Make extendedGaugeResident up to date with gaugePrecise.

[SaltyChiang]

Update extendedGaugeResident if gaugePrecise is modified. I add a function to do the work. After this change, we should expect that extendedGaugeResident is always the extended version of gaugePrecise.

Testings are required, hopefully CI/CD will do this for us.

[maddyscientist]

@weinbe2 can you test this with MILC when you get the chance please?

[maddyscientist]

cscs-ci run

[weinbe2]

Finally testing this, sorry for the hold-up @SaltyChiang . This technically needs develop merged in as well as a clang-format, plus some doxygen, but for simplicity I can also take care of that with one of my outstanding PR efforts...

[weinbe2]

There's a small correctness issue somewhere that I'm seeing running MILC RHMC, namely the fermion contribution to the action at the very start of the evolution disagrees relative to develop. The gauge action and the momentum action contributions are fine at the start, but then also deviate later in the evolution.

I'll have to dig further into this and try to understand what's going awry...

[SaltyChiang]

Thank you very much for the testing @weinbe2. I'll test the RHMC algorithm on my side.

[weinbe2]

Thank you very much for the testing @weinbe2. I'll test the RHMC algorithm on my side.

If you can take point on testing this with MILC RHM[C/D] for now, I would appreciate it @SaltyChiang , and if you remain stuck I'll loop back to it tomorrow (Thursday) my time, would that work?

[weinbe2]

@SaltyChiang fyi I'm getting back to testing your PR with MILC RHM[C/D] today

[weinbe2]

Okay it looks like everything is actually good @SaltyChiang . Can you re-merge in develop and confirm your PR is 100% good with your own (R)HMC tests? If so, I'll do a fresh CI kick-off and assuming no issues I can hit merge.

[SaltyChiang]

@weinbe2 I'm so sorry for taking so long to respond. I tested my HMC with HISQ yesterday but didn't see the difference on the delta H. I tried to configure the MILC code to check the issue but have not run it yet. You mean the issue disappears in your test with the current commit?

[weinbe2]

@SaltyChiang yes, sorry, everything is good with the current commit, there was an issue in the way I was verifying things. If you're also seeing agreement in your HMC, it sounds like we're good from a correctness standpoint. I just kicked off the CI and, assuming there are no issues there, I will merge this today.

[weinbe2]

@SaltyChiang you'll need to merge in the latest develop to grab the latest fixes to the CI tests