[ff3d-users] Improving precision for boundary conditions.

[Juzar Thingna]

Hello Everyone,
I'm trying to solve the modified Helmholtz equation. I need my boundary
conditions to be accurate to atleast 6 decimal places. In the verbose
printed by ff3d at the time of running I notice that ff3d takes the
precision of the boundary conditions as only 3 decimal places. In the input
file I have provided this value accurate till 6 decimal places. Is there
some way I can improve on the significant digits ff3d takes as input?

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
=========================
== Text mode execution ==
=========================
FreeFEM3D computing thread: started
Parsing data
Parsing data: done
Building Scene ... done
Parsing POVRay file: "domain.pov"
- the scene is composed of 2 object(s)
Parsing POVRay file: done
Building Mesh ... done
Building Domain ...
Domain is: inside((0,0,0))
Reference translation (POV-Ray -> FEM)
(0, 0, 0) -> 1
Done
Done
Computing degrees of freedom positions...
- number of degrees of freedom positions: 1250000
Computing degrees of freedom positions: done
Solving Problem:

Unknowns: [u:undefined]
Problem:
pde(u):
div(L*grad(u)) - 1*u = 0
dnu(u) = 0.016 on 3
u = 0 on 2
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Regards,

Juzar Thingna
Department of Physics,
Center for Computation Science and Engineering,
National University of Singapore.

[Stephane Del Pino]

Dear Juzar.

As far as I know there is no special precision limitation concerning "double"
variables in the data file. It might just be an output error. Could you please
send your data file so that I can investigate?

By the way, be careful with the 'dnu' keyword, it can be tricky to use (check
remark 14 in the user documentation).

Regards,
Stéphane.

PS : sorry for this late answer but I am on holidays.