search

hpc4cmb / toast

Time Ordered Astrophysics Scalable Tools
Other
44 stars 39 forks source link

Unit test failing in toast3: `Pointing weights do not agree` #744

Open ickc opened 8 months ago

ickc commented 8 months ago

Since around 5 days ago, I'm seeing Pointing weights do not agree in the toast3 branch. (which should probably narrow down the changes to these 3 commits: https://github.com/hpc4cmb/toast/compare/54132fb4ddd1e040909453ab88d4fb86b93fd96e...9ecc50ef0461a7332039b102246d541e0bf88885).

One example is in https://github.com/ickc/so-software-environment/actions/runs/8248173751/job/22557911328#step:7:1815, which basically is complaining

Pointing weights do not agree: [ 1 0 -1] != [ 1 0 1]...

Any ideas why? Thanks.

tskisner commented 8 months ago

Yep, I'm actively working on that since yesterday. I have been double-checking the HWP stokes weights (which were verified on real data) against what I get from the math document which describes our pointing model. Should have a fix today.