Conversation
|
I sorted out a bit of error and now the OMMBV vectors are even better. Uncertainty for multipole fields is at the same level as a pure dipole. |
…ield line tracing.
…o longer have any meaning at a pole.
|
I'm giving up on windows testing. OMMBC appears to compile everything just fine. However, pytest tests the pure python code without the compiled and installed Fortran. I also dropped MacOS testing way back since there is no installed Fortran compiler on the machines. Runs locally on Mac. |
|
|
||
| # Include extensions only when not on readthedocs.org | ||
| if os.environ.get('READTHEDOCS', None) == 'True': | ||
| extensions = [] |
There was a problem hiding this comment.
| extensions = [] | |
| from setuptools import setup | |
| extensions = [] |
There was a problem hiding this comment.
I've found the first setup needs to be at top level for setup.py develop to work.
| with open(os.path.join(here, version_filename)) as version_file: | ||
| version = version_file.read().strip() | ||
| import os | ||
| from setuptools import setup |
There was a problem hiding this comment.
| from setuptools import setup |
| else: | ||
| from numpy.distutils.core import setup, Extension | ||
| from numpy.distutils.core import Extension | ||
| from numpy.distutils.core import setup # noqa: F811 |
There was a problem hiding this comment.
| from numpy.distutils.core import setup # noqa: F811 | |
| from numpy.distutils.core import setup |
There was a problem hiding this comment.
flake8 or whatever complains about a redefinition of setup if the noqa isn't there. If I don't have the original setup import though then python setup.py develop doesn't work.
|
|
||
| - name: Run OMMBV setup.py | ||
| run: | | ||
| pip install -e ../OMMBV -t ../OMMBV |
There was a problem hiding this comment.
I have yet to find a windows compatible command that really installs the package. It goes through the motions, compiles the Fortran, etc, however, I can't get pytest to use whatever is installed.
There was a problem hiding this comment.
I think that it's skipping this part on windows. Github Actions will still try to load code locally if not installed.
Could probably set up os-dependent jobs here, one for unix, one for windows. But we'd need to test it on a windows machine locally, since I'm not sure if it's as easy as swapping directories. I'd bump this down the line since this is a big pull. Perhaps @JonathonMSmith can help?
There was a problem hiding this comment.
I agree that I need help from someone with a local windows machine. From the Actions tab windows does run the pip install stuff and it runs without error. I used to have python setup.py develop which I'll probably go back to.
| long_description = file: README.md, CHANGELOG.md | ||
| long_description_content_type = text/markdown | ||
| classifiers = | ||
| Development Status :: 5 - Production/Stable |
There was a problem hiding this comment.
I am prepping 1.0 here.
| Programming Language :: Python :: 3.7 | ||
| Programming Language :: Python :: 3.8 | ||
| Programming Language :: Python :: 3.9 | ||
| Operating System :: MacOS :: MacOS X | ||
| Operating System :: POSIX :: Linux |
There was a problem hiding this comment.
Update based on which tests are run
There was a problem hiding this comment.
I originally left Mac on there since I'm running local tests. I can update if the standard is to report what is publicly run.
Co-authored-by: Jeff Klenzing <jklenzing@gmail.com>
Co-authored-by: Jeff Klenzing <jklenzing@gmail.com>
…ex_location_info`
…into testing_sources
Description
Adds support for pure dipole and linear quadrupole sources to verify performance of OMMBV on multipole sources.
Addresses a variety of issues, including documented issues #50, #44.
Provides direct evidence that vectors are orthogonal for multipole fields.
Original Performance
There is an apparent reduction in performance for some linear quadrupoles in the equatorial plane. Current uncertainty on scaling values for drifts and electric fields is < 0.1%. In practical terms, given instrumentation errors are certainly higher, the observed uncertainty is fine for scientific applications. Still, looking to see if there is a way to improve this performance. Scaling uncertainty for dipole and linear quadrupole (along pole) is < .001%.
UPDATE
Performance improved and uncertainty with multipole fields is the same as dipole alone! Fixes are in here.
Also includes:
Type of change
Please delete options that are not relevant.
How Has This Been Tested?
Running unit tests on known source magnetic fields.
Test Configuration:
Checklist:
develop(notmaster) branchCHANGELOG.md, summarizing the changes