Add a MA demo solving Helmholtz PDE #53
Conversation
stephankramer
force-pushed
the
monge-ampere-helmholz-demo
branch
from
8c9ca1b to
f746f44
Compare
There was a problem hiding this comment.
Thanks very much for this @stephankramer!
A few fairly minor comments. Also, please could you put a link to this demo at the end of the previous one, with a short sentence? So that it's easy to read through the demos on the webpage.
demos/monge_ampere_helmholtz.py
Outdated
| # e.g. for testing purposes. | ||
| # | ||
| # Based on the code in the Firedrake demo, we first solve the PDE on a uniform mesh. | ||
| # Because our chosen solution does not satisfy homogenoues Neumann boundary conditions, |
| mover = MongeAmpereMover(mesh, monitor, method="quasi_newton") | ||
| mover.move() |
There was a problem hiding this comment.
What do you think about including what's printed to screen in a separate code block? i.e. showing the convergence of the different metrics. We do this in Thetis demos.
There was a problem hiding this comment.
good idea. Out of interest why do we refer to the standard deviation of cell areas as "equidistribution" ?
There was a problem hiding this comment.
Hm good point, not sure how that ended up in there. Standard deviation would be a better term. We can either fix it here or open an issue for it, whichever you prefer.
There was a problem hiding this comment.
Any objection to UTF -8 in terminal output? "Standard Deviation" is a bit long, and not entirely accurate - all other terminology (https://en.wikipedia.org/wiki/Coefficient_of_variation) is a bit obscure so I thought to clarify with σ/μ
There was a problem hiding this comment.
Fine by me. Nice! Even better.
|
I guess for the tests to pass we need to use the same Docker env as for Animate/Goalie. All the more reason to get mesh-adaptation/animate#47 working! |
Would you mind switching the Docker image used by Movement as part of this PR? If you switch it to the same one as we use for Animate and Goalie then the tests should pass. |
So that tests/demos can depend on animate.
equidistribution -> (Coefficient of) Variation
As previously discussed now with Hessian-based monitor