On Feb 21, 2022, at 7:48 PM, 'Guyer, Jonathan E. Dr. (Fed)' via fipy 
<[email protected]<mailto:[email protected]>> wrote:

On Feb 20, 2022, at 2:23 AM, Matan Mussel 
<[email protected]<mailto:[email protected]>> wrote:

I started examining this version of the code. I am still wondering about the 
equations since after ~500dt the circle takes an octagonal shape, and after 
~1100dt the solution diverges. I'll try to use parameters with a known 
solution. Since the solutions I know of were solved in cylindrical coordinates, 
I'll try first to convert the code to work with a cylindrical grid.

I haven’t observed either, but maybe I didn’t run long enough. The octagonal 
symmetry doesn’t surprise me; it’s almost certainly “feeling” the edges of the 
mesh.


OK, I ran longer and see what you mean. I still think the octagonal symmetry is 
arising because of the box size. As to the divergence, I’m only guessing, but 
there are three length scales in your equations:

- one associated with the terms 2nd-order in \phi. This is the \tanh solution 
that describes the diffuse interface. This length is of order \epsilon.
- one associated with the 4th-order term in \phi. I’m not positive, but I think 
this one may be order \epsilon, too.
- one associated with the term(s) in sigma and phi. It’s this one that I’m 
guessing is the issue. As the problem destabilizes, you can see undulations in 
\xi along the interface as sigma gets large.

My guess is that one or more of these length scales is under-resolved. The 
paper says that \sigma is small and essentially homogeneous, but they never 
quantify it.

-- 
To unsubscribe from this group, send email to [email protected]

View this message at https://list.nist.gov/fipy
--- 
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].

Reply via email to