Chen:
I'm not sure anyone still active in the project knows the program well, so I asked the author of the program, Abner Salgado, about the issue and this was his answer:
> What happens is that the pressure update equation, written right above
> the implementation of
>
> update_pressure
>
> is understood in "weak" form. We multiply by a test function in the
> pressure space and the divergence term is integrated by parts (hence the +).
>
> This is so because div u is not in the pressure space and so we cannot
> simply add the corresponding coefficient vectors.
>
> It involves solving a mass matrix problem, as it can be seen in the
> implementation.
>
> Hope this clarifies things.
I hope this helps!
Best
Wolfgang