Finite difference discretization for systems of higher ODEs

Click For Summary

Discussion Overview

The discussion focuses on the use of finite difference methods to discretize systems of higher-order ordinary differential equations (ODEs), specifically addressing a fourth-order ODE and its representation as a system of first-order ODEs for numerical integration.

Discussion Character

  • Technical explanation
  • Mathematical reasoning

Main Points Raised

  • One participant asks how to apply finite difference methods to discretize a system of fourth-order differential equations, providing an example of a single fourth-order ODE.
  • Another participant points out that the initial example provided is not a system of fourth-order ODEs but rather a single fourth-order ODE.
  • A later reply clarifies that a better representation of the system consists of three fourth-order equations with constant coefficients, suggesting that converting these to a set of first-order ODEs is a common numerical approach.
  • The same participant outlines a method to define the variables and their derivatives, leading to a system of 12 coupled first-order linear ODEs that can be solved using numerical methods such as Runge-Kutta or Euler methods.

Areas of Agreement / Disagreement

Participants do not reach a consensus on the initial representation of the equations, with some asserting it is a single ODE while others provide a more complex system. The discussion remains unresolved regarding the best approach to discretization.

Contextual Notes

The discussion involves assumptions about the representation of the equations and the methods for numerical integration, which may depend on the specific context and definitions used.

Ojo Saheed
Messages
5
Reaction score
0
How can I use finite difference to discretize a system of fourth order differential equations?
for example:
y(4)+5y(3)-2y''+3y'-y=0
 
Physics news on Phys.org
This doesn't look like a system of 4th order ODEs. It looks like a single 4th order ODE.

Chet
 
Chestermiller said:
This doesn't look like a system of 4th order ODEs. It looks like a single 4th order ODE.

Chet
Thanks for the reply Chet. Actually, I had thought that given an example of one the system of equations might suffice for a demonstration. In any case, a better representation of the system goes thus:
y1(4)=a1y1''+a2y2''+a3y3''+a4y1+a5y2+a6y3+a7
y2(4)=b1y1''+b2y2''+b3y3''+b4y1+b5y2+b6y3+b7
y3(4)=c1y1''+c2y2''+c3y3''+c4y1+c5y2+c6y3+c7
where a, b and c terms are constant coefficients
 
Ojo Saheed said:
Thanks for the reply Chet. Actually, I had thought that given an example of one the system of equations might suffice for a demonstration. In any case, a better representation of the system goes thus:
y1(4)=a1y1''+a2y2''+a3y3''+a4y1+a5y2+a6y3+a7
y2(4)=b1y1''+b2y2''+b3y3''+b4y1+b5y2+b6y3+b7
y3(4)=c1y1''+c2y2''+c3y3''+c4y1+c5y2+c6y3+c7
where a, b and c terms are constant coefficients
If you're trying to solve this numerically, one way it is often done is to convert to a set of first order ODEs. In this particular case, there would be 12 first order ODEs to integrate.

Define,
zj=yj (j = 1,3)
zj+3=y'j (j = 1,3)
zj+6=y''j (j=1,3)
zj+9=y'''j (j=1,3)
Then
z1' = z4
z4' = z7
z7' = z10
z10' = a1z7+a2z8+a3z9+a4z1+a5z2+a6z3+a7
etc.

You end up with 12 coupled first order linear ODEs explicit in the derivatives. You can use an automatic integrator to solve them or use your own coding of Runge Kutta or forward Euler, or backward Euler, or whatever.

Chet
 

Similar threads

  • · Replies 7 ·
Replies
7
Views
3K
  • · Replies 3 ·
Replies
3
Views
3K
Replies
1
Views
2K
  • · Replies 6 ·
Replies
6
Views
3K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 12 ·
Replies
12
Views
7K
  • · Replies 1 ·
Replies
1
Views
2K
Replies
8
Views
6K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 1 ·
Replies
1
Views
684