Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 531 Registered: 06-2003
| Posted on Thursday, February 02, 2006 - 06:00 pm: | |
Interchanging the boundary conditions does not result in the definition of a shrinking domain. 1) The parameter "lmove" is the end-point coordinate, and must have the sign of time reversed. 2) The initial values of velocity must be negated. 3) Value BC's on vx must be negated, but those on Xm must not. 4) in a shrinking domain, the boundaries encroach on the gaussian, so U=0 is no longer a valid BC on U. Value(U)=U0 works better. If you think about what the problem means, you can come up with a consistent conversion to a shrinking domain that works successfully. Note: U develops some saw-tooth structure at late time. This may be due to interpolation errors in the initial values, or it may be a result of incomplete treatment of Jacobian terms in velocity. I will investigate. |