MATH 525a SAMPLE FINAL EXAM SOLUTIONS FALL 2013 Prof. Alexander

MATH 525a
SAMPLE FINAL EXAM SOLUTIONS
FALL 2013
Prof. Alexander
(1)(a) Clearly φ, X ∈ M. Now E ∈ M implies either (i) A ⊂ E or (ii) E ∩ A = φ which
implies either (i) A ∩ E c = φ or (ii) A ⊂ E c . Either way this shows E c ∈ M. Suppose
E1 , E2 , ... ∈ M. If all Ei ∩ A = φ, then (∪i Ei ) ∩ A = φ. If instead some Ej ⊃ A then
(∪i Ei ) ⊃ A. Either way, ∪i Ei ∈ M.
(b) f is measurable if and only if f is constant on A. Proof: If f = c on A, then for every
D ⊂ R, A ⊂ f −1 (D) if c ∈ D, and A ∩ f −1 (D) = φ if c ∈
/ D, so f is measurable. Conversely
if f takes any two distinct values c1 , c2 somewhere on A then f −1 ({c1 }) is not measurable so
f is not measurable.
(2)(a) Let > 0. Then {x : |f (x)| ≥ |g(x)| + } ⊂ {x : |f (x) − fn (x)| ≥ } for all n
so µ({x : |f (x)| ≥ |g(x)| + }) ≤ µ({x : |f (x) − fn (x)| ≥ }) → 0 as n → ∞. Hence
µ({x : |f (x)| ≥ |g(x)| + }) = 0 for all . Take = 1/k, let k → ∞ to get µ({x : |f (x)| >
|g(x)|}) = 0.
(b) We have (fn − f )2 → 0 in measure. Using (a) we have (fn − f )2 ≤ (2g)2 = 4g 2 , and
2
4g ∈ L1 , so the result follows from Exercise 34b in Chapter 2.
Pn
|F (xj ) − F (xj−1 )| ≥ TF (x) − . Then
(3) Choose −∞ P
< x0 < ... < xn = x with
1P
TF (x) − ≤ limk n1 |Fk (xj ) − Fk (xj−1 )| = lim inf k n1 |Fk (xj ) − Fk (xj−1 )| ≤ lim inf k TFk (x).
Since is arbitrary, this shows TF (c) ≤ lim inf TFk (x).
(4), (5) See Ch. 3 #12, and (II), in Assignment 8 solutions.
(6)(a) There are 3 things to establish, as follows.
(i) χφ = 0 so ν(φ) = φ(0) = 0 by linearity.
(ii) |ν(E)| ≤ ||φ|| ||χE ||1 = ||φ||µ(E)P< ∞, forPall E.
P∞
P∞
(iii) For E1 , E2 , ... disjoint we have || n1 χEj − ∞
n+1 µ(Ej ) →
n+1 χEj ||1 = P
1 χEj ||1 = ||
0 as n P
→ ∞, since µP
is finite. Hence
since φ is continuous, ν(∪j Ej ) = φ( ∞
1 χEj ) =
P
ν(E
).
limn φ( n1 χEj ) = limn n1 ν(Ej ) = ∞
j
1
Together, (i) – (iii) show ν is a signed measure.
(b) If E is µ-null then χE = 0 as an element of L1 , so ν(E) = φ(χE ) = φ(0) = 0.
(c) Proof of hint: ⇒ is clear. Conversely, if dν/dµ > M on a set E of positive µ-measure
then the second inequality
R
R dν in the Rhint is false for that E. This proves the hint.
Now for every E, E dµ
dµ = E dν = ν(E) ≤ ||φ|| ||χE ||1 = ||φ||µ(E) = E ||φ|| dµ. By
the hint this means dν/dµ ≤ ||φ|| a.e.(µ).
≥ −||φ|| a.e.(µ).
R dνSimilarlyR dν/dµ
dν
(d) For all E, φ(χE ) = ν(E) = E dµ dµ = χE dµ dµ so the statement is true for
characteristic functions. By linearity it is true for simple functions. For general f take
1
simple fn → f pointwise Rwith |fn | % |f |. By Dominated
Convergence,
(µ) so
R dν
R dν
Rfn →dνf in L
dν
dν
φ(f ) = limn φ(fn ) = limn fn dµ dµ (*). Now | fn dµ dµ − f dµ dµ| ≤ |fn dµ − f dµ | dµ ≤
R
R dν
||φ|| |fn − f | dµ → 0 so the limit (*) is f dµ
dµ.