search

astro-group-bristol / Gradus.jl

Extensible spacetime agnostic general relativistic ray-tracing (GRRT).
https://astro-group-bristol.github.io/Gradus.jl/dev/
GNU General Public License v3.0
18 stars 2 forks source link

Interpolation changed in DataInterpolations #179

Closed phajy closed 11 months ago

phajy commented 11 months ago

The default value of extrapolate in DataInterpolations has been changed to false. See https://github.com/SciML/DataInterpolations.jl/pull/194.

This can, in some cases, break Gradus.

Adding ;extrapolate=true) to all of the DataInterpolations.LinearInterpolation calls in the following lines seems to fix the problem.

https://github.com/astro-group-bristol/Gradus.jl/blob/e9e863213a3bd6e67d87178141742d52acc27f98/src/orbits/orbit-solving.jl#L115-L121

I would guess that extrapolations near the edge of the domain are expected. Not sure if this needs to be looked at a bit more?

phajy commented 11 months ago

Fixed in https://github.com/astro-group-bristol/Gradus.jl/pull/180

fjebaker commented 11 months ago

Thanks for finding the source of this issue. I can see from the CI test suite failing that there are likely a few other places where there was extrapolation happening. I need to step through them and decide whether extrapolating those calculations is actually what we want to do. I think for the orbit solving that should be fine 👍