pulseArea GetS1 Gaussian width

[sophiafarrell]

In the NESTcalc::GetS1 function, there is a basic, no-timing block of code here:

nest/src/NEST.cpp

Lines 886 to 891 in 3322903

	Nphe = nHits + BinomFluct(nHits, fdetector->get_P_dphe());
	pulseArea = RandomGen::rndm()->rand_gauss(
	BinomFluct(Nphe, 1. - (1. - eff) / (1. + fdetector->get_P_dphe())),
	fdetector->get_sPEres() * sqrt(Nphe));
	//proper truncation not done here because this is meant to be approximation, quick and dirty
	spike = (double)nHits;

In these lines, pulseArea is smeared by sPEres * sqrt(Nphe) where the Nphe is defined:

nest/src/NEST.cpp

Line 886 in 3322903

Nphe = nHits + BinomFluct(nHits, fdetector->get_P_dphe());

To my understanding, this slightly (very, very slightly) overestimates the smearing effect, but it should probably not be ignored. My suggestion would be to replace the linked lines above with something like this to make the pulseArea calculation more statistically accurate. But, perhaps I am missing something.

    Nphe = nHits + BinomFluct(nHits, fdetector->get_P_dphe());
    Nphe_det = BinomFluct(Nphe, 1. - (1. - eff) / (1. + fdetector->get_P_dphe()));
    pulseArea = RandomGen::rndm()->rand_gauss(
					      Nphe_det,
					      fdetector->get_sPEres() * sqrt(Nphe_det));

This is in regards to the question I asked @mszydagis and then directed toward @riffard, but I figured that a git issue is more trackable for everyone. (Tagging @robertsjames here because he had the same question as well.)

[mszydagis]

thank you for catching this mistake. For >90% detection efficiency there is no difference, but who knows if we can maintain (like into G3 in the future) such high efficiencies. Therefore, it is important to do this right. The problem has been corrected

[sophiafarrell]

(Referencing 9363b96 so people can track the change). Thanks!