Open navidcy opened 3 years ago
To use periodic BCs, don't you just need a full sinusoid in the domain, and then amend the BCs? At the moment, the advection is advection in the BC from the left hand edge, which is not ideal. Why do you say that BC's can't be periodic?
To use periodic BCs, don't you just need a full sinusoid in the domain, and then amend the BCs? At the moment, the advection is advection in the BC from the left hand edge, which is not ideal. Why do you say that BC's can't be periodic?
I didn’t say they can’t be periodic. I iust said that I don’t see how Druv is imposing periodic boundary conditions. But I simply may have missed it.
Oh, sorry, I thought you were saying it couldn't be done...
@navidcy presently in the notebook there are no periodic boundary conditions, it was what I planned on doing. @AndyHoggANU you're right, I would need a full sine wave to be able to apply periodic boundary conditions.
I don’t understand why a full sin wave is needed to apply periodic bcs. Periodic bcs can be applied regardless of the initial condition/actual solution. I don’t understand your remark @dhruvbhagtani2105 or I disagree with @AndyHoggANU if he made such statement and mislead you.
Hi Navid - yes, you're correct. We were talking in the context of a sine wave only. My statement was misleading, theoretically, we can have any function and impose periodic boundary conditions, as long as the function is periodic.
In the 1D Notebook:
From the chat in Slack I understood that @dhruvbhagtani2105 and @AndyHoggANU agreed using periodic boundary conditions for these 1D tests. However:
I can’t see how we impose periodic boundary conditions.
We should make sure you get the convergence your scheme implies. That is we should integrate up to a fixed time
t
and check that the difference from the numerical and analytical solution diminishes asdt^p
and asdx^q
when you varydt
anddx
respectively. (p
andq
being the order of your temporal and spatial scheme).cc @AndyHoggANU