Error loading .wfx file created by Orca

[isolated-matrix]

Hi there,

I am trying to get the data from a calculation conducted in Orca into .fchk format so it can be read by another program. First I tried to do it by converting Orca's .gbw file into .molden input through Orca's orca_2mkl utility, but, after converting to .fchk the other program would not read the .fchk file (apparently something is missing from it).

Considering the molden input has far less information in it than I thought (according to the table in your documentation), I decided to try doing the conversion with a .wfx file instead, again converting the .gbw file through Orca's orca_2aim utility. However, for some reason, when I try to run the command iodata-convert apost-test-calc.wfx apost-test-calc.fchk, I get an error message I don't fully understand:

Traceback (most recent call last):
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/api.py", line 183, in load_one
    return IOData(**format_module.load_one(lit, **kwargs))
                    ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/formats/wfx.py", line 232, in load_one
    data = load_data_wfx(lit)
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/formats/wfx.py", line 126, in load_data_wfx
    data = parse_wfx(lit, required_tags)
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/formats/wfx.py", line 213, in parse_wfx
    data[section_start].append(line)
    ~~~~^^^^^^^^^^^^^^^
KeyError: None

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/bin/iodata-convert", line 8, in <module>
    sys.exit(main())
             ~~~~^^
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/__main__.py", line 146, in main
    convert(args.input, args.output, args.many, args.infmt, args.outfmt, args.allow_changes)
    ~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/__main__.py", line 137, in convert
    dump_one(load_one(infn, fmt=infmt), outfn, allow_changes=allow_changes, fmt=outfmt)
             ~~~~~~~~^^^^^^^^^^^^^^^^^
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/api.py", line 148, in inner
    result = func(*args, **kwargs)
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/api.py", line 189, in load_one
    raise LoadError("Uncaught exception while loading file.", lit) from exc
iodata.utils.LoadError: Uncaught exception while loading file. (apost-test-calc.wfx:1)

What exactly does this error message mean? Is there anything I can/need to edit in the .wfx file to make it work?

In case it's relevant, I'm running this on my MacBook Pro M1 Max using Python 3.13.5. I've also attached the Orca output, .gbw and generated .wfx files below.

Also, slightly related, IOData doesn't seem to recognise the .json_qcschema file type, as it gives the error:

Traceback (most recent call last):
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/bin/iodata-convert", line 8, in <module>
    sys.exit(main())
             ~~~~^^
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/__main__.py", line 146, in main
    convert(args.input, args.output, args.many, args.infmt, args.outfmt, args.allow_changes)
    ~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/__main__.py", line 137, in convert
    dump_one(load_one(infn, fmt=infmt), outfn, allow_changes=allow_changes, fmt=outfmt)
             ~~~~~~~~^^^^^^^^^^^^^^^^^
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/api.py", line 148, in inner
    result = func(*args, **kwargs)
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/api.py", line 180, in load_one
    format_module = _select_format_module(filename, "load_one", fmt)
  File "/Users/xxxxxxxxxxxxxx/Library/Python/3.13/lib/python/site-packages/iodata/api.py", line 86, in _select_format_module
    raise FileFormatError(f"Cannot find file format with feature {attrname}", filename)
iodata.utils.FileFormatError: Cannot find file format with feature load_one (apost-test-calc.json_qcschema)

Let me know if there's anything else you need!

apost-test-calc.gbw.txt
apost-test-calc.out.txt
apost-test-calc.wfx.txt

[PaulWAyers]

I think we more-or-less gave up on the qcschema after we realized that there were multiple versions being maintained at MolSSI without harmonization. The actual repository seems basically defunct/inactive. Given that there isn't a community standard seemingly, it was just hopeless. We believe it is a good idea; we tried; we can't do it singlehandedly.

For the other things, my best guess is that we do not support the pseudopotential you are using. We certainly have not done systematic testing with pseudopotentials. I'd suggest seeing if the workflows you are interested in work without the pseudopotential (try a molecule without any heavy elements and see if the workflow works). If that is the issue, then raise a specific issue related to the lack of support of pseudopotentials. Note that some file formats (.wfx, I suspect) do not support pseudopotentials at all.

There is some support for pseudopotentials within the QC-Devs ecosystem, but I'm unaware of work using pseudopotentials outside the framework of the density partitioning methods (and plane-wave codes).

[isolated-matrix]

Thanks for getting back to me so soon!

Shame to hear about qcschema. Apparently Orca is pursuing the JSON format exclusively for their output compatibility now, so maybe they'll be able to breathe some life back into it?

I did what you suggested and ran an equivalent calculation without a pseudopotential: no luck, unfortunately. Also, just FYI, apparently .wfx files do have support for ECPs (if you check the bottom of this page you'll find an example).

It also occurred to me that the reason might have been because orca used the RI approximation and, therefore, an additional basis set. I turned this approximation off, but this still did not help.

[PaulWAyers]

1. If Orca decides on a standard .json and it becomes adopted by other packages, we would support it. If not, if you would like to implement it we would be happy to review pull requests and advise.
2. The EDF at the end of the *.wfx are not usually output with the ECPs, I think; these are essentially "core densities" which something that we support in certain workflows. Your *.wfx file does not include them, so it is certainly not a valid *.wfx file for a ECP calculation. That's an issue you can raise with the Orca developers.
3. The *.molden file might have been a better choice, but it may also not include support for things like RI and ECPs. I suspect it does not.
4. If you share your *.wfx without ECPs or RI and your error messages, we may be able to diagnose the problem, provided it is a bug on our end and not a bug in the Orca *.wfx generator.

Probably, you will need to manually parse the Orca output to make a wrapper of the other program you wish to use. We are happy for you to contribute that parser to iodata. You can find detailed information on how to do that here. I'd probably recommend going through the Orca *.json if that is available and supports all the features/data you need.

[isolated-matrix]

1. Fair enough, that makes sense.
2. Thank you for letting me know, I will pass that onto them.
3. Yes, the molden file successfully converts, but still leads to an error with the other program.
4. Ah, sorry, I completely forgot to attach those files! They should be attached to this message. Let me know if you see anything strange

apost-test-calc2.out.txt
apost-test-calc2.wfx.txt
apost-test-calc2.gbw.txt

[PaulWAyers]

If the molden file converts but the other program doesn't read it, then the key thing for you to figure out is whether it is the other problem not supporting things or us. That's something we can deal with, but interface issues require very detailed descriptions of the files involved, workflows used, and errors obtained.

Ditto for the *.wfx file. We would need to know the error messages you are receiving.

[isolated-matrix]

I have submitted an issue report on the Github of the other program, and they are going to look into what the problem with the *.molden file might be (mgimferrer/APOST3D#11).

That being said, the problem with the *.wfx file is that IOData comes up with an error when trying to convert it to an *.fchk file—I don't get to the stage where I can test it on the other program. That's what I was hoping to get help with.

[PaulWAyers]

What is the error message?

You might tell the other program that *.molden files are very tricky, and there are several different conventions for basis sets. We put a lot of effort in iodata trying to support different options, but one has to be very careful.

[isolated-matrix]

It is the error that I included in my first post here.

That's annoying. That being said, the other program only accepts *.fchk files, and I am using iodata to generate those. I am generating the *.molden and *.wfx files in Orca.

[PaulWAyers]

I'm not sure exactly what is wrong, but there are definite problems with the Orca *.wfx file.

• </Primitive Exponents> should appear on a line by itself.

as per the official specification:
"The opening and closing tags of a data section MUST appear alone on separate lines before and after the corresponding data."'

There seem to be bigger problems that this, however.

[PaulWAyers]

It seems to parse if you:

1. Put <\Primitive Exponents> on a line by itself.
2. Remove all spurious carriage returns in the *.wfx file.

@tovrstra I think we could support files that have the extra carriage returns" and perhaps also make it more robust with respect to spurious spacing/capitalization within the tags. Alas, I don't think we can count on everyone who writes a *.wfx to be careful.

What do you think @tovrstra ?

P.S. @isolated-matrix I did some other revisions as I was trying to make this work, so there may be other issues. (E.g., I removed the intro content in the Orca file, removed leading spaces before GTO, etc.) You may need to play around a bit to get it working.

[isolated-matrix]

Thank you for the advice, that seemed to do the trick! It successfully converted to an .fchk file when I got rid of all those extra lines and moved the <\Primitive Exponents> to a new line.

Still isn't being read by the other program, but that's probably not your fault. I'm happy to close this issue now, if you like?

[PaulWAyers]

I'll close the issue. @tovrstra , I'm still curious about your thoughts as to whether we should make iodata more robust to malformed *.wfx files.