You are on page 1of 3

Rudin’s Real and Complex Analysis

Jia-Ming Liou (Frank)


Department of Mathematics, University of California, Davis.
E-mail: frankliou@math.ucdavis.edu
Website: http://www.math.ucdavis.edu/∼frankliou/

February 18, 2008

1 Abstract Integration
R
1. Suppose µ is a positive measure on X, f : X → [0, ∞] is measurable, X f dµ = c, where 0 < c < ∞
and α is a constant. Prove that

Z  
f (x)
α   ∞, if 0 < α < 1;
lim n log 1 + dµ = c, if α = 1;
n→∞ X n
0, if 1 < α < ∞.

Proof. It is not difficult to see that for each n ∈ N


  α 
f (x)
n log 1 + ≥ 0.
n

If α ≥ 1, we have by the elementary Calculus technique,


  α  
f (x) f (x), if α = 1;
n log 1 + ≤ f (x)α
n nα−1 , if α > 1.

Since f ∈ L1 (µ), we find by Lebesgue dominated convergence theorem,


Z   α  Z
f (x)
lim n log 1 + dµ = f (x)dµ.
n→∞ X n X

Since f ∈ L1 (µ), then F = {x ∈ X : f (x) = ∞} has measure zero. For each x 6∈ F , and α > 1,
α 
f (x)α
 
f (x)
0 ≤ n log 1 + ≤ α−1 .
n n
f (x)α
Since limn→∞ nα−1 = 0, then by Sandwich principle, we see that α > 1,
  α 
f (x)
lim n log 1 + = 0.
n→∞ n

Therefore Fatou’s lemma implies


Z   α 
f (x)
0 ≤ lim inf n log 1 + dµ
n→∞ X n
Z   α 
f (x)
≤ lim sup n log 1 + dµ
n→∞ X n
Z   α 
f (x)
≤ lim sup n log 1 + dµ
X n→∞ n
= 0.

1
Therefore Z   α 
f (x)
lim n log 1 + dµ = 0.
n→∞ X n
If 0 < α < 1, then by L Hospital Rule we see that
  α 
f (x)
lim n log 1 + = ∞.
n→∞ n

Hence Fatous Lemma implies


Z   α  Z   α 
f (x) f (x)
∞= lim inf n log 1 + dµ ≤ lim inf n log 1 + dµ
X n→∞ n n→∞ X n
Z   α 
f (x)
≤ lim sup n log 1 + dµ.
n→∞ X n

Then Z   α 
f (x)
lim n log 1 + dµ = ∞.
n→∞ X n

2 Elementary Hilbert Space


1. Compute Z 1
min |x3 − a − bx − cx2 |dx
a,b,c −1

and find Z 1
max x3 g(x)dx
−1

where g is subject to the restrictions


Z 1 Z 1 Z 1 Z 1
2
g(x)dx = xg(x)dx = x g(x)dx = 0; |g(x)|2 dx = 1.
−1 −1 −1 −1

Solution. L2 ([−1, 1]) is a Hilbert space and the set of polynomials {Pn (x)}, defined by

1 dn 2
Pn (x) = (x − 1)n
2n n! dxn
forms an orthogonal basis. To find the minimum of the integral is to find the distance of f (x) = x3
to span{1, x, x2 }. Since span{1, x, x2 } = span{P0 (x), P1 (x), P2 (x)}, then the problem is equivalent to
find Z 1
min |x3 − a0 P0 (x) − a1 P1 (x) − a2 P2 (x)|2 dx.
a0 ,a1 ,a2 −1

Write f in the linear combination of {Pn }, assume

x3 = c0 P0 (x) + c1 P1 (x) + c2 P2 (x) + c3 P3 (x),

where
3 2 1 5 3
x − , P3 (x) = x3 − x.
P0 (x) = 1, P1 (x) = x, P2 (x) =
2 2 2 2
Hence we find c3 = 2/5, c2 = 0, c1 = 3/5, c0 = 1, which implies that applying the Parserval identity
Z 1 2
3 2
|x3 − a0 P0 (x) − a1 P1 (x) − a2 P2 (x)|2 dx = |a0 |2 kP0 k22 + a1 − kP1 k22 + |a2 |2 kP2 k22 + kP3 k22

−1 5 5
4
≥ · kP3 k22 .
25

2
Since
2
kPn k22 = ,
2n + 1
then we find kP3 k22 = 2/7. Thus
Z 1
8
min |x3 − a0 P0 (x) − a1 P1 (x) − a2 P2 (x)|2 dx ≥ .
a0 ,a1 ,a2 −1 175

Note that when a0 = c0 , a1 = c1 , a2 = c2 ,


Z 1
8
|x3 − a0 P0 (x) − a1 P1 (x) − a2 P2 (x)|2 dx = ,
−1 175

which implies that


Z 1
8
min |x3 − a0 P0 (x) − a1 P1 (x) − a2 P2 (x)|2 dx = .
a0 ,a1 ,a2 −1 175

Since Z 1 Z 1 Z 1
g(x)dx = xg(x)dx = x2 g(x)dx = 0,
−1 −1 −1

then g ∈ span{1, x, x2 }⊥ = span{P0 (x), P1 (x), P2 (x)}⊥ , we may assume that



X
g(x) = bn Pn (x).
n=3

Hence Z 1
2
x3 g(x)dx = b3 .
−1 5
Since

X
kgk22 = b2n kPn k22 = 1,
n=3

we find b2n ≤ 1/kPn k22 for any n ≥ 3. Therefore


Z 1
r √
3 2 1 2 7 14
x g(x)dx ≤ · = · = .
−1 5 kP3 k2 5 2 5

Hence √
Z 1
14
max x3 g(x)dx ≤ .
−1 5
q
7
If we take g(x) = 2 P3 (x), then g satisfies the condition and
Z 1

3 14
x g(x)dx = .
−1 5

This shows that √


Z 1
3 14
max x g(x)dx = .
−1 5

You might also like