dolfin team mailing list archive

[Johan Hake <johan.hake@xxxxxxxxx>]

On Thursday April 29 2010 08:57:52 cutejeff wrote:
> Question #107819 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/107819
> 
>     Status: Answered => Open
> 
> cutejeff is still having a problem:
> Thank you for your help.
> 
> The PDE was derived using conservation of mass, but the results from
> Fenics showed the mass actually kept growing with time. Because my
> finite difference code doesn't grow, I consider my finite element code
> got some problems somewhere.
> 
> Interestingly, if I get rid of the (du/dx)^2 term in both Fenics and my
> finite difference code, the results match each other. Because I haven't
> seen any examples on manipulating the power of derivatives, I think it
> might be the cause. I don't think I can take integration by parts for
> v*(du/dx)^2*dx,  I don't know what to do with it, except keep it in this
> way.

Are you having (du/dx)^2 or d2u/dx2 as you state in your original question?

If it is the latter we do integrate that by part in several demos, Poisson for 
example. But you have u^3*d2u/dx2 which makes the integration by parts non 
trivial.

You need to get that correct in addition to Garth's suggestions, I think.

Johan