Example Sheet 2 Solutions

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Example Sheet 2 Solutions

1. (i)

h(L ) f, gi = hf, L gi (definition of adjoint)


= hL g, f i (property of inner product)
= hg, Lf i (definition of adjoint)
= hLf, gi (property of inner product)

(ii) (a)

h(L + L )f, gi = hLf, gi + hL f, gi (linear operator property)


= hf, L gi + hf, (L ) gi (definition of adjoint)
= hf, L gi + hf, Lgi (from Qu. 1(i))
= hf, (L + L)gi (linear operator property)

Therefore (L + L ) is self-adjoint.
(b)

h(L L )f, gi = hLf, gi hL f, gi (linear operator)


= hf, L gi hf, (L ) gi (definition of adjoint)
= hf, L gi hf, Lgi (from Qu. 1(i))
= hf, (L L )gi (linear operator)

Therefore (L L ) is skew-adjoint.
(iii) We can write
1 1
L = (L + L ) + (L L ).
2 2
Then from the properties shown in Qu 1(ii) this is the product of a self-adjoint
and skew-adjoint operator.

2. (i) We first apply the operator and take the inner product as follows
Z 1
hv, Lui = v(x)(u u + 2u) dx
Z0 1 Z 1 Z 1

= vu dx vu dx +2 vu dx
0 0 0
| {z } | {z } | {z }
(a) (b) (c)

11
Using integration by parts on (a) and (b) gives
Z 1  1 Z 1
(a) = vu dx = vu

u v dx
0 0 0
 1  1 Z !
1

= vu v u uv dx
0 0 0
 1 Z 1

= vu v u + v u dx
0 0
Z 1  1 Z 1
(b) = vu dx = vu uv dx
0 0 0

therefore
 1 Z 1

v u
(a) (b) + 2(c) = vu vu
(v + v + 2v)u dx
+
0 0
Z 1

= v(1)u (1) v(0)u (0) + (v + v + 2v)u dx
0

where the cancellation is due to applying the boundary conditions. This ex-
pression implies
d2 d
L = + + 2,
dx2 dx
and in order to satisfy the relation
hv, Lui = hL v, ui
we require the following adjoint boundary conditions
B = {v(0) = 0, v(1) = 0} .

(ii) If we use the partition found in Qu 1(iii) we have


1 1
L = (L + L ) + (L L )
2 2  2 
d d
= +2 + .
dx2 dx

3. (i) In order to prove that this is a necessary and sufficient condition we need to
prove the implication in both directions, i.e. we must prove
L is self-adjoint hf, Lf i R
Let us first prove the implication. Suppose L is self-adjoint so L = L and
so
hf, Lf i = hL f, f i
= hLf, f i
= hf, Lf i.

12
Therefore hf, Lf i is real valued (since z = z z R).
To prove the implication, suppose hf, Lf i R f S. Then m C and
some f, g S where f 6= g, we have hf + mg, L(f + mg)i R. Then

0 = [hf + mg, L(f + mgi]



=
[hf, Lf i] + [hf, L(mg)i] + [hmg, Lf i] +

[hmg, L(mg)i]
= [hf, L(mg)i] + [hmg, Lf i]
= [m hf, Lgi] + [m hg, Lf i]
h i
= [m hf, Lgi] + mhLf, gi

therfore
h i
[m hf, Lgi] = mhLf, gi
h i
= m hLf, gi
= [m hLf, gi]

since (z) = (z).


Let m = 1, then

[hf, Lgi] = [hLf, gi] .

Let m = i, then

[i hf, Lgi] = [i hLf, gi] ,


= [hf, Lgi] = [hLf, gi] .

Therefore both real and imaginary parts are in self-adjoint form so L is self-
adjoint.
(ii) An operator is positive if the inner product real and positive valued, and from
part (i) if the inner product is real then L is self-adjoint.

4. Let
 
d du
Lu = p(x) q(x)u,
dx dx

with p > 0, q 0 and u(a) = u(b) = 0. In order for L to be positive the inner
product hf, Lf i must be real and positive. Taking the inner product, we have
Z b    
d du
hu, Lui = u p(x) q(x)u dx
a dx dx
Z b   Z b
d du
= u p dx quu dx
a dx dx a
Z b   Z b
d du
= u p dx q |u|2 dx
a dx dx a

13
Using integration by parts on first term of RHS gives
 ! Z
b
 Z b b
hu, Lui = upu pu
u dx q|u|2 dx (applying BCs)

a a a
Z b Z b
= p|u|2 dx q|u|2 dx 0
a a

since p > 0 and q 0.


5. (iv) Using integration by parts for the inner product we have
Z 1 Z 1
d2 u
hv, Lui = v 2 dx + vu dx
0 dx 0
 1 Z 1 Z 1

= vu u v dx + vu dx
0 0 0
    Z ! Z
1 1 1 1

= vu uv uv dx + vu dx
0 0 0 0
 1 Z 1

= vu uv + (v + v) u dx
0 0
 1
= vu uv + hL v, ui .
0

Therefore L = L. In order to satisfy this condition we require


 1

0 = vu uv
0
   

= v(1)u (1) u(1)v (1) v(0)u (0) u(0)v (0)

   
= v(1)u (1) u(1)v (1) v(0)u (1) u(1)v (0)
   

= u (1) v(1) v(0) u(1) v (1) v (0) .

Therefore
B = {v(0) = v(1), v (0) = v (1)} .
Then, since L = L and B = B the operator is self-adjoint.
(v) Using integration by parts for the inner product we have
Z 1
hv, Lui = v (u + iu + u) dx
0
Z 1 Z 1 Z 1

= vu dx + i vu dx + vu dx
0 0 0
 1 Z 1 !  1 Z 1 ! Z
1
= vu uv + v u dx + i vu uv dx + vu dx
0 0 0 0 0
 1 Z 1
= vu + (v + iv + v)u dx
0 0
= hL v, ui .

14
Therefore L = L with B = {v(0) = 0, v(1) = 0} and so the operator is self-
adjoint.
(vi) Integrating by parts over the given interval gives
Z 2
hv, Lui = vx2 u + vu + vu dx
1
 2 Z 2 " 2 Z 2 " Z
 2
d d
u x2 v vx2 dx +
! 
= u xvu u (xv) dx + vu dx
1 1 dx 1 1 dx 1
2 2 Z 2 "  2
2 !
 
2 d ! 2
 d 2

= u x v vx u u 2 x v dx + xvu . . .
1 dx 1 1 dx 1
Z 2 Z 2

... u (v + xv ) dx + vu dx
1 1
 2 Z 2
2 d ! 2
u 2v + 4xv + x2 v dx . . .
! 
= u x v vx u + xvu +
dx 1 1
Z 2 Z 2

... (uv + xv ) dx + vu dx
1 1
  Z 1
2
u x2 v + 3xv + 2v dx
! 
! 
= u(2) 2v(2) 2xv + x v x=2 u (1)v(1) +
2

Therefore
d2 d
L = x2 2
+ 3x + 2,
dx dx

with

B = {v(2) + 2v (2) = 0, v(1) = 0} ,

so the operator is not self-adjoint.


(vii) Consider the limit as a and b . Then
Z b
hv, Lui = v(u + k 2 u) dx
a
 b Z b

= vu uv +
(v + k 2 v)u dx
a a
 b
= vu uv + hL v, ui .

Therefore L = L. In order to to satisfy this expression we require


   
0 = lim v(b)u (b) u(b)v (b) lim v(a)u (a) u(a)v (a)
b a
   
= lim v(b)iku(b) u(b)v (b) lim v(a)iku(a) u(a)v (a)
b a
   
= lim u(b)(v (b) ikv(b)) + lim u(a)(v (a) + ikv(a))

b a

15
(N.B. Here care must be taken with signs since complex conjugate over the
whole bracket changes the sign of the imaginary part.) We then have
lim (v (x) ikv(x)) = 0,
x
lim (v (x) + ikv(x)) = 0.
x

So L = L but B 6= B so the operator is only formally self-adjoint.


6. Following the usual process we have
Z b
hv, Lui = v(pu + ru + qu) dx
a
Z b Z b Z b

= (pv)u dx + (rv)u dx + (qv)u dx
a a a
 b Z b !  b Z b ! Z
b
d d
= pvu u (pv) dx + rvu u (rv) dx + (qv)u dx
a a dx a a dx a
 b  b Z b !  b
2
d d
= pvu u (pv) u 2 (pv) dx + rvu . . .
a dx a a dx a
Z b Z b
d
... u (rv) dx + (qv)u dx
a dx a

We now choose B so that the boundary terms are 0. We then have


Z b 2 
d d
hv, Lui = (pv) (rv) + qv u dx
a dx dx
d2 d
(pv) = (p v + pv ) = p v + 2p v + pv
dx dx
d
(rv) = r v + rv
dx
then
Z b
hv, Lui = (pv + (2p r)v + (p r + q)v) u dx
a
Z b
= (pv + (2p r)v + (p r + q)v)u dx
a
= hL v, ui

therefore
   2 
d2 dp d dp dr
L = p 2 + 2 r + +q
dx dx dx dx2 dx
as required.
7. Let L = vLu L vu. Then
L = v(pu + ru +
u(pv + (2p r)v + (p r + q)v)
qu)

= p(vu uv ) + rvu u [(2p r)v + (p r )v]

16
so pulling out derivatives we have
 
d




L= p(vu uv ) p (vu uv ) p(v u u v ) + . . .
dx
 
d
... (rvu) r vu rv u + . . .
dx
 
d


. . . 2p v u rv u + p r )vu (p r)v u (p r)vu
| {z } dx
leave alone for now
 
d
= p(vu uv ) + rvu (p r)vu + . . .
dx
 


. . . p ( vu uv ) r vu rv u

2p v u +
rv u + p v u
rv u +
p vu rvu

| {z }
d
=(r vurv urvu )= dx
(rvu)

thus
d 
L= p(vu uv ) + (r p )uv
dx
as required.
8. (i) For L to be formally self-adjoint we require L = L. Comparing the coefficients
of each term we have
d2
: p = p = p is real
dx2
d r + r
: 2p r = r = p =
dx 2  
r + r (r) i(r) + (r) + i(r)
= p = =
2 2
= p = (r)

constant : p r q = q = q = p r + q
but p = (r) so
d   d 

(r)
(r) + i(r) = q q
dx dx
d 
= i (r) = q q
dx
d  q q
= (r) =
dx i
i2(q)
=
i
= 2(q)

(ii) If L is real then p, q, r are real functions. From part (i) we already have that
p is real. Then from the constant term we had
d 
(r) = 2(q),
dx

17
but since q and r are real we have (r) = (q) = 0 so the above expression is
satisfied. That leaves us with the expression p = (r) but since r is real this
implies p = r as required.

9. (i) Since u, v are real we have u = u and v = v. Then

d4 u d4 v
vLu uLv = v u ,
dx4 dx4 
d2 d2 u d2 v
= v 2 u 2 ,
dx2 dx dx

which is an exact differential as required.


(ii)
Z 1 Z 1
d2
vLu uLv dx = 2
(vu uv ) dx
0 0 dx
 1
d
= (vu uv )
dx 0
 1

= v u + vu u v uv
0
= v (1)u (1) + v(1)u (1) u (1)v (1) u(1)v (1) . . .

. . . v (0)u (0) v(0)u (0) + u (0)v (0) + u(0)v (0)

(iii) First cancel terms in u:


Z 1
(1) . . .
u
vLu uLv dx = v (1) (1) + v(1)u(1) u(1)v (1)
u(1)v
0
(0) + u(0)v
(0)
u
. . . v (0)u (0) v(0)u (0) + (0)v

then in v:
Z 1
(1) u (1)v
(0) (0).

vLu uLv dx =
v(1)u v (0)u
(1) v(0)u
0

(iv) In order to give the required result, we choose the other set of boundary
conditions, i.e.

(0) = 0, (1) = 0,
(0) = 0, (1) = 0.

18

You might also like