[G12] parm files

[Michael C. Kunkel]

Greetings,

"

Would you mind explaining again why we have no idea how to tune Multi = false? My apologies if I am asking you to repeat yourself.

"
I think this answers you own question. Please reread Dustin Keller page 
https://userweb.jlab.org/~dustin/software/Makeshift/HowTo/HowToAddConstraints.txt

In particular the statement "

Option flase requires that you have tuned parameters
	  for you particle species.  These parms are called in KinLine.C and needs
	  to point to your copy of parms:
	  String fileName_base = "/your_dir/parms/";
	  */
"



The arrangement of the parm files is as follows, sector, theta bin (14 
bins), parameter.
When tuning multi = false you need to find the pulls for each of these 
bins and tune these pulls until they are proper gaussian with porper 
mean and sigma.

I have uploaded the histograms for the tune channel
https://jlabsvn.jlab.org/svnroot/clas/users/mkunkel/clas/kinematic_fitter/MULTI_FALSE_TUNE_HISTOGRAMS/


I hope this is more clear that you cannot just tune the multi=false case 
with res_scale, p_scale and r_scale.


BR
MK



On 5/22/14 6:48 AM, Volker Crede wrote:
> Dear Michael,
>
> Would you mind explaining again why we have no idea how to tune Multi = false? My apologies if I am asking you to repeat yourself.
>
> My understanding is that Dustin Keller extended the Williams g11a fitter, among other things by adding Multi = true. In other words, Multi = false was the g11a default and it worked great as we all know.
>
> Moreover, the g11a "parms" appear to be working for Christopher using the reaction p pi+ pi- and tweaking the overall "res_scale" factor slightly. We also slightly adjusted the proton momentum pull but the correction is well within the detector resolution (and also within the fluctuations among individual runs) and the procedure probably as justified as any other iterative approach you are suggesting. The remaining Unknown in this business still provides the biggest uncertainty.
>
> - Volker
>
>
>> On May 22, 2014, at 12:04 AM, "Michael C. Kunkel" <mkunkel at jlab.org> wrote:
>>
>> Greetings,
>>
>> I was afraid of that you somehow had g11 parms. My parms file look quite different. If I was asked to verify multi = false this problem would have been resolved last week. Regardless of that, I looked into the timestamp of when my parms were created with conjunction to when the beam corrections were performed. I have no clear indication that I tuned this with beam corrections, nor do I have verification that I tune it without beam corrections. I am re-running my tune program.
>>
>> With that said, I do suggest you use multi = true since you have no idea how to tune multi = false.
>>
>> BR
>> MK
>>
>>
>>
>>> On 5/21/14 5:49 PM, Pedersen, Christopher wrote:
>>> Hello,
>>>
>>>       I have attached the parms files you sent and the ones from Mukesh which apart from the additional electron parms are identical and are in fact identical to the g11 parms.
>>>
>>> Chris
>>> ________________________________________
>>> From: g12-bounces at jlab.org [g12-bounces at jlab.org] on behalf of Michael C. Kunkel [mkunkel at jlab.org]
>>> Sent: Wednesday, May 21, 2014 4:13 PM
>>> To: g12
>>> Subject: [G12] parm files
>>>
>>> Greetings,
>>>
>>> Would everyone using the fitter for pions please send me their
>>> g12_parms. I want to investigate something.
>>> For FSU, please send me Mukesh's original and the original I sent.
>>>
>>> BR
>>> MK
>>> _______________________________________________
>>> G12 mailing list
>>> G12 at jlab.org
>>> https://mailman.jlab.org/mailman/listinfo/g12
>>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/g12/attachments/20140522/07880807/attachment.html