SPRING 2011
JON AARONSON
Exercise 1, 23/2/2011
1. Semi-rings. Let X be a set and let S ` 2X be a collection of sets.
a) Show that if A, B > S
A 8 B, A B > S , then S is a ring.
b) Is it true that if A, B > S
A 8 B, A∆B > S , then S is a ring?
c) Let Γ ` R and for d C 1, let
d
Sd,Γ M ak , b k ak , b k > Γ ¦ 1 B k B d.
k 1
C "g>G g F .
1
2 Measure Theory
Exercise 2, 11/3/2009
1. Outer measures.
Let µ 2R 0, ª be Lebesgue outer measure defined by
ª ª
µE inf Q SIn S I1 , I2 , . . . intervals, E d In.
n 1 n 1
Let An ` 0, 1 n C 1.
b) Is it true that
ª
A1 ` A2 ` . . . An ` An1 ` . . . µAn µ An as n ?
ª
n 1
c) Is it true that
ª
A1 a A2 a . . . An a An1 a . . . µAn µ An as n ?
ª
n 1
Jon. Aaronson ©2000-2011 3
Exercise 3, 18/3/2009
1. Regularity.
Given a metric space X and a probability p B X 0, ª.
a) Show that p is regular in the sense that ¦ A > B X :
Show that
n 1
Exercise 4, 16/3/2011
3. Bernstein sets.
Let X be an uncountable, Polish space.
(a) Let p > P X, B X be non-atomic. Define p 2X 0, 1 by
pA inf pB A ` B > B X .
Show that if A ` X is a Bernstein set then pA pAc 1.
b) Show that there is a disjoint collection B of Bernstein sets so that
SBS c and "B >B B X.
4. Analytic sets.
Show that countable unions and intersections of analytic sets are
analytic.
Exercise No 5 23/3/2011
1. Riemann integrability I.
Let X, B , m 0, 1, B 0, 1, Lebesgue.
For a bounded function f X R show that the following are
equivalent:
(i) f is Riemann integrable;
(ii) if mX Cf 0 where
Cf x > X xn x f xn f x continuity points of f ;
r ¦ ε A 0,
(iii) r § f , f > C X with f B f B f and
X
f dm X f dm @ ε.
2. Borel-Cantelli lemma I.
Let X, B , m be a probability space, and let An > B , n C 1.
(i) Show that if PnC1 mAn @ ª, then Pª n 1 1An @ ª m-a.e.;
(ii) Exhibit a sequence An > B , n C 1 satisfying PnC1 mAn ª and
Pªn 1 1An @ ª m-a.e.;
(iii) Show that if An > B , n C 1 are independent in the sense that
mn>F An Ln>F mAn ¦ finite sets F ` N, and PnC1 mAn ª,
then Pª n 1 1An ª m-a.e..
8 Measure Theory
3. Convergence in measure.
Let X, B , m be a probability space, and suppose that fk X
R k C 1 are measurable functions. Recall that fn
m
0 means Ð
m Sfn S C ε Ð
0 ¦ ε A 0.
n ª
Ðm Q fk Ðm
n
1
fn 0 as n ª
n 0 as n ª ?
k 1
4. Non-decreasing re-arrangement.
Suppose that X, B , m is a probability space, and f X R is
measurable. Show that there is a non-decreasing function g 0, 1 R
such that m f A y λ g A y ¦ y > R where λ is Lebesgue measure
on I.
Exercise No 6 30/3/2011
Let X, B , m 0, 1, Borel, Leb. The rest of this exercise shows
§ a measurable function f X R for which ¨ g > B1 , f g m - a.e..
10 Measure Theory
4. Convolutions.
Let m denote Lebesgue measure on Rd .
(i) Show that if f, g Rd R are Lebesgue measurable, then so is the
function h Rd Rd R defined by hx, y f x y g y .
(ii) Prove that if f, g Rd 0, ª are measurable, then f g g f.
(iii) Show, using Hölder’s inequality (or otherwise), that if f > L1 Rd ,
and g > Lp Rd where 1 @ p @ ª, then Yf g Yp B Yf Y1 Yg Yp . When is
there equality?
r
(iv) Suppose that h Rd R is measurable, and suppose that Rd hxdx
1. Define, for t A 0, ht Rd R by ht x 1t h xt . Prove that for
1 B p @ ª,
Yf ht f Yp 0 ¦ f > Lp Rd . Ðt 0
5. Fourier transform on L1 Rd . For f > L1 Rd , define the Fourier
transform of f by
w
fÂt f xei`x,te dx t > Rd
Rd
g f Â
b) Show that fÅ Â g ¦ f, g > L1 Rd .
nsn
mJ a1 , . . . , an SJ ts n
1 t dt
14 Measure Theory
where
a1 , . . . , a n x1 , x2 , . . . > Ω xk ak 1 B k B n
and sn a1 an .
b) Show that
1
SI Ω An T 2 dm
6n
where T t, x t and An t, x n1 Pnk 1 xk .
c) Calculate E f SA for f I Ω 0, ª measurable, where A
I B Ω I A A > B Ω .
6. Martingale basics.
Suppose that X, B , m is a probability space. A filtration on X, B , m
is an increasing sequence of sub-σ-algebras Fn nC1 , Fn ` Fn1 Fª
σ nC1 Fn b B .
A sequence of measurable functions f0 , f1 , . . . is adapted to the filtra-
tion Fn nC0 (Fn nC0 -adapted) if fn is Fn -measurable ¦ n C 0.
Y A Fn nC0 -adapted sequence f1 , f2 , . . . with fn > L1 m ¦ n is a:
- Fn nC0 -martingale if E fn1 SFn fn ;
- Fn nC0 -submartingale if E fn1 SFn C fn .
Show that
(a) if f > L1 m and fn E f SFn , then fn nC0 is a Fn nC0 -martingale.
(b) if fn nC0 is a Fn nC0 -submartingale, then E fk SFj C fj ¦ k C j;
(c) if fn nC0 is a Fn nC0 -martingale, then Sfn SnC0 is a Fn nC0 - sub-
martingale.