MATHEMATICAL TRIPOS Part III
Monday, 1 June, 2009 1:30 pm to 4:30 pm
PAPER 9
METHODS IN ANALYSIS
Attempt no more than FOUR questions.
There are FIVE questions in total.
The questions carry equal weight.
STATIONERY REQUIREMENTS SPECIAL REQUIREMENTS
Cover sheet None
Treasury Tag
Script paper
You may not start to read the questions
printed on the subsequent pages until
instructed to do so by the Invigilator.
2
1 a) Let 2 6 p < ∞ and n ∈ N. Making use of the inequality
kf + gkpp + kf − gkpp 6 2p−1 kf kpp + kgkpp ,
valid for all f, g ∈ Lp (Rn ), prove the following statement.
Let K ⊂ Lp (Rn ) be closed and convex, and let f 6∈ K. Then there exists h ∈ K
such that
inf kg − f kp = kh − f kp . (1)
g∈K
b) Prove that if h satisfies (??), then
Z
Re |h(x) − f (x)|p−2 (h(x) − f (x)) (h(x) − g(x))dx 6 0 for all g ∈ K .
[You can make use of the fact that, if one defines N (t) = kf + tgkpp for f, g ∈ Lp (Rn ),
f 6= 0, and t ∈ R, then N (t) is differentiable in t and
Z
d
N (t)|t=0 = 2Re |f (x)|p−2 f (x) g(x).
dt
]
c) Let 2 6 p < ∞ and 1 < q 6 2 be such that 1/p + 1/q = 1. Using the results
from part a) and b), show that the q n p n ∗
q n
R map φ : L (R ) → (L (Rp )) n defined by φ(u) ≡ φu
for all u ∈ L (R ) and φu (v) = u(x)v(x) dx for all v ∈ L (R ) defines an isometric
isomorphism. [You have to prove that the map is well defined, that it is linear, that
kφu k(Lp )∗ = kukLq for every u ∈ Lq (Rn ) and that φ is surjective. You can make use of the
Hölder inequality without proving it.]
2 Suppose (Ω, Σ, µ) is a measure space.
a) For 1 6 p 6 ∞, prove the completeness of Lp (Ω, dµ). [You may use the fact that
k.kp is a norm as well as the monotone and the dominated convergence theorems without
proof.]
b) Suppose that 1 6 p < q < r 6 ∞. Prove that, for every Cp < ∞, Cr < ∞,
and Cq > 0, there exist ǫ > 0 and M > 0 such that µ({x : |f (x)| > ǫ}) > M for all
f ∈ Lp (Ω, dµ) ∩ Lr (Ω, dµ) such that kf kp 6 Cp , kf kq > Cq , and kf kr 6 Cr .
c) Find examples to show that the conclusion of b) does not necessarily hold true if
the condition kf kp 6 Cp or the condition kf kr 6 Cr is removed. [Hint: you can take Ω = R
and µ to be Lebesgue’s measure.]
Part III, Paper 9
3
3 a) Assume that n > 3. Suppose that f, fj ∈ H 1 (Rn ) for every j ∈ N are such that
fj → f weakly in H 1 (Rn ) as j → ∞. Suppose that Ω ⊂ Rn is a bounded set and that χΩ
denotes the characteristic function of Ω. Show that χΩ fj → χΩ f strongly in Lq (Rn ), for
all q < 2n/(n − 2).
b) State and prove the Poincaré inequality for f ∈ W 1,p (Ω), where Ω ⊂ Rn is a
bounded, connected, open set having the cone-property, and where p < n. [In the proof
you can make use of the Rellich-Kondrashov Theorem on general sets having the cone-
property.]
4 a) Suppose that Ω ⊂ Rn is open. Suppose that Tn ∈ D ′ (Ω) for all n ∈ N, and
T ∈ D ′ (Ω). What does Tn → T in D ′ (Ω) mean? How is the distribution ∂xj T defined?
Show that the derivative of a distribution is a distribution. Show that Tn → T implies
that ∂xj Tn → ∂xj T in D ′ (Ω).
b) Explain the meaning of the equation
1
−∆ = 4πδ in D ′ (R3 ) . (1)
|x|
Prove (??).
c) Consider the sequence of functions on R2 :
cn (1 − |x|4 )n if |x| 6 1
gn (x) :=
0 if |x| > 1
with Z
c−1
n = dxdy (1 − |x|4 )n
|x|61
Prove that gn → δ in D ′ (R2 ). [Hint: it may be useful to start by finding an upper bound for
the constants cn .]
Part III, Paper 9 [TURN OVER
4
5 a) Suppose that Ω ⊂ Rn is open, and let f ∈ L1loc (Ω) be a real valued function.
What does it mean for f to be subharmonic on Ω? What does it mean for f to be
superharmonic on Ω? What does it mean for f to be harmonic on Ω?
b) Show that, if fn is a sequence of subharmonic functions on Ω, then g(x) =
supn>1 fn (x) is also subharmonic.
If f ∈ L1loc (Ω) is subharmonic, we proved in class that there exists a unique function
fe : Ω → R ∪ {−∞} such that f (x) = fe(x) for a.e. x ∈ Ω, fe is upper semicontinuous, and
fe is subharmonic for all x ∈ Ω. This function then satisfies the mean-value inequality for
all x ∈ Ω. You can make use of these facts to answer the two following questions.
c) Show that if f ∈ L1loc (Ω) is harmonic on Ω and f = fe, then f ∈ C ∞ (Ω).
d) State and prove the strong maximum principle.
END OF PAPER
Part III, Paper 9
