Fake energy in hypathia's PMT noise simulation

[gondiaz]

Fake energy in hypathia's PMT noise simulation

[gonzaponte]

I've just checked and for NEW, the value of pmt_pe_rms ranges from 0.354 to 0.434. Do we expect higher values for N100? For pmt_pe_rms = 0.434 the mean of the distribution is 1.0016 and for pmt_pe_rms = 0.354, 1.0002. The expected fake energy overall is < 0.1%, so this might help for now, but we need a longer-term solution.

We could think of something like redefining the distribution so it is not defined below 0 (like cutting it away instead of clipping). This requires cutting away the tail towards high values too, something like the attached picture. This would keep the mean in the right place.

[paolafer]

I think that @andLaing wrote the charge_fluctuation function, maybe he can give some input here.

[andLaing]

So the function was written to simulate the fluctuation in the amplification stages of the PMTs which can't produce negative charge alone, only reduce or increase the current (in the case of PMTs) seen at the anode. It's not simulating electronics noise which is bipolar and in the full simulation (diomira?) would be added by convoluting with the Pedestal distributions. Like in a lot of these toy simulations the Gaussian is an approximation meaning we clip the negative part to zero.

The values stored in the database for NEW are the 1 pe peak positions that we extract in the calibration where the fit is to a function that uses the distribution without light and a sum of related Gaussians (with the clipping taken into account, I think, but look at the functions in IC) to fit for the relevant parameters.

If I've understood the plot correctly, @gondiaz, what you call 'sampling mean' is the original test value? Or always 1s? and the other value is the mean of everything but the zeros? The relevance of the shift depends on your definitions. Really the 1 should be the peak, not the mean of the values. Moreover, surely you need to add the noise before worrying?