17.05.2015 Views

Numerical Integration Over Polygonal Domains using Convex ... - Ijecs

Numerical Integration Over Polygonal Domains using Convex ... - Ijecs

Numerical Integration Over Polygonal Domains using Convex ... - Ijecs

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

x<br />

y<br />

x<br />

y<br />

x<br />

y<br />

(1)<br />

(1)<br />

(2)<br />

(2)<br />

(3)<br />

(3)<br />

( , )<br />

<br />

( , )<br />

<br />

( , )<br />

<br />

( , )<br />

<br />

( , )<br />

<br />

( , )<br />

<br />

1<br />

x<br />

p<br />

xq<br />

xr<br />

1<br />

y<br />

p<br />

yq<br />

yr<br />

1<br />

xq<br />

xr<br />

x<br />

p<br />

1<br />

yq<br />

yr<br />

y<br />

p<br />

1<br />

xr<br />

x<br />

p<br />

xq<br />

1<br />

yr<br />

y<br />

p<br />

yq<br />

with ( , )<br />

, (<br />

, )<br />

as given in eqns. (19) and (24).<br />

(26)<br />

(27)<br />

(28)<br />

6 <strong>Numerical</strong> <strong>Integration</strong> Formulas<br />

6.1 <strong>Numerical</strong> integration over an Arbitrary Triangle PQR<br />

We could use either of the formulas in eqn.(16) or eqns.(26-28). We prefer to use eqn.(25), since it requires<br />

(1) (1)<br />

the computation of just one set of ( u,<br />

v)<br />

( u , v ) for all the three quadrilaterals. The transformation<br />

formulas of eqns. (26-28) are easy to implement as a computer code, since the coordinates of PQR<br />

are to<br />

( e)<br />

( e)<br />

(1) (1)<br />

be used in cyclic permutation in ( x , y ), e 1,2,3)<br />

. Note that in ( x , y )<br />

T<br />

T<br />

T<br />

( x<br />

p,<br />

y<br />

p<br />

) , ( xq,<br />

yq<br />

) , ( xr<br />

, yr<br />

) respectively. In<br />

T<br />

T<br />

T<br />

( x<br />

q,<br />

yq<br />

) , ( xr<br />

, yr<br />

) , ( xp,<br />

y<br />

p<br />

) respectively and in<br />

(2) (2)<br />

( x , y )<br />

(3) (3)<br />

( x , y )<br />

H. T. Rathod a IJECS Volume 2 Issue 8 August, 2013 Page No.2576-2610 Page 2582<br />

T<br />

T<br />

T<br />

the coefficients of w , u,<br />

v are<br />

the coefficients of w , u,<br />

v are<br />

the coefficients of w , u,<br />

v are<br />

T<br />

T<br />

T<br />

( x<br />

r<br />

, yr<br />

) , ( x<br />

p,<br />

y<br />

p<br />

) , ( xq,<br />

yq<br />

) respectively. We can use Gauss Legendre quadrature rule to evaluate eqn.(25).<br />

The resulting numerical integration formula can be written as<br />

PQR<br />

NN<br />

<br />

3<br />

( e)<br />

( N ) ( N ) ( e)<br />

( N ) ( N )<br />

f ( x ( U<br />

k<br />

, Vk<br />

), y ( U<br />

k<br />

, Vk<br />

)) <br />

( N )<br />

I I ( f ) 2<br />

W<br />

(29)<br />

pqr<br />

k 1<br />

k<br />

e1<br />

The weights and sampling points in the above formula satisfy the relation<br />

N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( W , U , V ) (4 s s ) w w 96, u(<br />

s , s ), v(<br />

s , s<br />

( ( N)<br />

) <br />

k<br />

k<br />

k<br />

i<br />

j<br />

i<br />

j<br />

k 1 ,2,3,......<br />

N N , i , j 1,2,3 ,...... N<br />

(30)<br />

and<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

u(<br />

s , s ) (1 s )(1 s ) 12 (1 s )(1 s ) 8,<br />

i<br />

j<br />

i<br />

j<br />

i<br />

j<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

( N)<br />

v(<br />

si<br />

, s<br />

j<br />

) (1 si<br />

)(1 s<br />

j<br />

) 12 (1 si<br />

)(1 s<br />

j<br />

) 8<br />

(31)<br />

( N ) ( N )<br />

for a N- point Gauss Legendre rule of order N with (<br />

wn , sn<br />

), n 1, 2,3,......<br />

N as the weights and<br />

sampling points respectively.<br />

( N ) ( N ) ( N )<br />

2<br />

We can compute the arrays ( Wk , U<br />

k<br />

, Vk<br />

), k 1, 2,3,...... N for any available Gauss Legendre<br />

quadrature rule of order N. We have listed a code to compute the arrays<br />

( N ) ( N ) ( N )<br />

2<br />

( Wk , U<br />

k<br />

, Vk<br />

, k 1, 2,3,...... N ) for N = 5, 10, 15, 20, 25, 30, 35, 40. This is necessary since explicit list<br />

( N ) ( N ) ( N )<br />

of ( Wk , U<br />

k<br />

, Vk<br />

, N = 5, 10, 15, 20, 25, 30, 35, 40)<br />

will generate a large amount of values, viz: 25, 100,<br />

( N ) ( N ) ( N )<br />

225, 400, 625, 900, 1225, and 1600 for each of W<br />

k<br />

, U<br />

k<br />

, Vk<br />

. The computer code will be simple with<br />

( N ) ( N )<br />

few statements and it requires the input values of (<br />

w , s , n 1,2,..., N),<br />

N 5,10,15, 20,25,30,35,40<br />

6.2 Composite <strong>Integration</strong> over a polygonal Domain P N<br />

<br />

<br />

i<br />

j<br />

i<br />

j<br />

n n<br />

.<br />

We now consider the evaluation of II<br />

<br />

( f ) f ( x,<br />

y)<br />

dxdy , where f C()<br />

and is any polygonal<br />

2<br />

domain in . That is a domain with boundary composed of piecewise straight lines. We then write<br />

M<br />

M<br />

<br />

2<br />

<br />

<br />

<br />

P<br />

( f ) <br />

T<br />

( f ) <br />

<br />

Q<br />

( f )<br />

<br />

(32) in which, we define<br />

N<br />

n<br />

3 n p<br />

n1<br />

n1<br />

p0<br />

<br />

P is a polygonal domain of N oriented edges ( k i 1,<br />

i 1,2,3,..., N),<br />

with end points x , y ), x , y )<br />

N<br />

l ik<br />

(<br />

i i<br />

(<br />

k k

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!