Differences in distribution of number of DLA per forest between data and mocks

[londumas]

Same as issue igmhub/LyaCoLoRe#33.
The number of DLA per forest is quite different between data DLA and mock DLA and random mock DLA.

[TEtourneau]

I guess for randoms it's normal, since we need to have more DLA at the end in the catalog.

[jmarclegoff]

The difference between mock and data is tricking on log scale, but actually Javier just needs to increase the DLA proba in the mock by ~10% to solve it.
Thomas, we indeed want more stat in the randoms, to make the statistical error due to random negligible. I think we should rather reach this by having random catalogues with the same DLA proba as for mocks, but having several random catalogues for 1 mock catalogue, and we compute RR,DR and RD for each random catalogue, then make the average. This can make a difference, e.g. in the mocks most quasars have only one DLA, so there are few DLAs (<20%) at the same (ra,dec), whereas in the random catalogue you have typically 3 or 4 DLAs for a single quasar.

[londumas]

@jmarclegoff, sorry for not being clear enough, the difference is not in the raw number. These histograms give the normalized distribution. The raw number is perfect.
The difference I want to point at in this plot is in the shape of the distribution of number of DLA per forests.
This is really not an important difference, and we don't have to get that perfectly ok.

[londumas]

Still different in v4.2, this is really not important and also we don't have a clear idea of what is the proper distribution. Let's not fix that for the moment.