Revert "Add calibration uncertainty model"

[ahnitz]

Reverts #54

Thanks for the work here @ColmTalbot. One minor niggle is that the recalibration function should live in the pycbc.strain package. There is a similarly implemented physical model there already.

https://github.com/gwastro/pycbc/blob/master/pycbc/strain/recalibrate.py

For now we should revert, split up, and then reapply.

[coveralls]

Pull Request Test Coverage Report for Build 228

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.9%) to 40.648%

---

Totals
Change from base Build 218:	-0.9%
Covered Lines:	1067
Relevant Lines:	2625

---

💛 - Coveralls

[ahnitz]

@ColmTalbot To expand a bit. PyCBC already has a package designed to condition and handle strain. Recalibration is conceptually squarely in that wheelhouse and in fact there is already an implementation of a physical model there which this PR imitates the interface of. The fact that we do want to do studies (and have done so in the past) for calibration on waveforms and searches indicates that this also should be in the core package, not the PE-only one. Even the ability to do a simply jupyter notebook with recalibrated data and / or waveform indicates this should be accessible in the core package.

I think the right path forward here is to revert this commit, and then add it as another option to the strain function which selects the calibration model. At the moment, this patch breaks existing functionality to call the physical calibration model, but it should be possible to do both. A good option would be to add this function to pycbc/strain/recalibrate alongside the physical model.

[cmbiwer]

I agree with Alex. The calibration is used much more widely than just parameter estimation. I think it should be access through pycbc.strain as well, and support the reversion of this commit.

But this is nice work and look forward to it being available!

[cdcapano]

I talked to @ColmTalbot and @cplb, and they're ok moving the recalibration module to the strain module in pycbc. However, rather than reverting, I'd prefer that that be done in the following order: a commit to pycbc/strain/recalibration is made adding the spline calibration to that; then a commit is made to only remove the redundant calibration code in gwin. I don't want to revert this because there are other useful things in this commit. I'll keep this PR open in the mean time with a "WIP" tag until that is completed.

[ahnitz]

I think it would actually be much cleaner to revert and then refactor the original PR. Especially, since the vast majority of changes you would now add will just be a manual revert. Won't it also be easier to see what will be changing from the original working state? I'd suggest the following, Revert Add PR to add relevant code to strain package Add PR to add changes to gwin docs/config/etc. That will be a much clearer history, no? However, if you have a strong preference, we should go your route as you are there in person to coordinate.

…

On Fri, Jul 20, 2018, 16:40 Collin Capano ***@***.***> wrote: I talked to @ColmTalbot <https://github.com/ColmTalbot> and @cplb <https://github.com/cplb>, and they're ok moving the recalibration module to the strain module in pycbc. However, rather than reverting, I'd prefer that that be done in the following order: a commit to pycbc/strain/recalibration is made adding the spline calibration to that; then a commit is made to only remove the redundant calibration code in gwin. I don't want to revert this because there are other useful things in this commit. I'll keep this PR open in the mean time with a "WIP" tag until that is completed. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#63 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACGrRoXWretDs-HW6oc8kejwxa8z1Ogdks5uIkBQgaJpZM4VYNQ2> .

[cdcapano]

No, I don't like reverts, and it'll make it harder to make sure the docs, example config file, etc. are put back.

[ahnitz]

OK, no problem. Please close this then when that's ready.

…

On Fri, Jul 20, 2018 at 5:05 PM, Collin Capano ***@***.***> wrote: No, I don't like reverts, and it'll make it harder to make sure the docs, example config file, etc. are put back. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#63 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACGrRrVRk4gBNEjF7DVkU9Dkh3XVRvyzks5uIkYvgaJpZM4VYNQ2> .

-- Dr. Alexander Nitz Max Planck Institute for Gravitational Physics (Albert Einstein Institute) Callinstrasse 38 D-30167 Hannover, Germany Tel: +49 511 762-17097