§1 - Smooth Functions on a Euclidean Space
1.1. A function that is C2 but not C3
Let g:R→R be the function in Example 1.2(iii). Show that the function h(x)=∫0xg(t)dt is C2 but not C3 at x=0.
Solution
First, recall that g:R→R is defined as
g(x)=43x4/3
So
g′(x)=x1/3
And
g′′(x)={31x−2/3undefinedfor x=0,for x=0.
By the fundamental theorem of calculus h′(x)=g(x), so this shows that h is C2 but not C3 at x=0 (as h′′′=g′′ ). Note that h is a polynomial and therefore continuous.
1.2.* A C∞ function very flat at 0
Let f(x) be the function on R defined in Example 1.3.
(a)
Show by induction that for x>0 and k≥0, the k-th derivative f(k)(x) is of the form p2k(1/x)e−1/x for some polynomial p2k(y) of degree 2k in y.
Solution
First recall that f(x) is defined on R by
f(x)={e−1/x0for x>0,for x≤0
The base case of the induction is for k=0. By definition when x>0 we have f(x)=e−1/x=1⋅e−1/x, proving the hypothesis via the polynomial p0(y)=1 which is of degree 2k=0.
Now assume that f(k)(x)=p2k(1/x)e−1/x. We need to show that f(k+1)(x)=p2k+2(1/x)e−1/x for some polynomial p2k+2 of degree 2k+2. Proceed by differentiating
f(k+1)(x)=dxd(p2k(x1)e−1/x)=dxd(p2k(x1))e−1/x+p2k(x1)dxde−1/x=p2k′(x1)⋅(−x21)e−1/x+p2k(x1)⋅(−x21)e−1/x=−x21(p2k′(x1)+p2k(x1))e−1/x
Since p2k is of degree 2k, p2k′ is of degree 2k−1, so (p2k′(x1)+p2k(x1)) is of degree 2k, and p2k+2=−x21(p2k′(x1)+p2k(x1)) is of degree 2k+2.
(b)
Prove that f is C∞ on R and that f(k)(0)=0 for all k≥0.
Solution
To prove that f is C∞ on R it is sufficient to show that it is smooth at x=0, as we handled that case x>0 in part (a) and the case x<0 is trivial as we have a whole epsilon ball where f=0. Take k≥0, we need to show that the limit
h→0limhf(k)(h)−f(k)(0)
exists, and for the second part of the question we need to show that it equals to zero. We proceed by induction. For the base case k=0 we have that
h→0limhf(k)(h)−f(k)(0)=h→0limhf(h)−f(0)
This splits into two cases: approaching from the left and from the right. From the left we always have h<0 so
h→0−limhf(h)−f(0)=h→0−limh0−0=0
From the right we always have h>0 so
h→0+limhf(h)−f(0)=h→0+limhe−1/h−0
Let y=1/h so as h→0+ we have y→∞
h→0+limhe−1/h−0=y→∞limye−y=y→∞limeyy=0
The last equality follows from L'Hospital's Rule. Now it's time for the induction. Assume that
h→0limhf(k)(h)−f(k)(0)=0
We need to show that (note that the following limit is not even assumed to exist)
h→0limhf(k+1)(h)−f(k+1)(0)=0
From the assumption it is clear that f(k+1)(0)=0 so all that remains to show is that
h→0limhf(k+1)(h)=0
If h approaches from the left than this is obvious as the left side of the function is identically zero (since the left side of f is zero so are all of its derivatives). So without loss of generality approach from the right, and use part (a)
h→0+limhf(k+1)(h)=h→0+limhp2k+2(1/h)e−1/h=h→0+lime1/hp2k+3(1/h)
Once again substitute y=1/h
h→0+limhf(k+1)(h)=y→∞limeyp2k+3(y)=0
The last equality following from repeated applications of L'Hospital's Rule.