Re: Speeding up an RD system

Hi Erik, I did a few more efficient improvements to FiPy on a branch. I have also attached and updated script. The new script (500x500) and gives the following timings with OLD being old script + trunk and NEW being new script + branches/e

Re: variable convection term

On Tue, Oct 12, 2010 at 1:27 PM, BIN ZHANG wrote: > > Dear Daniel: > > Thanks a lot for your reply. Greatly appreciated. > > The trick you used to transform the equation is quite clever ;-). I guess > now my question is how to represent $\partial_i V$ in the convection term. Try, >>> Convec

Re: variable convection term

Dear Daniel: Thanks a lot for your reply. Greatly appreciated. The trick you used to transform the equation is quite clever ;-). I guess now my question is how to represent $\partial_i V$ in the convection term. Though you confirmed to me that it's possible to have a convection term wit

Re: variable convection term

Hi Bin, Sorry for the delayed reply. I think the best way to deal with the equation you proposed is to convert to the following form: \begin{equation} 0 = a q \partial_i^2 V - a \partial \left(q \partial_i V \right) + b \partial_i^2 q \end{equation} That way, you'll have an implicit sourc