosh2 t
so we have
′ 1 1
F (t) = −(2 + tanh t) sin t + cos t, (2 + tanh t) cos t + sin t , t ∈ R,
cosh2 t cosh2 t
and thus
1
∥F ′ (t)∥2 = (2 + tanh t)2 + > 0 for all t ∈ R,
cosh4 t
which implies that F ′ (t) ̸= 0 for every t ∈ R, as desired.
We now deal with (ii). Consider the open annulus
and note that F (t) ∈ U for every t ∈ R. (Incidentally, the image of F |[−4π,4π] has been
plotted below.)
3
Thus, F may be viewed as an injective smooth immersion F : R → U . Since the inclusion
map ι : U ,→ R2 is a smooth embedding, in view of part (a)(iii) of Exercise 1, Sheet 6 and
Proposition 4.6, to prove (ii), it suffices to show that F : R → U is a proper map. To this
end, given a compact subset K of U , we have to show that F −1 (K) is a compact subset
of R, or equivalently that it is closed and bounded. Since K ⊆ U is compact and U ⊆ R2
is Hausdorff, K is a closed subset of U , and since F is continuous, F −1 (K) is a closed
subset of R. Now, denote by m (resp. M ) the minimum (resp. the maximum) norm of the
points of K, and observe that [m, M ] ⊆ (1, 3). Denote also by ℓ (resp. L) the preimage
of m (resp. M ) under the strictly increasing function
and note that F −1 (K) ⊆ [ℓ, L], which shows that F −1 (K) is a bounded subset of R. This
finishes the proof of (ii).
and thus F |U1 is injective. Now, consider a precompact neighborhood U of p such that
U ⊆ U1 . The restriction of F to U 1 is an injective continuous map with compact domain
4
and Hausdorff codomain, so it is a topological embedding; see Claims 1 and 2 in the proof
of Proposition 4.6. Since any restriction of a topological embedding is again a topological
embedding, F |U is both a topological embedding and a smooth immersion, and hence F |U
is a smooth embedding.
Solution: Denote by IdN , respectively IdNe , the identity map of N , respectively N e , with
the smooth structure of N , respectively N e , on both the source and the target. Denote
also by IdN,Ne , respectively IdNe ,N , the identity map, where on the source, respectively on
the target, we put the smooth structure of N , and where on the target, respectively on
the source, we put the smooth structure of N e . In addition, denote by πN , respectively
πNe , the surjective smooth submersion with the smooth structure of N , respectively of N e,
on the target. Now, note that
IdN,Ne ◦πN = πNe ,
which is smooth, so by the assumption on N applied to P = N e and F = Id e we
N,N
conclude that IdN,Ne is smooth. On the other hand, we also have
IdNe ,N ◦πNe = πN ,
π : R2 → R, (x, y) 7→ xy.
Show that π is surjective and smooth, and that for each smooth manifold P , a map
F : R → P is smooth if and only if F ◦ π is smooth; but π is not a smooth submersion.
(Therefore, the converse of Theorem 4.12 is false.)
Solution: Both the smoothness and the surjectivity of π is clear. Therefore, if a map
F : R → P is smooth, then F ◦ π is also smooth by Exercise 3, Sheet 3. Now, assume that
we have a smooth manifold P and a map of sets F : R → P such that F ◦ π is smooth.
Consider the map
ι : R → R2 , x 7→ (x, 1),
which is clearly smooth and additionally satisfies π ◦ ι = IdR . Hence, the map
F = F ◦ IdR = (F ◦ π) ◦ ι
is smooth. Finally, note that the Jacobian of π is given by (y x), which vanishes at
(x, y) = 0, so π is not a smooth submersion.
5
Exercise 6 (Uniqueness of smooth quotients): Let π1 : M → N1 and π2 : M → N2 be
surjective smooth submersions that are constant on each other’s fibers. Show that there
exists a unique diffeomorphism F : N1 → N2 such that F ◦ π1 = π2 :
M
π1 π2
F
N1 N2
M
π1 π2
G1
N1 N2
By reversing now the roles of π1 and π2 , we see that there exists a unique smooth map
G2 : N2 → N1 such that G2 ◦ π2 = π1 :
M
π1 π2
N1 G2
N2
M
π1 π1
N1 N1
and observing that IdN1 ◦π1 = π1 , we deduce by (the uniqueness part of) Theorem 4.13
and (∗) that
G2 ◦ G1 = IdN1 .
Considering now the corresponding diagram for π2 and using (∗∗) instead, we infer simi-
larly that
G1 ◦ G2 = IdN2 .
Hence, F := G1 : N1 → N2 is a diffeomorphism such that F ◦ π1 = π2 , which is unique
(with this property) by construction.