Pineappl FkTable reader fails for new Jet grids

[comane]

When trying to load a theory 600 pineappl grid such as ATLAS_1JET_8TEV_R06_PTY_BIN1.pineappl.lz4 using:

pineappl.fk_table.FkTable("ATLAS_1JET_8TEV_R06_PTY_BIN1.pineappl.lz4")

I get the following error

---------------------------------------------------------------------------
PanicException                            Traceback (most recent call last)
File /var/folders/h8/rpdyzfjj6bv49nxrm1s22g1c0000gn/T/ipykernel_48719/2477570161.py:1
----> 1 FkTable.read("/Users/markcostantini/local/nnpdf_data/theory/theory_600/fastkernel/ATLAS_1JET_8TEV_R06_PTY_BIN1.pineappl.lz4")

File ~/miniconda3/envs/nnpdf-dev/lib/python3.9/site-packages/pineappl/fk_table.py:35, in FkTable.read(cls, path)
     18 @classmethod
     19 def read(cls, path):
     20     """
     21     Load an existing grid from file.
     22 
   (...)
     33             grid object
     34     """
---> 35     return cls(PyFkTable.read(path))

PanicException: called `Result::unwrap()` on an `Err` value: NonUniqueGrids

Maybe @scarlehoff and @t7phy can help with this?

[scarlehoff]

Create a PR against #1813 so that I can have a look because there are many possible reasons for this error:

1. Is the metadata.yml on the new dataset correct?
2. Is the theory part of the metadata.yml of the new dataset correct?
3. Does the fktable exist at all?
4. Is your version of theory 600 up to date?
5. Is this an error of current theory 600?

[comane]

I am not sure that the metadata.yaml matters at all for this issue. I am trying to load directly the pineappl file contained In the most up to date version of theory 600

from pineappl.fk_table import FkTable
FkTable.read("/path_to_theory_600/fastkernel/ATLAS_1JET_8TEV_R06_PTY_BIN1.pineappl.lz4")

[scarlehoff]

Then it is a problem with the FkTable.
But if you are using old datasets I would not recommend the usage of theory 600, I think @t7phy only added there the new datasets.

[t7phy]

@scarlehoff No I actually added the fktables for the old jets too because these are the new grids (at NNLO)

[scarlehoff]

And can you read those locally using the code @comane put in his comment?

[t7phy]

I haven't tried with validphys. I know they were with fine pineapl cli though. So they should be fine.

[scarlehoff]

No, that's not validphys, that's the python interface of pineappl.

If they work with the cli and not with the python interface it's a bug of pineappl.

However, another possibility is that either @t7phy or @comane are not using the latest pineappl version.

[comane]

The version I am using is:

import pineappl
pineappl.version
'0.5.9'

[t7phy]

@scarlehoff tbh I did not look at the above messages thoroughly, I was told my @comane that he was using it with validphys
@comane thats an old versionm try updating it

[comane]

@scarlehoff @t7phy , thank you, indeed upgrading solved the issue.

pineappl.version
'0.6.0'