Scribd
Upload
2 views

sp_23_-_analysis

Uploaded by

sahlewel weldemichael
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
2 views

sp_23_-_analysis

Uploaded by

sahlewel weldemichael
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Analysis Qualifying Examination

Department of Mathematics and Statistics


University of Massachusetts

January 2023

This exam consists of eight equally weighted problems (ten points each): a
passing grade is 65% (52/80), including at least five "essentially correct"
problems (≈ 7.5/10).
Clearly show your work, explicitly stating or naming results that you use;
justify the use of named theorems by verifying the necessary conditions.
Please work legibly and clearly. Put your name at the top of this page.
Unless otherwise stated, the measure in every problem is the Lebesgue mea-
sure. Unless otherwise specified, the underlying space is ℝd .

1
1. Let f : ℝ → ℝ. Prove that f satisfies the Lipschitz condition

∣f (x) − f (y)∣ ≤ M ∣x − y∣

for some M and all x, y ∈ ℝ, if and only if f satisfies the following two
properties

(a) f is absolutely continuous.


(b) ∣f ' (x)∣ ≤ M for a.e. x.

2
2. Let T : H → H be a compact operator on a Hilbert space H, and let λ be
a nonzero number.

(a) Prove that that there is c > 0 such that ∥(T − λI)f ∥ ≥ c∥f ∥ for all f
that are orthogonal to Ker(T − λI). Hint: Argue by contradiction.
(b) Prove that the range of T − λI defined by

{g ∈ H : g = (T − λI)f, for some f ∈ H}

is closed. Hint: Suppose gj → g where gj = (T − λI)fj . Argue that


fj may be taken to be orthogonal to Ker(T − λI) and that under this
assumption {fj } is a bounded sequence. You may use part (a).
(c) Prove that the range of T − λI is all of H if and only if the kernel of
λI − T * is trivial. Hint: You may use part (c).

3
3. Suppose ν, ν1 , ν2 are signed measures on (M, F) and μ a positive measure
on (M, F). Prove:

(a) If ν1 ⊥ μ and ν2 ⊥ μ then ν1 + ν2 ⊥ μ.


(b) If ν1 ≪ μ and ν2 ≪ μ then ν1 + ν2 ≪ μ.
(c) ν1 ⊥ ν2 implies ∣ν1 ∣ ⊥ ∣ν2 ∣.
(d) ν ≪ ∣ν∣.
(e) If ν ⊥ μ and ν ≪ μ then ν = 0.

4
4. You may use the conclusion of part (a) in proving part (b).

(a) Let φ be a compactly supported smooth function supported in the ball


of radius one, and let φk (ξ) = φ(2k ξ). With F and F −1 denoting the
Fourier transform and the inverse Fourier transform respectively, for
any Schwartz function f let T (f ) := F −1 (φk F(f )). Prove that there
is a constant C, independent of f and k, such that for 1 ≤ p ≤ q < ∞
k( pd − dq )
∥T (f )∥Lq ≤ C2 ∥f ∥Lp .

Hint: You may use Young's inequality ∥f * g∥Lr ≤ ∥f ∥Lp ∥g∥Lq , when
1 ≤ p, r, q ≤ ∞ satisfy p1 + 1q = 1 + 1r .
(b) With the same notation as in the previous part prove that there is a
constant C, independent of f and k, such that for 1 ≤ p ≤ q < ∞
k( dp − dq )
∥T (f )∥Lq ≤ C2 ∥T (f )∥Lp .

Hint: Think of another function like φ but with slightly different sup-
port properties (applied to T on the Fourier side).

5
5. Prove that if f is integrable on ℝd and f is not identically zero, then (here
f * denotes the maximal function)
c
f * (x) ≥ , for some c > 0 and all ∣x∣ ≥ 1.
∣x∣d

Conclude that f * is not integrable on ℝd . Then prove that the weak type
estimate

m({x : f * (x) > α}) ≤ c/α


∫
for all α > 0 whenever ∣f ∣ = ∫1 is optimal in the following sense: if f is
supported in the unit ball with ∣f ∣ = 1then

m({x : f * (x) > α}) ≥ c' /α

for some c' > 0 and all sufficiently small α.

6
6. Let E be a subset of ℝd . Egorov's theorem states that if m(E) < ∞ and
f , {fk }k≥1 are measurable functions on E such that fk → f a.e. in E then
for every ε > 0 there exists a closed set Aϵ such that m(E \ Aϵ ) ≤ ε and
fk → f uniformly on Aϵ . The type of convergence involved in the conclusion
of Egorov's theorem is sometimes called almost uniform convergence. For
the questions below we assume that we are under the hypothesis of Egorov.

(a) Give an example to show that under the hypothesis of Egorov one
cannot conclude that fk → f uniformly on E \ Z where m(Z) = 0.
(b) Prove that if fk → f almost uniformly in E then fk → f a.e. in E.
(c) Prove that if fk → f almost uniformly in E then for every δ > 0 we
have that m({x : ∣fk (x) − f (x)∣ ≥ δ}) → 0 as k → ∞. Hint: Fix a
given δ > 0 and use the hypothesis to prove that for every ε > 0 there
exists a K such that for all k ≥ K, m({x : ∣fk (x) − f (x)∣ ≥ δ}) ≤ ε.

7
7. Suppose that {fn }n≥1 is a sequence of non-negative measurable functions
that converges in measure to f . Show that
∫ ∫
f dx ≤ lim inf fn (x)dx.
n

Hint: In the course of the proof you may use the fact that if a sequence of
measurable functions {gn } converges in measure to another measurable func-
tion g, then there is a subsequence {gnj } that converges almost everywhere
to g.

8
8. We say that the family of integrable functions Kδ : ℝd → ℝ, indexed by δ,
is an approximation to the identity if
∫
• ℝd Kδ (x)dx = 1.
• ∣Kδ (x)∣ ≤ Cδ −d for some C > 0 and all δ > 0.
• ∣Kδ (x)∣ ≤ Cδ/∣x∣d+1 for some C > 0 and all δ > 0 and all x ∈ ℝ3 .

(a) Prove that if {Kδ }δ > 0 is an approximation to the identity and f is


integrable of ℝ3 , then

(f * Kδ )(x) → f (x) as δ → 0

for every x in the Lebesgue set of∫ f . Hint: First show that for all x in
the Lebesgue set of f , A(r) = r1d ∣y∣≤r ∣f (x − y) − f (x)∣dy, defined for
r > 0, is continuous, bounded, and satisfies A(r) → 0 as r → 0.
1 y
(b) Verify that {Py (x)}y>0 defined by Py (x) = П x2 +y 2 , x ∈ ℝ, is an ap-
proximation to the identity.

You might also like