Author |
Message |
Noah Tracy (noahsark)
New member Username: noahsark
Post Number: 1 Registered: 11-2003
| Posted on Monday, November 10, 2003 - 03:15 pm: | |
Does anyone know how to get FlexPDE to solve backwards? I am working with voltage and Laplace's equation in a rectangle. I know the voltage profile I want to see on the right hand side and need to solve for the voltage profile on the left hand side that generated it. Thanks. |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 58 Registered: 06-2003
| Posted on Monday, November 10, 2003 - 07:37 pm: | |
The solution to such a problem is not unique - there may be many left-hand distributions that result in the same right-hand distribution. What other information about the system can you provide to help distinguish the true solution (derivatives, intermediate voltages, etc.)?
|
Noah Tracy (noahsark)
New member Username: noahsark
Post Number: 2 Registered: 11-2003
| Posted on Monday, November 10, 2003 - 08:03 pm: | |
I know that the voltage on the left hand side can't be greater than 5000V and the voltage at any height on the left side should be greater than the voltage at the same height on the right side. That's about it. I don't mind if the solution isn't unique. Does flexpde? I fed it a profile on the r.h.s, but it solved it like an initial value problem. I would really like to treat it as a final value problem, and do something like a shooting method. |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 59 Registered: 06-2003
| Posted on Tuesday, November 11, 2003 - 03:30 pm: | |
Even with a shooting method you need a value and a slope to start the integration. |