phockett / pemtk Goto Github PK

Check line

Should also change csphase to an option.

Currently some issues with classifiers (tested on v1.2.3) which don't drop unclassified data in this case. Working OK on v1.0.1.

To do:

• Update testing envs.
• Test and debug - likely change in merge or pivot_table routines in _setWide or classifyFits? May be able to fix with dropna() prior to reformatting.

Now mostly in place, using libmsym base routines and trimmings.

• Main docs: working with symmetry .
• Commits 18th - 20th April 2023 (cfd7b40 and few preceding) for main changes.

Todo:

• Mapping to ePS term symbols, see previous work in ePSman for this.
• Verification of m and mu mappings, inc. degeneracies and phases. May still have issues here, should check phase terms and also dipole mapping (

  PEMtk/pemtk/sym/_dipoleTerms.py

  Line 72 in d498762

  def dipoleTermsSymHarm(self, reCartMap = {-1:'y',0:'z',1:'x'}, dimMap = {'C':'Cont', 'mu':'it'}):

  ) is complete. (See also #8.)
• Application of ang mom coupling limits, could apply these to m <> mu mappings.

Things to address:

• Consolidate/split methods more coherently over files/subgroups.
• Finish and implement class options storage and reuse (for plotters/calcs).
• Simplify defaults and calls, maybe with functools? https://docs.python.org/3/library/functools.html#functools.partialmethod

New bug, trying to run a fit currently results in "RecursionError: maximum recursion depth exceeded while calling a Python object".

Upon inspection, this was due to the use of .drop(inplace=True) in symCheck(), which was not working and resulting in all fitting params being reset, with vary=False (reference case should be as shown at https://pemtk.readthedocs.io/en/latest/fitting/PEMtk_fitting_demo_multi-fit_tests_130621-para_010922.html#Setup)

This is now changed to .pop() in 5df6a0e, which seems to work as expected. This may be Pandas and/or Python version specific - tested OK with PD 1.4.1, Python 3.9.10.

TODO:

• test other versions.
• check Docker builds, why is PD 1.4.1 installed, possibly Xarray dependency? May need to added explicit conda pd install line to pull newer versions.

Mainly working but:

• Possible m=+/-1 ordering switch on some tabulations? (Arose during N2 results testing.) May be due to relabelling routines assuming ordering? TBC.
• Var names for paramPlot, need setting otherwise get linked 'value' plot y-axes in layouts (although setting .opts(shared_axes=False) will fix at plot).
• fit._aggUtil debug. OK, see cf0ac4b

Currently doing this by converting to magnitude-phase form, but lmfit can also support complex data: https://lmfit.github.io/lmfit-py/faq.html#how-can-i-fit-complex-data

Might be worth investigating.

Updated with arg parser and better module and data options in 34b4b7e

Todo:

• add automatic data download option
• tidy up!

Quick-fix is Pickle, but quite brittle to package version changes (e.g. recently broke after updating XR & Pandas versions).

Otherwise need to wrap something per-dataset/dict/dataframe/dataarray...? This is probably OK for aggregate cases, but a bit unwieldy for batch fit cases.

Basics at https://docs.xarray.dev/en/stable/user-guide/io.html

For more on this (for IO in ePSproc) see phockett/ePSproc#8