Add switch weight to normalized weights. #305
Conversation
Thanks Marcel! I think you have to re-run only the DY sample during the weekend :) |
Marcel can you please hold this for a moment? I am just checking the output of the skims after adding the stitchWeight to EvtW and the DY yield is non-sense. I had a doubt but now I am sure it is not needed, here you have my explanation. The stitch weight that was computed before was there to stitch different samples with different XS in a single sample compatible with the inclusive one. Now we are doing something quite different: we have three macro samples (inclusive, jet-binned, pt-binned) and they all have the same XS by construction. The weight we are applying now is quite different: it basically cuts by 1/2 or 1/3 the event yield for events with Z pT larger or greater than 50 GeV so it is not a shape weight but a rough cut we have to apply coming from the fact that we have 2 or 3 time the statistics. I don't know if I made myself clear... |
|
You're right I think. I forgot that the meaning of the I think we can close this one then. |
|
Sorry I missed this entire discussion. I also agree that the stitchWeight should not be included in the sum of weight. Closing now. (Thanks for looking into it though!) |
This PR adds the
stitchWeightto theEvtWweight in the producers for all years, which effectively renormalizes particular weights and only preserves their shape-changing effect as reported by @dzuolo in #304.Closes #304.