Issues for consideration by the CAL SW group

Toby Burnett 09/22/2004 07:58 -0700

1: The energy estimate EvtEnergySumOpt is biased!

This plot is a comparison of the relative error in the energy estimate between GlastRelease v4r4 (blue) and v4r3p1 (red), generated by the System Test interface for normal 100 MeV photons. It shows that v4r4 is 10% low. The recent v4r5 is the same as v4r4, and cannot be validated until this is fixed. It is a matter of agreeing about the threshold. Let's do it, please!

2: It is very awkward for CalRecon to depend on summary code in AnalysisNtuple.

The class CalRecon::CalValsCorrTool was written as a temporary kluge 18 months ago by Richard and myself to invoke Bill Atwood's code in AnalysisNtuple::CalValsTool to correct the measured energy with the geometry of the cracks, etc. It is invoked by the following:

double correctedEnergy;
StatusCode sc = m_calValsTool->getVal("CalEnergyCorr",correctedEnergy);

where m_calValsTool points to the AnalysisNtuple::CalValsTool tool. The getVal method causes the class to execute its calculate method, setting all values like CalEnergyCorr.

I have copied, and suitably adapted, the current version of this code into this class, and added the following lines just after the above:

sc = calculate();
double correctedEnergy2 = CAL_Energy_Corr;
if( fabs(correctedEnergy-correctedEnergy2) > 1e-6 ){
throw std::runtime_error("CalValsCoorTool recaculation different from CalValsTool");
}

where calculate() is the method from CalValsTool that sets a bunch of data members, including CAL_Energy_Corr, as above.

I ran this for 10 K 1 GeV vertical muons, the exception was never thrown.

This code really belongs in CalRecon, and the dependency of CalRecon on AnalysisNtuple has caused problems. Can I put it there and remove the getVal?