Pollack-Cap1 1
Pollack-Cap1 1
Roy Howie
Exercise 1
Let k < l and let f be a smooth function on Rk . Let F be a function on Rl such that F (a1 , · · · , ak , 0, · · · , 0) =
f (a1 , · · · , ak ). Consider π : Rk → Rk defined by x 7→ x and note that F = f ◦π. Hence, F is the composition
of smooth functions and is therefore smooth itself.
On the other hand, suppose F is smooth on {(a1 , · · · , ak , 0, · · · , 0)}. Let i : Rk → Rl be the smooth map
defined by (a1 , · · · , ak ) 7→ (a1 , · · · , ak , 0, · · · , 0) and let f be a function on Rk such that f (a1 , · · · , ak ) =
F (a1 , · · · , ak , 0, · · · , 0). Then F is smooth, as it is the composition of the smooth maps f and i.
Exercise 2
Exercise 4
Note that f −1 (0) = 0 and limy→∞ f −1 (y) = a. That is, f −1 maps [0, ∞)k to [0, a)k , which makes
intuitive sense as f did the opposite.
b Since X is a manifold, for every x ∈ X, there is a parameterization p : U → O(x) where U ⊂ Rk
and O(x) is an open neighborhood of x. But U can be the ball Ba of radius a, as there is always one
small enough inside of U such that x ∈ p(Ba ) ⊂ V . So consider f −1 ◦ p restricted to Ba , which is a
parameterization of an open neighborhood of x with all of Rk as its domain.
Exercise 6
Let h(x) = x1/3 . Note that f ◦ h = h ◦ f = id and that h0 (x) = 31 x−2/3 . However limx→0 h(x) does not exist,
so h is not smooth and f is not a diffeomorphism.
1
Exercise 8
Let a > 0 and let H be the hyperboloid {(x, y, z) | x2 + y 2 − z 2 = a}. Let Ba be the ball of radius a centered
at theporigin. The upper half of H can then be parameterized via φ : R2 − Ba → R3 defined p by (x, y) 7→
(x, y, x2 + y 2 − a). Similarly, the lower half of H can be parameterized by (x, y) 7→ (x, y, − x2 + y 2 − a).
Intuitively speaking, this involves lifting the plane minus Ba so that it “covers” the given half of H.
When a = 0, the point (0, 0, 0) becomes a problem. Removing the origin from R2 leaves one component,
whereas removing the origin from H leaves two components, so H is not a manifold.
Exercise 12
Let N = (0, 0, 1) and let p be a point on S 2 . The line through points N and p then has the equation
l(t) = (0 + t(x − 0), 0 + t(y − 0), 1 + t(z − 1))
= (tx, ty, 1 + t(z − 1))
1
This line hits the xy-plane when z = 0, or when 1 + t(z − 1) = 0, implying t = 1−z . Hence,
x y
π(x, y, z) = ( , )
1−z 1−z
To find π −1 , note that
π −1 (0, 0) = −N
π −1 (1, 0) = (1, 0, 0)
π −1 (0, 1) = (0, 1, 0)
and ||(x, y)|| = 1 ⇐⇒ z = 0. I couldn’t think of a function z = f (x, y) which satisfied these conditions, but
2 2
−1
google gave me z = xx2 +y
+y 2 +1 , which definitely works. This makes finding π
−1
easy:
2x 2y x2 + y 2 − 1
π −1 (x, y) = ( , , )
1 + x + y 1 + x + y 1 + x2 + y 2
2 2 2 2
Exercise 14
Let (x, y) ∈ X × Y and let U × V be an open neighborhood of (x, y) such that F is smooth on U , G is smooth
on V , F restricted to U ∩ X equals f , and G restricted to V ∩ Y equals g. Note that (U × V ) ∩ (X × Y ) =
(U ∩X)×(V ∩Y ). Hence, since F ×G is smooth on (U ∩X)×(V ∩Y ), it is also smooth on (U ×V )∩(X ×Y ),
so f × g is too.
Exercise 18
2
a From class, we had that f (n) (x) = Pn e−1/x , where Pn is a polynomial of order n or less. Thus
limx→0 f (n) (x) = 0 for all n ∈ N, so f is smooth.
b Subtraction and (x, y) 7→ xy are smooth functions, so g, the composition
R ∞of smooth functions, is too.
Since g is smooth and positive function on (a, b), we have that c = −∞ g dx is nonzero. Hence,
Rx
h(x) = 1c −∞ g dx = 1c G(x) by the Second Fundamental Theorem of Calculus. As G0 = g and g was
smooth, h must be too, with h(n) = 1c G(n) for all n ∈ N.
c Consider the function r(x) = 1 − h(||x||).