Test whether CATKE works with the baroclinic instability model#212
Test whether CATKE works with the baroclinic instability model#212
Conversation
|
|
|
@wsmoses I believe thre will be a raising error from the latest commit |
| include_halos = true | ||
| rtol = sqrt(eps(Float64)) | ||
| atol = 0 | ||
| atol = 1e-16 |
There was a problem hiding this comment.
The problem with absolute tolerance is that you need to know the scale of the numbers you're dealing with, and when the reference is exactly 0 there's no generally meaningful scale, any number can be both quite far or very close depending on how you look at it.
There was a problem hiding this comment.
ah yeah, was just seeing failures for really small numbers so wanted to get past that
There was a problem hiding this comment.
"Small" is relative to something, related to 0 any number can be anything 😁
There was a problem hiding this comment.
related to me, 1e-16 is huge 😢
There was a problem hiding this comment.
Is there a way not to have data with zeros? That'd make things easier
wsmoses
force-pushed
the
glw/catke-baro-instability
branch
from
f74970a to
da61a92
Compare
giordano
force-pushed
the
glw/catke-baro-instability
branch
from
da61a92 to
b75b0bd
Compare
3 participants
No description provided.