todo.org

feff85exafs to do list

The feff85exafs to do list

feffpath library [10/12]

1. [X] write header into f3ffNNNN.dat files
2. [X] deg –> degen to match larch’s _feffdat group
3. [X] use polarization correctly in feffpath library (fix test_materials.py around line 56)
4. [X] capture rat and ipot from STRUCT, accumulate ipot and rat wrappers as part of call to add_scatterer, also geom for python wrapper
5. [X] capture a list of Z numbers so ipots can be readily translated to atomic species
6. [X] capture header information in C struct, propagated into wrappers
7. [X] check for existance/readability of phase.pad at level of feffpath.c
8. [X] python wrapper / Larch plugin
9. [X] nose testing for feffpath generated f3ffNNNN.dat files
10. [X] SWIG-less perl library (uses Inline::C)
11. [ ] SWIG-less python library (Matt has started this)
12. [ ] _wrapper tests not currently working

libpotph library [4/6]

1. [X] get stand-along potph program to run all the way through
2. [X] test that stand-alone works all the way through
3. [X] turn it into a library with content of libpotph.json as it’s input
4. [X] write a C wrapper with the content of libpotph.json as its struct
5. [ ] write a stand-alone feff.inp –> libpotph.json converter
6. [ ] write a larch plugin which is a feff.inp –> struct wrapper converter, i.e. take a feff.inp as input and give a libpotph object as its output

libpotph issues and things to test

• does NOHOLE work?
• does SPIN work? is it needed for exafs?
• how much of AFOLP, FOLP, ION, RGRID, UNFREEZEF, INTERSTITIAL, JUMPRM do we want to support?
  • AFOLP: yes
  • FOLP: yes
  • ION: reluctantly, yes
  • RGRID: no idea…
  • UNFREEZEF: no idea…
  • INTERSTITIAL: no idea…
  • JUMPRM: probably not
• does the iafolp parameter need to be carried around? seems that it serves to indicate that folp was filled with the value given with AFOLP.
• what is the point of capturing polarization and ellipticity at this stage? is there any effect on the potantials? or can they be put off until libfeffpath?
• are we safe ignoring CFAVERAGE, MULTIPOLE, RPHASES
• the following seem outside the scope of feff85exafs and outside what the FP wanted to share: PMBSE, TDLDA, PLASMON
• the following are outside the scope of feff85exafs: XANES, FMS
• Bruce wants OVERLAP to be retired – the concept is better done in the sense of a Demeter::SSPath using libfeffpath
• might want flags for triggering the various output files controled by feff’s PRINT card…

big jobs [0/6]

1. [ ] build on Mac
2. [ ] make scatteringpath a drop in replacement for feffdat in larch
3. [ ] nosetests on result of scatteringpath with nnnn=False
4. [ ] fit with different feff versions – comparison framework (begun in Bruce’s privat SCFtests git repo)
5. [ ] capture information about muffin tin radii, norman radii, ionization, a lá Larch _feffdat’s potentials attribute

build on Windows [1/4]

1. [ ] f85ut.py needs to be installed into a place where larch will find it. (see larch> show _sys.path, but plugins rather than modules)
2. [X] need to replace ~bin/f85e~ with something that works on windows. made a runfeff method in f85ut instead
3. [ ] platforms/windows/prep.sh needs aliases for running nosetests and larch
4. [ ] perl wrapper cannot be linked successfully under Strawberry until a version of Strawberry using gcc 4.9 become available.

compilation issues

• src/PAR/parallel.f cannot be copmpiled – mpif.h is missing
• It seems as though the NOHOLE option won’t work. rdinp allows it, but src/XSPH/xsph.f wants, when nohole.eq.2, some data files (see lines 209 to 224) that are nowhere written in f85e
• In file src.EXCH/mpse.f, mpse.f did not compile with rank mismatches at lines 37 and 57. Variable edens is defined as rank 2 and called as rank 1. I “fixed” this by adding “,1” to the argument of edens. This does not, in fact, get exercised in feff85exafs – triggering this bit of code was disabled when the code base was delivered to us.
• src/EXCH/xcpot writes to numbered output, but in places that shouldn’t get exercised by feff85exafs.