Sie sind auf Seite 1von 17

Elliptic Partial Differential Equations)

Iterative Methods

Examples

An International Workshop on Advances in Computational Partial Differential Equations


Dr. P. Dhanumjaya
BITS-Pilani, Goa Campus

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

General Second Order PDE We consider second order linear partial differential equation of the form: a(x, y )uxx + 2b(x, y )uxy + c(x, y )uyy + d(x, y )ux +e(x, y )uy + g(x, y )u(x, y ) = f (x, y ), in a domain , where the coefcients are independent of u(x, y ).

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Classication Elliptic if b2 ac < 0, for all (x, y ) . Parabola if b2 ac = 0, for all (x, y ) . Hyperbola if b2 ac > 0, for all (x, y ) .

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Examples

ut uxx = f (x, t). uxx + uyy + u = f (x, y ). utt 2 uxx = f (x, t). uxx + yuyy = 0.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Elliptic PDE Consider the Poisson equation uxx + uyy = f (x, y ), (x, y ) = (a, b) (c, d) with Dirichlet boundary conditions u(x, y )| = u0 (x, y ). If f L2 () then the solution exists and it is unique. Note that the analytic solution is rarely available.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

We discuss how to use nite difference method to solve Poisson equation Generate a grid. A uniform Cartesian grid can be used xi yj = a + ihx , i = 0, 1, 2, , m, hx = ba , m d c = c + jhy , j = 0, 1, 2, , n, hy = . n

We want to nd an approximate solution Uij . Here, there are (m 1)(n 1) unknowns for Dirichlet boundary condition.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Finite Difference Method Substitute the partial derivatives with a nite difference formula in terms of the function values at grid points to get u(xi1 , yj ) 2u(xi , yj ) + u(xi+1 , yj ) (hx )2 u(xi , yj1 ) 2u(xi , yj ) + u(xi , yj+1 ) (hy )2

= fij + Tij ,

where i = 1, 2, 3, , m 1, j = 1, 2, 3, , n 1 and fij = f (xi , yj ).

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Local Truncation Error The local truncation error satises Tij (hx )2 4 u (hy )2 4 u + . 12 x 4 12 y 4

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Let Uij be an approximation to uij , then Ui1,j + Ui+1,j (hx )


2

Ui,j1 + Ui,j+1 (hy )


2

2 (hx )
2

2 (hy )2

Uij = fij .

The nite difference scheme at a grid point (xi , yj ) involves ve grid points, east, north, west, south and the center. The center is called the master grid point.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Solve the linear system of equations to get an approximate solution at grid points. Error analysis and visualization. We can solve the system of equations using the following three iterative methods.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Jacobi Method This is based on the transformation of the linear system Ax = b, into the system x = Cx + d, in which the matrix C has zeros on the diagonal.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Jacobi Method The vector x is updated using the previous estimate for all components of x to evaluate the right-hand side of the equation Uij
(n+1)

1 2 (n) 2 (n) {hx Ui,j1 + hx Ui,j+1 2 2 2 hx + hy


2 2 2 2 +hy Ui1,j + hy Ui+1,j hx hy fij .} (n) (n)

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

The Gauss-Seidel Method This method of solving linear system is a simple iterative technique obtained by transforming the linear system Ax = b, into the system x = Cx + d, in which the matrix C has zeros on the diagonal. However, in contrast to the Jacobi method, each component of the vector x on the right hand side of the transformed equation is updated immediately as each iteration progresses.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

The Gauss-Seidel Method This procedure is called sequential updating. Uij


(n+1)

1 2 (n) 2 (n+1) {hx Ui,j1 + hx Ui,j+1 2 2 2 hx + hy


2 2 2 2 +hy Ui1,j + hy Ui+1,j hx hy fij .} (n+1) (n)

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Successive Over Relaxation (SOR) Method It is possible to modify the Gauss-Seidel method by introducing an additional parameter that may accelerate the convergence of the iterations. The idea is to take a combination of the previous values of x and the current update form the Gauss-Seidel method. The parameter controls the proportion of the update that comes from the previous solution and the proportion that comes from the current calculation.

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Successive Over Relaxation (SOR) Method For 0 < < 1, the method is called successive under relaxation. For 1 < < 2 then the method is called successive over relaxation (SOR). For = 1 then SOR reduces to the Gauss-Seidel method. Uij
(n+1)

2 (n+1) 2 (n) {hx Ui,j1 + hx Ui,j+1 2 2 2 hx + hy


(n) (n)

2 2 2 2 +hy Ui1,j + hy Ui+1,j hx hy fij } + (1 )Uij .

(n+1)

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Elliptic Partial Differential Equations)

Iterative Methods

Examples

Example 1 We consider the following elliptic equation uxx + uyy = 0, (x, y ) (0, 1) (0, 1) with the boundary conditions u(x, 0) = 0, u(x, 1) = x, u(0, y ) = 0, u(1, y ) = y . The exact solution is given by u(x, y ) = xy .

Dr. P. Dhanumjaya

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE-Pilani, Goa Campus

An International Workshop on Advances in Computational Partial Differential Equations

Das könnte Ihnen auch gefallen