Add converter for 'new' APPLgrids

[cschwan]

For the ATLAS and CMS single-inclusive jet and dijet NNLO grids available here:

• ATLAS 7 TeV dijet
• CMS 7 TeV dijet
• ATLAS 8 TeV single-inclusive jets
• CMS 8 TeV single-inclusive jets

Right now we can't convert them into PineAPPL grids because those are in a new file format, with a .appl suffix, which is incompatible with our version of APPLgrid: https://github.com/NNPDF/external/tree/master/applgrid-1.4.70-nnpdf.

[cschwan]

This depends on #161.

[cschwan]

With the new converters these work out of the box. Here's an example:

cschwan@wtpp123 grids $ pineappl --silence-lhapdf import applfast-atlas-dijets-appl-arxiv-1312.3524-xsec000.appl applfast-atlas-dijets-appl-arxiv-1312.3524-xsec000.pineappl.lz4 NNPDF40_nnlo_as_01180
appl::grid() version 1.6.27 compiled Jul 30 2022
appl::grid() reading grid from file applfast-atlas-dijets-appl-arxiv-1312.3524-xsec000.appl     :  applgrid version 1.6.29 (transformed to 1.6.27)      :  grid generated with NNPDF31_nnlo_as_0118 : set 0
+----------------------------------------------+
| ATLAS dijet production at 7 TeV              |
| arxiv: 1312.3524                             |
| d2sigma/dm12/dys [fb/GeV]                    |
| anti-kT R=0.6                                |
| Table 10:  0.0 <= y* < 0.5                   |
|                                              |
| https://ploughshare.web.cern.ch              |
|                                              |
| please cite:                                 |
|  PoS RADCOR2017, 074 (2018)                  |
|  PRL 118, 072002 (2017)                      |
|  JHEP 10, 155 (2018)                         |
|  EPJ. C79, 845 (2019)                        |
|  EPJ. C66, 503 (2010)                        |
|  Proc. DIS 2012, 217 (2012)                  |
+----------------------------------------------+

appl::grid() read grid, size 1.44429 MB in 213.615 ms   trim in 1.479 ms
b    PineAPPL     APPLgrid     rel. diff
--+------------+------------+--------------
0   9.5741685e5  9.5741685e5  1.5543122e-15
1   3.7061744e5  3.7061744e5 -6.6613381e-16
2   1.4077513e5  1.4077513e5  1.1102230e-15
3   5.6582832e4  5.6582832e4 -1.4432899e-15
4   2.3159408e4  2.3159408e4 -1.1102230e-16
5   1.0422135e4  1.0422135e4 -3.3306691e-16
6   4.7894678e3  4.7894678e3  1.3322676e-15
7   2.2842040e3  2.2842040e3  2.2204460e-16
8   1.1109483e3  1.1109483e3    0.0000000e0
9   5.3681270e2  5.3681270e2 -1.8873791e-15
10  2.5508164e2  2.5508164e2 -5.5511151e-16
11  1.1910298e2  1.1910298e2  4.4408921e-16
12  5.8051300e1  5.8051300e1  2.2204460e-16
13  2.7500268e1  2.7500268e1 -1.6653345e-15
14  1.2987431e1  1.2987431e1 -5.5511151e-16
15  5.9537060e0  5.9537060e0  6.6613381e-16
16  2.7115592e0  2.7115592e0    0.0000000e0
17  1.2028974e0  1.2028974e0 -3.3306691e-16
18 5.0109534e-1 5.0109534e-1  2.2204460e-16
19 2.0040917e-1 2.0040917e-1  8.8817842e-16
20 2.0618094e-2 2.0618094e-2 -1.2212453e-15

cschwan@wtpp123 grids $ pineappl obl --orders applfast-atlas-dijets-appl-arxiv-1312.3524-xsec000.pineappl.lz4
o  order
-+-------
0 O(as^2)
1 O(as^3)
2 O(as^4)

So they do contain the NNLO grids, but no scale-variation grids.

[alecandido]

So they do contain the NNLO grids, but no scale-variation grids.

I'm already happy with NNLO grids :D (finally we're getting closer to retire K-factors)

[felixhekhorn]

So they do contain the NNLO grids, but no scale-variation grids.

I'm already happy with NNLO grids :D (finally we're getting closer to retire K-factors)

In principle they should be calculable from the previous orders, right? as they are for DIS (of course now everything more complicated due to two PDFs)

[cschwan]

They should be, yes.

[alecandido]

In principle they should be calculable from the previous orders, right? as they are for DIS (of course now everything more complicated due to two PDFs)

Everything -> factorization scale variation (renormalization scale is the exact same). I wonder if it is actually possible from the grid itself, since they are events and you don't have the full analytic structure (I'm only confident that you can do it giving the analytic expression of lower orders), but maybe it is possible.

[cschwan]

@AleCandido let's talk here: https://github.com/N3PDF/pineappl/discussions/163.