VC.

07

Change of Variables to Compute

Double Integrals
 Example 1: A Familiar Double
Integral
Use a double integral to calculate the area of a circle of radius 4 centered
at the origin:
 1 dA
R
The region R is given by x 2  y 2  16 :
4 16  x 2

 
4  16  x 2
1 dA

Now dA means "a small change in area" in

xy-coordinates. We can measure a small change
in area with a little rectangle.
4 16  x 2
dy dA
  1 dy dx
4  16  x 2
So dA = dy dx:
dx
Created by Christopher Grattoni. All rights reserved.
 Example 1: A Familiar Double
Integral
This integral is actually pretty complicated to evaluate:

 
4 16  x 2 4

 
4  16  x 2
1 dy dx 
4
2 16  x 2 dx
4
 1  x 

  16  x  16 sin   
2

  4   4
 16 
What a mess! There must be a better way to
use integration to do this...

What coordinate system would you prefer to

use to when dealing with a region like R?
Created by Christopher Grattoni. All rights reserved.
 Example 2: An Attempt at a Change
of Variables
Right! Let's try polar coordinates:
 x(r, t)  r cos(t) r(x, y)  x 2  y 2
 
 y(r, t)  r sin(t)  t(x, y)  tan  y x  (quadrant adjusted)
1

In this coordinate system, the region R is

described quite elegantly with 0  r  4
and 0  t  2. We will try to (naively) apply
this to our double integral:
2 4

 1 dA    1 dr dt
R 0 0
2 4

   1 dr dt
0 0
 8

Oh no! 8 is not the right answer!! We wanted 16...

Created by Christopher Grattoni. All rights reserved.
 Detour: Working Out the Change of
Variables the Right Way
Our misconception was that dA  dr dt. Our dA refers to a small change
in xy-area, not rt-area. We need to delve a bit deeper to sort this out:
4 16  x 2 2 4

For  
4  16  x 2
1 dy dx our region was a circle, while   1 dr dt looks more
0 0

like the integral you'd take for a rectangle:

x(r ,t)  r cos( t)

y(r ,t)  r sin( t)


r( x ,y )  x 2  y 2
t( x ,y )  tan1  y x 


Created by Christopher Grattoni. All rights reserved.
 Detour: Working Out the Change of
Variables the Right Way
Let's analyze the transformation T(r, t)   r cos t,r sin t  piece by piece
and verify that the picture shown below makes sense:

For
Forthe
For theorange
the red
blue
green line, line,
line,
line,
tLet's map a few points
rrt0,
4,00,00
0,
2 rttr422:4
 ::
just for practice:
x(r ,0)
x(0,t)
x(
x(r ,2
4,t) r0
) 4rcos t
y(r
y(,0)
y(0,t),20)
040sin t

y(r
 


4,t)

This whole process can be called a mapping, a transformation,

a change of variables, or a change of coordinates.
Created by Christopher Grattoni. All rights reserved.
 Detour: Working Out the Change of
Variables the Right Way
Try the Mathematica applet I made to see how a small region
in rt-space maps into xy-space:
This is called a non-area-preserving map.

rt-rectangles map into

xy-sectors. Notice that
this size of the sectors
varies while the size of
the rectangles doesn't...

Now we are ready to start thinking about how a small change in area on
the rectangular region relates to a change in area on the circular region.
Created by Christopher Grattoni. All rights reserved.
 Detour: Working Out the Change of
Variables the Right Way
Now imagine a very small rectangle in the rt-plane: T(r  r, t  t)
(r, t  t) (r  r, t  t)

T(r, t  t)

(r, t) (r  r, t) T(r  r, t)

T(r, t)
That is, consider the limit as A, r, and t tend to 0:
T(r  r, t)  T(r, t)
T(r  r, t)  T(r, t)  r T(r, t  t)  T(r, t)
r T(r, t  t)  T(r, t)  t
t
T T
 r  t
r t
These approximations using partial derivatives let us approximate
our sector using a parallelogram: Created by Christopher Grattoni. All rights reserved.
 Detour: Working Out the Change of
Variables the Right Way
Now imagine a very small rectangle in the rt-plane: T(r  r, t  t)

(r, t  t) (r  r, t  t)

T(r, t  t)

T(r  r, t)
T
(r, t) (r  r, t) t
t T
T(r, t) r
Theorem: The area of the parallelogram formed by two r
vectors V and W equals V  W .

T T T  T (We can take a scalar out of

A  r  t   rt either vector in a cross product)
r t r t

i j k
T1 T2
 r r rt
T1 T2
 0 rt
r r T1 T2
T1 T2
0
t Created t reserved.t
t by Christopher Grattoni. Allrights
 The Area Conversion Factor:
If we let r andt tend to zero, we can us e this to get
T1 T2
dA   r r drdt
T1 T2
t t
 T1 T2  T1 T2
 
 r r  is called the Jacobian matrix, and r r is called the Jacobian determinant.
 T1 T2  T1 T2
 t t  t t

T1 T2
You can let A xy (r, t)  r r . Think of the Jacobian determinant as an
T1 T2
t t
Area Conversion Factor that lets us compute an xy-space integral in rt-space:

Hence,  f(x, y)dA   f(x(r, t), y(r, t))

R xy CreatedRby
A xy (r, t) dr dt
rtChristopher Grattoni. All rights reserved.
 The Area Conversion Factor:

 f(x, y)dA   f(x(r, t), y(r, t))

R xy R rt
A xy (r, t) dr dt

Let T(r, t) be a transformation from rt-space to xy-space.

That is, T(r,t)   T1 (r, t), T2 (r, t)   (x(r, t), y(r, t)).

T1 T2
Then A xy (r, t )  r r .
T1 T2
t t

Note: Since we derived A xy (r, t) as the magnitude of a cross product, we need

it to be positive. This is why weCreated
put absolute
by Christopher value
Grattoni. All rights reserved. bars into the formula above.
 Example 3: Fixing Example 2
Remember we wanted  1 dA for a circle of radius 4 centered at (0,0).
R
T(r, t)  (x(r, t), y(r, t))  (r cos(t),r sin(t))
x  y 2 4

A xy (r, t)  r r  1 dA
R
   1 r d r dt
x  y 0 0
4
t  t 2
r  2


cos(t) sin(t)  0
  dt
 2 0
r sin(t) r cos(t)
2

 r cos2 (t)  r sin2 (t)   8 dt

0


 r cos2 (t)  sin2 (t)   16
r Phew! We did it!
Created by Christopher Grattoni. All rights reserved.
Summary of Change of Variables for
Polar Coordinates

T(r, t)  (x(r, t), y(r, t))  (r cos(t),r sin(t))

A xy (r, t)  r

 f(x, y)dA   f(x(r, t), y(r, t)) r dr dt

R xy Rrt

Created by Christopher Grattoni. All rights reserved.

 Example 4: More With Polar
Coordinates
• By now you probably already asked yourself why this
change of variables is useful. This was just a circle of
course! We could have used A = πr2 or the Gauss-
Green formula.

• Hence, we should look at an example where a double

integral in xy-coordinate space would be horribly
messy, the boundary region is hard to parameterize
for Gauss-Green, and we can’t just plug into a
familiar area formula…
Created by Christopher Grattoni. All rights reserved.
 Example 4: More With Polar
Coordinates
Compute  x y dA for the region satisfying the following inequalities:
2 2

R

x 2  y 2  25, x 2  y 2  4, y  0
This would have been HORRIBLE in rectangular coordinates.
But using polar coordinates, x  r cos(t) and y  r sin(t) this
is the rt-rectangle with 0  t   and 2  r  5 :

   r dr dt
 5

 r cos(t)    r sin(t)
2 2

 x  y dA 
2 2

R 0 2
 5

  dr dt
r 3

0 2

609

Substitute x  r cos(t) and y  r sin(t)
 dt
0
4

609
 
4 Created by Christopher Grattoni. All rights reserved.
 Analogy Time:
This whole change of variables thing isn't just for polar coordinates.
We can change from xy-space to any uv-coordinate space we want:

 f(x, y)dA   f(x(u, v), y(u, v))

R xy R uv
A xy (u, v) du dv

This should be reminiscent of change of variables for single-variable calculus:

x(b) b


x(a)
f(x) dx   f(x(u)) x'(u) du
a

So you can think of the Jacobian Determinant, A xy (u, v) , as a

higher-dimensional analogue of x'(u) in our old friend, u-substitution.

Let's try one in a single variable:

Created by Christopher Grattoni. All rights reserved.
 Example 5: Change of Variables for
Single Variable Calculus
e6 x(b) b
ln(x)
Compute  dx.  f(x) dx   f(x(u)) x '(u) du
e2
x x(a) a
6
ln(x)
  f(x(u)) x '(u) du f(x) 
2 x
6 x(u)  eu
  f(eu ) eu du x'(u)  eu
2
6
u u It works! This is the same as
  u e du
2 e
your more familiar method:
6 u  ln(x)
u  2 6
 
 2 2
 u du  16 1
du  dx
2
x
 16 Created by Christopher Grattoni. All rights reserved.
 Next Up:
• As I said, a change of variables isn’t just good for
polar coordinates. Next we’ll try a few regions
tomorrow that require a different change of variables
with a different Jacobian determinant (area
conversion factor).

Created by Christopher Grattoni. All rights reserved.

 Adjusting the rt-space argument for
general uv-space:
Now imagine a very small rectangle in the uv-plane: T(u  u, v  v)

(u, v  v) (u  u, v  v)

T(u, v  v)

T(u  u, v)
T
(u, v) (u  u, v) (v  b)
v T
(a,b) (u  a)
Theorem: The area of the parallelogram formed by two u
vectors V and W equals V  W .

T T T T (We can take a scalar out of

A  u  v   uv either vector in a cross product)
u v  u v

i j k
T1 T2
 u  u  u v
T1 T2
 0  u v
u u T1 T2
T1 T2
0
v v v v
Created by Christopher Grattoni. All rights reserved.
 An Important Connection
Now imagine a very small rectangle in the uv-plane: T(u  u, v  v)

(u, v  v) (u  u, v  v)

T(u, v  v)

T(u  u, v)
T
(u, v) (u  u, v) (v  b)
v T
(a,b) (u  a)
u

We are using linearizations to approximate the

area of the sector by creating a parallelogram.
You will be responsible for explaining how this
works on your literacy sheet. Make sure you read
B.2 and B.3 in the Basics.
 Example 6: Beyond Polar
Coordinates
2
x
Compute R y dA for the region given by the ellipse y  1.
2 2
  
2
Let's let x  2u cos(v ) and y  usin(v) for 0  v  2 and
0  u  1. These are NOT polar coordinates, so we need
to compute A xy (u, v) :

x y
2cos(v) sin(v)
A xy (u, v )  u u

x y 2u sin(v) ucos(v)
v v
 2ucos2 (v)  2usin2 (v)


 2u cos2 (v)  sin2 (v) 
 2u Created by Christopher Grattoni. All rights reserved.
 Example 6: Beyond Polar
Coordinates
If x(u, v)  2u cos(v), then
  x x   x  2
Compute  y 2 dA for the region given by the, ellipse
   x  gradx[u,
 y 2
 1. v].
 u v   2 
R

Let's let x  2u cos(v ) and y  usin(v) for 0  v  2 and

If y(u, v)  usin(v), then
0  u  1. These are NOT polar coordinates, so we need
to compute A xy (u, v) :  y  y 
 ,   y  grady[u, v].
 u  v 
x y
2cos(v) sin(v)
A xy (u, v)  u u
 So you can think of the area conversion
x y 2u sin(v) ucos(v)
factor as a determinant of a pair of
v v
gradient vectors from x(u,v) and y(u,v).
 2ucos (v)  2usin (v)
2 2

In cos
 2u general,
2 you
(v)  sin2
(v)will start to notice that a

change of variables can often be found
by2u
parameterizing your curve/surface/etc.
Created by Christopher Grattoni. All rights reserved.
 Example 6: Beyond Polar
Coordinates
2
x
Compute R y dA for the region given by the ellipse y  1.
2 2
  
2
2 1

 y dA 
2
 
0 0

u2 sin2 (v )  2u du dv
Substitute x  2ucos( v)
R
2 1
and y  usin(v).
 2   u gsin (v) du dv
3 2

0 0
2 u 1
 u4 
 2   gsin (v)  dv
2

0  4  u 0
2
1
  sin2 (v)dv
20



2 Created by Christopher Grattoni. All rights reserved.
Example 7: Mathematica-Aided Change
of Variables (Parallelogram Region)
Use Mathematica to compute  dA for R given by the parallelogram:
e y

R
These lines are given by y  x  1, y  x  4 ,
1 13 1
y x , and y   x  6.
4 4 4
We can rewrite them as 1  y  x , 4  y  x ,
1 13 1
xy , and x  y  6.
4 4 4
1 13
Then we can let u  y  x and v  x  y for  4  u  1 and  v  6!
4 4
But to compute our integral, we need the map from uv-space to xy-space...
That is, we have u(x, y) and v(x, y ), but we need x(u, v ) and y(u, v).

Let Mathematica do the work: Solve[{u  u[ x , y ], v  v[ x , y ]},{ x , y }]

4 1
x   Created
(u by Christopher
v ) and y  (u  4v )
5 5
Grattoni. All rights reserved.
Example 7: Mathematica-Aided Change
of Variables (Parallelogram Region)
Use Mathematica to compute  dA for R given by the parallelogram:
e y

R
These lines are given by y  x  1, y  x  4 ,
1 13 What is this
1 Mathematica command doing?
y  x , and y   x  6.
4 4 What will4YOU do on a quiz/test without

Mathematica?
We can rewrite them as 1  y  x , 4  y  x ,
1 13 1
xy , and x  y  6.
4 4 4
1 13
Then we can let u  y  x and v  x  y for  4  u  1 and  v  6!
4 4
But to compute our integral, we need the map from uv-space to xy-space...
That is, we have u(x, y) and v(x, y ), but we need x(u, v ) and y(u, v).

Let Mathematica do the work: Solve[{u  u[ x , y ], v  v[ x , y ]},{ x , y }]

4 1
x   Created
(u by Christopher
v ) and y  (u  4v )
5 5
Grattoni. All rights reserved.
Example 7: Mathematica-Aided Change
of Variables (Parallelogram Region)
Use Mathematica to compute  dA for R given by the parallelogram:
e y

4 1
x   (u  v ) and y  (u  4v )
5 5

x y 4 1
 4
A xy (u, v)  u u
 5 5 
x y 4 4 5
v v 5 5

4
So we use
5
Created by Christopher Grattoni. All rights reserved.
Example 7: Mathematica-Aided 4 Change
If x(u, v)   (u  v ), then
of Variables (Parallelogram5 Region)
 x  x 
 ,   x  gradx[u, v].
Rugiven
Use Mathematica to compute  e dA for 
y
v  by the parallelogram:
R

4 1 1
x   (u  v ) and y  (u  4v ) If y(u, v)  (u  4v), then
5 5
5
x y  y y 
4 1  ,   y  grady[u, v].
 4
A xy (u, v)  u u
 5 5    u v 
x y 4 4 5
v v 5
So you can think of the area conversion
5
factor as a determinant of a pair of
gradient vectors from x(u,v) and y(u,v).
4 Again, you can see that our change
So we use
5 of variables is found by parameterizing
our region (in this case, a parallelogram).
Created by Christopher Grattoni. All rights reserved.
Example 7: Mathematica-Aided Change
of Variables (Parallelogram Region)
Use Mathematica to compute  dA for R given by the parallelogram:
e y

4 4 1
A xy (u, v)  , x   (u  v), y  (u  4v),
5 5 5
13
for  4  u  1 and  v6:
4

6 1
4
 e dA 
y
 e   du dv
(u  4v )/5

R 13/ 4 4 5

 417.1 (from Mathematica)

Created by Christopher Grattoni. All rights reserved.