[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: "nan" in the energy of the DBCALPhoton
Hi Alex,
I'll look at this today.
-Blake
Alexander Somov wrote:
> Hi Matt,
>
> Thanks for the prompt reply and explanations.
> Blake, could you please take a look on this
> reconstruction feature.
>
> Thanks,
> Alex
>
>
> On Tue, 30 Jun 2009, Matthew Shepherd wrote:
>
>>
>> Hi Alex,
>>
>> This particular section of code was added by Blake recently to try to
>> improve reconstruction at the end of the module. I'm not familiar
>> with it -- Blake, can you help Alex?
>>
>> Indeed it looks like the problem is trying to call pow with a
>> negative base. That will certainly generate nan.
>>
>> -Matt
>>
>> On Jun 30, 2009, at 5:04 PM, Alexander Somov wrote:
>>
>>>
>>> Hi Matt,
>>>
>>> I have found a few events in the BCAL reconstruction
>>> where the photon energy is not calculated properly.
>>> Below is a typical example:
>>>
>>> DBCALPhoton:
>>> x: y: z: px: py: pz: E:
>>> ----------------------------------------------------
>>> 56.62 37.97 133.08 0.12 0.08 0.15 0.21
>>> -9.08 67.51 211.40 -0.06 0.46 1.01 1.11
>>> -64.92 23.81 306.27 -0.45 0.16 1.67 1.74
>>> 19.78 -62.68 405.37 nan nan nan nan <---
>>> 61.69 -31.96 397.37 0.01 -0.00 0.05 0.05
>>>
>>> DBCALShower:
>>> x: y: z: t: E:
>>> --------------------------------------
>>> 56.62 37.97 133.08 6.11 0.18
>>> -9.08 67.51 211.40 5.40 1.04
>>> -64.92 23.81 306.27 8.38 1.61
>>> 19.78 -62.68 405.37 11.77 0.51 <---
>>> 61.69 -31.96 397.37 12.55 0.03
>>>
>>> The problem appears at the edge of the BCAL, specifically
>>> in the DBCALPhoton_factory.cc file where the energy of a
>>> photon is calculated (for zEntry >= 370.0 ):
>>>
>>> double energy = pow( ((**showItr).E - lin ) / scale, 1 / ( 1 +
>>> nonlin ) );
>>> lin -0.08144
>>> nonlin -0.616129
>>> scale -0.0563619
>>> **showItr 0.505658 (shower energy)
>>> energy nan
>>>
>>> pow(x,y) is called here with the negative x.
>>>
>>> Could you please check the callibration coefficients in the
>>> DBCALPhoton_factory.cc
>>> factory (I have no clue how these coefficients have been obtained).
>>> I would like to finish with the acceptance studies shortly.
>>>
>>> Thanks,
>>> Sascha
>>>
>>>
>>> P.S. Just in case, I put the hddm file with "p eta pi0 pi0" events
>>> where i have observed the problem to ~/public_html/debug.hddm
>>>
>>
>
>