Author |
Message |
Takeyuki Ojima (tac)
Member Username: tac
Post Number: 16 Registered: 07-2005
| Posted on Monday, February 25, 2008 - 09:07 pm: | |
Execution of my attached program was halted with "Bad Basis Selector!" To avoid second-order derivative of time, two functions of w and gx are used in a moving-mesh application on an electromagnetic problem. Please give me advice. |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 1067 Registered: 06-2003
| Posted on Tuesday, February 26, 2008 - 12:25 am: | |
FlexPDE assumes that all equations are presented in the Eulerian (Lab) frame. If the mesh is moving, it symbolically constructs correction terms for the node motion, and adds these terms to the equations as stated. Your ax and V equations both involved crossed time-space derivatives. The FlexPDE symbolic analyzer is not smart enough to construct moving-mesh corrections for crossed time-space derivatives. You will have to see if your system can be reformulated in terms of some other primary variables that do not require these cross derivatives. There is a provision for entering Lagrangian equations (ie, equations in the moving frame), for which no corrections are computed. Declare your Equations section as LAGRANGIAN EQUATIONS ... (I am not aware of any testing on this option. It may not work.)
|
Takeyuki Ojima (tac)
Member Username: tac
Post Number: 17 Registered: 07-2005
| Posted on Tuesday, February 26, 2008 - 06:21 pm: | |
"Lagrangian Equations" seems to work for disappearing the message but another kind of error took place like "0=0"-error, which is due to my problem. Thanks. |