programming:continue_solution

*Matthew Salewski, 2009-09-21 18:10 *

~~DISCUSSION~~ Hello,

I would like to know how to continue a solution at a given Re to another Re.

Currently i am running

./continuesoln -eqb -contRe -R 450 -o LBNext -log LBLog LB.ff

I've got 22 continued solutions, but i'm not quite sure if these are what i was expecting.

I assumed that the -R option was the target Re to get to(?) thinking that the program would `know' the Re of the original solution; is this correct? Later, i tried 400, and it crashed (from the log):

124994 equations 124994 unknowns Initial integration time T == 15 Initial symmetry sigma == 1 1 1 1 0 0 Search for solutions of sigma f^T(u) - u == 0 with sigma == 1 1 1 1 0 0 Reynolds == 0 Computing G(x) = 1/T (sigma f(u,T) - u) ....10..L2Norm(f^t(u)) == nan error in f: f^t(u) is not finite. exiting.

What does this mean?

And when does it stop? The 22 continued solutions go from 220 to 585. If i'm feeding Re as a parameter that is different from the Re of my initial solution, how small/big should it be?

Also, does this require there to be more than one initial Re or fields at different Re as described for continuefields?

Finally, there was some mention about using changegrid and findorbit to compute new solutions. What is the procedure for this?

Thanks.

programming/continue_solution.txt · Last modified: 2010/02/02 07:55 (external edit)