Positivity #146
Positivity #146
Conversation
Codecov Report
@@ Coverage Diff @@
## develop #146 +/- ##
===========================================
+ Coverage 68.74% 68.82% +0.08%
===========================================
Files 82 82
Lines 3996 4007 +11
===========================================
+ Hits 2747 2758 +11
Misses 1249 1249
Flags with carried forward coverage won't be shown. Click here to find out more.
|
This part should be implemented as a suitable runcard, not in |
actually, yes and no: of course everything will eventually end up in runcardsrunner (as said): NNPDF/pinecards#143 - but it will be just a plain copy (as for the other DIS experiments); first I'd like to take advantage of the data package here which allows me an easy benchmark |
|
@enocera thanks for your info! I hope I implemented the python equivalent here - however when running a simple, LO benchmark against apfel it is complaining " Invalid value of x = 4.9999999999999998E-007" (full output below) ... do you know if they (=apfelcomb) hacked apfel in some way to run it (i.e. extend it's default grids)? in any case I try to benchmark against the old FK table ... WARNING: ZM-VFNS is a VFN scheme
... setting VFNS PDF evolution
Allowed evolution range [ 1.0000 : 10000.0000 ] GeV Initialization of the evolution completed in 0.140 s Report of the electroweak parameters: Mass of the Z = 91.188 GeV Report of the DIS parameters: Computation in the ZM-VFNS mass scheme Initialization of the DIS module completed in 0.018 s In F2light.f: |
|
@felixhekhorn This is strange. If I run apfelcomb it runs just fine. And as far as I know, apfel has not been hacked for the purpose of positivity observables. Have you tried to sample the x grid, instead from x=5e-7, from 5.000001e-7? |
|
@enocera yes I tried, and this does not work either - in fact I believe the minimum x-grid value in apfel is 1e-5 as defined here: https://github.com/scarrazza/apfel/blob/4426f689a95fbde94012083dd55a9af0edcc5928/src/Evolution/initParameters.f#L89 and indeed this works PS: since we're speaking of it and you might be interested in the future: our default x_min is 2e-7 as defined by the PineAPPL paper https://github.com/N3PDF/yadism/blob/ecc16f965504435423e44868790cf5d154e0d7b2/extras/data/machinery/generate/utils.py#L12 |
|
alright, there are strange things happening ... someone has an idea? @alecandido @cschwan @scarlehoff @enocera the summary is:
check against apfelif I fix the x_min problem, by either putting the cut temporarily to 5e-5 or by forcing APFEL to use my grid (which has support in that region) I'm getting something reasonable:
the old FK tablewhen I look into the NNPDF FK table I can see problems:
comparing new against old inside vp
comparing new against old with pineappl
|
Then I might doubt this statement: the version of APFEL used to produce old FkTables is not the same as yours. In the meanwhile a few things have been changed... (in principle not so relevant)
On the other side, APFEL is Evolution + DIS. You agree with DIS, so the problem might still be inside evolution (at the level of FkTables). |
|
@felixhekhorn (cc @scarlehoff @alecandido). Let me make some remarks.
|
|
Since the comparison seems alright - I'm ready to merge; @alecandido whenever you have time, please review and we merge
PS: in the end the non-matching was just due to not using the correct version inside pinefarm 🙈 thanks @scarlehoff for making me realize |
|
Overall, there are only minor corrections and a few clarifications to be provided. |
|
🎉 |
F2up(x=?,Q2=?)