GENARAL ANALYSIS
(PhD Program Qualify Exam: Feb. 23, 2001 )
I. (15%) Given an∈ {0, 1, 2, 3, 4} and bn∈ {0, 1, 2, 3, 4, 5, 6} and define the set P by
P = n
(x, y)∈ [0, 1] × [0, 1]¯¯x = X∞ n=1
an
5n, y = X∞ n=1
bn
7n an∈ {0, 2} , bn∈ {1, 3, 5}o
What does P look like? Compute the Lebesgue measure of P . Is P open, closed, compact, perfect?
Can you evaluate the Hausdorff dimension of P ?
II. (15%) Given x = (x1, x2,· · · , xn)∈ Rn with|x| = (x21+ x22+· · · + x2n)1/2. Show that the function ρ(x) defined in Rn by ρ(x) = exp(|x|21−1) if|x| < 1 and ρ(x) = 0 if |x| > 1 belongs to Cc(Rn), the space of infinitely differentiable with compact support. Compute the integralR
Rnρ(x)dx.
III. (15%) If the function f (x) is absolutely continuous on [a, b], then the length s of the curve y = f (x) can be computed according to the formula
s = Z b
a
p1 +|f0|2(x)dx
(You need to start from the definition of arc-length!)
IV. (20%) Let Ω be a bounded domain in Rn, If u is a measurable function on Ω such that|u|p∈ L1(Ω) for some p∈ R, we define
Φp(u)≡
· 1
|Ω|
Z
Ω
|u|pdx
¸1/p
where|Ω| denotes the measure of Ω. Show that (a) lim
p→∞Φp(u) = sup
Ω
|u|;
(b) lim
p→−∞Φp(u) = inf
Ω |u|;
(c) lim
p→0Φp(u) = exp
· 1
|Ω|
Z
Ω
log|u|dx
¸ .
(d) Φ is logarithmically convex in 1/p, i.e. if p≤ q ≤ r and
1/q = λ/p + (1− λ)/r then log Φq ≤ λ log Φp+ (1− λ) log Φr
1
V. (15%) Give 1≤ p < ∞ and a sequence {fn}∞n=1⊂ Lp(Ω) and f∈ Lp(Ω).
(a) Can you define the weak convergence of fn to f in Lp(Ω) (denoted by fn* f ).w (b) For 1 < p <∞ show that fn
* f in Lw p(Ω) if and only if supnkfnkLp(Ω)<∞ and R
Efndx→R
Ef dx
for all bounded measurable set E ⊂ Ω. Is it true for p = 1?
VI. (20%) Let (X, µ) be a measure space, and let 1 ≤ p ≤ ∞ and C > 0. Suppose K is a measurable function on X× X such thatR
X|K(x, y)|dµ(y) ≤ C for all x ∈ X and R
X|K(x, y)|dµ(x) ≤ C for all y∈ X. Define the function T f by
T f (x)≡ Z
X
K(x, y)f (y)dµ(y) .
(a) Show that T f is well defined almost everywhere and is in Lp(X), andkT fkLp(X) ≤ CkfkLp(X). (Hint: H¨older inequality)
(b) Use (a) to show that if f ∈ L1(Rn) and g∈ Lp(Rn), 1≤ p ≤ ∞, then the convolution f ∗g ∈ Lp(Rn) andkf ∗ gkLp(Rn)≤ kfkL1(Rn)kgkLp(Rn)
(c) Let ρε(x)≡ ε−nρ(x/ε), ρ being the same as problem II, show that if f ∈ Lγ(Rn) (1≤ γ < ∞) then f∗ ρε→ f in Lγ(Rn) strongly.
2