(837, #11) Estimate

where R = [0,1] x [0,1]. Use the midpoint rule and the following number of squares:

1, 4 16, 64, 256, 1024.

Solution: Let N be a nonnegative integer. Partion R into squares by means of the points

for i = 0..2N and j = 0..2N. Note that the corresponding partition has 2N 2N = 22N rectangles. Furthermore If Ri, j is the rectangle whose lower left and upper right

corners are, respectively,

then

is the midpoint of Ri, j .

Thus, the Maple code for eestimating the double integral here is

> f := (x,y)-> exp(-x^2-y^2):

S := N -> sum(sum(f( (i-0.5)/2^N, (j-0.5)/2^N), i = 1..2^N), j = 1..2^N)*(1/4^N);

Evaluating S(N) for N = 0..5 we get the following tabulation.

N

# of squares

S(N)

0

1

.6065306597

1

4

.5694180568

2

16

.5606222675

3

64

.5584626388

4

256

.5579252098

5

1024

.5577910063

A graph of the underlying solid shows that the estimate is reasonable: The average value of the function f(x,y) is about 1/2; the domain has area 1. The estimates should be in the neighborhood of 1/2, which they are.


by:rjm