New baseline 26/03/13

[kamillaurent]

I updated the example-nnpdf41.yml runcard with the new baseline (26-03-13).

Here is a comparison with the old baseline (260202-jk-nnpdf41-mhou):
https://vp.nnpdf.science/l2adpDSrRt2lGJK78Ncc1g==

[scarlehoff]

One problem I see (cc @tgiani) is that using the hyperscan trials might need more than 100 replicas to reach the kind of gaussianity we are used to 🤔

How many replicas did you do? Could you double that?

[kamillaurent]

I did 100 replicas, I will add other 100 and update the result

[kamillaurent]

Here is the comparison to the previous baseline, this time including 200 replicas:
https://vp.nnpdf.science/kmKtsmmVSQiTecvB1UXEOA==

[scarlehoff]

I have the feeling we might need to go all the way up to 1000 replicas :'(

[kamillaurent]

Doing it now ;)

[scarlehoff]

I'll use this PR to regenerate the data before merging.

I'm going to merge it, but please copy here as well the report with the 1000 replicas once it is done (and then also in the wiki of course) but it is always useful to have it associated to the code itself.

[kamillaurent]

Here is the 1000 replcas fit, compared to the previous baseline 260202-jk-nnpdf41-mhou:
https://vp.nnpdf.science/F0cBTV7JReSfXuy3F_Pvhw==