Finite difference method
Download as PPTX, PDF11 likes13,857 views
The document discusses finite difference methods for solving differential equations. It begins by introducing finite difference methods as alternatives to shooting methods for solving differential equations numerically. It then provides details on using finite difference methods to transform differential equations into algebraic equations that can be solved. This includes deriving finite difference approximations for derivatives, setting up the finite difference equations at interior points, and assembling the equations in matrix form. The document also provides an example of applying a finite difference method to solve a linear boundary value problem and a nonlinear boundary value problem.
![Finite Difference Method for Linear Problem
The finite difference method for the linear second-order BVP
y = p(x)y + q(x)y + r(x) for a x b with y(a) = 留 and y(b) = 硫
we select an integer N > 0 and divide the interval [a, b] into
(N+1) equal subintervals whose endpoints are the mesh points
xi = a + ih for i = 0, 1, . . . , N+1 where h = (ba)/(N+1)
xi are called collocation points, we find the solution at these points.
5/10/2015 3
x1 xN+1x2 x3 . . .
h h
xN](https://image.slidesharecdn.com/finitedifferencemethod-150508193228-lva1-app6891/85/Finite-difference-method-3-320.jpg)
![5/10/2015 4
Finite Difference Method for Linear Problem
At the interior mesh points, xi, for i = 1, 2, . . . , N, the differential
equation to be approximated is y(xi) = p(xi)y(xi) + q(xi)y(xi) + r(xi)
Expanding y in a third Taylor polynomial about xi evaluated
at xi+1 and xi1, assuming that y C4[xi-1,xi], we have,
y(xi+1) = y(xi + h) = y(xi ) + h.y(1)(xi) + (h2/2).y(2) (xi) + (h3/6).y(3) (xi)
+ (h4/24).y(4)(両 i
+) where 両i
+ (xi,xi+1) (I)
y(xi-1) = y(xi - h) = y(xi ) - h.y(1)(xi) + (h2/2).y(2) (xi) - (h3/6).y(3) (xi)
+ (h4/24).y(4)(両 i
-) where 両i
- (xi,xi+1) (II)](https://image.slidesharecdn.com/finitedifferencemethod-150508193228-lva1-app6891/85/Finite-difference-method-4-320.jpg)
![5/10/2015 5
Finite Difference Method for Linear Problem
Adding I and II , we get
y(xi + h) + y(xi - h) = 2y(xi) + h2.y(2) (xi) + (h4/24).[y(4)(両 i
+) + y(4)(両 i
-)]
(By intermediate value theorem, there exists 両i (両 i
+,両i
-) such that
y(4)(両 i) = [y(4)(両 i
+) + y(4)(両 i
-)] / 2 or 2y(4)(両 i) = [y(4)(両 i
+) + y(4)(両 i
-)] )
y(xi + h) + y(xi - h) = 2y(xi) + h2.y(2) (xi) + (h4/24).[2y(4)(両 i)]
y(2) (xi) = [ [y(xi + h) + y(xi - h) - 2y(xi) ] / h2 ] - (h2/12).[y(4)(両 i)]
Subtracting II from I , we get
y(xi + h) - y(xi - h) = 2hy(1) (xi) + (2h3/6).y(3) (xi)
y(1) (xi) = [ [y(xi + h) - y(xi - h)] / 2h ] - (h2/6).y(3) (xi)](https://image.slidesharecdn.com/finitedifferencemethod-150508193228-lva1-app6891/85/Finite-difference-method-5-320.jpg)
![5/10/2015 6
Finite Difference Method for Linear Problem
Now substituting the value of y(2) (xi) and y(1) (xi) in original
differential equation, we get
[ [y(xi + h) + y(xi - h) - 2y(xi) ] / h2 ] - (h2/12).[y(4)(両 i)]
= p(xi) [ [y(xi + h) - y(xi - h)] / 2h ] - (h2/6).y(3) (xi) + q(xi)y + r(xi)
Simplifying the above equation, we get
-(1+h.p(xi)/2)yi+1 + (2+h2.q(xi))yi (1- h.p(xi)/2)yi = -h2ri
For i=1, -(1+h.p(x1)/2)y2 + (2+h2.q(x1))y1 (1- h.p(x1)/2)y1 = -h2r1
For i=2, -(1+h.p(x2)/2)y3 + (2+h2.q(x2))y2 (1- h.p(x2)/2)y2 = -h2r2
.
.
.
For i=N, -(1+h.p(xN)/2)yN+1 + (2+h2.q(xN))yN (1- h.p(xN)/2)yN = -h2rN](https://image.slidesharecdn.com/finitedifferencemethod-150508193228-lva1-app6891/85/Finite-difference-method-6-320.jpg)
![5/10/2015 14
Example for Non-Linear BVP
Now, Jacobian J =
=
J-1 = 1/[(2y1+12)(2y2+12)-36]
F1 /y1 F1 /y2
F2 /y1 F2 /y2
2y1 +12 -6
-6 2y2 +12
2y1 +12 -6
-6 2y2 +12](https://image.slidesharecdn.com/finitedifferencemethod-150508193228-lva1-app6891/85/Finite-difference-method-14-320.jpg)
![5/10/2015 15
Example for Non-Linear BVP
Method : = - J-1 ((y1)N , (y2)N ) F ((y1)N , (y2)N )
for N=0
= - J-1 ((y1)0 , (y2)0 ) F ((y1)0, (y2)0 )
Now choose,
=
So we have, J
-1
= [1/(144-36)]
(y1)N+1
(y2)N+1
(y1)N
(y2)N
(y1)1
(y2)1
(y1)0
(y2)0
(y1)0
(y2)0
0
0
12 6
6 12](https://image.slidesharecdn.com/finitedifferencemethod-150508193228-lva1-app6891/85/Finite-difference-method-15-320.jpg)
More Related Content
What's hot (20)
Similar to Finite difference method (20)
Recently uploaded (20)
Finite difference method
- 1. Finite Difference Methods 5/10/2015 1 Gaurav Mallik SAU/AM(M)/2014/22 South Asian University Rupali Sharma SAU/AM(M)/2014/27 South Asian University Divyansh Verma SAU/AM(M)/2014/14 South Asian University
- 2. 5/10/2015 2 Finite Difference Methods The most common alternatives to the shooting method are finite-difference approaches. In these techniques, finite differences are substituted for the derivatives in the original equation, transforming a linear differential equation into a set of simultaneous algebraic equations.
- 3. Finite Difference Method for Linear Problem The finite difference method for the linear second-order BVP y = p(x)y + q(x)y + r(x) for a x b with y(a) = 留 and y(b) = 硫 we select an integer N > 0 and divide the interval [a, b] into (N+1) equal subintervals whose endpoints are the mesh points xi = a + ih for i = 0, 1, . . . , N+1 where h = (ba)/(N+1) xi are called collocation points, we find the solution at these points. 5/10/2015 3 x1 xN+1x2 x3 . . . h h xN
- 4. 5/10/2015 4 Finite Difference Method for Linear Problem At the interior mesh points, xi, for i = 1, 2, . . . , N, the differential equation to be approximated is y(xi) = p(xi)y(xi) + q(xi)y(xi) + r(xi) Expanding y in a third Taylor polynomial about xi evaluated at xi+1 and xi1, assuming that y C4[xi-1,xi], we have, y(xi+1) = y(xi + h) = y(xi ) + h.y(1)(xi) + (h2/2).y(2) (xi) + (h3/6).y(3) (xi) + (h4/24).y(4)(両 i +) where 両i + (xi,xi+1) (I) y(xi-1) = y(xi - h) = y(xi ) - h.y(1)(xi) + (h2/2).y(2) (xi) - (h3/6).y(3) (xi) + (h4/24).y(4)(両 i -) where 両i - (xi,xi+1) (II)
- 5. 5/10/2015 5 Finite Difference Method for Linear Problem Adding I and II , we get y(xi + h) + y(xi - h) = 2y(xi) + h2.y(2) (xi) + (h4/24).[y(4)(両 i +) + y(4)(両 i -)] (By intermediate value theorem, there exists 両i (両 i +,両i -) such that y(4)(両 i) = [y(4)(両 i +) + y(4)(両 i -)] / 2 or 2y(4)(両 i) = [y(4)(両 i +) + y(4)(両 i -)] ) y(xi + h) + y(xi - h) = 2y(xi) + h2.y(2) (xi) + (h4/24).[2y(4)(両 i)] y(2) (xi) = [ [y(xi + h) + y(xi - h) - 2y(xi) ] / h2 ] - (h2/12).[y(4)(両 i)] Subtracting II from I , we get y(xi + h) - y(xi - h) = 2hy(1) (xi) + (2h3/6).y(3) (xi) y(1) (xi) = [ [y(xi + h) - y(xi - h)] / 2h ] - (h2/6).y(3) (xi)
- 6. 5/10/2015 6 Finite Difference Method for Linear Problem Now substituting the value of y(2) (xi) and y(1) (xi) in original differential equation, we get [ [y(xi + h) + y(xi - h) - 2y(xi) ] / h2 ] - (h2/12).[y(4)(両 i)] = p(xi) [ [y(xi + h) - y(xi - h)] / 2h ] - (h2/6).y(3) (xi) + q(xi)y + r(xi) Simplifying the above equation, we get -(1+h.p(xi)/2)yi+1 + (2+h2.q(xi))yi (1- h.p(xi)/2)yi = -h2ri For i=1, -(1+h.p(x1)/2)y2 + (2+h2.q(x1))y1 (1- h.p(x1)/2)y1 = -h2r1 For i=2, -(1+h.p(x2)/2)y3 + (2+h2.q(x2))y2 (1- h.p(x2)/2)y2 = -h2r2 . . . For i=N, -(1+h.p(xN)/2)yN+1 + (2+h2.q(xN))yN (1- h.p(xN)/2)yN = -h2rN
- 7. 5/10/2015 7 Finite Difference Method for Linear Problem The system of equations can be expressed in Tri-diagonal nXn matrix form Aw=b, where =
- 8. 5/10/2015 8 Finite Difference Method for Linear Problem let yi=wi
- 9. 5/10/2015 9 Example for Linear BVP Solve (d2y/dx2) = xy with y(0)+y(0)=1 and y(1)=1 such that 0x1 Solution : Here let h= 1/3 So by the formula discussed earlier we have, (yi-1 -2yi +yi+1)/h2 = xi yi (yi-1 -2yi +yi+1) = h2 xi yi (yi-1 -2yi +yi+1) = (1/9)xi yi Now, for i=0 we have, (y-1 -2y0 +y1) = (1/9)x0 y0 (y-1 -2y0 +y1) = 0 1
- 10. 5/10/2015 10 for i=1 we have, (y0 -2y1 +y2) = (1/9)x1 y1 (y0 -2y1 +y2) = (1/9)(1/3)y1 (y0 -2y1 +y2) = (1/27) y1 for i=2 we have, (y1 -2y2 +y3) = (1/9)x2y2 (y1 -2y2 +y3) = (1/9)(2/3)y2 (y1 -2y2 +y3) = (2/27)y2 The unknowns are y-1 , y1 , y2 and y0 Using (yi ) = (yi+1 yi-1 +o(h3))/2h we get, y(0)= (y1 - y-1)/2h 1+ y0 = (y1 - y-1)/2h y-1 = y1 (2/3)(1- y0) putting Eqn (4) in (1), we get, -2 y0 + 3y1 =1 Example for Linear BVP 2 3 4
- 11. 5/10/2015 11 So, -2 y0 + 3y1 =1 (y0 -2y1 +y2) = (1/27) y1 (y1 -2y2 +y3) = (2/27)y2 The matrix will be : The Soln. is y1 = - 0.9879518, y2 = -0.3253012, y3 = 0.3253012 Example for Linear BVP -2 3 0 1 -2-(1/27) 1 0 1 -2-(2/27) y0 y1 y2 1 0 -1 =
- 12. 5/10/2015 12 Finite Difference Method for Non- Linear Problem General form of Non linear BVP: y= f(x,y,y) for axb such that y(a)=留 , y(b)=硫 ie (yi+1 -2yi + yi-1)/h2 = f(xi ,yi , (yi+1- yi-1 )/2h (h2)/6 y(侶))- (h2)/12 y(n)(両 i) y0 = GIVEN and yN+1 = GIVEN For i=1 y2 -2y1 = h2 * f(x1 ,y1 , (y2 留 )/2h) 留 i=2 y3 -2y2 + y1 = h2 * f(x2 ,y2 , (y3- y2 )/2h) i=N -2yN + yN-1)/h2 = f(xN ,yN , (硫- yN-1 )/2h )-硫
- 13. 5/10/2015 13 Example for Non-Linear BVP Solve y=(3/2)y2 with y(0)=4, y(1)=1 such that 0x1 using Newton Method Solution : yi+1 -2yi + yi-1 = (3/2) h2 (yi )2 = (3/2)(1/9)(yi )2 for i=1 we have, y2 -2y1 + y0 =(1/6)(y1)2 for i=2 we have, y3 -2y2 + y1 =(1/6)(y2 )2 So we get, (y1)2 +12y1 -6y2 -24=0 F(y1,y2) (y2 )2 -6y1 + 12y2 -6 =0 F(y1,y2)
- 14. 5/10/2015 14 Example for Non-Linear BVP Now, Jacobian J = = J-1 = 1/[(2y1+12)(2y2+12)-36] F1 /y1 F1 /y2 F2 /y1 F2 /y2 2y1 +12 -6 -6 2y2 +12 2y1 +12 -6 -6 2y2 +12
- 15. 5/10/2015 15 Example for Non-Linear BVP Method : = - J-1 ((y1)N , (y2)N ) F ((y1)N , (y2)N ) for N=0 = - J-1 ((y1)0 , (y2)0 ) F ((y1)0, (y2)0 ) Now choose, = So we have, J -1 = [1/(144-36)] (y1)N+1 (y2)N+1 (y1)N (y2)N (y1)1 (y2)1 (y1)0 (y2)0 (y1)0 (y2)0 0 0 12 6 6 12
- 16. 5/10/2015 16 Example for Non-Linear BVP = (1/108) And F((y1)0, (y2)0 ) = = - (1/108) = 12 6 6 12 (y1)1 (y2)1 0 0 12 6 6 12 -24 -6 3 2 -24 -6
- 17. 5/10/2015 17 References Numerical Analysis (9th Edition) Richard L. Burden, J. Douglas Faires, 2010