This weeks material is mostly conceptual, although the statement and proof in Pugh are all based on concrete formula.
Consider the generalized angular forms
Ωn−1 defined on
Rn\0
For
n=2, we define
Ω1=∣x∣−2(x1dx2−x2dx1)
For
n=3, we define
Ω2=∣x∣−3(x1dx2∧dx3−x2dx1∧dx3+x3dx1∧dx2)
Can you prove that
dΩ1=0,
dΩ2=0?
Can you write down the expression for the general
n? Or just prove the general case?
Consider the following 2-cell in
R3 (it parametrized the unit sphere),
γ:[0,1]2→R3,(s,t)↦(sin(πs)cos(2πt),sin(πs)sin(2πt),cosπs) What is
∫γΩ2?
Suppose we use a different parametrization of
S2,
the stereographic projection γ:R2↦R3,(a,b)↦(1+a2+b22a,1+a2+b22b,1+a2+b2−1+a2+b2) Can you explain why
∫γΩ2 is the same as the previous one?
(Optional) Let
γ1,γ2,[0,1]→R3 be two smooth loops, i.e.
γi(0)=γi(1) and
γi′(0)=γi′(1). Suppose they have disjoint images. Define a 2-cell
ϕ:[0,1]2→R3 by
ϕ(s,t)=γ1(s)−γ2(t). Prove that
(4π)−1∫ϕΩ2 is an integer (hence insensitive to small perturbation of
γ1,γ2). This is called the
linking number of two knots, and is a topological invariant of links. Can you compute some examples? Can you see its topological meaning?
Here is one example (maybe a bit degenerate):
imagine two loops with very large radiu, one is lying on the xy plane, one is lying on the xz plane. Say, γ1 is given by z=0,(x+R−1)2+y2=R2; and γ2 is given by y=0,(x−R+1)2+z2=R2. Can you compute the linking number? Can you picture what happens to the two circles if R→∞?
That wasn't very easy to compute. Here is a simpler one:
γ1 has the image of a loop
z=0,x2+y2=r2, with
r very small. And
γ2 is the circle
y=0,(x−R)2+z2=R2, with
R very large.
1. Read appendix F about Littlewood's three principles, and write some comments about it in your webpage (for example, a summary of what this is about, or questions)
2. Do Pugh Ex 83
3. Let (Rn,∣⋅∣1) be the normed vector space where ∣(x1,⋯,xn)∣1:=∑i∣xi∣. Let T:Rn→Rn be a linear operator, given by the matrix Tij, that sends (xi) to (yj), where yi=∑jTijxj. How to compute ∥T∥?
optional: if we use
∥−∥max norm on
Rn, how to compute the operator norm
∥T∥?
4. Read about Hölder inequality and Minkowski inequality. In the simplest setting, we have
(Hölder inequality), for
p,q≥1 that
1/q+1/p=1, we have
(i=1∑n∣xiyi∣)≤(i∑∣xi∣p)1/p(i∑∣yi∣q)1/q
(Minkowski inequality) for any
p≥1,
(i=1∑n∣xi+yi∣p)1/p≤(i∑∣xi∣p)1/p+(i∑∣yi∣p)1/p
Read about the proof (in wiki, or any textbook about functional analysis, say Folland). Why it works?