Gauss hit zeroes fit

[cmargalejo]

This pull request aims to solve some issues that were found with the observables GaussSigmaX, GaussSigmaY and GaussSigmaZ, which are computed by fitting a gaussian to each event.

Some of the events were found to have sigmas that were too large. After a deep study, the hypothesis is that the fit fails especially for small events that have too few hits. Not seeing a hit doens't necesarily mean there is no energy on that channel, but that it is below the threshold. Thus, the proposed solution is to add one hit to the left and one hit to the right, at a value 0+/-error.

The main changes are:
-Adding two hits to each hits event.
-Define the energy of these hits as 0 +/- 70 ADC. This error is defined based on typical for the micromegas detectors.

To illustrate the changes, the image below shows the same event before the changes in the code in the left, and the new fit in the right.

[jgalan]

Hi @cmargalejo , the problem with the pipeline is surely related to the fact that your branch is not updated with master.

You should do locally:

git checkout master
git pull
git checkout gaussHitZeroesFit
git merge master
git push

So that this PR gets updated

[jgalan]

It would be good to add a description to this method. Perhaps reusing some text from the PR description. I see this class is not following the standard headers. I will just quickly updated that

[cmargalejo]

Description added in 154df75

[jgalan]

-Define the energy of these hits as 0 +/- 70 ADC. This error is defined based on typical for the micromegas detectors.

What it happens if you set the new hits on the sides to equal zero? If you add the new hits, then the energy of the event will be affected, or the hits are added just for the gaussian fit, then removed for further processing?

[jgalan]

Perhaps it would be interesting to add the image you attached to the PR description to the documentation.

[cmargalejo]

-Define the energy of these hits as 0 +/- 70 ADC. This error is defined based on typical for the micromegas detectors.

What it happens if you set the new hits on the sides to equal zero? If you add the new hits, then the energy of the event will be affected, or the hits are added just for the gaussian fit, then removed for further processing?

Yes, the hits are only added to a TGraph to perform the fit, so they don't go into any further processing. Still, their energy is 0.