SHUANGLIN SHAO
1. P5. #1.
Proof. Equations 1, 2, 4, 7, 8 are linear equations. Equations 3, 5, 6 are
nonlinear equations.
2. P5. #2.
Proof. (a). linear.
L(au + bv) = (au + bv)x + x(au + bv)y
= aux + bvx + xauy + xbvy
= a(ux + xuy ) + b(vx + xvy )
= aL(u) + bL(v).
This proves that L is linear.
(b). nonlinear.
L(au + bv) = (au + bv)x + (au + bv)(au + bv)y
= aux + bvx + (au + bv)(auy + bvy )
= aux + bvx + a2 uuy + b2 vvy + abuvy + bavuy .
On the other hand,
aL(u) + bL(v) = a(ux + uuy ) + b(vx + vvy )
= aux + bvx + auuy + bvvy .
If a = 2, b = 0, these two identities above not equal; otherwise
uuy = 0, implies that (u2 )y = 0, i.e., u2 (x, y) = u2 (x, 0).
The last condition put a constraint on u, which is not known. So the linearity
fails.
1
(c). nonlinear.
L(au + bv) = (au + bv)x + ((au + bv)y )2
= aux + bvx + (auy + bvy )2
= aux + bvx + a2 u2y + b2 vy2 + 2abuy vy .
On the other hand,
aL(u) + bL(v) = a(ux + u2y ) + b(vx + vy2 )
= aux + bvx + au2y + bvy2 .
Similarly as in (b), if a = 2, b = 0, if L is linear, u2y = 0. So
uy = 0.
This is not known. So linearity fails.
(d). nonlinear.
L(au + bv) = (au + bv)x + (au + bv)y + 1
= a(ux + uy ) + b(vx + vy ) + 1.
On the other hand
aL(u) + bL(v) = a(ux + uy + 1) + b(vx + vy + 1)
= a(ux + uy ) + b(vx + vy ) + a + b.
So if a = b = 1, if L is linear, we have 1 = 2. So linearity fails. Another way
to see this is that L does not map zero functions to zero.
(e). linear. The proof is similar to that in (a). So we omit it.
3. P5. # 3.
Proof. (a). It is a linear inhomogeneous equation. Define
L(u) = ut uxx , g = 1.
(b). It is a linear homogeneous equation. Define
L(u) = ut uxx + xu.
(c). nonlinear. The proof is similar to that in (b) in #2; so we omit it.
(d). It is a linear inhomogeneous equation. Define
L(u) = ut uxx , g = x2 .
2
1
u.
x
p
p
1 + u2x
1 + 4u2x
= 0.
+ 2uy q
q
2
2
1 + uy
1 + 4uy
L(u) = ut + uxxxx + 1 + u,
then L(0) = 1 6= 0. So linearity fails.
4. P5. # 4.
Proof. Let u and v solve the linear homogeneous equation L(u) = g. Then
we have
L(u) = g,
L(v) = g.
So we have
L(u v) = 0.
So the difference u v solves the equation L(u) = 0.
3
5. P5. #10.
000
00
000
00
6. P5. #11.
Proof. To the equation, uuxy = ux uy ,
LHS = f (x)g(y) (f (x)g(y))xy = f (x)fx (x)g(y)gy (y).
On the other hand,
RHS = ux uy = (f (x)g(y))x (f (x)g(y))y = f (x)fx (x)g(y)gy (y).
Thus u(x, y) = f (x)g(y) is a solution.
7. P9. #1.
Proof. By formula (2) in Section 1.2,
u(x, t) = f (2x 3t),
where f is a function of one variable. Given the condition, we have
u(x, 0) = f (2x) = sin x.
So f (x) = sin x3 . Thus
u(x, t) = sin
2x 3t
3
= sin(x t).
2
2
4
8. P9. #2.
Proof. Let v = uy . Then 3v + vx = 0. We multiply both sides by e3x ,
vx e3x + 3e3x v = 0.
By using the chain rule of differentiation,
(ve3x )x = 0.
Thus by the fundamental theorem of calculus,
v(x, y)e3x v(0, y) = 0.
Set g(y) = v(0, y). Then
v(x, y) = e3x g(y).
To solve for u,
uy = e3x g(y).
Again by the fundamental theorem of calculus,
Z y
3x
g(t)dt.
u(x, y) u(x, 0) = e
0
Ry
0
g(t)dt. We have
9. P10. # 3.
Proof. We follow the method in solving Equation (4) in section 1.2. The
equation is
1
ux +
uy = 0.
1 + x2
1
So the directional derivative of u along the direction (1, 1+x
2 ) is zero. We
look for the equations of these characteristic curves y = y(x) in the plane
1
such that their directions are (1, 1+x
2 ). Their equations are
dy(x)
1
=
.
dx
1 + x2
Thus integrating from 0 to x, we obtain
Z x
1
y(x) y(0) =
dt = tan1 x.
2
1
+
t
0
Hence
y(x) = tan1 x + y(0);
5
y 0 (x)
y(x)
= x1 . Thus
d(ln y(x))
1
= .
dx
x
11. P10. # 8.
Proof. The equation is
(1)
aux + buy + cu = 0.
Thus
uy
ux
+b
= c.
u
u
Let v = ln u. The equation above is
a
avx + bvy + c = 0.
Thus without loss of generality, we assume that a 6= 0. Thus
c
c
a(v + x)x + b(v + x)y = 0.
a
a
Therefore we have
c
v + x = f (bx ay),
a
where f is a function of one variable. Since u = ev ,
c
(2)
Another way is to use the coordinate method. From the equation a uux +b
c. Let v = ln u. Then
avx + bvy = c.
0
We change variables: x = bx ay and y 0 = ax + by. Then
vx = bvx0 + avy0 ,
vy = avx0 + bvy0 .
Thus
c = avx + bvy = (a2 + b2 )vy0 .
Hence we have
vy 0 =
c
.
a2 + b2
cy 0
.
a2 + b2
Therefore
cy 0
a2 +b2
i.e.,
(3)
c
(ax+by)
a2 +b2
e a x = e
c
bc
(ax+by)
(bxay)
a2 +b2
a(a2 +b2 )
uy
u
Then we write
ac x+f (bxay)
=e
c
(ax+by)
a2 +b2
bc
12. P10. # 9.
Proof. Let v = u x. Then the original equation is
vx + vy = 0.
Thus we see that
v(x, y) = f (x y),
where f is a function of one variable. Therefore v(x, y) = x + f (x y).
13. P10. # 10.
Proof. Let x0 = x + y and y 0 = x + y. Then
ux = ux0 + uy0 ,
uy = ux0 + uy0 ,
y 0 x0
,
2
x0 + y 0
y=
.
2
Thus the equation that ux + uy + u = ex+2y becomes
x=
1 x0 3 0
1
uy0 + u = e 2 + 2 y .
2
2
y0
Hence
Z
0
y0
1 x0
0
e2t dy 0 = e 2 (e2y 1).
4
y0
1 x0 y0
0
u(x0 , y 0 ) = e 2 (e2y 1) e 2 u(x0 , 0).
4
In terms of (x, y)-variables,
x+y
1 x+2y
u(x, y) =
e
ex e 2 u(y x, 0).
4
One can verify that u(x, y) satisfies that ux +uy +u = ex+2y . But if imposing
the condition that u(x, 0) = 0, we obtain a contradiction. Indeed, since
u(x, 0) = 0 for any x,
u(y x, 0) = 0.
Thus
1
(4)
u(x, y) = (ex+2y ex ).
4
This is also a solution to the equation. Let y = 0 in (4). Thus
1
u(x, 0) = (ex ex ),
4
which is generally not zero. Therefore there is no solution for this equation
with u(x, 0) = 0.