[bug] Load of results from RVFWHMmodel is broken

[Kamuish]

Hi,

I am trying to load kima products using the following MWE

with chdir(selected_folder):
    with open(selected_folder / "kima_model_setup.txt") as to_read:
        for line in to_read:
            if "files" in line:
                input_files = line.split(":")[-1].split(",")[:-1]
                input_files = [i.strip() for i in input_files]
                break

    data = RVData(
        filenames=input_files,
        units="kms",
        skip=0,
        indicators=["fwhm", "s-fwhm", "cont", "s-cont", "bis", "s-bis"],
    )
    model = RVFWHMmodel(fix=False, npmax=2, data=data)

    res = kima.load_results(model, diagnostic=False)

It is failing with

Traceback (most recent call last):
  File "/home/amiguel/work/RV_analysis/EpsIndi/kima_runs/teste_joint.py", line 47, in <module>
    res = kima.load_results(model, diagnostic=False)
  File "/home/amiguel/work/RV_analysis/venv/lib64/python3.9/site-packages/kima/pykima/results.py", line 245, in load_results
    res = KimaResults(model_or_file, data,
  File "/home/amiguel/work/RV_analysis/venv/lib64/python3.9/site-packages/kima/pykima/results.py", line 602, in __init__
    self._read_GP()
  File "/home/amiguel/work/RV_analysis/venv/lib64/python3.9/site-packages/kima/pykima/results.py", line 850, in _read_GP
    self.kernel = GPmodel.KernelType(int(self.setup['kima']['kernel']))
ValueError: invalid literal for int() with base 10: 'standard'

[j-faria]

Can you check if this is still causing an error with the current version on main (after commit 0e8e5dd)?

[Kamuish]

The new version does not even run the RVFWHM model with the default configurations

 model = RVFWHMmodel(
fix=fix,
        npmax=NPmax,
        data=data,
    )
model.enforce_stability = True
model.trend = True
model.degree = 1
kima.run(
    model,
    steps=n_steps,
    num_threads=30,
    new_level_interval=4000,
)

This fails with

(venv) ➜  EpsIndi python kima_runs/apply_kima.py
# Seeding random number generators. First seed = 1736503611.
# Generating 30 particles from the prior...Traceback (most recent call last):
  File "/home/amiguel/RV_analysis/EpsIndi/kima_runs/apply_kima.py", line 98, in <module>
    run_kima(
    ~~~~~~~~^
        (results_path / "jointModel").as_posix(),
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    ...<6 lines>...
        run_joint_model=True,
        ^^^^^^^^^^^^^^^^^^^^^
    )
    ^
  File "/home/amiguel/RV_analysis/EpsIndi/kima_runs/apply_kima.py", line 84, in run_kima
    kima.run(
    ~~~~~~~~^
        model,
        ^^^^^^
    ...<2 lines>...
        new_level_interval=4000,
        ^^^^^^^^^^^^^^^^^^^^^^^^
    )
    ^
ValueError: LogUniform distribution must have lower < upper limits

I can't seem to find which prior is setting the LogUniform distribution with the incorrect parameters, as a breakpoint() before the kima.run shows all priors to be None

[j-faria]

ValueError: LogUniform distribution must have lower < upper limits

This is probably coming from the default prior for eta1_fwhm, which uses the span of the observed FWHM. Can you check if the FWHM is in the right units?