Re: My updated fairtools library not loading [message #13666 is a reply to message #13665] |
Wed, 27 June 2012 09:12 |
Florian Uhlig
Messages: 424 Registered: May 2007
|
first-grade participant |
From: *gsi.de
|
|
Hi Raghav,
Quote: |
I think that i finally understand what the problem here is. So i need to create a digifile first using a macro and some parameter file and then call the reconstruction run that i am doing now.
|
This is correct. The normal sequence
1. Simulation which creates perfect MC Data.
2. Digitization which creates out of the MC data a data format which would be
produced by the real detector. For example out of the position of a MCPoint the
fired detector channel is calculated and written to output.
3. Reconstruction means that you come from your detector data to something like
particles (tracks with particle ID)
4. In the Analysis step you do your physics analysis, whatever this means for you.
Quote: |
Can you guys tell me if there is a prescription or a template on how to create the .par file. I see that it is very important for the digi and reco runs because it contains detector and individual specific cuts which vary from person to person.
|
There is no such prescription because the parameters very much depend on the procedure how you come from MC data to detector data. This you can do in a first step with a simple smearing of the MCPoint, where you probably don't need
parameters from a parameter file. You can simply hard code them. If you then have a
running chain you can identify the parameters in the task and the create a parameter file which contain the needed information. In my opinion you can't copy everything from Panda. You have to think what you need and implement only this stuff. This also helps to understand how the framework works
Quote: |
Right now when i look at the all.par file (pandaroot/macro/params/all.par) i see a lot of things which make sense right of the bat but a lot more which i really have no idea how to make for my detector.
|
As I said. You can't do copy and paste from Panda. You have to define what you need and then you can look how such a thing is implemented in Panda and try to learn from the example. Even if the frameworks makes many things very easy and also reusable this is not true for everything. Best is to start with simple tasks as all the experiments did and then go step by step to a more realistic chain of tasks when this is needed. If you want to go one step further then it is a good idea to look if someone else had a similar problem and how he solved it.
Ciao
Florian
[Updated on: Wed, 27 June 2012 09:12] Report message to a moderator
|
|
|