Strauss PDEs 2e: Section 1.

2 - Exercise 1 Page 1 of 4

Exercise 1
Solve the first-order equation 2ut + 3ux = 0 with the auxiliary condition u = sin x when t = 0.

Solution

The Geometric Method: Characteristic Curves

Start by rewriting the PDE as

3
ut + ux = 0
2
and then apply the method of characteristics to solve for u. On the paths defined by
dx 3
= , x(ξ, 0) = ξ, (1)
dt 2
the PDE reduces to an ODE,
du
= 0. (2)
dt
That is, u = u(x, t) is constant on the characteristics defined by (1). Integrating (2), we find that
u(ξ, t) = f (ξ),
where f is an arbitrary function of the characteristic coordinate, ξ. Integrating (1), we see that
3
x = t + ξ.
2
Solving for ξ gives
3
ξ = x − t.
2
Therefore,  
3
u(x, t) = f x − t .
2
We can check that this is the solution to the PDE.
ux = f 0
3
ut = − f 0
2
2ut + 3ux = 0, so this is the solution to the PDE. We’re told that u(x, 0) = sin x, though, so we
can determine this unknown function, f .
u(x, 0) = f (x) = sin x
This implies that f (w) = sin w, where w is any expression. Thus,
 
3
u(x, t) = sin x − t .
2
The function is shown below in Figure 1. Shown below that in Figure 2 are the characteristic
curves in the tx-plane for various values of ξ along with the line t = 0 (where the auxiliary
condition is defined). Note that because t = 0 intersects each of the characteristics exactly once,
the solution we obtained for u(x, t) is valid everywhere in the tx-plane, that is, for all x and t.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 1 Page 2 of 4

Figure 1: Plot of u(x, t) for −5 < t < 5 and −5 < x < 5.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 1 Page 3 of 4

Figure 2: Plot of the characteristic curves along with the data curve in the tx-plane.

The Coordinate Method: Change of Variables

To solve this PDE with the coordinate method, start by making the change of variables,

t0 = 2t + 3x
x0 = 3t − 2x.

Solving for the old variables in terms of the new ones gives us
1
t= (2t0 + 3x0 )
13
1
x = (3t0 − 2x0 ).
13

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 1 Page 4 of 4

To find what ut and ux are in terms of these new variables, it’s necessary to use the chain rule.

∂u ∂t0 ∂u ∂x0
ut = + = 2ut0 + 3ux0
∂t0 ∂t ∂x0 ∂t
∂u ∂t0 ∂u ∂x0
ux = 0 + 0 = 3ut0 − 2ux0
∂t ∂x ∂x ∂x
Now we substitute these expressions into the PDE. The transformed equation is

2(2ut0 + 3ux0 ) + 3(3ut0 − 2ux0 ) = 0.

Simplifying this gives

13ut0 = 0
ut0 = 0.

Solve for u by partially integrating both sides with respect to t0 .

u(x0 , t0 ) = g(x0 ),

where g is an arbitrary function of x0 . Now we return to the original variables, x and t.

u(x, t) = g(3t − 2x)

We can check that this is the solution.

ut = 3g 0
ux = −2g 0

2ut + 3ux = 0, which means this is the solution to the PDE. Now plug in the initial condition,
u(x, 0) = sin x to determine g.

u(x, 0) = g(−2x) = sin x

This implies that

 w
g(w) = sin − ,
2
where w is any expression. Therefore,
 
3t − 2x
u(x, t) = sin −
2
 
3
u(x, t) = sin x − t ,
2
and we get the same answer as with the method of characteristics.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 2 Page 1 of 1

Exercise 2
Solve the equation 3uy + uxy = 0. (Hint: Let v = uy .)

Solution

Following the given hint, let v = uy . The PDE transforms to

3v + vx = 0

This is a first-order linear differential equation that can be solved with an integrating factor.
´
3 dx
I=e = e3x

Multiplying both sides of the equation by this integrating factor gives

e3x vx + 3e3x v = 0.

But the left side is simply the x-derivative of the product, e3x v.

∂x (e3x v) = 0

Integrating both sides partially with respect to x gives

e3x v = f (y),

where f is an arbitrary function of y. Solving for v gives

v(x, y) = e−3x f (y)

uy (x, y) = e−3x f (y).

To solve for u, integrate both sides partially with respect to y.

ˆ y
u(x, y) = e−3x f (s) ds + g(x)
ˆ y
−3x
=e f (s) ds + g(x)

Therefore,
u(x, y) = e−3x F (y) + g(x),
where F is an arbitrary function of y and g is an arbitrary function of x. We can check to see if
this solves the PDE.

uy = e−3x F 0 (y)
uyx = uxy = −3e−3x F 0 (y)

And so, 3uy + uxy = 0. This is indeed the correct solution.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 3 Page 1 of 2

Exercise 3
Solve the equation (1 + x2 )ux + uy = 0. Sketch some of the characteristic curves.

Solution

Start by rewriting the PDE as

1
ux + uy = 0
1 + x2
and then apply the method of characteristics to solve for u. On the paths defined by
dy 1
= , y(ξ, 0) = ξ, (1)
dx 1 + x2
the PDE reduces to an ODE,
du
= 0. (2)
dx
That is, u = u(x, y) is constant on the characteristics defined by (1). Integrating (2), we find that

u(x, ξ) = f (ξ),

where f is an arbitrary function of the characteristic coordinate, ξ. Integrating (1) gives

y = tan−1 x + ξ.

Solving for ξ gives

ξ = y − tan−1 x.

Therefore,
u(x, y) = f y − tan−1 x .

We can check that this is the solution of the PDE.

1
ux = − f0
1 + x2
uy = f 0

(1 + x2 )ux + uy = 0, so this is the correct solution. Shown in Figure 1 are the characteristic curves
in the xy-plane for various values of ξ.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 3 Page 2 of 2

Figure 1: Plot of the characteristic curves in the xy-plane.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 4 Page 1 of 1

Exercise 4
Check that (7) indeed solves (4).

Solution

The problem is asking us to check that the solution in (7) satisfies the PDE in (4).

ux + yuy = 0 (4)

u(x, y) = f (e−x y) (7)

Evaluating the first derivatives in x and y gives

ux = (ye−x )x f 0 = −ye−x f 0
uy = (ye−x )y f 0 = e−x f 0 .

Multiplying uy by y gives ye−x f 0 . Therefore,

ux + yuy = −ye−x f 0 + ye−x f 0 = 0,

and (7) indeed solves (4).

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 6 Page 1 of 3

Exercise 6
√
Solve the equation 1 − x2 ux + uy = 0 with the condition u(0, y) = y.

Solution

Start by rewriting the PDE as

1
ux + √ uy = 0
1 − x2
and then apply the method of characteristics to solve for u. On the paths defined by
dy 1
= , y(ξ, 0) = ξ, (1)
dx 1 − x2
the PDE reduces to an ODE,
du
= 0. (2)
dx
That is, u = u(x, y) is constant on the characteristics defined by (1). Integrating (2), we find that

u(x, ξ) = f (ξ),

where f is an arbitrary function of the characteristic coordinate, ξ. Integrating (1) gives

y = sin−1 x + ξ.

Solving for ξ gives

ξ = y − sin−1 x.

Therefore,
u(x, y) = f y − sin−1 x .

We’re told that u(0, y) = y, though, so we can determine this unknown function, f .

u(0, y) = f (y) = y

This implies that f (w) = w, where w is any expression. Thus,

u(x, y) = y − sin−1 x.

We can check that this is the solution of the PDE.

1
ux = − √
1 − x2
uy = 1
√
1 − x2 ux + uy = 0, so this is the correct solution. The function is shown below in Figure 1.
Shown below that in Figure 2 are the characteristic curves in the xy-plane for various values of ξ
along with the line x = 0 (where the boundary condition is defined). Note that because x = 0
intersects each of the characteristics exactly
√ once, the solution we obtained for u(x, y) is valid for
−1 < x < 1 and all y. It’s because of the 1 − x2 term in the PDE that x is restricted.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 6 Page 2 of 3

Figure 1: Plot of u(x, y) for −1 < x < 1 and −1 < y < 1.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 6 Page 3 of 3

Figure 2: Plot of the characteristic curves along with the data curve in the xy-plane.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 7 Page 1 of 3

Exercise 7
2
(a) Solve the equation yux + xuy = 0 with u(0, y) = e−y .

(b) In which region of the xy plane is the solution uniquely determined?

Solution

Part (a)

Start by rewriting the PDE as

x
ux + uy = 0
y
and then apply the method of characteristics to solve for u. On the paths defined by
dy x
= , (1)
dx y
the PDE reduces to an ODE,
du
= 0. (2)
dx
That is, u = u(x, y) is constant on the characteristics defined by (1). Integrating (2), we find that

u(x, ξ) = f (ξ),

where f is an arbitrary function of the characteristic coordinate, ξ. Solving (1) with separation of
variables gives

y dy = x dx
1 2 1 2
y = x + C.
2 2
Multiply both sides by 2 to get rid of the fractions, and let ξ be the constant of integration.

y 2 = x2 + ξ

Solving for ξ gives

ξ = y 2 − x2 .

Therefore,
u(x, y) = f y 2 − x2 .

2
We’re told that u(0, y) = e−y , though, so we can determine this unknown function, f .
2
u(0, y) = f (y 2 ) = e−y

This implies that f (w) = e−w , where w is any expression. Thus,

2 −x2 )
u(x, y) = e−(y .

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 7 Page 2 of 3

Figure 1: Plot of the characteristic curves along with the data curve in the xy-plane.

Part (b)

The data curve x = 0 only intersects the characteristics with a nonnegative ξ value exactly once.
All the characteristics with a negative value for ξ, that is, those in the dark blue and purple
regions never touch the data curve. Therefore, the solution we obtained in the previous part is
only valid for ξ = y 2 − x2 ≥ 0. In the part of the xy-plane where y 2 − x2 < 0, the function f is
undetermined. ( 2 2
e−(y −x ) for y 2 − x2 ≥ 0
u(x, y) =
f y 2 − x2 for y 2 − x2 < 0

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 7 Page 3 of 3

Figure 2: Plot of u(x, y) for −2 < x < 2 and −2 < y < 2. Note that this graph is defined only for
y 2 − x2 ≥ 0.

We can check that this is the solution of the PDE.

ux = −2xf 0
uy = 2yf 0

yux + xuy = 0, so this is the correct solution.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 8 Page 1 of 3

Exercise 8
Solve aux + buy + cu = 0.

Solution

The Geometric Method: Characteristic Curves

Start by rewriting the PDE as

b c
ux + uy = − u
a a
and then apply the method of characteristics to solve for u. On the paths defined by
dy b
= , y(ξ, 0) = ξ, (1)
dx a
the PDE reduces to an ODE,
du c
= − u. (2)
dx a
That is, u = u(x, y) is constant on the characteristics defined by (1). Solving (2) by separation of
variables, we find that
du c
= − dx
u a
c
ln |u| = − x + C(ξ)
a
− ac x C(ξ)
|u| = e e
c
u(x, ξ) = ±eC(ξ) e− a x
c
u(x, ξ) = f (ξ)e− a x ,

where f is an arbitrary function of the characteristic coordinate, ξ. Integrating (1) gives

b
y= x + ξ.
a
Solving for ξ gives
b
ξ = y − x.
a
Therefore,  
b c
u(x, y) = f y − x e− a x .
a
We can check that this is the solution to the PDE.
b c c c
ux = − e− a x f 0 − e− a x f
a a
c
uy = e− a x f 0
c c c c
−a x f 0 − ceX
 − aX
x − a
x 0 − aX
x

−be
X X
aux + buy + cu =  Xf +
be f + ceX Xf
= 0,

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 8 Page 2 of 3

The Coordinate Method: Change of Variables

To solve this PDE with the coordinate method, start by making the change of variables,

x0 = ax + by
y 0 = bx − ay.

Solving for the old variables in terms of the new ones gives us

ax0 + by 0
x=
a2 + b2
bx0 − ay 0
y= 2 .
a + b2
To find what ux and uy are in terms of these new variables, it’s necessary to use the chain rule.

∂u ∂x0 ∂u ∂y 0
ux = + = aux0 + buy0
∂x0 ∂x ∂y 0 ∂x
∂u ∂x0 ∂u ∂y 0
uy = + = bux0 − auy0
∂x0 ∂y ∂y 0 ∂y
Now we substitute these expressions into the PDE. The transformed equation is

a(aux0 + buy0 ) + b(bux0 − auy0 ) + cu = 0.

Simplifying this gives

(a2 + b2 )ux0 + cu = 0.
Now isolate ux0 .
c
ux0 = − u
a2 + b2
Solve for u by separating variables and partially integrating both sides with respect to x0 .
∂u c
0
=− 2 u
∂x a + b2
∂u c
=− 2 ∂x0
u a + b2
c
ln |u| = − 2 x0 + g(y 0 )
a + b2
c
− x0 +g(y 0 )
|u| = e a2 +b2

c
0 − x0
u(x0 , y 0 ) = ±eg(y ) e a2 +b2

c
− x0
u(x0 , y 0 ) = h(y 0 )e a2 +b2 ,

where h is an arbitrary function of y 0 . Now return back to the original variables, x and y.
c
− (ax+by)
u(x, y) = h(bx − ay)e a2 +b2

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 8 Page 3 of 3

We can check that this is the solution.

− c ac
(ax+by) − c (ax+by)
ux = be a2 +b2 h0 −
e a2 +b2 h
+b 2 a2
− c (ax+by) 0 bc − c (ax+by)
uy = −ae a2 +b2 h − 2 2
e a2 +b2 h
a +b
And so

− c
(ax+by)
(( 0 a2 c − a2 +b
(( c
2 (ax+by)
aux + buy + cu = (
abe
((a2(
+b2
(
2
h −2
e h
a +b
c
− 2( ((((0
((ax+by) b2 c − a2 +b c
2 (ax+by) − 2 c 2 (ax+by)
−(abe
((a +b2 h − 2 e h + ce a +b h
a + b2
X2XX2
a + bX − c (ax+by) − c (ax+by)
aux + buy + cu = −cX2X 2 e a2 +b2 h + ce a2 +b2 h
a +XbX
aux + buy + cu = 0.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 9 Page 1 of 3

Exercise 9
Solve the equation ux + uy = 1.

Solution

The Geometric Method: Characteristic Curves

On the paths defined by

dy
= 1, y(ξ, 0) = ξ, (1)
dx
the PDE reduces to an ODE,
du
= 1. (2)
dx
That is, u = u(x, y) is constant on the characteristics defined by (1). Integrating (2), we find that

u(x, ξ) = x + f (ξ).

where f is an arbitrary function of the characteristic coordinate, ξ. Integrating (1) gives

y =x+ξ

Solving for ξ gives

ξ = y − x.

Therefore,
u(x, y) = x + f (y − x).
We can check that this is the solution.

ux = 1 − f 0
uy = f 0

ux + uy = 1, which means this is the solution to the PDE.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 9 Page 2 of 3

Figure 1: Plot of the characteristic curves in the xy-plane for −5 < x < 5 and −5 < y < 5.

The Coordinate Method: Change of Variables

To solve this PDE with the coordinate method, start by making the change of variables,

x0 = x + y
y 0 = x − y.

Solving for the old variables in terms of the new ones gives us
1
x = (x0 + y 0 )
2
1
y = (x0 − y 0 ).
2
To find what ux and uy are in terms of these new variables, it’s necessary to use the chain rule.

∂u ∂x0 ∂u ∂y 0
ux = + = ux0 + uy0
∂x0 ∂x ∂y 0 ∂x
∂u ∂x0 ∂u ∂y 0
uy = + = ux0 − uy0
∂x0 ∂y ∂y 0 ∂y

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 9 Page 3 of 3

Now we substitute these expressions into the PDE. The transformed equation is

(ux0 + uy0 ) + (ux0 − uy0 ) = 1.

Simplifying this gives

2ux0 = 1
1
ux0 = .
2
Solve for u by partially integrating both sides with respect to x0 .
1
u(x0 , y 0 ) = x0 + g(y 0 ),
2
where g is an arbitrary function of y 0 . Now we return to the original variables, x and y.
1
u(x, y) = (x + y) + g(x − y)
2
We can check that this is the solution.
1
ux = + g0
2
1
uy = − g 0
2
ux + uy = 1, which means this is the solution to the PDE.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 10 Page 1 of 5

Exercise 10
Solve ux + uy + u = ex+2y with u(x, 0) = 0.

Solution

The Geometric Method: Characteristic Curves

On the paths defined by

dy
= 1, y(ξ, 0) = ξ, (1)
dx
the PDE reduces to an ODE,
du
+ u = ex+2y . (2)
dx
That is, u = u(x, y) is constant on the characteristics defined by (1). (2) is a first-order linear
ODE that can be solved with an integrating factor. Multiply both sides by
´
1 dx
I=e = ex .

We find that
du
ex + ex u = ex ex+2y
dx
d x
(e u) = e2x+2y
dx
Now integrate both sides.
ˆ
x
e u= e2(x+y) dx + f (ξ),

where f is an arbitrary function of the characteristic coordinate, ξ. Before we go any further, it is

critical to note that y is a function of ξ and x, so we have to solve (1) to proceed. Integrating (1)
gives
y = x + ξ. (3)
Substituting this into the integral yields
ˆ
x
e u= e2(2x+ξ) dx + f (ξ)
1
ex u = e2(2x+ξ) + f (ξ)
4
1 3x+2ξ
u(x, ξ) = e + e−x f (ξ),
4
Solving (3) for ξ gives
ξ = y − x.
And so
1
u(x, y) = e3x+2(y−x) + e−x f (y − x).
4
Simplifying this gives us
1
u(x, y) = ex+2y + e−x f (y − x).
4

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 10 Page 2 of 5

We’re told that u(x, 0) = 0, though, so we can determine this unknown function, f .
1
u(x, 0) = ex + e−x f (−x) = 0
4
Solving for f gives
1
f (−x) = − e2x .
4
This implies that f (w) = − 14 e−2w , where w is any expression. Thus,
 
1 x+2y −x 1 −2(y−x)
u(x, y) = e +e − e
4 4
1 x+2y
− ex−2y .


u(x, y) = e (4)
4
Another form for u(x, y) can be obtained by factoring out ex .

ex 1 2y 1 −2y
 
u(x, y) = e − e
2 2 2

And so
1
u(x, y) = ex sinh 2y.
2
We can check that (4) is the solution to the PDE.
1 x+2y
− ex−2y


ux = e
4
1
2ex+2y + 2ex−2y


uy =
4
Hence,
1 x+2y
1
1 x+2y
− ex−2y + 2ex+2y + 2ex−2y + − ex−2y


ux + uy + u = e e
4
  4  4
1 1 1 x+2y 1 1 1 x−2y
ux + uy + u = + + e + − + − e
4 2 4 4 2 4
ux + uy + u = ex+2y ,

which means that (4) is the solution to the PDE. Shown below in Figure 1 are the characteristic
curves in the xy-plane. Since the data curve (y = 0) intersects each of the characteristics exactly
once, the solution we obtained for the PDE is valid everywhere in the xy-plane, that is, for all x
and y. u(x, y) is shown in Figure 2.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 10 Page 3 of 5

Figure 1: Plot of the characteristic curves in the xy-plane along with the data curve for −5 < x < 5
and −5 < y < 5.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 10 Page 4 of 5

Figure 2: Plot of u(x, y) for −1 < x < 1 and −1 < y < 1.

The Coordinate Method: Change of Variables

To solve this PDE with the coordinate method, start by making the change of variables,

x0 = x + y
y 0 = x − y.

Solving for the old variables in terms of the new ones gives us
1
x = (x0 + y 0 )
2
1
y = (x0 − y 0 ).
2
To find what ux and uy are in terms of these new variables, it’s necessary to use the chain rule.

∂u ∂x0 ∂u ∂y 0
ux = + = ux0 + uy0
∂x0 ∂x ∂y 0 ∂x
∂u ∂x0 ∂u ∂y 0
uy = + = ux0 − uy0
∂x0 ∂y ∂y 0 ∂y
Now we substitute these expressions into the PDE. The transformed equation is
1 0 0 1 0 0
(ux0 + uy0 ) + (ux0 − uy0 ) + u = e 2 (x +y )+2[ 2 (x −y )] .

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 10 Page 5 of 5

Simplifying this gives

1 0 0
2ux0 + u = e 2 (3x −y )
1 1 1 0 0
ux0 + u = e 2 (3x −y ) .
2 2
This is a linear inhomogeneous equation for u, so we have to multiply both sides by the
integrating factor,
´ x0 1 1 0
I(x0 ) = e 2 ds = e 2 x .
1 0 1 1 0 1 1 0 0
e 2 x ux0 + e 2 x u = e 2 (4x −y )
2 2
0
Now the left side can be written as the x -derivative of Iu.
∂  1 x0  1 1 (4x0 −y0 )
e2 u = e2
∂x0 2
Partially integrate both sides with respect to x0 .
ˆ x0
1 0 1 1 (4s−y0 )
x
e u=
2 e2 ds + g(y 0 ),
2
where g is an arbitrary function of y 0 .
1 0 1 1 0 0
e 2 x u = e 2 (4x −y ) + g(y 0 )
4
1 1 0 0 1 0
u(x0 , y 0 ) = e 2 (3x −y ) + e− 2 x g(y 0 )
4
Now we return to the original variables, x and y.
1 1 1
u(x, y) = e 2 [3(x+y)−(x−y)] + e− 2 (x+y) g(x − y)
4
Simplifying this result gives
1 1
u(x, y) = ex+2y + e− 2 (x+y) g(x − y).
4
We’re told that u(x, 0) = 0, though, so we can determine this unknown function, g.
1 1
u(x, 0) = ex + e− 2 x g(x) = 0
4
1 1
e− 2 x g(x) = − ex
4
1 3
g(x) = − e 2 x
4
This implies that
1 3
g(w) = − e 2 w ,
4
where w is any expression. Thus,
 
1 x+2y − 12 (x+y) 1 3 (x−y)
u(x, y) = e +e − e 2
4 4
1 x+2y
− ex−2y ,


u(x, y) =
e
4
which is the same answer we obtained using the method of characteristics.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 11 Page 1 of 3

Exercise 11
Solve aux + buy = f (x, y), where f (x, y) is a given function. If a 6= 0, write the solution in the form
ˆ
u(x, y) = (a2 + b2 )−1/2 f ds + g(bx − ay),
L

where g is an arbitrary function of one variable, L is the characteristic line segment from the y
axis to the point (x, y), and the integral is a line integral. (Hint: Use the coordinate method.)

Solution

Following the hint, we will use the coordinate method to prove this result. Start by making the
change of variables,

x0 = ax + by
y 0 = bx − ay.

Solving for the old variables in terms of the new ones gives us
ax0 + by 0
x=
a2 + b2
bx0 − ay 0
y= 2 .
a + b2
To find what ux and uy are in terms of these new variables, it’s necessary to use the chain rule.
∂u ∂x0 ∂u ∂y 0
ux = + = aux0 + buy0
∂x0 ∂x ∂y 0 ∂x
∂u ∂x0 ∂u ∂y 0
uy = + = bux0 − auy0
∂x0 ∂y ∂y 0 ∂y
Now we substitute these expressions into the PDE. The transformed equation is
 0
ax + by 0 bx0 − ay 0

a(aux0 + buy0 ) + b(bux0 − auy0 ) = f , .
a2 + b2 a2 + b2
Simplifying this gives
ax0 + by 0 bx0 − ay 0
 
2 2
(a + b )ux0 = f , .
a2 + b2 a2 + b2
Dividing both sides by a2 + b2 isolates ux0 .
ax0 + by 0 bx0 − ay 0
 
1
ux0 = 2 f ,
a + b2 a2 + b2 a2 + b2
To solve for u, integrate both sides partially with respect to x0 .
ˆ x0
ar + by 0 br − ay 0
 
0 0 1
u(x , y ) = f , dr + g(y 0 ),
a2 + b2 a2 + b2 a2 + b2
where g is an arbitrary function of y 0 . Now we return to the original variables, x and y.
ˆ ax+by  
1 ar + b(bx − ay) br − a(bx − ay)
u(x, y) = 2 f , dr + g(bx − ay)
a + b2 a2 + b2 a2 + b2

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 11 Page 2 of 3

Simplifying this result gives us

ˆ ax+by  2
b x + a(r − by) a2 y + b(r − ax)

1
u(x, y) = 2 f , dr + g(bx − ay).
a + b2 a2 + b2 a2 + b2
To write this in the form given in the problem statement, we need to have a factor of
(a2 + b2 )−1/2 in front. This implies that we need to make a substitution in the integral such that
dr
√ = ds,
a2+ b2
which means s has to be
r p
√ =s → r= a2 + b2 s.
a2 + b2
This results in
ˆ √ax+by
" √ √ #
1 a2 +b2 b2 x + a( a2 + b2 s − by) a2 y + b( a2 + b2 s − ax)
u(x, y) = √ f , ds
a + b2
2 a2 + b2 a2 + b2
+ g(bx − ay), (1)
which we can write compactly as
ˆ
2 2 −1/2
u(x, y) = (a + b ) f ds + g(bx − ay).
L
Note that the upper limit of integration can be thought of as the dot product between the unit
vector pointing in the direction of the characteristics and the position vector of (x, y). That is, it
is the component of the position vector that lies in the direction of the characteristic curves.
 
ax + by a b
√ = √ , √ · hx, yi
a2 + b2 a2 + b2 a2 + b2
Is (1) the solution to the PDE? We will check to see that it does satisfy the equation. The easiest
way to do this is to notice that (a∂x + b∂y ) of any function of bx − ay is 0. If we apply this
operator to both sides of (1), we get
1
(a∂x + b∂y )u(x, y) = √ ×
a2 + b2
ˆ √ax+by " 2 √ √ #
a2 +b2 b x + a( a2 + b2 s − by) a2 y + b( a2 + b2 s − ax)
(a∂x + b∂y ) f , ds
a2 + b2 a2 + b2
+ (a∂x + b∂y )g(bx − ay).
Since both arguments of f are functions of bx − ay, applying the operator to the integral just
gives us the function inside with s evaluated at the upper limit of integration. But don’t forget
that this has to be multiplied by the operator applied to the upper limit of integration (the chain
rule). (a∂x + b∂y )g(bx − ay) is just abg 0 − abg 0 = 0. Therefore,
1
aux + buy = √ ×
a + b2
2
" √ √ #
b2 x + a( a2 + b2 s − by) a2 y + b( a2 + b2 s − ax) 
 
ax + by
f , (a∂x + b∂y ) √
a2 + b2 a2 + b2


s= √ax+by
a2 + b2
2 2
a +b

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 11 Page 3 of 3

 2
b x + a(ax) a2 y + b(by)
    
1 a b
aux + buy = √ f , a √ +b √
a2 + b2 a2 + b2 a2 + b2 a2 + b2 a2 + b2
 2
a + b2

1
aux + buy = √ f (x, y) √
a2 + b2 a2 + b2
aux + buy = f (x, y).
Hence, (1) is indeed the solution to the PDE, and it can be represented compactly as
ˆ
2 2 −1/2
u(x, y) = (a + b ) f ds + g(bx − ay).
L

Before we call it a night, let’s have a little fun with this formula. Let’s use it to solve Exercise 9,
ux + uy = 1. Here f (x, y) = 1 and a = b = 1. If we plug these values into (1), we get
ˆ x+y
√
1 2
u(x, y) = √ ds + g(x − y).
2
And so
1 x+y
u(x, y) = √ √ + g(x − y)
2 2
x y
= + + g(x − y).
2 2
Although this form is different from the one derived in Exercise 9, it does satisfy the PDE.
1
ux = + g0
2
1
uy = − g 0
2
ux + uy = 1, so this solution we obtained using formula (1) is valid. As another example, let’s find
the solution of ux + uy = xy. Here f (x, y) = xy and a = b = 1. So formula (1) gives us
ˆ x+y
√
" √ #" √ #
1 2 x + ( 2s − y) y + ( 2s − x)
u(x, y) = √ ds + g(x − y).
2 2 2

Performing the integration,

(   )
1 s 2s2 − 3(x − y)2 
u(x, y) = √ √  + g(x − y).
2 12 2  x+y
s= √
2
√
Plugging in (x + y)/ 2 for s,
1
u(x, y) = − (x + y)(x2 − 4xy + y 2 ) + g(x − y).
12
And this is the solution to the example PDE. We can check it very quickly.
1
ux = (−x2 + 2xy + y 2 ) + g 0
4
1
uy = (x2 + 2xy − y 2 ) − g 0
4
Thus, ux + uy = xy. In conclusion, formula (1) works.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 12 Page 1 of 1

Exercise 12
Show that the new coordinate axes defined by (3) are orthogonal.

Solution

The coordinate axes defined by (3) are

x0 = ax + by y 0 = bx − ay. (3)

Treat these coordinate axes as vectors in the plane. We can tell if they are orthogonal by seeing if
the dot product of the two is 0. Recall that the dot product is the sum of the product of the
x-components and the product of the y-components.

x0 = [a, b]
y 0 = [b, −a]

x0 · y 0 = ab + b(−a) = ab − ab = 0
Because the dot product is 0, the coordinate axes defined by (3) are orthogonal.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 13 Page 1 of 5

Exercise 13
Use the coordinate method to solve the equation

ux + 2uy + (2x − y)u = 2x2 + 3xy − 2y 2 .

Solution

The Coordinate Method: Change of Variables

Because we’re using the coordinate method, we have to start by making the change of variables,

x0 = ax + by = x + 2y
y 0 = bx − ay = 2x − y.

The original variables in terms of these new variables are

1
x = (x0 + 2y 0 )
5
1
y = (2x0 − y 0 ).
5
To find what ux and uy are in terms of these new variables, it’s necessary to use the chain rule.

∂u ∂x0 ∂u ∂y 0
ux = 0
+ 0 = ux0 + 2uy0
∂x ∂x ∂y ∂x
∂u ∂x0 ∂u ∂y 0
uy = + = 2ux0 − uy0
∂x0 ∂y ∂y 0 ∂y
Now we substitute these expressions into the PDE. The transformed equation is
 
1 0 0 1 0 0
(ux0 + 2uy0 ) + 2(2ux0 − uy0 ) + 2 (x + 2y ) − (2x − y ) u
5 5
 2     2
1 0 0 1 0 0 1 0 0 1 0 0
= 2 (x + 2y ) + 3 (x + 2y ) (2x − y ) − 2 (2x − y ) .
5 5 5 5

Surprisingly, simplifying this gives

5ux0 + y 0 u = x0 y 0 .

Divide both sides by 5.

y0 1
ux0 + u = x0 y 0
5 5
This is a first-order linear inhomogeneous equation. To solve this equation for u, multiply both
sides by the integrating factor,
´ x0 y0 x0 y 0
I(x0 ) = e 5
ds
=e 5 .

x0 y 0 y 0 x0 y 0 1 x0 y 0
e 5 ux0 + e 5 u = x0 y 0 e 5
5 5

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 13 Page 2 of 5

Now write the left side as the x0 -derivative of Iu.

 0 0 
∂ x y 1 x0 y 0

0
e 5 u = x0 y 0 e 5
∂x 5

Integrate both sides partially with respect to x0 .

ˆ x0
x0 y 0 1 0 sy0
e 5 u= sy e 5 ds + f (y 0 ),
5

where f is an arbitrary function of y 0 .

 
x0 y 0 5x0 y 0
e u=e
5 x − 0 + f (y 0 )
0 5
y
5 x0 y 0
u(x0 , y 0 ) = x0 − 0 + e− 5 f (y 0 )
y
All that’s left to do now is to plug back in the original variables.
5 (x+2y)(2x−y)
u(x, y) = (x + 2y) − + e− 5 f (2x − y)
(2x − y)

Therefore,
5 −2x2 −3xy+2y 2
u(x, y) = x + 2y + +e 5 f (2x − y).
y − 2x
We can check that this is the solution to the PDE.
10 1 −2x2 −3xy+2y2
−(4x + 3y)f + 10f 0
 
ux = 1 + + e 5
(y − 2x)2 5
5 1 −2x2 −3xy+2y2
(−3x + 4y)f − 5f 0
 
uy = 2 − + e 5
(y − 2x)2 5

Hence,
10 1 −2x2 −3xy+2y2  0

ux + 2uy + (2x − y)u = 1 +  2 + e 5 −(4x + 3y)f + 10f
H
(y− 2x)
H
 5 H
10  2 −2x2 −3xy+2y2 
5fZ0

+ 4 −  2 + e 5 (−3x + 4y)f − Z
(y − 2x)
 5
 
5 −2x2 −3xy+2y 2
+ (2x − y) x + 2y + +e 5 f
y − 2x
1 −2x2 −3xy+2y2
ux + 2uy + (2x − y)u = 5
+ e 5 (−10x + 5y)f + (2x − y)(x + 2y) − 5

5
−2x2 −3xy+2y 2
+ (2x − y)e 5 f
( (((2(
−2x2 −3xy+2y 2
(((( 2 −3xy+2y
−2x(
ux + 2uy + (2x − y)u = −( (− y)e
(2x (( ( ( 5 f + (2x − y)(x + 2y) + ( (−
(2x ( y)e
( ( ( 5 f
ux + 2uy + (2x − y)u = 2x2 + 3xy − 2y 2 ,

which means that this is the solution to the PDE.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 13 Page 3 of 5

The Geometric Method: Characteristic Curves

Suppose that we wish to solve the PDE using the method of characteristics. We know that it
collapses to an ODE
du
+ (2x − y)u = 2x2 + 3xy − 2y 2 (1)
dx
on the curves in the xy-plane defined by
dy
= 2, y(ξ, 0) = ξ, . (2)
dx
That is, u = u(x, y) is constant on the characteristics defined by (2). (2) implies that y is a
function of x, so before we can solve (1), we must solve (2) first. Integrating (2) yields

y = 2x + ξ. (3)

Plugging this form for y into (1), we get

du
+ [2x − (2x + ξ)]u = 2x2 + 3x(2x + ξ) − 2(2x + ξ)2 ,
dx
which simplifies to
du
− ξu = −ξ(5x + 2ξ).
dx
This is a linear inhomogeneous differential equation for u. That means we have to multiply both
sides by an integrating factor in order to solve for u.
´x
−ξ ds
I(x) = e = e−ξx
du
e−ξx − ξe−ξx u = −ξ(5x + 2ξ)e−ξx
dx
The left side is just the x-derivative of Iu.
d  −ξx 
e u = −ξ(5x + 2ξ)e−ξx
dx
Now partially integrate both sides with respect to x.
ˆ x
−ξx
e u= −ξ(5s + 2ξ)e−ξs ds + g(ξ),

where g is an arbitrary function of ξ.

1
e−ξx u = e−ξx (5 + 5ξx + 2ξ 2 ) + g(ξ)
ξ
1
u(x, ξ) = (5 + 5ξx + 2ξ 2 ) + eξx g(ξ)
ξ
Solving (3) for ξ gives
ξ = y − 2x.
Plugging this in to u(x, ξ) yields the solution, u(x, y).
1
u(x, y) = [5 + 5(y − 2x)x + 2(y − 2x)2 ] + e(y−2x)x g(y − 2x)
y − 2x

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 13 Page 4 of 5

Simplifying this gives

5
u(x, y) = x + 2y + + ex(y−2x) g(y − 2x).
y − 2x
We can check to see whether this is the solution to the PDE.
10
− ex(y−2x) (4x − y)g + 2g 0
 
ux = 1 + 2
(y − 2x)
5
+ ex(y−2x) xg + g 0


uy = 2 − 2
(y − 2x)

Hence,
10 0
 2 − ex(y−2x) (4x − y)g + Z
 
ux + 2uy + (2x − y)u = 1 + 2g
(y − 2x)
  Z
10 
+ 4 −  2 + 2ex(y−2x) xg + S gS0
 
(y − 2x)
  
5
+ (2x − y) x + 2y + + ex(y−2x) g
y − 2x
ux + 2uy + (2x − y)u = 5
+ ex(y−2x) (−2x + y)g + (2x − y)(x + 2y) − 5
+ (2x − y)ex(y−2x) g
(
(( (
((
ux + 2uy + (2x − y)u = −( (−
(2x (x(y−2x)
y)e
(( (
g + (2x − y)(x + 2y) + ( (−
(2x (x(y−2x)
y)e
(( (
g
ux + 2uy + (2x − y)u = 2x2 + 3xy − 2y 2 ,

which means that this is the solution to the PDE.

www.stemjock.com
Strauss PDEs 2e: Section 1.2 - Exercise 13 Page 5 of 5

Figure 1: Plot of the characteristic curves in the xy-plane for −10 < x < 10 and −10 < y < 10.

www.stemjock.com