On Apr 7, 2014, at 8:53 AM, Joakim Odqvist <[email protected]> wrote:

> This is a real problem we have at the moment, and the "coeff" doesn't (as far 
> as we can see) update itself.
> No, you are right, the specific solver shouldn't have any bearing on this 
> problem I guess.

I apologize that it's taken me this long to get to this. When I run your 
script, I get:

diffusion coefficient....         PHI value at 4 = 0.596937242845
 
Solving for dt= 0.00673794699909  PHI value at 4 = 0.596937242845
Solved  for dt= 0.00673794699909  PHI value at 4 = 0.539379662528
dt:  0.00673794699909 0.00673794699909
 
Solving for dt= 0.00680566449223  PHI value at 4 = 0.539379662528
diffusion coefficient....         PHI value at 4 = 0.539379662528
Solved  for dt= 0.00680566449223  PHI value at 4 = 0.533992598283
dt:  0.00680566449223 0.0135436114913
 
Solving for dt= 0.0068740625575  PHI value at 4 = 0.533992598283
diffusion coefficient....         PHI value at 4 = 0.533992598283
Solved  for dt= 0.0068740625575  PHI value at 4 = 0.532785404054
dt:  0.0068740625575 0.0204176740488
 
Solving for dt= 0.00694314803475  PHI value at 4 = 0.532785404054
diffusion coefficient....         PHI value at 4 = 0.532785404054
Solved  for dt= 0.00694314803475  PHI value at 4 = 0.532306418791
dt:  0.00694314803475 0.0273608220836
 
Solving for dt= 0.00701292783259  PHI value at 4 = 0.532306418791
diffusion coefficient....         PHI value at 4 = 0.532306418791
Solved  for dt= 0.00701292783259  PHI value at 4 = 0.532150957698
dt:  0.00701292783259 0.0343737499161

so the diffusion coefficient is being recalculated at every time step, during 
the solve. 

I guess I don't know what you mean by "at the beginning of the time-step" and 
"at the end of the time-step". You call eq.solve() one time per loop and the 
diffusion coefficient is updated at the beginning of that solve. Since the 
diffusion coefficient is only used by eq.solve(), what purpose would be served 
by recalculating diffusion coefficient at other times? If it were used, it 
*would* be recalculated.
_______________________________________________
fipy mailing list
[email protected]
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Reply via email to