Issue converting NNLO ttbar fastnlo grids

[LucianHL]

Hi,

I have run the fastnlo to pineappl converter without issue for NLO grids, but for the NNLO grids for ttbar production provided here there is a small but non-negligible difference between the original and converted grids. E.g. for LHC8-Mtt-HT4-173_3-bin1.tab from here I get:

b   PineAPPL     fastNLO      rel. diff
-+------------+------------+-------------
0  1.0817194e0  1.0821781e0  4.2405530e-4
1  1.0725099e0  1.0722009e0 -2.8814110e-4
2 6.5609663e-1 6.5619054e-1  1.4314243e-4
3 3.3629315e-1 3.3631982e-1  7.9301309e-5
4 1.3767895e-1 1.3750424e-1 -1.2689380e-3
5 3.4284160e-2 3.4298089e-2  4.0628058e-4
6 4.1848976e-3 4.1888985e-3  9.5603763e-4

which is not awful but a lot more than usual, and indeed it fails the default accuracy test. I don't think this is the same issue as in #228 as the scale is not flexible - and anyway the differences are too small for that. Perhaps this is due the grids being NNLO?

[cschwan]

Hi @LucianHL,

I can reproduce the behaviour and that's definitely a bug. I don't think it's directly related to NNLO, as we have successfully converted other fastNLO tables A. Mitov's website.

EDIT: It seems I can no longer convert any table from Mitov's site, so maybe there's a regression.

[LucianHL]

Hi @cschwan, thank you - yes indeed I was going to say that I had tried others and they had the same issue.

[cschwan]

This was a genuine bug that was caused by having accidentally swapped the x1 and x2, see commit 213e68a. This has now been fixed in master and v0.8.x.

[LucianHL]

Hi @cschwan thank you yes that has worked!

EDIT: I closed this with a comment but should probably leave that to do you to do in case there is more to be done here, so have re-opened

[cschwan]

No worries and thank you for the report, @LucianHL, this should've been discovered a long time ago!

[felixhekhorn]

EDIT: I closed this with a comment but should probably leave that to do you to do in case there is more to be done here, so have re-opened

just for future reference: I would say the person who opens an issue should preferably also close it to confirm it is actually resolved. If further actions are required on our side we reopen it 🙃 (and close again)