🐛 Issues with formatting in XML output for occupations

[mbercx]

Very small values in the occupations output in the XML (possible also elsewhere) are not printed properly:

        <occupations size="11">
   1.000000000000000E+00   1.000000000000000E+00   1.000000000000000E+00
   1.000000000000000E+00   1.000000000000000E+00
   1.000000000000000E+00   9.513213811999113E-26   2.475414006339448-265
   0.000000000000000E+00   0.000000000000000E+00
   0.000000000000000E+00
        </occupations>

Note the missing E in 2.475414006339448-265. This seems to happen quite often, and breaks the parsers of aiida-quantumespresso since this produces a NoneType that subsequently fails to interoperate with e.g. floats:

  File "/Users/mbercx/project/sirius/git/aiida-quantumespresso/src/aiida_quantumespresso/parsers/pw.py", line 575, in build_output_bands
    occupations = 2. * numpy.array(parsed_bands['occupations'][0])
TypeError: unsupported operand type(s) for *: 'float' and 'NoneType'

[mbercx]

Just a quick note: I ran into this issue as well for the forces:

    <forces rank="2" dims="           3           6">
   3.220681500679849E-86   4.347898683069749-103  -3.696810758148386E-49
   0.000000000000000E+00   0.000000000000000E+00   0.000000000000000E+00
  -2.030916305444379E-63   5.726077947344848E-45  -3.696810758148386E-49
  -2.030916305444379E-63   5.726077947344848E-45  -3.696810758148386E-49
   2.527523082442126E-55  -7.126238607164159E-37  -2.320596130163241E-91
   2.468601466883413E-37   2.850495442865664E-37  -2.190254762280840E-74
    </forces>

[simonpintarelli]

Thanks @mbercx, I'll add a sensible threshold for force/stress too.

[mbercx]

Another one I'm afraid...

        <starting_magnetization>-7.481963004351236-179</starting_magnetization>
        <starting_magnetization>-1.436278500171930-177</starting_magnetization>

[mbercx]

It seems the atomic positions can also have this issue

      <atomic_positions>
        <atom name="Mn" index="1">
   9.492715137401231E-57   1.919671551571680E-36   9.854966977617403E+00
        </atom>
        <atom name="Mn" index="2">-7.906128800136544-150   1.919174004571837E-36   9.691860452671211E-77</atom>
        <atom name="Mo" index="3">4.746357568700615E-57   3.067994052813471E+00   4.927483488808702E+00</atom>
        <atom name="Mo" index="4">2.656960788448982E+00   1.533997026497647E+00   1.478245046642611E+01</atom>
        <atom name="N" index="5">1.659450767641205E-56   3.067994052813471E+00   1.722777127447125E+01</atom>
        <atom name="N" index="6">1.188363773579164E-56   3.067994052813471E+00   1.233712965838096E+01</atom>
        <atom name="N" index="7">2.656960788448982E+00   1.533997026497647E+00   7.372804296853846E+00</atom>
        <atom name="N" index="8">2.656960788448982E+00   1.533997026497647E+00   2.482162680763558E+00</atom>
      </atomic_positions>

See on LUMI

/scratch/project_465000416/mabercx/manual/sirius-tests/xml-positions

[simonpintarelli]

@mbercx , thanks for the input. I'll use it to test and let you know when it's done:
electronic-structure/SIRIUS#943

I think the problem was just the occupation numbers and then it can propagate via force > atomic positions.