Professional Documents
Culture Documents
GSM 143 Fa Solutions
GSM 143 Fa Solutions
GSM 143 Fa Solutions
Alberto Bressan
1
`1 , `p , `∞ spaces of sequences of real (or complex) numbers.
L1 (Ω), Lp (Ω), L∞ (Ω) Lebesgue spaces.
W k,p (Ω) Sobolev space of functions whose weak partial derivatives up to order k lie in Lp (Ω),
for some open set Ω ⊆ Rn .
H k (Ω) = W k,2 (Ω) Hilbert-Sobolev space.
C k,γ (Ω) Hölder space of functions u : Ω 7→ R whose derivatives up to order k are Hölder
continuous with exponent γ ∈ ]0, 1].
k · k = k · kX norm on a vector space X.
(·, ·) = (·, ·)H inner product on a Hilbert space H.
X ∗ is the dual space of X, i.e. the space of all continuous linear functionals x∗ : X 7→ K.
hx∗ , xi = x∗ (x) duality product of x∗ ∈ X ∗ and x ∈ X.
xn → x strong convergence in norm; this means kxn − xk → 0.
xn * x weak convergence.
∗
ϕn * ϕ weak-star convergence.
f ∗g convolution of two functions f, g : Rn 7→ R.
∇u = (ux1 , ux2 , . . . , uxn ) gradient of a function u : Rn 7→ R.
α1 α2 αn
∂ ∂ ∂
Dα = ∂x1 ∂x2 · · · ∂xn = ∂xα11 ∂xα22 · · · ∂xαnn a partial differential operator of
.
order |α| = α1 + α2 + · · · + αn .
meas(Ω) the Lebesgue measure of a set Ω ⊂ Rn .
Z Z
. 1
− f dx = f dx the average value of f over the set Ω.
Ω meas(Ω) Ω
2
Solutions to Homework Problems
Chapter 2
Then the sequence of continuous functions (fn )n≥1 is a Cauchy sequence without any limit in
1
the space X. Namely, kf − fn kL1 ([0,1]) = 2n → 0 as n → ∞, but f ∈
/ X.
(vi) For p < 1 this is not a norm, because (N3) fails. For example, in R2 take x = (1, 0),
y = (0, 1). Then kxk = kyk = 1, but kx + yk = 21/p > 2.
3
3. (i) Let (x̄, ȳ) ∈ X × X and ε > 0 be given. Choose δ = ε/2. If k(x, y) − (x̄, ȳ)k ≤ δ then
This proves that the map (x, y) 7→ x + y is continuous at the point (x̄, ȳ).
(ii) Let α ∈ K, x ∈ X, and ε > 0 and be given. Choose
. ε/2 ε/2
δ = min 1, , .
1 + kxk 1 + |α|
Assume |β − α| ≤ δ, ky − xk ≤ δ. Then kyk ≤ 1 + kxk and
ε ε
kβy − αxk ≤ kβy − αyk + kαy − αxk ≤ |β − α| (1 + kxk) + |α| ky − xk ≤ + = ε.
2 2
4. If V is a subspace of X it is clear that the properties (N1)–(N3) of the norm remain valid.
Now assume that X is a Banach space and that V is a closed subset of X. If (vn )n≥1 is a
Cauchy sequence in V , then it is also a Cauchy sequence in X. Since X is complete, one has
the convergence xn → x for some point x ∈ X. Since V is closed, x ∈ V , proving that V is
complete as well.
P
5. Let X be a Banach space and assume that n≥1 kxn k < ∞. Consider the sequence of
. Pn
partial sums yn = k=1 xk . This is a Cauchy sequence. Indeed, for m < n we have
X
X
kym − yn k =
xk
≤ kxk k → 0 as m, n → ∞.
m<k≤n
m<k≤n
Viceversa, assume that every absolutely convergent series of elements of X has a sum. Let
(yn )n≥1 be a Cauchy sequence of elements in X. For each integer k ≥ 1, choose an index Nk
such that
kym − yn k ≤ 2−k whenever m, n ≥ Nk .
P
It is not restrictive to assume N1 < N2 < N3 < · · · . Consider the series k xk , where
6. (i) Let x0 , x00 ∈ coA, and assume λ ∈ [0, 1]. We need to prove that x = λx0 + (1 − λ)x00 ∈ A.
By assumption we have
N 0 N00
X X
0
x = θk0 a0k , 00
x = θk00 a00k .
k=1 k=1
4
Choosing
λθk0 if k = 1, . . . , N 0 ,
0 00
N =N +N , θk = 00
(1 − λ)θk−N 0 if k = N 0 + 1, . . . , N 0 + N 00 ,
(ii) Let x = N
P
k=1 θk ak ∈ coA and let K be any convex set containing A. Then K contains
all elements a1 , . . . , aN . Being convex, K must also contain all convex combinations of these
elements. In particular, x ∈ K.
The converse inclusion is trivial: since coA is a convex set containing A, the intersection of all
convex sets that contain A cannot be larger than coA.
n N
X o
B(x, θ1 r) = θ1 y + θk ak ; y ∈ B(a1 , r) ⊂ coA
k=2
This shows that x lies in the interior of coA, hence coA is open.
(ii) Let A be bounded, say A ⊆ B(0, R) for some radius R > 0. Then coA ⊆ B(0, R) because
B(0, R) is a bounded convex set containing A. Hence coA is bounded.
(iii) If A ⊆ B(0, r1 ) and B ⊆ B(0, r2 ), then A + B ⊆ B(0, r1 + r2 ), showing that A + B is
bounded.
(iv) Let (xn )n≥1 be a sequence of points in A + B, with xn → x as n → ∞. We need to prove
that x ∈ A + B. By assumption xn = an + bn , for some an ∈ A and bn ∈ B. Since B is
compact, by possibly choosing a subsequence and relabeling we can assume bn → b for some
b ∈ B. We now have
lim sup kam − an k = lim sup kxm − bm − (xn − bn )k ≤ lim sup kxm − xn k + kbm − bn k = 0.
m,n→∞ m,n→∞ m,n→∞
Indeed, the sequences (xn )n≥1 and (bn )n≥1 are both convergent, hence they are Cauchy se-
quences. This shows that the sequence (an )n≥1 is Cauchy as well. Hence there exists the limit
an → a for some a ∈ A, because X is a Banach space and A ⊆ X is closed. We conclude that
x = a + b ∈ A + B, proving that the sum A + B is closed.
(v) In R2 , consider the two closed subsets
. .
A = {(x, y) ; x > 0, y > 0, xy ≥ 1} , B = {(x, y) ; x < 0, y > 0, xy ≤ −1} .
5
(vii) Assume that A is closed and A + A = 2A. Let a, b ∈ A. By assumption there exists
c ∈ A such that a + b = 2c. This implies a+b
2 ∈ A. Repeating the argument, we obtain that
1 a+b 1 1 3 1 1 a+b 3 1
· + · b = a + b ∈ A, ·a+ · = a + b ∈ A.
2 2 2 4 4 2 2 2 4 4
By induction, we obtain
k k
a+ 1− n b ∈ A for all k = 0, 1, . . . 2n . (1)
2n 2
Since the set of all convex combinations of the form (1) are dense in the set of all convex
combinations, and we are assuming that A is closed, we conclude that A is convex.
Notice that the result would fail if A were not closed. For example, let A be the set of all
rational numbers contained inside [0, 1].
8. (i) Let S be convex. Let x, y ∈ S and let θ ∈ [0, 1]. We need to show that θx + (1 − θ)y ∈ S.
By assumption, there exist sequences xn → x, yn → y, with xn , yn ∈ S for every n ≥ 1. By
convexity, θxn + (1 − θ)yn ∈ S. We now have the convergence θxn + (1 − θ)yn → θx + (1 − θ)y,
showing that S is convex.
(ii) Let S be symmetric and assume x ∈ S. Then there exists a sequence xn → x , with xn ∈ S
for every n ≥ 1. Therefore −xn ∈ S and −xn → −x. This shows that −x ∈ S, hence S is
symmetric.
9. (i) Let S be convex. Assume y1 , y2 ∈ Λ(S) and θ ∈ [0, 1]. Then there exist x1 , x2 ∈ S such
that y1 = Λ(x1 ), y2 = Λx2 . Since Λ is linear, this implies
because S is convex.
(ii) Let S be symmetric. Assume y ∈ Λ(S). Then y = Λ(x) for some x ∈ S. This implies
−y = Λ(−x) ∈ Λ(S) because S is symmetric.
10. If Λ is not continuous, then it is not bounded. Hence one can find a sequence of points
.
xn ∈ X with kyn k ≤ 1 but kΛ(yn )k ≥ n for every n ≥ 1. Defining the sequence xn = n−1/2 yn
we have xn → 0 but kΛ(xn )k ≥ n1/2 → ∞, against the assumption.
11. (i) If S is a finite set, then span(S) is a finite dimensional space. By Theorem 2.20, it is
complete. Being complete, it must be a closed subset of X.
.
(ii) Since X is a Banach space, the closure Y = span(S) is complete, hence it is a Banach space
as well. For each n ≥ 1, consider the finite dimensional subspace Vn = span{v1 , . . . , vn } ⊂ Y .
Being finite dimensional, this subspace is complete, hence it must be closed. It is easy to
see that Vn has empty interior. Indeed, since by assumptions the vectors {v1 , . . . , vN +1 } are
1
linearly independent, for every y ∈ Vn the sequence y + m vn+1 does not lie in Vn and converges
to y as m → ∞.
6
According to Baire’s category theorem, the complement
[ \
Y \ Vn = (Y \ Vn )
n≥1 n≥1
is everywhere dense in Y . Observing that span(S) = ∪n≥1 Vn , this shows that span(S) 6= Y =
span(S).
13. If X is finite dimensional, let {e1 , . . . , en } be a basis. Then every linear functional is
continuous and is a linear combination of the n linear functionals e∗1 , . . . , e∗n where e∗i (ej ) = δij .
On the other hand, if X is infinite dimensional, let {e1 , e2 , . . .} be a countable family of
linearly independent vectors. For each n, consider the subspace Vn = span{e1 , . . . , en }. Let
ϕn : Vn 7→ R be the linear functional such that
ϕn (en ) = 1 , ϕn (ek ) = 0 for all k < n .
Since Vn is finite dimensional, ϕn is continuous. By Hahn-Banach, it can be extended to a
continuous linear functional ϕn : X 7→ R.
It remains to check that the countable set of functionals {ϕn ; n ≥ 1} is linearly independent.
If
N
X
ak ϕk (x) = 0 for all x ∈ X ,
k=1
let j be the smallest index such that aj 6= 0. Then
N
X
0 = ak ϕk (ej ) = aj ,
k=1
reaching a contradiction.
Hence x ∈ `∞
and the embedding `p ⊆ `∞ is bounded, with operator norm 1. Moreover, if
p < q < ∞, then
∞
!1/q ∞
!1/q ∞
!1/q
kxkq−p
X X X
q p q−p p
kxk`q = |xn | = |xn | |xn | ≤ |xn | · `∞
n=1 n=1 n=1
7
15. Consider first the case 1 ≤ p < ∞. Let x = (x1 , x2 , . . .) ∈ `p . Define the sequence of
linear combinations
n
. X .
yn = (x1 , x2 , . . . , xn , 0, 0, . . .) = xk ek ∈ V = span{ek ; k ≥ 1} . (2)
k=1
Then
kx − yn kp`p =
X
|xk |p → 0 as n → ∞ .
k>n
This shows that kΛk ≤ M . On the other hand, given any ε > 0, choose k such that |λk | >
M − ε. Then
kΛek k`p = kλk ek k`p = |λk |kek k`p ≥ (M − ε) kek k`p .
This shows that kΛk ≥ M − ε. Since ε > 0 was arbitrary, we conclude that kΛk = M .
To prove (ii), assume that the sequence (λk )k≥1 is unbounded. Then
P∞ −k e .
In this case, we can extract a subsequence with |λnk | ≥ 3k . Consider the vector v = k=1 2 nk
Observe that this sum is well defined, because the series is absolutely convergent:
X X
k2−k enk k`p = |2−k | = 1 .
k≥1 k≥1
8
We now have X
Λv = λnk 2−k enk ∈
/ `p
k
Indeed,
m
X m
X
−k
lim |λnk 2 | ≥ lim (3/2)k = ∞ .
m→∞ m→∞
k=1 k=1
Hence the norm of the multiplication operator satisfies kMg k ≤ K. On the other hand, for
any ε > 0 there exists a bounded measurable subset A ⊆ Ω with strictly positive measure
such that |g(x)| ≥ K − ε for all x ∈ A. Let f = χ be the characteristic function of the set
A
A. Namely, f (x) = 1 if x ∈ A and f (x) = 0 if x ∈
/ A. Then
Z 1/p Z 1/p
p p
kMg f kLp = kf gkLp = |g| dx ≥ (K − ε) dx
A A
1/p
= (K − ε) · meas(A) = (K − ε) · kf kLp .
/ L∞ (Ω \ ∪nj=1 Aj ).
3) For every n ≥ 1 one has g ∈
9
18. For any given a < b, the map
Z b
.
f →
7 ϕ(f ) = f (x) dx
a
definition implies
2 1
ϕ(fn2k ) ≥ , ϕ(fn2k+1 ) ≤ .
3 3
Therefore the subsequence ϕ(fnk ) has no limit. As a consequence, the original subsequence
does not converge weakly.
Choosing a
,0 if |a| ≥ |b| ,
|a|
(x̄1 , x̄2 ) =
b
if |a| ≤ |b| ,
0,
|b|
10
one obtains the converse inequality
Choosing
a b
(x̄1 , x̄2 ) = , ,
|a| |b|
one obtains the converse inequality
Choosing
1−q
q
(x̄1 , x̄2 ) = |a|q + |b|q · |a|q−1 sign a , |b|q−1 sign b ,
one obtains the converse inequality
1
kφkp ≥ |ax̄1 + bx̄2 | = (|a|p + |b|p ) p .
22. By assumption, B ⊂ X is a convex set with the following property. For every non-zero
vector x ∈ X, there exists ηx > 0 such that
B ∩ {t x ; t ∈ K} = {t x ; |t| ≤ ηx } . (3)
It remains to prove the properties (N1)–(N3) of a norm. From (3)-(4) one immediately obtains
11
Hence
ηx
θ|λ|x = , kλxk = |λ| kxk .
|λ|
x ∈ kxk B , y ∈ kyk B .
x + y ∈ (kxk + kyk)B ,
.
kx + yk = min{r ; x + y ∈ rB} ≤ kxk + kyk .
23. (i) =⇒ (iii) Let (fj )j≥1 be a Cauchy sequence for the distance d. Let K ⊂ Ω be any
compact subset and let ε > 0 be given. Since the open sets Aj cover K, there exists N large
enough such that K ⊂ N
S
k=1 Ak . This implies
lim sup sup |fj (x) − f` (x)| ≤ 2N lim d(fj , f` ) = 0.
j,`→∞ x∈K j,`→∞
Therefore, the sequence (fj )j≥1 converges uniformly on the compact set K.
An identical argument shows that (ii) =⇒ (iii).
(iii)=⇒ (ii) Assume that the sequence (fj )j≥1 converges uniformly on compact subsets of Ω.
.
Let ε > 0 be given. Choose N so that 2−N < ε. Then the set K = ∪1≤k≤N Ak is a compact
subset of Ω. We thus have the estimate
N ∞
X pk (fj − f` ) X pk (fj − f` )
lim sup d(fj , f` ) = lim sup 2−k + lim sup 2−k
j,`→∞ j,`→∞ k=1 1 + pk (fj − f` ) j,`→∞ 1 + pk (fj − f` )
k=N +1
≤ 0 + 2−N < ε .
Hence the sequence (fj )j≥1 is Cauchy w.r.t. the distance d(·, ·).
An identical argument shows that (iii) =⇒ (ii).
(ii) Choose N large enough so that 2−N < ε. Then, if p1 (f ) = · · · = pN (f ) = 0, then for any
λ > 0 we have
∞ ∞
X
−n pn (λ−1 f ) X pn (λ−1 f )
2 ≤ 2−n < 2−N < ε .
1 + pn (λ−1 f ) 1 + pn (λ−1 f )
n=1 n=N +1
12
Hence λ−1 f ∈ Bε for every λ > 0. By definition, this implies kf kε = 0, showing that the
property (N1) of the norm can fail.
with
Xk
cni xni
≤ 2kyn+1 − yn k ≤ 21−n for every k ∈ {1, . . . , N (n)}. (5)
i=1
∞ N
X X (n)
cni xni
n=1 i=1
c11 x11 + · · · + c1,N (1) x1,N (1) + c21 x21 + · · · + c2,N (2) x2,N (2) + c31 x31 + · · ·
13
27. (i) Let V ⊂ `∞ be the subspace of all sequences x = (x1 , x2 , . . .) such that limn→∞ xn
.
exists. For x ∈ V define the linear operator F (x) = limn→∞ xn . Observing that |F (x)| =
| limn→∞ xn | ≤ supn |xn | = kxk∞ ` . it is clear that f : V 7→ R is a continuous linear
functional with norm kF k = 1. Using the Hahn-Banach extension theorem, we can extend F
to the entire space `∞ , still with the same norm kF k = 1.
(ii) Assume m = lim inf xn < lim sup xn = M . If F (x) = M + δ for some δ > 0, a
contradiction is achieved as follows. Choose N large enough so that m − 2δ < xn < M + 2δ for
all n > N . Consider the sequence y = (x1 , x2 , . . . , xN , m, m, m, . . .). Then
δ
F (x) − F (y) = (M + δ) − m > M + − m ≥ kx − yk`∞ .
2
This is a contradiction, because kF k = 1, hence we should have F (x) − F (y)| ≤ kx − yk`∞ .
The case F (x) < m is handled in a similar way.
(iii) Using linearity and then (ii), we obtain
F (y) − F (x) = F (y − x) ≥ lim inf yn − xn ≥ 0 .
n→∞
(iv) If a = (a1 , a2 , . . .) ∈ `1 , then there exists N large enough so that n>N |an | < 1. Consider
P
the sequence x = (0, 0, . . . , 0, 1, 1, 1, . . .), where the first N entries are zero and all other entries
are equal to 1. Then
∞
X X
an xn = an < 1 = lim xn = F (x).
n→∞
n=1 n>N
(v) The functional Fe satisfies (2.43)-(2.44), but it is not linear. For example, take x =
(1, 0, 0, 1, 0, 0, 1, 0, 0, . . .), y = (0, 0, 1, 0, 0, 1, 0, 0, 1, . . .). Then x + y = (1, 0, 1, 1, 0, 1, 1, 0, 1, . . .)
and
1 1 1
F (x + y) = 6= F (x) + F (y) = + .
2 2 2
By the Hahn-Banach theorem it can be extended to a linear functional Λ defined on the entire
space L∞ (R), with norm kΛk = 1.
Next, assume that there exists g ∈ L1 (R) such that Λf = f g dx for all f ∈ L∞ (R). The
R
14
29. (i) Assuming that x ∈ λ1 Ω, y ∈ λ2 Ω for some λ1 , λ2 > 0, we claim that x+y ∈ (λ1 +λ2 )Ω.
Indeed, by assumption there exist ω1 , ω2 ∈ Ω such that x = λ1 ω1 , y = λ2 ω2 . Using the
assumption that Ω we obtain
λ1 λ2
x + y = λ1 ω1 + λ2 ω2 = (λ1 + λ2 ) · ω1 + ω2 ∈ (λ1 + λ2 )Ω .
λ1 + λ2 λ1 + λ2
The identity p(tx) = tp(x) for t > 0 follows from the fact that x ∈ λΩ if and only if tx ∈ tλΩ.
(ii) We have
kxk
p(x) = inf{λ > 0 ; x ∈ λΩ} ≤ inf{λ > 0 ; x ∈ λBr } = .
r
(iii) If Ω = B1 , then
30. (i) If (fn )n≥1 is a sequence of continuous functions such that fn → f uniformly on [0, 1]
and fn (0) = 0 for every n ≥ 1, then f (0) = 0. Hence X is a closed subspace.
. R1
(ii) The map f 7→ Λf = 0 f (x) dx is a linear functional on X. for every f ∈ X with
R1
kf k = maxx∈[0,1] |f (x)| ≤ 1 we have |Λf | = | 0 f (x) dx| < 1, because f (0) = 0, hence
|f (x)| < 1/2 for x in some interval [0, δ]. On the other hand, choosing fn = x1/n we have
n
fn ∈ X, kfn k = 1 and Λfn = n+1 → 1 as n → ∞. Hence
1
Z
.
kΛk = sup
f (x) dx = 1,
f ∈X , kf k≤1 0
31. Being the intersection of a family of closed, convex sets, the set Ω∗∗ is also closed and
convex. It remains to show that Ω∗∗ is contained in the closure of Ω. Assume, on the contrary,
that there exists x0 ∈ Ω∗∗ and a radius r > 0 such that the open ball B(x0 , r) does not
intersect Ω. By the separation theorem, there exists a bounded linear functional F : X 7→ R
and c > 0 such that
32. On the one dimensional space U = {tξ ; t ∈ R} define the functional F (tξ) = t. This
is a bounded linear functional with norm kF k = kξk. Using the Hahn-Banach Theorem, we
extend F to a bounded linear functional F : X 7→ R with the same norm. Let V = ker(F ).
We claim that all conclusions are satisfied.
15
Since F is continuous linear functional, V is a closed subspace. Given x ∈ X, consider the
projections
. .
πU x = F (x) ξ , πV x = x − F (x)ξ .
These are bounded linear operators, and satisfy
πU x ∈ U , πV x ∈ V , πU x + πV x = x .
|ϕ(f )| a
kϕk = sup = sup 1 = 1
kf kL1
Z
f ∈Y, f 6=0 (a,b)6=(0,0)
|ax + b| dx
−1
34. It is clear that each pk (·) is a seminorm. If f ∈ C m (Ω), f 6= 0, then f (x) 6= 0 for some
point x ∈ Ω. If x ∈ Ak , then pk (f ) > 0.
To show that the space C m (Ω) is complete for the corresponding distance
∞
. X −k pk (f − g)
d(f, g) = 2 ,
1 + pk (f − g)
k=1
let (fj )j≥1 be a Cauchy sequence. Then for every multi-index α with |α| ≤ m the sequence
of continuous functions (Dα fj )j≥1 is uniformly convergent on each open set Ak . Therefore, it
has a continuous limit, say
for some continuous functions g, gα . Since the limit is uniform on compact sets, it follows
gα = Dα g for every |α| ≤ m. Hence d(fj , g) → 0, proving that this Cauchy sequence has a
limit.
.
35. Consider the subspace Ye = {ỹ = y + tx0 ; y ∈ Y, t ∈ R} be the space spanned by Y
.
together with x + 0. Let α = d(x0 , Y ) > 0. Define the functional ϕ : Ye 7→ R by setting
.
ϕ(y + tx0 ) = αt .
16
We claim that ϕ is a bounded operator with norm kϕk = 1. Indeed, assume that ky0 + tx0 k <
αt for some choice of y0 ∈ Y and t 6= 0. This leads to a contradiction, because
Using the Hahn-Banch theorem, the functional ϕ can then be extended to the entire space X.
37. In the case 1 ≤ p < ∞, assume φ0 (x) ≥ c > 0 for all x ∈ R. For every f ∈ Lp (R), setting
y = φ(x) one obtains
Z 1/p Z 1/p
p p dy 1
|f (φ(x))| dx = |f (y)| 0 ≤ · kf kLp .
φ (x) c1/p
Hence, in this case, Λ is a bounded linear operator on Lp with norm kΛk ≤ c−1/p .
In the case p = ∞, it suffices to assume that φ0 (x) > 0. Then Λ is a bounded
n operator with
∞ . .
norm kΛk = 1. Indeed, if f ∈ L , let M = kf kL∞ . Consider the set A = x ; f (φ(x)) >
o
M . If meas(A) > 0, we obtain a contradiction by writing
Z
0 = meas φ(A) = φ0 (x) dx > 0.
A
Hence kΛf kL∞ ≤ M . This proves that kΛk ≤ 1. On the other hand, taking the function
g(x) ≡ 1, one has kΛgkL∞ = kgkL∞ = 1.
38. If the Banach space X is infinite dimensional, the closed unit ball is not precompact. In
particular, we can find a sequence of points xn ∈ X such that kxn k ≤ 1, kxm − xn k ≥ 1/2
whenever m 6= n.
17
Choose x̄ ∈ Ω0 and r > 0 such that the ball B(x̄, r) is contained in Ω0 . By assumption, this
implies 0 < µ(B(x̄, r)) ≤ µ(Ω0 ) < ∞.
.
Consider the countably many open balls Bn = B 2r xn , 8r . These are mutually disjoint, and
all contained in the open ball B(0, r). Hence, by countable additivity we have
[ X
µ(B(0, r)) ≥ µ Bn = µ(Bn ) = ∞
n≥1 n≥1
because all balls Bn have the same radius and the same strictly positive measure. This yields
a contradiction, because by translation invariance µ(B(0, r)) = µ(B(x̄, r)) < ∞.
39. (i) This is an immediate consequence of Theorem 2.33 (ii), taking A = {y} and B = S.
/ S. Then there exists a bounded linear functional ϕ ∈ X ∗ such that ϕ(y) <
(ii) Assume y ∈
inf x∈S ϕ(x).
The weak convergence xn * y implies ϕ(xn ) → ϕ(y). But this yields a contradiction because
lim ϕ(xn ) ≥ inf ϕ(x) > ϕ(y).
n→∞ x∈S
40. (i) Let V ⊂ L∞ (R) be the subspace of all functions f such that ess-limx→0 f (x) exists. We
.
claim that the functional Φ(f ) = ess-limx→0 f (x) is a bounded linear functional on V . The
linearity is obvious. If Φ(f ) = λ, then for any ε > 0 there exists δ > 0 such that |f (x) − λ| < ε
for a.e. x ∈ [−δ, δ]. In particular, this implies kf kL∞ ≥ |λ| − ε. Since ε > 0 is arbitrary, we
conclude
|Φ(f )| ≤ kf kL∞ (6)
for all f ∈ V . By the Hahn-Banach theorem, we can extend the linear functional Φ to the
entire space L∞ (R), still satisfying the bound (6) for all f ∈ L∞ (R).
(ii) In the space L1 (R) the conclusion fails because the functional Φ : V 7→ R is unbounded.
Indeed, consider the sequence of continuous functions:
. n(1 − nx) if |x| ≤ 1/n ,
fn (x) =
0 otherwise.
Then fn ∈ V and kfn kL1 = 1 for every n ≥ 1, but Φ(fn ) = fn (0) = n.
Chapter 3
2. (i) True. For every n ≥ 1, let pn be a polymomial such that maxx∈[−n,n] |f (x) −
pn (x)| < 2−n . Such a polynomial exists by the Stone-Weierstrass theorem. The sequence of
polynomials (pn )n≥1 converges to f uniformly on bounded sets.
18
2
(ii) False. If p is a polynomial such that supx∈R |p(x) − e−x | < ∞, then p must be bounded,
2
hence constant. But there is no way to approximate the function f (x) = e−x with constant
functions, uniformly on R.
3. (i) Extend g to an even function defined on [−π, π], and then by periodicity to an even
function defined on the whole real line, periodic of period 2π. By Corollary 3.9 there exists a
trigonometric polynomial of the form
N
X N
X
p(x) = ak sin kx + bk cos kx
k=1 k=0
such that |p(x) − g(x)| < ε for all x ∈ R. Since g is even, this implies
N
p(x) + p(−x) g(x) + g(−x) X
ε > − = b cos kx − g(x). for all x ∈ R .
k
2 2
k=0
(ii) Extend g to an odd function defined on [−π, π], and then by periodicity to an odd func-
tion defined on the whole real line, periodic of period 2π. By Corollary 3.9 there exists a
trigonometric polynomial of the form
N
X N
X
p(x) = ak sin kx + bk cos kx
k=1 k=1
such that |p(x) − g(x)| < ε for all x ∈ R. Since g is odd, this implies
N
p(x) − p(−x) g(x) − g(−x) X
ε > − = a sin kx − g(x) . for all x ∈ R .
k
2 2
k=1
4. By the Stone-Weierstrass theorem, for any n ≥ 1 there exists a polynomial qn such that
|qn (x) − f 0 (x)| ≤ 2−n for every x ∈ [a, b]. Define the polynomial
Z x
.
pn (x) = f (a) + qn (y) dy.
a
5. The properties (i) and (ii) are easily checked. However, (iii) fails. For example, the
function f (θ) = e−iθ cannot be uniformly approximated with functions in the algebra A.
Indeed, assume that
N
X 1
f (θ) − cn einθ ≤ for all θ ∈ [0, 2π].
2
n=0
19
This leads to a contradiction because
Z
N
!
2π X Z 2π
1 iθ
inθ iθ
f (θ) − c e e dθ ≤ |e | dθ ≤ π .
n
2
0 0
n=0
Since Ω is bounded, its closure Ω is compact. By Corollary 3.6, the continuous function f˜ can
be uniformly approximated by polynomials on the compact set Ω. In particular, there exists
a polynomial p(x) = p(x1 , . . . , xn ) such that
7. (i) Choose δ > 0 such that |f (x, y) − f (x0 , y)| ≤ ε whenever |x − x0 | ≤ δ. Construct a
continuous partition of unity {g1 , g2 , . . . , gN } on the interval [0, a] such that each gi is supported
on some interval Ii ⊆ [a, b] of length ≤ δ. That means
N
X
gi (x) = 1 for each x ∈ [0, a], gi (x) = 0 for x ∈
/ Ii .
i=1
For each i ∈ {1, . . . , N }, choose a point xi ∈ Ii . For every (x, y) ∈ Q = [0, a] × [0, b] we now
have
XN N
X
f (x, y) − gi (x) f (xi , y) ≤ gi (x)f (x, y) − f (xi , y) ≤ ε. (7)
i=1 i=1
Indeed, the only nonzero terms in the second summation are those for which x ∈ Ii , hence
|x − xi | ≤ δ. By construction, this implies |f (x, y) − f (xi , y)| ≤ ε.
.
Defining hi (y) = f (xi , y), all requirements are satisfied.
(ii) Let 0 < ε ≤ 1 be given. If f vanishes on the boundary of Q, then in the above construction
we can choose xi = 0 if 0 ∈ Ii and xi = a if a ∈ Ii . In both cases, f (xi , y) = 0. Therefore, in
the sum (7) such terms gi (x)f (xi , y) can be discarded, because these vanish identically. As a
result, it is not restrictive to assume that in (7) every gi is a continuous function with support
contained strictly inside the open interval ]0, a[.
20
Since all functions gi vanish at x = 0 and at x = a, and all functions f (xi , y) vanish at y = 0
and at y = b, using problem 3.(ii), we can now find coefficients Aim , Bim and an integer M
such that
M
X πmx
gi (x) − Aim sin ≤ ε for all x ∈ [0, a] ,
a
m=1
M
X πny
f (xi , y) − Bin sin ≤ ε for all y ∈ [0, b] .
b
n=1
This implies
M
gi (x)f (xi , y) −
X πmx πny
Aim Bin sin sin
a b
m,n=1
M
M M
X πmx X πmx X πny
≤ gi (x) − Aim sin |f (xi , y)| + Aim sin · f (xi , y) − Bin sin
a a b
m=1 m=1 n=1
≤ ε · kf kC 0 + (1 + kgi kC 0 ) · ε ≤ (2 + kf kC 0 ) ε .
. P
Therefore, setting cmn = N i=1 Aim Bin , we obtain
N M
X X πmx πny
gi (x)f (xi , y) − cmn sin sin ≤ N (2 + kf kC 0 ) ε . (8)
a b
i=1 m,n=1
Since ε > 0 was arbitrary, by (7) and (8) the result is proved.
8. Let (fn )n≥1 be a sequence of functions in the Hölder space C 0,γ (Ω), with kfn kC 0,γ ≤ C for
some constant C and every n ≥ 1.
Being Hölder continuous, each function fn is uniformly continuous and hence can be uniquely
extended by continuity to the closure Ω, which is a compact set.
Since the functions fn are uniformly bounded and equicontinuous on the compact set Ω, by
Ascoli’s theorem we can extract a subsequence converging to some continuous function f ,
uniformly for x ∈ Ω.
On the other hand, choosing an = n−2 , bn = n−1 , for every γ > 1/3 we have
1/3 1/3
bn − an 1
lim = lim = ∞.
n→∞ |bn − an |1/3 · |bn − an |γ−1/3 n→∞ |bn − an |γ−1/3
21
/ C 0,γ for any γ > 1/3.
Therefore f1 ∈
1 1
(ii) f2 ∈ / C 0,γ for any γ > 1. Indeed, choose an = (n+1)π bn = nπ . observing that sin a1n =
− sin b1n = ±1, we compute
p p
|f2 (bn ) − f2 (an )| 1/(n + 1) + 1/n 1
= ≥ .
|bn − an |1/4 (n + n2 )−1/4 2
/ C 0,γ for any γ > 1/4. The fact that f2 ∈ C 0,1/4 is proved by
This already shows that f2 ∈
estimating the ratio
|f2 (b) − f2 (a)|
|b − a|1/4
separately in two cases:
h i h i
1 1 1
CASE 1: b ∈ nπ , (n−1)π , a ∈ (n+1)π , b .
h i
1 1 1
CASE 2: b ∈ nπ , (n−1)π , a< (n+1)π .
The first case is handled simply by integrating |f20 | on the interval [a, b], observing that
|f20 (x)| ≤ x−1/2 + x−3/2 . Case 2 follows from Case 1, using the inequality
i
h 1
|f2 (b) − f2 (a)| ≤ sup |f2 (b) − f2 (x)| ; x ∈ , b .
(n + 1)π
(iii) f3 ∈ C 0,γ for every γ ∈ ]0, 1[ . Indeed, for any 0 < a < b < 1 one has
Z b Z b Z b
γ−1 Cγ
f3 (b) − f3 (a) = (| ln x| − 1) dx ≤ Cγ x dx ≤ Cγ (x − a)γ−1 dx ≤ (b − a)γ ,
a a a γ
10. Ascoli’s theorem guarantees the existence of a subsequence (fnk )k≥1 which converges to
some limit function f ∈ C 0,γ ([0, 1]), uniformly on [0, 1]. This means kfnk −f kC 0 → 0. However,
in general it is not true that kfnk − f kC 0,γ → 0. In other words, the convergence takes place
only in the norm of C 0 , not in the stronger norm of C 0,γ .
11. To show that k · kϕ is a norm, we need to check that it satisfies the conditions (N1)–(N3)
in the definition. The first two conditions are clear. To prove (N3) we observe that, for any
x, y ∈ Ω, x 6= y, one has
|(f + g)(x)| ≤ |f (x)| + |g(x)| ,
|f (x) + g(x) − f (y) − g(y)| |f (x) − f (y)| |g(x) − g(y)|
≤ + .
ϕ(|x − y|) ϕ(|x − y|) ϕ(|x − y|)
Taking the supremum over all x, y we obtain kf + gkϕ ≤ kf kϕ + kgkϕ .
It remains to show that the space C ϕ is complete. For this purpose, let (fn )n≥1 be a Cauchy
sequence w.r.t. the norm k · kϕ . Then this sequence is Cauchy w.r.t. the C 0 norm, hence it
22
converges uniformly to a continuous function f : Ω 7→ R. To show that kfn −f kϕ → 0, observe
that, for any ε > 0 there exists N large enough so that
fn (x) − fm (x) − fn (y) + fm (y)
sup < ε for all m, n ≥ N .
x6=y |x − y|
Hence kfn − f kϕ → 0.
12. One simply needs to retrace all steps in the proof of Ascoli’s theorem. Since E is compact,
given ε > 0 there exists δ > 0 such that
Chapter 4
1. (1) The operator Λ is linear and bounded, with kΛk = 1, but not compact.
(2) This operator is not linear: (Λ(2f ))(x) = sin(2f (x)) 6= 2 sin(f (x)) = 2(Λf )(x).
(3) Λ is linear and bounded, with kΛk = 1, but not compact.
(4) Λ is linear, bounded, and compact. Indeed, Λf is always a polynomial of degree ≤ 1.
Hence the range of Λ is two-dimensional.
(5) This operator Λ is linear, bounded, and compact. Indeed,
Z x
(Λf )(x) = ey−x f (y) dy ,
0
To see that Λ is compact, let (fn )n≥1 be a sequence of continuous functions with kfn kC ≤ 1
for every n. Then all functions Λ(fn ) are Lipschitz continuous with constant 2. Indeed, from
the differential equation it follows
23
By Ascoli’s theorem, the sequence (fn ) admits a uniformly convergent subsequence.
2. To show that En is closed, assume kfk − f kL1 → 0 and kfk k2L2 ≤ n for every k. By possibly
taking a subsequence we can assume fk (x) → f (x) for a.e. x ∈ [0, 1]. Fatou’s lemma now
yields Z 1 Z 1 Z 1
|f (x)|2 dx = lim inf |fk (x)|2 dx ≤ lim inf |fk (x)|2 dx ≤ n .
0 0 n→∞ n→∞ 0
Hence f ∈ En as well.
To prove that En has empty interior in L1 ([0, 1]), consider any f ∈ En . For any ε > 0, define
f (x) + εx−1/2
. if f (x) ≥ 0 ,
fε (x) = −1/2
f (x) − εx if f (x) < 0 .
Then kfε − f kL1 = 2ε but |fε (x)| ≥ εx−1/2 for every x ∈ ]0, 1], and hence fε ∈
/ L2 ([0, 1]).
The uniform boundedness principle thus implies kΛk = supn≥1 kΛn k < ∞.
4. By the closed graph theorem, it suffices to prove that Λ has closed graph. Toward this
goal, consider a sequence of functions fn ∈ Lp ([0, 1]) such that
for some functions f, g ∈ Lp ([0, 1]). We need to show that Λf = g. By choosing a subsequence
and relabeling, we can assume the pointwise convergence
By assumption, this implies (Λfn )(x) → (Λf )(x) for a.e. x. Hence (Λf )(x) = g(x) for a.e.
x ∈ [0, 1].
5. Assume that the range Y = K(X) is a closed subspace of X. Then Y itself is complete.
The linear operator K is a continuous bijection between X and the Banach space Y . Hence
it has a continuous inverse K −1 .
Being the composition of a compact operator and a continuous one, the identity map ι(x) =
x = K −1 K(x) is a compact operator from X into itself. But this is impossible, because X is
infinite dimensional.
24
CASE 2: Λ2 is compact. Since Λ1 is continuous, the sequence (Λ1 (xn ))n≥1 is bounded in Y .
By the compactness of Λ2 , the sequence (Λ2 ◦ Λ1 (xn ))n≥1 admits a convergent subsequence.
.
7. If Y = Range(Λ) is an infinite dimensional normed space, then the unit ball in Y is not
precompact. As shown in the proof of Theorem 2.22, there exist a sequence of points yn ∈ Y
such that kyn k ≤ 1, kyn − ym k ≥ 1/2 for all m, n ≥ 1 m 6= n. Hence from this sequence
(yn )n≥1 one cannot extract any convergent subsequence.
This leads a contradiction, because the assumptions imply that Λ is compact and Λ(y) = y
for every y ∈ Y .
.
8. First of all, observe that the assumptions imply 0 ∈ U . Call V = −U ∩ U . Since V ⊆ U ,
it suffices to prove that V contains a neighborhood of the origin. Observe that
(i) V is closed, convex.
(ii) V = −V , 0 ∈ V .
S
(iii) n≥1 nV = X.
Indeed, (i)-(ii) are clear. To prove (iii), consider any x ∈ X. Then there exist integers n1 , n2
.
large enough so that x ∈ n1 U , −x ∈ n2 U . Calling n = max{n1 , n2 } we have nx ∈ U , − nx ∈ U .
By convexity, this implies
nx xo 1o
.
n
A = co , − = θx ; |θ| ≤ ⊆ U.
n n n
Since A = −A, we conclude that A ⊆ V . Hence x ∈ nA ⊆ nV .
S
Since X is a Banach space, X = n≥1 nV , and each set nV is closed, by Baire’s category
theorem at least one of the sets nV must have non-empty interior. By homogeneity, V itself
has non-empty interior. If B(x, r) ⊆ V , then (ii) implies B(−x, r) ⊆ V . By convexity,
1 1
B(0, r) = B(x, r) + B(−x, r) ⊆ V.
2 2
10. Fix any x ∈ R and consider a sequence xn → x. Since f is bounded, there exists the
upper and lower limits
. .
m = lim inf f (y) ≤ lim sup f (y) = M .
y→x y→x
We claim that m = M = f (x). Indeed, assume m 6= f (x). Then we can find a sequence
xn → x such that f (xn ) → m. Hence the point (x, m) lies in the closure of the graph of f . If
25
f (x) 6= m, this graph is not closed and we reach a contradiction. The proof that f (x) = M is
entirely similar.
To show that the result fails without the boundedness assumption, consider the function
f (x) = x−1 if x 6= 0, f (0) = 0. Then f has closed graph but is not continuous at x = 0.
12. Assume that limk→∞ λk → 0. For each n ≥ 1 consider the truncated operator
Λn (x1 , x2 , . . .) = (λ1 x1 , λ2 x2 , . . . , λn xn , 0, 0, . . .). (9)
Each Λn is continuous and has finite dimensional range, hence is compact. Observing that
kΛ − Λn k = supj>n |λj | → 0 as n → ∞, by Theorem 4.10 we conclude that Λ is compact.
On the other hand, assume that lim supn→∞ |λn | > 0. Then there exists ε > 0 and a subse-
quence (λnk )k≥1 such that |λnk | ≥ ε for every k ≥ 1. Consider the unit vectors
e1 = (1, 0, 0, 0, . . .), e2 = (0, 1, 0, 0, . . .), e3 = (0, 0, 1, 0, . . .), ...
Then the sequence (enk )k≥1 is bounded, but the sequence Λ(enk ) = λnk enk does not admit
any convergent subsequence. Indeed,
kΛenj − Λenk k = kλnj enj − λnk enk k ≥ kλnk enk k ≥ ε .
Hence Λ is not compact.
13. Let B be continuous at the origin. Then there exists δ > 0 such that |B(x, y)| ≤ 1
whenever kxk ≤ δ and kyk ≤ δ.
For any (x, y) ∈ X × Y we now have
1 δx δy
|B(x, y)| = 2 · kxk kyk · B , ≤ C · kxk kyk ,
δ kxk kyk
with C = δ −2 .
26
Hence the map Λp : q 7→ B(p, q) is a bounded linear operator from X into itself, with norm
kΛp k = kpkC([0,1]) .
To prove that the bilinear map B is not continuous on the product space X ×X, we proceed as
follows. Choose a sequence of numbers xn > 0 such that | ln xn | > n2 . Define the continuous
functions 1
√ if x ∈ [xn , 1] ,
nx
.
fn (x) =
1
if x ∈ [0, xn ] .
√
nxn
Observe that Z 1 Z 1
√
1 2
fn (x) dx < √ x dx = √ .
0 n 0 n
Z 1 Z 1
1 1
fn2 (x) dx > dx > | ln xn | ≥ n .
0 xn nx n
By the Stone-Weierstrass theorem, each function fn can be uniformly approximated by a
polynomial. We can thus construct a sequence of polynomials (pn )n≥1 such that
Z 1 Z 1
2
pn (x) dx < √ , p2n (x) dx > n
0 n 0
15. Since S is bounded, there exists M such that kxk ≤ M for all x ∈ S. Let ε ∈ ]0, 1] be
given. Let Yε be a finite dimensional space such that d(x, Yε ) ≤ ε for all x ∈ S.
.
In the finite dimensional space Yε , the closed ball B = {y ∈ Yε ; kyk ≤ 1 + M } centered at
the origin with radius 1 + M is compact. Hence it can be covered with finitely many balls of
radius ε, say B(yn , ε), n = 1, . . . , N . We claim that
N
[
S ⊆ B(yn , 2ε) .
n=1
Indeed, for every x ∈ S there exists yx ∈ Yε such that kx − yx k ≤ ε. Clearly, we must have
kyx k ≤ M + ε. Hence yx ∈ B(yn , ε) for some n. This implies x ∈ B(yn , 2ε).
The previous analysis shows that S is precompact. Namely, for every ε > 0 is can be covered
by finitely many balls of radius ε. By assumption, S is a closed subset of a Banach space,
hence it is complete. We thus conclude that S is compact.
27
Using the Hahn-Banach extension theorem, this functional can be extended to a bounded
linear functional, still called ϕ, defined on the entire space X.
.
We then define the operator K : X 7→ X by setting K(x) = ϕ(x) g. Notice that this implies
17. Since V is an infinite dimensional normed space, we can find a sequence of points (xn )n≥1
in V such that kxn k ≤ 1, kxm − xn k ≥ 1/2 for all m 6= n. Then the sequence (Λxn )n≥1 does
not admit any convergent subsequence. Indeed, by assumption
ε
kΛxm − Λxn k ≥ εkxm − xn k ≥ for all m 6= n.
2
Therefore, the operator Λ cannot be compact.
By assumption,
lim ϕn (x∗ ) = ϕ(x∗ )
n→∞
exists for every x∗ ∈ X ∗ . Hence supn≥1 |ϕn (x∗ )| < ∞ for every x∗ ∈ X ∗ .
.
By the uniform boundedness principle, M = supn≥1 kϕn k < ∞. Therefore
.
kϕk = sup |ϕ(x∗ )| ≤ sup sup |ϕn (x∗ )| ≤ sup kϕn k = M,
kx∗ k≤1 kx∗ k≤1 n≥1 n≥1
19. For every continuous function f : [0, 1] 7→ R, the mean value theorem implies
1 t
Z
lim f (s) ds = f (0) .
t→0 t 0
28
we conclude that Λ is a bounded operator, with norm kΛk ≤ 1.
Rt
(ii) If (Λf )(x) = 0 for every x ∈ [0, 1], then 0 f (s) ds = 0 for every t > 0 and hence f (s) = 0
for every x ∈ [0, 1]. This proves that Λ is one-to-one.
If g = Λf for some continuous function f , taking derivatives we obtain
f (t) − g(t)
g 0 (t) = for all t ∈ ]0, 1[ .
t
Hence g is continuously differentiable. Any continuous function g : [0, 1] 7→ R which is not
continuously differentiable cannot be in the range of Λ.
(iii) To prove that Λ is not compact, consider the sequence of functions
Observe that Λfn (x) = 0 for x ≥ 21−n π. Hence, for m < n a direct computation yields
2 2
kΛfm − Λfn kC 0 ≥ Λfm (2−m π) − Λfn (2−m π) = − 0 = .
π π
As a consequence, the sequence (Λfn )n≥1 does not admit any uniformly convergent subse-
quence.
20. (i) The multiplication operator Mf is one-to-one if and only if g(x) 6= 0 for a.e. x ∈ Ω.
(ii) In general, the range of Mg is not closed. For example, let Ω = ]0, 1[ and g(x) = x. Then
the function h(x) ≡ 1 is not in the range of Mg , because the function f (x) = h(x)/g(x) = 1/x
.
does not lie in Lp (Ω). However, if 1 ≤ p < ∞, then the characteristic functions fn = χ
[1/n , 1]
lie in the range of Mg . Moreover, kfn − f kLp → 0 whenever 1 ≤ p < ∞.
To see a positive result, assume that there exists ε > 0 such that the domain Ω can be
decomposed as Ω = Ω0 ∪ Ωε , with Ω0 , Ωε disjoint measurable sets, and
(iii) If g does not coincide a.e. with the zero function, then the operator Mg is not compact.
.
To see this, assume that, for some ε > 0, the set Aε = {x ∈ Ω ; |g(x)| ≥ ε} has strictly
positive measure. Then we can find countably many disjoint measurable sets An such that
29
choosing the constants cn so that kfn kLp = 1. By construction, the sequence (fn )n≥1 is
bounded. However, the sequence (Mg fn )n≥1 does not admit any convergent subsequence.
Indeed, for any m 6= n, since fn and fm have disjoint support, one has
22. (i) By the open mapping theorem, the functional Λ has a continuous inverse. Hence there
exists a constant C > 0 such that kΛ−1 yk ≤ C kyk for every y ∈ Y . Taking β = C −1 , this
implies βkxk ≤ kΛxk for all x ∈ X.
.
(ii) We check that, if γ = kΨk < 1/β, then the map u 7→ Λ−1 (f − Ψu) is a strict contraction.
Indeed, for any u, v ∈ X one has
kΛ−1 (f −Ψu)−Λ−1 (f −Ψv)k ≤ kΛ−1 (Ψv)−Λ−1 (Ψu)k ≤ kΛ−1 k·kΨk ku−vk ≤ β · γku−vk .
Since β · γ < 1, this is a strict contraction and therefore has a unique fixed point.
(iii) Let Λ : X 7→ Y be a continuous bijection, and let β > 0 be such that kΛxk ≥ βkxk for all
x ∈ X. We claim that, if Ψ : X 7→ Y is any bounded linear operator with kΨk < β −1 , then
the sum Λ + Ψ is still a bijection. Indeed, by (ii) for any f ∈ Y there exists a unique element
u ∈ X such that u = Λ−1 (f − Ψu). This implies Λu = f − Ψu, hence (Λ + Ψ)u = f . showing
that the operator ΛΨ is a bijection.
30
(ii) Λis not a bounded operator from `1 into `1 . For example, if x = (1, 0, 0, . . .) ∈ `1 , then
Λx = 1, 21 , 13 , . . . ∈/ `1 .
24. (i) If Y = X, let ι : Y 7→ X be the identity map. This is a bijective, continuous, linear
map from Y onto X. By Corollary 4.5, the inverse map ι−1 : X 7→ Y is continuous as well.
Hence there exists a constant C 0 such that kxkY ≤ C 0 kxkX for all x ∈ X.
.
(ii) Assume Y 6= X. Consider the sets Sn = {y ∈ Y ; kykY ≤ n}, and let S n be the closure
of Sn in the space X. We claim that
Hence Y is of first category, being contained in the union of countably many closed sets with
empty interior.
To prove (C), we argue by contradiction. If some S n has nonempty interior, by positive
homogeneity the set S 1 has nonempty interior as well. To fix the ideas, assume that the open
ball B(x, r) ⊂ S 1 for some x ∈ S 1 and r > 0. Since S 1 = −S 1 , we also have B(−x, r) ⊂ S 1 .
By convexity, this implies
1 1
B(0, r) = B(x, r) + B(−x, r) ⊂ S 1 .
2 2
In turn, by positive homogeneity this yields
.
B(0, 2−n r) ⊂ S 2−n = {y ∈ Y ; kykY ≤ 2−n } .
Repeating the argument in step 3. of the proof of Theorem 4.4 (the Open Mapping Theorem),
we conclude that B(0, r/2) ⊂ S1 .
For any x ∈ X we thus have
Chapter 5
31
(x, αy) = (αy, x) = α(y, x) = ᾱ(x, y) .
Saying that the vectors x, y are mutually orthogonal means that (x, y) = 0. This implies
kx + yk2 = (x + y, x + y) = (x, x̄) + (x, ȳ) + (y, x̄) + (y, ȳ) = kxk2 + 0 + 0 + kyk2 .
.
2. Let x, y ∈ H and ε > 0 be given. Let M = max{kxk, kyk}. Choose 0 < δ < 1, such that
δ = ε/(1 + 2M ). If kx0 − xk ≤ δ and ky 0 − yk ≤ δ, then
|(x0 , y 0 ) − (x, y)| = (x0 − x, y 0 − y) + (x0 − x, y) + (x, y 0 − y)
(ii) Assume that the norm k · k satisfies the parallelogram identity. We claim that the assign-
ment
. 1 1
(x, y) = kx + yk2 − kxk2 − kyk2 = kx + yk2 − kx − yk2 (11)
2 4
satisfies all properties of an inner product. Note that the equality between the last two terms
in (11) is an easy consequence of (10).
By (11), the identity (x, y) = (y, x) is obvious.
To check that (x + x0 , y) = (x, y) + (x0 , y), using the parallelogram identity we obtain
1 1
= kxk2 + kx0 k2 + kx + yk2 + kx0 + yk2 − kx + y − x0 k2 − kx0 + y − xk2 .
2 2
This implies
1
(x + x0 , y) = kx + x0 + yk2 − kx + x0 − yk2
4
1 2 0 2 2 0 2 1 0 2 1 0 2
= kxk + kx k + kx + yk + kx + yk − kx + y − x k − kx + y − xk
4 2 2
1 2 0 2 2 0 2 1 0 2 1 0 2
− kxk + kx k + kx − yk + kx − yk − kx − y − x k − kx − y − xk
4 2 2
1 1
= kx + yk2 − kx − yk2 + kx0 + yk2 − kx0 − yk2 = (x, y) + (x0 , y) .
4 4
32
In particular, from the above identity by induction it follows
(nx, y) = (x + x + · · · + x, y) = n(x, y) .
1
In turn, this implies n x, y = n1 (x, y), and hence
4. (i) Choose x = (1, 0), y = (0, 1). Then kxkp = kykp = 1, kx + ykp = kx − ykp = 21/p . If
p 6= 2, then
(ii) Choose f (x) = sin πx, g(x) = 1 − sin πx. Then kf k = kgk = kf + gk = kf − gk = 1 and
the parallelogram identity fails.
5. Since the monomials 1, x are mutually orthogonal in L1 ([−1, 1]), the first two polynomials
in an orthonormal basis are
r
1 1 x 3
p0 (x) = = √ , p1 (x) = = x.
k1kL2 2 kxkL2 2
Therefore r
q(x) 45 2 1
p2 (x) = = x − .
kqkL2 8 3
P∞ 2
6. For every x ∈ H, Bessel’s inequality (5.10) implies n=1 |(x, en )| ≤ kxk2 . Hence
limn→∞ |(x, en )| = 0. By definition, this means that en * 0.
33
7. Since H is infinite dimensional, for a given vector x ∈ Hpwe can find an orthonormal
.
sequence (ek )k≥1 such that (x, ek ) = 0 for every k. Choose λ = 1 − kxk2 . Then the sequence
.
xn = x+λen satisfies all requirements. Indeed, by Pythagora’s theorem kxn k2 = kxk2 +λ2 = 1.
Moreover, for any y ∈ H by the previous problem 6. we have
P
8. (i) By assumption, k kαk vk k < ∞, hence the series is absolutely convergent. Since H is
complete, the series converges.
(ii) Let (vk )k≥1 be an orthonormal sequence. Then (assuming m ≤ n)
X X
|αk |2 .
lim sup
αk vk = lim sup
m,n→∞ m,n→∞
m<k≤n m<k≤n
|αk |2 < ∞.
P
This shows that the sequence of partial sums is Cauchy if and only if k
9. For any f ∈ L2 (Rn ), performing the change of variable y = φ(x) and using the fact that
detDφ ≡ 1, we compute
Z Z Z
2 2
|(Λf )(x)| dx = |f (φ(x))| dx = |f (y)|2 dy.
Rn Rn Rn
Hence kΛf kL2 = kf kL2 for every f ∈ L2 (Rn ), proving that kΛk = 1.
The adjoint operator Λ∗ is defined by the identity
Z Z Z Z
∗
f (y)·(Λ g)(y) dy = (Λf )(x)·g(x) dx = f (φ(x))·g(x) dx = f (y)·g(φ−1 (y)) dy .
Rn Rn Rn Rn
Hence Λ∗ (y1 , y2 , y3 , . . .) = (0, y1 , y2 , . . .). Clearly, Λ is onto but not one-to-one, while Λ∗ is
one-to-one but not onto.
11. (i) Assume x = θx1 + (1 − θ)x2 , for some x1 6= x2 , with kx1 k ≤ 1, kx2 k ≤ 1 and 0 < θ < 1.
By the parallelogram’ identity
34
Therefore
kxk2 = θx1 + (1 − θ)x2 , θx1 + (1 − θ)x2
≤ θ2 + (1 − θ)2 + 2θ(1 − θ) = 1.
This proves that every point x in the closed unit ball of a Hilbert space which is not extremal
must have norm kxk < 1. Hence all points with kxk = 1 are extremal.
(ii) Next, assume f ∈ L1 ([0, 1]) with kf kL1 = 1. Then we can find a set A ⊂ [0, 1] with
positive measure such that |f (x)| ≥ 1/2 for every x ∈ A. To fix the ideas, assume f (x) ≥ 1/2
for all x ∈ A, with meas(A) > 0. Choose two disjoint measurable subsets A1 , A2 ⊂ [0, 1] with
A1 ∪ A2 = A, meas(A1 ) = meas(A2 ). Consider the function
1/2 if x ∈ A1 ,
g(x) = −1/2 if x ∈ A2 ,
0 if x ∈
/ A1 ∪ A2 .
Then
f +g f −g
kf + gk1L = kf − gkL1 = 1 , f =
+ .
2 2
Hence f is not an extreme point of the unit ball in L1 , being a convex combination of the two
functions f + g and f − g.
12. By taking a subsequence and relabeling, one can assume kxn k → C as n → ∞. Since this
sequence is bounded, by Theorem 5.14 it admits a weakly convergent sequence, say xnj * x
for some x ∈ H. We now compute
kxk2 = (x, x) = lim (x, xnj ) ≤ lim sup kxk kxnj k ≤ Ckxk .
j→∞ j→∞
13. By assumption, H admits a countable, everywhere dense set {x1 , x2 , x3 , . . .}. Using
the Gram-Schmidt procedure, from this set we construct a countable orthonormal basis
{e1 , e2 , e3 , . . .}.
Defining
. .
Λx = a = (a1 , a2 , a3 , . . .) with ak = (x, ek ),
we obtain the desired bijection. Indeed,
∞
X ∞
X
kΛxk`2 = |ak |2 = |(x, ek )|2 = kxk .
k=1 k=1
.
14. (i) Assume x ∈ V = span{v1 , . . . , vn }, so that
n
X
x = θk v k
k=1
35
v2 v3
h3
h
2
v2
v1 v1
Figure 1: Computing the area of a parallelogram and the volume of a parallelepiped. Here h2 =
d(v2 , span{v1 }) while h3 = d(v3 , span{v1 , v2 }).
for some coefficients θk . To actually compute these coefficients, we observe that, for every
j = 1, . . . , n, one must have
n
X
(x, vj ) = θk (vk , vj ) .
k=1
Therefore the numbers θ1 , . . . , θn are obtained by solving the system of n linear equations
(v1 , v1 ) · · · (vn , v1 ) θ1 (x, v1 )
.. .. .. .. = .. .
. . . . . (13)
(v1 , vn ) · · · (vn , vn ) θn (x, vn )
Take x to be the zero vector in (13). Then that the vectors v1 , . . . , vn are linearly independent
if and only if θ1 = θ2 = · · · = θn = 0 is the only solution of (13) when x = 0. This is the case
if and only if the determinant G((v1 , . . . , vn ) of the n × n symmetric matrix in (13) is not zero.
(ii) If x ∈ H and y ∈ V then G(x, v1 , v2 , . . . , vn ) = G(x + y, v1 , v2 , . . . , vn ). In particular,
taking y = PV (x) and z = x − y = PV ⊥ (x), we obtain
Therefore s
1/2 G(x, v1 , v2 , . . . , vn )
d(x, V ) = kzk = (z, z) = .
G(v1 , v2 , . . . , vn )
(iii) The volume of the parallelepiped with edges v1 , . . . , vn can be computed as a product:
kvn k · d vn−1 ; span{vn } · d vn−2 ; span{vn−1 , vn } · · · d v1 ; span{v2 , . . . , vn }
s s
p G(vn−1 , vn ) G(v1 , v2 , . . . , vn ) p
G(vn ) · ··· = G(v1 , v2 , . . . , vn ),
G(vn ) G(v2 , . . . , vn )
15. Let U ⊂ L2 (R) be the subspace of all even functions. This is a closed subspace, hence for
every f ∈ L2 the perpendicular projection g0 = πU f is well defined. We claim that
f (x) + f (−x)
g0 (x) = .
2
36
Indeed, g0 (x) = g0 (−x) for every x ∈ R, hence g0 ∈ U . It remains to check that f − g0 is
perpendicular to every function g ∈ U . This is indeed the case because, if g is even,
Z Z Z
f (x) + f (−x) f (x) f (−x)
g(x) f (x) − dx = g(x) dx − g(x) dx
2 2 2
Z Z
f (x) f (x)
= g(x) dx − g(−x) dx = 0 .
2 2
16. To prove that (i)=⇒(ii), assume that K is compact. Let B1 = {x ∈ X ; kxk ≤ 1} be the
closed unit ball in X. Given any ε > 0, by assumption the image K(BS1 ) is precompact, hence
it can be covered with finitely many balls of radius ε. Say, K(B1 ) ⊆ ni=1 B(yi , ε).
.
Call Y = span{y1 , . . . , yn } and set
.
Kε (x) = πY ◦ K(x) .
Indeed,
d(K(x), Y ) ≤ min kK(x) − yi k ≤ ε .
1≤i≤n
P
17. Assume that n kun − vn k < 1. Let (un )n≥1 be complete. If (vn )n≥1 is not complete,
then there exists a unit vector x ∈ H which is perpendicular to every vn . A contradiction is
obtained by writing
∞
X ∞
X ∞
X ∞
X
2 2 2 2 2
1 = kxk = (x, un ) = (x, un − vn ) ≤ kxk kun − vn k ≤ kun − vn k .
n=1 n=1 n=1 n=1
18. (i) Let (un )n≥1 be mutually orthogonal. Assume θ0 y + θ1 u1 + · · · + θN uN = 0, for some
coefficients θj not all zero. Since the vectors u1 , . . . , uN , uN +1 are linearly independent, we
must have θ0 6= 0 and uj 6= 0 for every j. This yields a contradiction because
. P .
(ii) Let (un )n≥2 be an orthonormal set, and define v = ∞n=2 2
−n u , u =
n 1 v/kvk. Observe
that in this case the two vectors y, u1 are parallel.
19. Assume that the strong convergence kxn − xk → 0 holds. Then for every y ∈ H one has
37
Hence one has the weak convergence xn * x as well. Moreover, if kxn k ≥ kxk, then
kxn k − kxk ≤ kxn − xk → 0 as n → ∞.
Z 1
2
20. The orthogonal subspace is U⊥ = W = w ∈ L (Q) ; w(x, y) dx = 0 for a.e. y ∈ [0, 1] .
0
In this case, the function g ∈ U which has minimum distance from f is the perpendicular
projection: g = πU (f ) = u.
22. (i) The closure and convexity of Ω ⊂ L2 (R) are straightforward. If kfn − f kL2 → 0, we
can find a subsequence such that fn (x) → f (x) for a.e. x ∈ R. If fn (x) ≤ ex for every n ≥ 1
and a.e. x ∈ R, then also f (x) ≤ ex for a.e. x. Hence Ω is a closed subset of L2 (R).
Moreover, if f (x) ≤ ex and g(x) ≤ ex for a.e. x ∈ R, the same is true for any convex
combination: θf (x) + (1 − θ)g(x) ≤ ex for every θ ∈ [0, 1] and a.e. x.
(ii) Clearly the function g(x) = min{f (x), ex } lies in the convex set Ω. By the result proved
in problem 21, it suffices to check that
Z
(ω(x) − g(x)) · (g(x) − f (x)) dx ≥ 0 for all ω ∈ Ω . (14)
38
In both cases, (ω(x) − g(x)) · (g(x) − f (x)) ≥ 0. Hence (14) holds.
√
23. (i) The operator Λ has norm kΛk = 2. Indeed,
Z 0 Z ∞ 1/2 Z ∞ 1/2
kΛf kL2 = |f (−x)|2 dx + |f (x)|2 dx = 2 |f (x)|2 dx
−∞ 0 0
1/2 √
≤ 2kf k2L2 = 2 · kf kL2 ,
This implies
0 if x < 0,
(Λ∗ g)(x) =
g(−x) + g(x) if x > 0.
Hence kΛf kL2 ≤ kf kL2 and kΛk ≤ 1. To prove the converse inequality consider the sequence
√
of functions fn = n · χ[1,1+1/n] . Then
Z 1+1/n
2 n 1
kfn kL2 = 1 , kΛfn k = dy = n ln 1 + → 1 as n → ∞.
1 y n
Therefore kΛk = 1.
n o
(ii) Ker(Λ) = f ∈ L2 ([0, ∞[) ; f (x) = 0 for a.e. x ∈ [0, 1] .
Z ∞
n
2 |g(ln y)|2 o
Range(Λ) = g ∈ L ([0, ∞[) ; dy < ∞ .
1 y
(iii) The adjoint operator Λ∗ is characterized by the identity
Z ∞ Z ∞ Z ∞ Z ∞
g(ln y)
f (x)(Λ∗ g)(x) dx = (Λf )(x)g(x) dx = f (ex )g(x) dx = f (y) dy .
0 0 0 1 y
Therefore
0 if y ∈ [0, 1] ,
∗
(Λ g)(y) =
g(ln y)
if y > 1 .
y
39
25. Assume fn * f . Then for any b ∈ [0, T ], taking g = χ[0,b] ∈ L2 ([0, T ]) we have
Z b Z T Z T Z b
lim fn (x) dx = lim fn (x)g(x) dx = f (x)g(x) dx = f (x) dx .
n→∞ 0 n→∞ 0 0 0
Viceversa, assume that kf kL2 ≤ M , kfn kL2 ≤ M for all n ≥ 1, and that
Z b Z b
lim fn (x) dx = f (x) dx for every b ∈ [0, T ] .
n→∞ 0 0
Now consider an arbitrary function g ∈ L2 ([0, T ]). Given any ε > 0, we can find a piecewise
constant function ϕ of the form (15) such that kg − ϕkL2 < ε. This yields the estimate
Z T Z T Z T
lim sup
(fn − f )g dx ≤ lim sup
(fn − f )ϕ dx + sup
(fn − f )(ϕ − g) dx
n→∞ 0 n→∞ 0 n≥1 0
√
26. Consider first the case fn (x) = n · cos nx . Then for any b ∈ [0, 1] we have
Z b
1
fn (x) dx = √ sin nb → 0 as n → ∞ .
0 n
By taking linear combinations, it is clear that
Z 1
lim fn (x)g(x) dx = 0 (16)
n→∞ 0
for any piecewise constant function g. In spite of the fact that piecewise constant functions
are dense in L2 ([0, 1]), the weak convergence fn * 0 FAILS, because the sequence (fn )n≥1 is
not bounded. Indeed,
Z 1 Z 1
2
|fn (x)| dx = n cos2 nx dx → ∞ as n → ∞ .
0 0
40
For each fixed 0 < b ≤ 1, and all n > b−1 we have
Z b Z 1/n
fn (x) dx = n2/3 dx = n−1/3 → 0 as n → ∞ .
0 0
Hence, also in this case (16) holds for every piecewise constant function g. However,
Z 1 Z 1/n
2
|fn (x)| dx = |n2/3 |2 dx = n1/3 → ∞ as n → ∞ .
0 0
Also in this case, the sequence (fn )n≥1 is not bounded in L2 ([0, 1]), hence it cannot converge
weakly. Notice that in this case the result proved in problem 25 cannot be used.
For example, take g(x) = x−3/7 . Then g ∈ L2 ([0, 1]) but
Z 1 Z 1/n
7
fn (x)g(x) dx = n2/3 x−3/7 dx = n2/3 · n−4/7 → ∞ as n → ∞ .
0 0 4
27. We only need to prove the implication (ii) =⇒ (i). By linearity, it is clear that the
convergence (xn , y) → (x, y) must hold for every y ∈ span(S). By assumption, span(S) is a
subspace whose closure has non-empty interior. Hence span(S) = H.
Let y ∈ H be given. For any ε > 0 we can find a point ỹ ∈ span(S) with ky − ỹk < ε. By
assumption, there exists a constant M such that kxk ≤ M and kxn k ≤ M for all n ≥ 1. This
yields the estimate
lim sup (xn − x, y) ≤ lim sup (xn − x, y − ỹ) + lim sup (xn − x, ỹ)
n→∞ n→∞ n→∞
28. Use the result in problem 27, choosing S as the set of all characteristic functions χQ ,
where Q = [a1 , b1 ] × · · · × [aN , bN ]. Observe that the subspace spanned by all these functions
is dense on L2 (Ω).
29. If y ∈
/ S = co{xn ; n ≥ 1}, then by Theorem 2.33 there exists a linear functional φ : H 7→
R that strictly separates the compact, closed convex set {y} from the closed convex set S. By
the Riesz’ representation theorem, there exists an element z ∈ H and constants c1 < c2 such
that
(y, z) = φ(y) ≤ c1 < c2 ≤ φ(xn ) = (xn , z) for all n ≥ 1 .
This leads to a contradiction, because the weak convergence xn * y implies (xn , z) → (y, z).
Chapter 6
41
1. (i) Define
. f (x − 1) if x > 1 ,
(Λ1 f )(x) = f (x + 1) , (Λ2 f )(x) =
0 if x ∈ [0, 1] .
Then
f (x) if x > 1 ,
(Λ1 ◦ Λ2 f )(x) = f (x) , (Λ2 ◦ Λ1 f )(x)
0 if x ∈ [0, 1] .
(ii) If Λ(I − K) = I, then Ker(I − K) = {0} hence the Fredholm operator I − K is one-to-one.
By Theorem 6.1, I −K is onto. We conclude that I −K is a bijection, and hence Λ = (I −K)−1
commutes with I − K.
On the other hand, if (I − K)Λ = I, then I − K must be onto. Hence by Theorem 6.1 the
Fredholm operator I − K is one-to-one. We conclude that I − K is a bijection, and hence
Λ = (I − K)−1 commutes with I − K.
Therefore
∗ 2g(y − 1) if y > 1 ,
(Λ g)(y) =
0 if y ∈ [0, 1] .
n o
(iii) One has KerΛ = f ∈ L2 ([0, ∞[ ; f (x) = 0 for a.e. x ∈ [0, 1] , while KerΛ∗ = {0}.
Of course, this implies that Λ is not a Fredholm operator. In particular, it cannot be written
as λI + K, with λ ∈ R and K compact.
An application of Theorem 5.12, with β = η − kΛk, yields the existence of a continuous inverse
(ηI − Λ)−1 , having norm k(ηI − Λ)−1 k ≤ 1/β. Therefore, η ∈ σ(Λ).
The case η < −M is entirely similar.
42
.
4. (i) Let M = max(x,y)∈[0,1]×[0,1] |K(x, y)| . Then, by Hölder’s inequality,
Z 1 Z 1
|Λf (x)| =
K(x, y)f (y) dy ≤ M
|f (y)| dy ≤ M kf kL2 ([0,1]) .
0 0
Z 1 Z 1
= K(y, x)g(x)dx f (y)dy = (f, Λg)L2 .
0 0
To show that Λ is compact, follow the proof of Theorem 4.12. Observe that, if a sequence
of continuous functions gn = Λfn converges uniformly on [0, 1], then it also converges in
L2 ([0, 1]).
(ii) Consider the special case where
Z x Z 1
u(x) = (Λf )(x) = (1 − x)y f (y) dy + (1 − y)xf (y) dy .
0 x
Differentiating twice w.r.t. x one obtains
Z x Z 1
0
u (x) = (−y)f (y) dy + (1 − y)f (y) dy ,
0 x
00
u (x) = − f (x) .
6. (i) Assume that u ∈ L2 ([0, 1]), say with kukL2 = M . Then, for any 0 ≤ t1 < t2 ≤ 1, using
Hölder’s inequality we obtain
Z t2
√
Λu(t2 ) − Λu(t1 ) ≤ 1 · |u(s)| ds ≤ k1kL2 ([t1 ,t2 ]) · kukL2 = t2 − t1 · kukL2 .
t1
43
Hence Λu ∈ C 0,1/2 .
(ii) Consider any sequence (fn )n≥1 with kfn kL2 ≤ 1 for every n. Then the corresponding func-
tions Λfn are uniformly bounded and Hölder continuous, hence equicontinuous. By Ascoli’s
theorem, one can extract a subsequence (fnk )k≥1 that converges uniformly on [0, 1] to some
continuous function f . In turn, this implies kfnk − f kL2 → 0, proving that the operator Λ is
compact.
(iii) Writing
Z 1 Z 1 Z 1 Z x
∗
u(x) (Λ v)(x) dx = (Λu)(x) v(x) dx = u(s) ds v(x) dx
0 0 0 0
Z 1 Z 1 Z 1 Z x
= u(s) ds v(s) ds − u(x) v(s) ds dx ,
0 0 0 0
we conclude Z 1
∗
(Λ v)(x) = v(s) ds .
x
Rx
(iv) The operator (I−K) is a Fredholm operator on L2 ([0, 1]). If u = Ku, then u(x) = 0 u(s) ds .
hence u is an absolutely continuous solution of the Cauchy problem
d
u(x) = u(x) , u(0) = 0 .
dx
By Gronwall’s inequality, the only solution is u(x) = 0 for every x. We conclude that the
operator I − K is one-to-one. By Fredholm’s theorem, I − K is onto. Hence for every
g ∈ L2 ([0, 1]) there exists one and only one function u such that u − Ku = g. The inverse
map g 7→ u = (I − K)−1 is also a bounded linear operator.
If g is continuously differentiable, then u is an absolutely continuous function that satisfies
the Cauchy problem
v 0 (x) = v + g , v(0) = 0 ,
Z x
v(x) = ex−y g(y) dy .
0
This provides an explicit formula for the inverse operator:
Z x
−1
u(x) = (I − K) g (x) = g(x) + ex−y g(y) dy .
0
44
If (17) holds, then u is a solution if and only if
N
X X (f, wk )
u = ck wk + wk ,
1−λ
k=1 k>N
9. In the case where H is finite dimensional, the result is trivial: every operator is compact
and (ii) is satisfied simply because the set of orthonormal sequences is empty. In the following
we thus assume that H is infinite dimensional.
(i) =⇒ (ii). Assume that Λ : H 7→ H is compact, and let {v1 , v2 , . . .} be an orthonormal
sequence of vectors in H.
If the sequence Λvn does not converge to zero, by taking a subsequence we can assume kΛvn k ≥
ε > 0 for all n ≥ 1. A contradiction is obtained as follows.
By compactness, there exists a subsequence such that Λvnk → w for some vector w. Our
previous assumption implies kwk ≥ ε.
By choosing a further subsequence, we can assume
kwk
kΛvnk − wk < for all k ≥ 1 . (18)
2
have
N N
!
X Λvn k
X vn k
= Λ → Λ(v) as N → ∞.
k k
k=1 k=1
45
We now inductively construct an orthonormal sequence {v1 , v2 , . . .}, as follows.
• Assume that the orthonormal set {v1 , . . . , vn } has been constructed. Again by (Pε ),
.
the set S is not contained in the ε-neighborhood of the finite dimensional space Vn =
span{Λv1 , . . . , Λvn }. Hence there exists a vector w such that kwk ≤ 1 and d(Λw, Vn ) ≥ ε.
Define the vectors
n
. X . wn+1
wn+1 = w− (w, vn )vn , vn+1 = .
kwn+1 k
k=1
By induction, we thus obtain an orthonormal sequence {vn ; n ≥ 1} such that kΛvn k ≥ ε > 0
for every n. Hence (ii) fails.
This is a bounded linear operator with norm kΛk = 1. It is not compact, because the
image of the unit ball contains the entire orthonormal sequence {en ; n ≥ 1}. However,
limk→∞ Λek = 0. Indeed, for 2n < k ≤ 2n+1 we have
46
Since K is compact, there exists a subsequence such that Kunj → z for some z ∈ H. This
yields
lim un = lim (yn + Kun ) = z ∈ V .
n→∞ n→∞
Hence
(I − K)z = lim (I − K)un = 0 ,
n→∞
in contradiction with the assumption that (I − K) is one-to-one restricted to V .
Now assume vn ∈ V for every n ≥ 1, and
yn = (I − K)vn → y.
Therefore the sequence (vn )n≥1 is Cauchy, and converges to some limit v ∈ V . By continuity,
y = (I − K)v.
11. Let {w1 , w2 , . . .} be an orthonormal basis of a real Hilbert space H, consisting of eigen-
vectors of a linear, compact, self-adjoint operator K. Let λ1 , λ2 , . . . be the corresponding
eigenvalues.
Assume that t 7→ u(t) provides a solution to the Cauchy problem
d
u(t) = Ku(t) , u(0) = f , (20)
dt
for some f ∈ H. Taking the inner product of u(t) with wk we obtain
d
(u(t), wk ) = (Ku(t), wk ) = (u(t), Kwk ) = λk (u(t), wk ) , (u(0), wk ) = (f, wk ) .
dt
(21)
Therefore, the solution can be written as
∞
X
u(t) = ck (t)wk , (22)
k=1
where each coefficient ck (·) is obtained by solving the linear scalar Cauchy problem
12. In the more general case of a non-homogenous equation, the coefficients ck (·) satisfy the
equations
c0k (t) = λk ck (t) + (g(t), wk ) , ck (0) = (f, wk ) .
47
Hence the formula (22) remains valid, with
Z t
ck (t) = eλk t (f, wk ) + eλk (t−s) (g(t), wk ) ds .
0
13. Taking the inner product of the solution with wk , in this case we find that the function
.
ck (t) = (u(t), wk ) satisfies the second order initial value problem
14. Assume that ηI − Λ is a continuous bijection. Then there exists β > 0 such that, for
every bounded linear operator Ψ with norm kΨk < β, the operator ηI − Λ + Ψ is a continuous
bijection. Hence every η̃ ∈ R with |η̃ − η| < β lies in the resolvent set ρ(Λ).
Chapter 7
(ii) Similarly,
kAn St k = sup λnk eλk t ≤ sup (|ω| + r)n e(ω−ηr)t < ∞ .
k r≥0
48
3. The computation of the various integrals is an exercise in basic Calculus.
From the identities
Z ∞ ∞ ∞
T2
Z Z
Wn (t) dt = 1 , tWn (t) = T , (t − T )2 Wn (t) dt =
0 0 0 n
we deduce r
T2
Z
Wn (t) dt ≤ → 0 as n → ∞.
|t−T |>ε n
This proves (7.39).
4. The formula Z ∞
−n
(I − hA) u = wn (t)St u dt (23)
0
is proved by induction on n. The case n = 1 is already known. Now assume (23) is valid.
Using the variable τ = t + s we obtain
Z ∞
−(n+1) −1 −n
(I − hA) u = (I − hA) (I − hA) u = w(t)St (I − hA)−n u dt
0
Z ∞ Z ∞ Z ∞
= w(t)St wn (s)Ss u ds dt = (w ∗ wn )(τ )Sr u dτ .
0 0 0
is proved by using (7.39). Indeed, let ε > 0 By the continuity of the map t 7→ St u there exists
δ > 0 such that
∞
Z
Z Z
≤
S
T u − Wn (t)ST u dt
+ Wn (t)kSt u − ST uk dt + Wn (t)kSt u − ST uk dt
0 |t−T |≤δ |t−T |>δ
≤ 0 + ε + ε 2kuk .
49
5. (i) =⇒ (ii). Assume that St u ∈ Ω for every u ∈ Ω and t ≥ 0. Then
∞
e−t/h
Z
−1
(I − hA) u = St u dt .
0 h
−t/h
This showsRthat (I −hA)−1 u is an integral average of points St u ∈ Ω, with weight w(t) = e h
∞
satisfying 0 w(t) dt = 1. Approximating the integral with a finite sum, we see that (I −
hA)−1 u can be approximated by convex combinations of elements Sti u ∈ Ω. Since Ω is closed
and convex, we conclude (I − hA)−1 u ∈ Ω.
eγh Sh u − u eγh Sh u − Sh u Sh u − u
lim = lim + lim = γu + Au .
h→0+ h h→0+ h h→0+ h
Calling Aγ the generator of the semigroup {eγt St ; t ≥ 0}, this proves that Dom(Aγ ) ⊇
Dom(A) and Aγ (u) = Au + γu.
Inverting the roles of A, Aγ , we see that Dom(A) ⊇ Dom(Aγ ) and Au = Aγ u − γu. This
concludes the proof.
In the following, we first assume 1 ≤ p < ∞. By definition, the function f ∈ Lp (R) lies in the
domain of the generator A if and only if the following limit exists in Lp (R):
f (x + h) − f (x)
= − 2f (x) + lim .
h→0+ h
(26)
50
Hence
n o
Dom(A) = u ∈ Lp (R) ; u is absolutely continuous and ux ∈ Lp (R)
Hence the map t 7→ St u is not continuous from [0, ∞[ into the space L∞ (R).
8. The generator A of the semigroup must be a linear operator on Rn with dense domain.
The only dense subspace of Rn is Rn itself. Hence Dom(A) = Rn , and A must be described
by an n × n matrix. In particular, the linear operator A is continuous, hence for every u ∈ Rn
and t ≥ 0 we must have
d
St u = ASt u, S0 u = 0 .
dt
Since t 7→ St u is the solution to the above Cauchy problem, we conclude that St u = etA u.
9. (i) On the space of all bounded continuous functions w : [0, T ] 7→ X, consider the Picard
operator defined as Z t
.
Φ(w)(t) = St ū + St−s f (s, w(s)) ds .
0
Following the proof of Theorem 7.1, one checks that Φ is a strict contraction w.r.t. the equiv-
alent norm
.
kwk† = max e−2Lt kw(t)k .
t∈[0,T ]
Hence Ψ has a unique fixed point, which by definition provides a mild solution of (7.40).
(ii) Assume that the generator A is a bounded linear operator. Then we can differentiate
(7.41) w.r.t. time, and obtain
d+ St+h ū − St ū 1 t+h
Z
u(t) = lim + lim St−s f (s, u(s)) ds
dt h→0+ h h→0+ h t
t
St+h−s f (s, u(s)) − St−s f (s, u(s))
Z
+ lim ds
h→0+ 0 h
Rt
= ASt ū + f (t, u(t)) + 0 ASt−s f (s, u(s)) ds
51
Observing that the forward derivative (obtained by taking the limit as h → 0+) is a continuous
function of time, we conclude that it coincides with the backward derivative (obtained by
letting h → 0−). This completes the proof.
11. Assume Sτ is compact. If t > τ then St = St−τ Sτ is the composition of the continuous
operator St − τ with the compact operator Sτ , hence it is compact.
To see that the converse may not hold in general, observe that the semigroup (27) is trivially
compact for t ≥ T but not compact for 0 ≤ t < T .
12. (i) (St u)(x) = u(x + t). Not that this is well defined also for t < 0. This is a group of
isometries: kSt ukL1 = kukL1 .
.
(ii) If v = Eh− u = (I −hA)−1 u, then v = u+hvx . We are thus looking for a function v ∈ L1 (R)
such that v − hv 0 = u, hence
v(x) u(x)
v 0 (x) = − x ∈ R.
h h
The explicit solution is provided by
∞
e(x−y)/h
Z
v(x) = u(y) dy.
x h
Setting t = y − x, it is interesting to observe that the above formula is equivalent to
Z ∞ −t/h Z ∞ −t/h
e e
v(x) = u(x + t) dt = (St u)(x) dt .
0 h 0 h
(iii) If u ∈ Cc∞ with support contained in the interval [a, b], then the same is true of the
derivative ux . Hence (Eh+ u)(x) = u(x) + hux (h) is a smooth function supported inside [a, b].
By induction on n we see that the same holds for (Eh+ )n u.
(iv) Take T ≥ b − a. If u vanishes outside [a, b], then ST u is supported inside [a − T, b − T ].
In particular, (ST u)(x) = u(x + T ) = 0 for every x ∈ [a, ∞[ .
On the other hand, every forward Euler approximation vanishes outside the interval [a, b],
hence the same must be true for the limit (if it exists). We conclude that, if the function u is
not identically zero, the forward Euler approximations cannot converge to ST u.
Chapter 8
52
1. (i) This is a distribution of infinite order.
(ii)-(iii) These linear functionals are not distributions on R. They are both distributions on
the open half line ]0, ∞[ .
(iv) This is a distribution of order zero on Ω = ]0, ∞[ .
2. By assumption, f has a weak derivative Df ∈ Lp (]a, b[) ⊂ L1 (]a, b[). By Corollary 8.17, f
coincides a.e. with an absolutely continuous function.
If f is continuously differentiable, for any a < x < y < b, using Hölder’s inequality on the
interval [x, y] we obtain
Z y
1 1
|f (x) − f (y)| ≤ 1 · |f 0 (s)| ds ≤ k1kLq ([x,y]) · kf 0 kLp ([x,y]) + = 1.
x q p
1− p1
k1kLq ([x,y]) = |x − y|1/q = |x − y| , kf 0 kLp ([x,y]) ≤ kf 0 kLp ([x,y]) ≤ kf kW 1,p (]a,b[) .
By approximation, the same inequality holds for every f ∈ W 1,p .
.
5. (i) Consider the mollifications uε = Jε ∗ u. The assumption u ∈ W 1,∞ (Ω) implies that
.
|∇uε (x)| ≤ kukW 1,∞ for all x ∈ Ωε = {x ∈ Ω ; B(x, ε) ⊂ Ω} .
53
Therefore, each uε is Lipschitz continuous with Lipschitz constant L = kukW 1,∞ . Taking the
limit as ε → 0 we obtain the result.
(ii) Consider the open set
.
Ω = {(x1 , x2 ) ; 1 < x21 + x22 < 4} \ {(x1 , x2 ) ; x2 = 0 , x1 > 0}.
6. (i) Consider first the case where Ω is a bounded, open, convex set. Let u be Lipschitz
continuous with constant C. For every fixed (n − 1)-tuple (x1 , x2 , . . . , xi−1 , xi+1 , . . . , xn ), the
function
s 7→ u(x1 , x2 , . . . , xi−1 , s, xi+1 , . . . , xn )
is defined for s in some open interval ]a, b[ (possibly empty). Moreover, it is Lipschitz continu-
ous of the same constant C. Being absolutely continuous, it is differentiable almost everywhere.
We thus conclude that the partial derivative
u(x + hei ) − u(x)
uxi (x) = lim .
h→0 h
exists for a.e. x ∈ Ω. An integration by parts shows that this function provides the weak
derivative Dxi u. Since u is bounded and |uxi (x)| ≤ C at every point x where the partial
derivative exists, it is clear that u ∈ W 1,∞ (Ω).
The result can be easily extended to a general open set Ω, observing that
Z 1
1
≤ cn · dr < ∞ .
n
0 r ln 1 + 1r
54
8. (i) Let f be continuously differentiable. Then
Z
|f (0)| ≤ |f (x)| + |f 0 (y)| dy.
[0,x]
Since C 1 (Ω) is dense in W 1,1 (Ω), this functional can be extended by continuity to a bounded
linear functional on the whole space W 1,1 (Ω).
(ii) Next, assume Ω = B(0, 1) ⊂ R2 . For p > 2, setting γ = 1 − p2 , Morrey’s inequality yields
for some constant C and every f ∈ W 1,p ∩ C 1 . Hence the map T : f 7→ f (0) can be uniquely
extended to a bounded linear functional on the entire space W 1,p (Ω).
To see that this functional is not continuous on W 1,p (Ω) for p ∈ [1, 2], we prove that
|f (0)|
sup = ∞. (28)
f ∈W 1,2 (Ω) kf kW 1,2
Observing that, for every p ∈ [1, 2], one has kf kW 1,p ≤ Cp kf kW 1,2 , from (28) we conclude that
the functional f 7→ f (0) is unbounded in the space kf kW 1,p as well.
To prove (28), consider the function
.
1
f (x) = ln ln 1 +
|x|
and the decreasing sequence of continuous functions
. f (x)
fn (x) = min 1 , .
n
9. Assume p > 2 and set γ = 1 − p2 . Let f ∈ C0∞ (R3 ). Then for every t ∈ R the restriction of
f to the plane
.
Σt = {(t, x2 , x3 ) ; x2 , x3 ∈ R}
lies in Cc∞ (Σt ). Morrey’s inequality yields
we have Z
|f (x̄1 , x2 , x3 )|p + |∇f (x̄1 , x2 , x3 )|p dx2 dx3 < ∞ .
R2
55
If p is suitably large, then f coincides a.e. with a Hölder continuous function on the plane
{(x1 , x2 , x3 ) ; x1 = x̄1 } .
10. Construct a sequence of smooth approximations uν ∈ Cc∞ (Rn ) such that kuν −ukW 1,p (Rn ) ≤ 2−ν−1
.
for every ν ≥ 1. Setting u0 = 0, vν = uν+1 − uν , we have
∞
X
u = w0 + wν , kwν kW 1,p ≤ 2−ν forall ν ≥ 1 .
ν=1
for a.e. z ∈ V ⊥ . If (29) holds, then the sequence of partial sums is absolutely convergent in
W 1,p (y + V ), hence also in C 0,γ (y + V ), by Morrey’s embedding theorem, with γ = 1 − m p.
Hence, restricted to the affine subspace y + V , the function u coincides a.e. with a Hölder
continuous function.
⊥
V
x = v+y
y y+V
V
0 v
Figure 2: Each affine subspace y + V has dimension m < p, hence Morrey’s inequality can be applied.
(ii) For any smooth function w ∈ Cc∞ (Rn ) with compact support, one has
Z Z Z
p
p
|w(y)| dy ≤ kwkC 0,γ (y+V ) dy ≤ C kwkpW 1,p (y+V ) dy = CkwkpW 1,p (Rn ) .
V⊥ V⊥ V⊥
By approximation, we conclude that kukLp (V ⊥ ) ≤ CkwkpW 1,p (Rn ) for every u ∈ W 1,p (Rn ).
is valid for every g ∈ Cc∞ (Ω), by approximation the same holds for every f ∈ W01,p (Ω), after a
modification on a set of measure zero.
56
12. Consider any open set Ω ⊂ Rn . Consider the open subsets
.
n o
Ω1/n = x ∈ Ω ; |x| < n , B(x, 1/n) ⊂ Ω . (30)
By the dominated convergence theorem, for any f ∈ Lp (Ω), 1 ≤ p < ∞, the approximations
. f (x) if x ∈ Ω1/n
fn (x) =
0 otherwise
converge to f in Lp . In turn, for ε < 1/n the mollifications Jε ∗ fn lie in Cc∞ (Ω) and converge
to fn as ε → 0. Hence f can be approximated in Lp (Ω) by smooth functions with compact
support.
Next, consider any sequence of functions fn ∈ Cc∞ (Ω). and assume kfn − f kL∞ → 0. This
implies that f is the uniform limit of a sequence of continuous functions which vanish on the
boundary of Ω. Hence f is a bounded continuous function which satisfies
Viceversa, every bounded continuous function f satisfying (31) lies in W00,∞ (Ω). Indeed, if
this holds, then we can approximate f with a sequence of functions fn,εn ∈ Cc∞ (Ω), choosing
1
fn,εn = Jεn ∗ fn = Jεn ∗ (f · χ ), εn << .
Ω1/n n
.
13. Let fk (x) = f (x)ϕ(k − |x|). Then fk is supported on the set where |x| ≤ k and coincides
with f for |x| ≤ k − 1. By the dominated convergence theorem
Z p !
∂f
kfk − f kpW 1,p ≤ C
X
|f |p +
∂xi dx → 0 as k → ∞.
|x|>k−1 i
Performing the mollifications fk,εk = Jεk ∗ fk with εk << 1/k we obtain a sequence of smooth
functions with compact support which converge to f in W 1,p (Rn ).
.
14. If u ∈ W 2,p (R+ ) then u is absolutely continuous, hence the limit u(0) == limx→0+ u(x)
is well defined. The even extension Eu(x) = u(|x|) is absolutely continuous, with derivative
(Eu)x (x) = (sign x) ux (|x|) for a.e. x ∈ R. Our assumptions imply
Z ∞ Z ∞
|Eu|p + |(Eu)x |p dx = 2 u|p + |ux |p dx < ∞.
−∞ 0
57
15. (i) A direct computation yields
Z 1/q Z 1/q
−n
kuλ k
Lq = q
|u(λx)| dx = q
|u(y)| λ dy = λ−n/q kukLq .
16. Since H 1 (Ω) is a Hilbert space and the sequence (um )m≥1 is bounded, we can extract
a convergent subsequence umk * ũ in H 1 (Ω). Moreover, since the embedding H 1 (Ω) ⊂
L2 (Ω) is compact, by possibly extracting a further subsequence we obtain the convergence
kumk − ukL2 → 0. This clearly implies ũ = u, hence u ∈ H 1 (Ω).
To prove a bound on kukH 1 , set
.
M = lim inf kum kH 1 .
m→∞
M = lim kum kH 1 .
m→∞
Writing
0 ≤ (um − u, um − u)H 1 = (um , um )H 1 + (u, u)H 1 − 2(um , u)H 1 ,
and letting m → ∞, we obtain
M 2 + kuk2H 1 − 2kuk2H 1 ≥ 0 .
Hence kukH 1 ≤ M .
Show that each fn has compact support in Ω. Moreover, kfn −f kW 1,p → 0. Choose a sequence
εn → 0 so that the mollified functions Jεn ∗ fn have compact support and converge to f in H 1 .
(ii) To show that f ∈ W01,2 (Ω0 ), let ϕ(x) = log log 1 + |x|1
. Notice that ϕ ∈ W 1,2 (Ω) as
claimed in problem 6. For each integer n, define the function
o
n n+1
gn (x) = max 0 , min 1 − |x| , n − ϕ(x) .
n
58
Show that each gn has compact support in Ω0 . Moreover, kgn − f kW 1,p → 0. Choose a
sequence εn → 0 so that the mollified functions Jεn ∗ fn have compact support contained in
Ω0 and converge to f in H 1 .
(iii) If the functions ϕn ∈ Cc∞ (Ω0 ) form a Cauchy sequence w.r.t. the W 1,p norm, with p > 2,
then by Morrey’s inequality the sequence (ϕn )n≥1 converges to some continuous function g,
uniformly on the closed disc Ω. In particular, g = 0 at the origin. Derive a contradiction,
showing that f cannot coincide a.e. with a continuous function that vanishes at the origin.
18. (i) Since Ω is connected, by Corollary 8.16 the assumption implies that u coincides a.e. with
a constant. Since the set where u = 0 has positive measure, we conclude that u(x) = 0 for
a.e. x ∈ Ω.
(ii) If (8.82) fails, consider a sequence of functions un such that
1
kun kL2 = 1 , k∇un kL2 < ,
n
and meas {x ∈ Ω ; un (x) = 0} ≥ α for every n.
Since the embedding H 1 (Ω) ⊂ L2 (Ω) is compact, we can extract a subsequence (unk )k≥1 such
that kunk − ukL2 → 0 for some limit function u ∈ L2 (Ω), and moreover unk * u in H 1 (Ω).
By taking a further subsequence, we can assume the pointwise convergence unk (x) → u(x) for
a.e. x ∈ Ω.
By assumption, the weak gradients ∇unk converge to zero in L2 (Ω). Hence ∇u ≡ 0 and
u(x) = c for some constant c and a.e. x ∈ Ω. We claim that c = 0, because otherwise
Z Z
kunk − uk2L2 = |unk (x) − c|2 dx ≥ c2 dx ≥ αc2
Ω {x ; unk (x)=0}
for every k.
On the other hand, the strong convergence in L2 implies kukL2 = limk→∞ kunk kL2 = 1,
providing a contradiction.
19. Fix i ∈ {1, . . . , n} We claim that the function v = uxi , defined a.e. as the classical
derivative of u, provides the weak derivative of u on Rn .
In the following, without loss of generality we assume i = 1 and use the notation x = (x1 , x0 ).
Moreover, we consider the set
.
n o
N = x0 = (x2 , . . . , xn ) ∈ Rn−1 ; (t, x0 ) ∈ K for some t ∈ R .
59
In turn, the assumptions u, ∇u ∈ Lp (Ω) imply u ∈ W 1,p (Ω).
21. Assume 1 ≤ p < ∞. If the result were not true, we could find a sequence un ∈ W 1,p (Ω)
such that
By the compact embedding W 1,p (Ω) ⊂⊂ Lp (Ω) there exists a subsequence unk such that
kunk − ukLp → 0 for some u ∈ Lp (Ω)
The above conditions imply kukLp (Ω0 ) = 0, k∇ukLp (Ω) = 0. Since Ω is connected, we conclude
that u ≡ 0. However, this is in contradiction with kukLp (Ω) = limn→∞ kukLp (Ω) = 1.
imply that v(x) = f 0 (x). Since this limit does not depend on the particular subsequence, we
conclude that the entire sequence ∂x fn converges to ∂x f (x), uniformly on ]0, 1[.
Notice that this implies that ∂x f is also Lipschitz continuous with constant M . The above
shows that f ∈ S, hence S is a closed subset of C 0 (]0, 1[). The above arguments prove both
(ii) and (i).
23. Let Eu ∈ W 1,p (Rn ) be an extension of u to the entire space Rn , with kEukW 1,p (Rn ) ≤
.
CkukW 1,p Ω) . Consider the mollifications uk = J1/k ∗ Eu.
24. The assumption implies that f is absolutely continuous on any bounded interval. Hence
f differentiable at a.e. point. Writing
Z x
f (x) = f (a) + g(s) ds,
a
we have Z x+1/n
1
gn (x) = g(s) ds .
n x
60
This implies the pointwise convergence gn (x) → g(x) at every Lebesgue point x of g, hence
almost everywhere.
To prove that kgn − gkL1 ([a,b]) → 0, let ε > 0 be given. Choose a continuous function ϕ such
that kg − ϕkL1 ([a, b+1]) < ε and define
Z x+1/n
1
ϕn (x) = ϕ(s) ds .
n x
Then
lim sup kgn − gkL1 ([a,b])
n→∞
≤ lim sup kgn − ϕn kL1 ([a,b]) + lim sup kϕn − ϕ|L1 ([a,b]) + lim sup kϕ − gkL1 ([a,b])
n→∞ n→∞ n→∞
≤ ε + 0 + ε.
Chapter 9
1. The PDE can be rewritten as Lu = −f , where Lu = − (ux )x + (xux )y + (uy )y . To
prove that the operator L is uniformly elliptic, it suffices to check that the quadratic form
(ξ1 , ξ2 ) 7→ ξ12 + xξ1 ξ2 + ξ22 is strictly positive definite for (x, y) ∈ Ω. The conclusion is then
obtained by applying Theorem 9.8.
2. (i) The fact that B[u, v] = hu, vi♦ is a continuous bilinear map on H01 is clear. We need to
show that it is strictly positive definite. For |y| ≤ 1 one has
√ √ √
a2 + 2b2 + 2yab ≥ (2 − 2)(a2 + 2b2 ) + 2(a2 + 2b2 ) + 2yab ≥ (2 − 2)(a2 + 2b2 ) .
61
integration by parts, we find that this function u provides a weak solution to the elliptic
boundary value problem
−uxx − 2uyy − (2 + y)uxy = f on Ω ,
u = 0 on ∂Ω .
The operator L is uniformly elliptic on a bounded domain Ω if and only if the closure Ω is
entirely contained in the open set Ω+ .
By assumption,
kφk2L2 (Ω)
β2 = sup ,
u∈H01 ,u6=0 k∇φk2L2 (Ω)
hence µ ≥ 1/β 2 .
(ii) Using the representation formula (9.57), the solution of the parabolic equation with initial
data u(0) = g is provided by
∞
X
u(t) = e−µk t (g, φk )L2 φk , (32)
k=1
where φk ∈ H01 (Ω), the set {φk ; k ≥ 1} is an orthonormal basis of L2 (Ω) consisting of
eigenfunctions of the Laplace operator, and µk are the corresponding eigenvalues. Since µk ≥
β −2 for every k ≥ 1, from (32) it follows
∞ 2 ∞ 2
X −µk t X −t/β 2 2
ku(t)k2L2 = e (g, φk )L2 ≤ e (g, φk )L2 = e−2t/β kgk2L2 .
k=1 k=1
62
By continuity, the same inequality remains valid for every v ∈ H01 (Ω). On the other hand, the
eigenfunction φ1 ∈ H01 (Ω) satisfies
Z Z
2
B[φ1 , φ1 ] = |∇φ1 | dx = − φ1 · ∆φ1 dx = µ1 = µ1 kφ1 k2L2 .
Ω Ω
Similarly,
|∇v|2 dx
R
B[v, v]
µ̃1 = inf = inf RΩe .
06=v∈H01 (Ω) kvk2L2 06=v∈H01 (Ω) Ω
2
e |v| dx
.
6. Choose γ = maxt∈[0,T ] q(t) and consider the operator
.
Lγ u = − (p(t)u0 )0 + (γ − q(t))u .
Since p(t) ≥ θ > 0, the operator Lγ is uniformly elliptic. We claim that the bilinear form
Z T
Bγ [u, v] = p(t)u0 (t)v 0 (t) + (γ − q(t))u(t)v(t) dt
0
for some eigenvalue µk and all v ∈ H01 (Ω). Assume j 6= k and choose v = φj . Then
Z Z
∇φk · ∇φj dx = µk φk φj dx = 0 .
Ω Ω
63
Hence (φk , φj )H 1 = 0.
contradicting the assumption that the bilinear form B[·, ·] is strictly positive definite.
(ii) If the identity
B[um , v] = (f, v)L2
holds for v = ϕ1 , . . . , ϕm , then by linearity it remains valid for every v ∈ span{ϕ1 , . . . , ϕm }.
In particular, since B is strictly positive definite, this implies
βkum k2H 1 ≤ B[um , um ] = (f, um )L2 ≤ kf kL2 kum kL2 ≤ kf kL2 kum kH 1 ,
kum kH 1 ≤ β −1 kf kL2 .
Thanks to the uniform bound, we can extract a subsequence (unj )j≥1 and achieve the weak
convergence unj * u in H01 (Ω). We now have
By linearity
B[u, v] = lim B[um , v] = (f, v) (34)
m→∞
for every v ∈ span{ϕk ; k ≥ 1}. By approximation, (34) remains valid for every v ∈
span{ϕk ; k ≥ 1} = H01 (Ω). This proves that u is a weak solution to the elliptic problem.
9. According to (9.36), the solution u of the boundary value problem satisfies an abstract
equation of the form u − Ku = h, where K : L2 (Ω) 7→ L2 (Ω) is a compact operator and
64
h = L−1 −1
γ f , where Lγ is another compact operator. Moreover, we are assuming that the
operator I − K is a continuous bijection. By the open mapping theorem, the inverse operator
(I − K)−1 is continuous. Being the composition of a compact operator and a continuous one,
the map f 7→ u = (I − K)−1 L−1γ f is compact operator.
10. (i) By Problem 9 in Chapter 2, every function ϕ ∈ Cc∞ (Q) can be uniformly approximated
by a finite linear combination of the functions φm,n . Since Cc∞ (Q) is dense in L2 (Q), we
conclude that span{φm,n ; m, n ≥ 1} is dense in L2 (Q).
A straightforward computation shows that
mπ 2 nπ 2
−∆φm,n = µm,n φm,n , µm,n = + .
a b
(ii) If µ1 is the smallest eigenvalue of −∆ on Ω, then by the result proved in problem 5. the
2 2
assumption Ω ⊆ Q implies µ1 ≥ µ1,1 = πa2 + πb2 . The conclusion now follows from (33).
11. This is a special case of problem 8. Consider the 2 × 2 matrix A = (aij ) and the vector
b = (b1 , b2 ) with entries
Z 3 Z 3
aij = B[ϕi , ϕj ] = ϕ0i (x)ϕ0j (x) dx , bi = 1 · ϕi dx .
0 0
65
On the open domain Ω = {(x, y) ; x2 + y 2 < 1}, the operator L is uniformly elliptic. Indeed,
for |y| ≤ 1 we have
Differentiating w.r.t. time, and using the boundary condition u = ut = 0 on ∂Ω, performing
an integration by parts we obtain
d
E(t) = (ut , utt )L2 + B[u, ut ]
dt
Z Z
y
= ut utt dx + 2ux uxt + (uy uxt + ux uyt ) + 3uy uyt dx
Ω Ω 2
Z Z ux
= ut utt dx − ut 2uxx + yuxy + 2uyy + dx = 0 .
Ω Ω 2
13. (i) The fact that on H01 (Ω) the norm k·k♦ is equivalent to the H 1 norm is a straightforward
consequence of the fact that the bilinear form B[·, ·] is strictly positive definite.
(ii) Repeat the arguments used in the proof of Lemma 9.7.
We claim that, as N → ∞, the above sum converges to a well defined limit, uniformly for t
in compact subsets of ]0, ∞[ . Since the φk form an orthonormal sequence, it suffices to prove
that X 2
lim (−µk )n e−µk t (g, φk )L2 = 0
N →∞
n>N
66
Given any ε > 0, choose N so large that
X
M |(g, φk )|2 < ε .
n>N
X 2 X
n −µk t
(−µk ) e (g, φk )L2 ≤ M |(g, φk )|2 < ε .
n>N n>N
satisfies
.
Bγ [u, u] ≥ β kuk2H 1 , β = min{1, γ}.
Hence for every f ∈ L2 (Ω) there exists a unique u ∈ H 1 (Ω) such that
u − L−1 −1
γ γu = Lγ f .
This is the case if and only if (h, v)L2 = 0 for every solution v of v − Kv = 0, i.e. for every
weak solution of the homogeneous Neumann problem
−∆u = 0
on Ω ,
(37)
∂u = 0
on ∂Ω .
∂ν
By definition, v ∈ H 1 (Ω) is a weak solution of the above problem if
Z
∇v · ∇w dx = 0 for every w ∈ H 1 (Ω) .
Ω
67
Since the open set Ω is connected, we conclude that the solutions of (37) are precisely the
constant functions.
Going back to the original problem,Rby the previous analysis the Neumann boundary value
problem has a solution ifRand only if Ω f · v dx = 0 for every constant function v. Of course,
this holds if and only if Ω f dx = 0.
16. On the space H02 (Ω) consider the continuous bilinear form
Z
.
B[u, v] = ∆u ∆v dx .
Ω
We claim that there exists β > 0 such that B[u, u] ≥ β kuk2H 2 , namely
Z X X Z X Z X Z
uxi xi uxj xj dx ≥ |uxi xj |2 dx + |uxi |2 dx + |u|2 dx . (38)
Ω i j Ω ij Ω i Ω
Hence Z X X Z X
uxi xi uxj xj dx = |uxi xj |2 dx . (39)
Ω i j Ω ij
Together, (39)–(41) yield (38), whenever u ∈ Cc∞ (Ω). By an approximation argument, the
same estimate is valid for every u ∈ H02 (Ω).
Since B[·, ·] is strictly positive definite, the Lax-Milgram theorem yields the existence of a
unique u ∈ H02 (Ω) such that
B[u, v] = (f, v)L2 for every v ∈ H02 (Ω) .
By definition, u is the unique weak solution of the boundary value problem
∆2 u x ∈ Ω,
= f
u = ∂u
= 0 x ∈ ∂Ω ,
∂ν
Chapter 10
68
1. Since K is compact, it can be covered by finitely many of the sets Ai , say K ⊆ A1 ∪ A2 ∪
· · ·∪An . Consider the functions ρi (x) = d(x, Aci ), where Aci = K \Ai denotes the complement
of the set Ai . These functions are all Lipschitz continuous. Namely |ρi (x) − ρi (y)| ≤ d(x, y).
Hence the function r(x) = maxi ρi (x) is also Lipschitz continuous.
We now observe that, if x ∈ Ai then ρ(x) ≥ ρi (x) > 0. Hence the minimum of the continuous
function r(·) on the compact set K is strictly positive. By construction, it is now clear that
.
the constant ρ = minx∈K r(x) > 0 satisfies the requirement.
After Ik−1 ⊂ N has been constructed, we choose an infinite set Ik ⊂ Ik−1 such that
After all the sets Ik have been constructed, we choose a sequence n1 < n2 < n3 < · · · such
that nk ∈ Ik for every k. We claim that the sequence (xnk )k≥1 is Cauchy. Indeed, if m < n,
then d(xm , xn ) ≤ 2−m . Since the space E is complete, this subsequence has a limit.
Next, observe that the mollifier Jε is supported on the interval −ε, ε] and strictly positive on
the subinterval [−ε/2, ε/2]. Namely, it satisfies
δ ε
if 0 < x < ,
.
Jε (x) ≥ φε (x) = ε 2
0 otherwise ,
69
for some constant δ > 0. For 0 < x < 1/6, choosing ε = 2x we obtain
Z 2x Z ln 2x
δ 1 δ 1 δ
F (x) ≥ (J2x ∗f )(x) ≥ (φ2x ∗f )(x) = 2
ds = 2
dy = .
0 2x s(ln s) 2x −∞ y 2x| ln(2x)|
/ L1 (R).
Hence F ∈
We claim that, on the compact interval [−M, M ], the functions fn are equicontinuous. Indeed,
R x+δ
Fix x ∈ [−M, M ] and > 0. Choose δ > 0 such that x−δ g(s) ds < . Then |fn (y)−fn (x)| <
for all y ∈ [x − δ , x + δ] and n ≥ 1.
Given any subsequence, we can thus apply Ascoli’s theorem and extract a further subsequence
(fnk )k≥1 which converges uniformly on the interval [−M, M ]. We now estimate (42) by writing
n
lim sup kfnj − fnk kC(R) = max lim sup kfnj − fnk kC(]−∞,−M ])
j,k→∞ j,k→∞
(43)
+ lim sup kfnj − fnk kC([−M,M ]) + lim sup kfnj − fnk kC([−M,M ])
j,k→∞ j,k→∞
We now have
lim sup kfnj − fnk kC([−M,M ]) = 0 .
j,k→∞
Moreover
Z −M
lim sup kfnj − fnk kC(]−∞,−M ]) ≤ lim sup |fnj (−M ) − fnk (−M )| + 2 g(x) dx ≤ 0 + ε ,
j,k→∞ j,k→∞ −∞
Z ∞
lim sup kfnj − fnk kC([M,∞[) ≤ lim sup |fnj (M ) − fnk (M )| + 2 g(x) dx ≤ 0 + ε .
j,k→∞ j,k→∞ M
.
5. Set gn (x) = arctan fn (x) and consider the functions
. .
a(x) = lim inf gn (x) ≤ lim sup gn (x) = b(x) .
n→∞ n→∞
70
.
The functions a, b are Lebesgue measurable, and so is the set A = {x ∈ R ; − π < a(x) =
.
b(x) < π}. By replacing each function fn with f˜n = χA · fn , it is not restrictive to assume
that the sequence (fn )n≥1 converges pointwise at every point x ∈ R. By Fatou’s lemma,
Z Z
kf kL1 (R) = lim |fn (x)| dx ≤ lim inf |fn (x)| dx ≤ C .
n→∞ n→∞
6. Let f : [a, b] 7→ R be absolutely continuous, and let A ⊂ [a, b] be a set with measure zero.
Fix any ε > 0. We need to show that meas(f (A)) ≤ ε.
Toward this goal, let δ > 0 be such that N
P
i=1 |f (bi ) − f (ai )| < P
ε for any finite family of
disjoint intervals [ai , bi ] ⊂ [a, b], i = 1, . . . , N , with total length ni=1 (bi − ai ) < δ. Since
meas(A) = 0, there exists an open set V ⊇ A such that meas(V ) < δ. Since V ⊂ R is open,
we can write V as S a countable union of its connected components, which are disjoint open
intervals: V = k≥1 ]ck , dk [ . We now have
n
X n
X
meas(f (A)) ≤ meas(f (V )) ≤ sup meas f ([ck , dk ]) = sup f (Mk )−f (mk ). (44)
n n
k=1 k=1
Here mk , Mk ∈ [ck , dk ] are points where the continuous function f attains respectively its
minimum and maximum, restricted to the interval [ck , dk ]. Observing that
n
X n
X
|mk − Mk | ≤ (dk − ck ) < δ for all n ≥ 1,
k=1 k=1
7. (i) Choose a subsequence such that kfnk kL1 ≤ 2−k for every k ≥ 1. Then for any N ≥ 1
we have Z 1 Z 1X
sup |fn (x)| dx ≤ |fn (x)| dx ≤ 2−N .
0 n>N 0 n>N
Calling
.
n o
Aε = x ∈ [0, 1] ; lim sup |fn (x)| ≥ ε ,
n→∞
Since N is arbitrary, this yields meas(Aε ) = 0. The above shows that lim supn→∞ |fn (x)| = 0
for a.e. x ∈ [0, 1]. Hence limn→∞ fn (x) = 0 for a.e. x.
(ii) For 2k < n ≤ 2k+1 , define
n − 2k − 1 n − 2k
1 if x∈ , ,
2k 2k
fn (x) =
0 otherwise .
71
8. Let B(xj , rj ), j ≥ 1 be a countable family of disjoint open balls contained in Ω. For each
j, let gj be the characteristic function of the set B(xj , rj ). Clearly, all these functions lie in
Lp (Ω) and are linearly independent, hence the space Lp (Ω) is infinite dimensional.
Moreover, kgj kL∞ = 1, kgi − gj kL∞ = 1 for i 6= j.
To cover the case 1 ≤ p < ∞, define
1/p
. 1
fj (x) = cj gj (x) , cj = .
meas(B(xj , rj ))
Then
kfj kLp (Ω) = 1 , kfi − fj kLp (Ω) = 21/p for all i 6= j .
9. It is clear that the set S is totally ordered: if (t1 , f (t1 )) and (t2 , f (t2 )) lie in S, with t1 ≤ t2 ,
then (t1 , f (t1 )) (t2 , f (t2 )). To prove that S is maximal, assume (x, y) ∈ / S. To fix the ideas,
assume f (x) < y, the other case being similar. Then there exists t > x such that f (t) < y.
The set S ∪ {(x, y)} is not totally ordered, because the two relations
(x, y) (t, f (t)), (t, f (t)) (x, y)
are both false.
A maximal totally ordered subset of different kind is
.
S = {(−1, y) ; y ≤ 0} ∪ {(x, 0) ; x ∈ [−1, 1]} ∪ {(1, y) ; y ≥ 0} .
10. The proof is by induction. When m = 2, this is the standard Hölder’s inequality. Assume
that the result is true for m = 2, 3, . . . , N − 1. Let fk ∈ Lpk (Ω), with p11 + p12 + . . . + p1N = 1.
If pN = ∞, then by the inductive hypothesis we immediately get
Z Z N
Y
|f1 · · · fN −1 · fN | dx ≤ |f1 · · · fN −1 | dx · kfN kL∞ ≤ kfk kLpk .
Ω Ω k=1
72