[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Minutes of the June 17, 2009 GlueX Offline Meeting
Hi Mark,
Sorry to miss the mtg. -- we were getting our radiation safety
training so we can nuke lead glass blocks at IUCF.
On Jun 19, 2009, at 9:41 AM, Mark M. Ito wrote:
> We spent some time discussing Matt's suggestion (from two
> meetings
> ago, actually) of having all random-number related processes,
> such as
> smearing, occur in a stage before reconstruction and having
> the
> results of that stage written to disk. David pointed out that
> there is
> already a program, mcsmear that takes the hdgeant output and
> produces
> smeared hits in HDDM format, suitable for the reconstruction
> program.
> Right now only tracking is smeared but it would be
> straightforward to
> do smearing of the calorimeters as well.
My suggestion pertains specifically to how we choose to model detector
response in the MC. To try to give a concrete example: suppose that
using data we measure the efficiency for every straw in the CDC (e.g.,
by finding tracks that go through the straw and then asking whether or
not the straw saw a hit). We can load this info into the database for
various run number regions and then for any given run we can get the
actual straw tube efficiency. We then want to be sure that our Monte
Carlo models this efficiency properly so at generation time we pull
the info from the database for the run we are measuring and throw
random numbers to determine which straw hits get suppressed. In order
to do this, we need to be able to use the database API. As far as I
know, it is not foreseen to have this available in HDGeant. The
natural place to put it (in my opinion) is in some factory that sits
between HDGeant and the tracking code since there is a database API
there. The goal should be to make the output of this factory
identical to what comes in from the raw detector data stream. Right
now, I don't know how to do this and save the intermediate output.
What this means is that every time I run my code, hits in the CDC
could appear and disappear which could potentially change which tracks
are found in an event, their momenta, etc.
There are going to be aspects of the simulation like this that we want
to tune in a run dependent way. In E852 we saw clear evidence for
time dependent deadening of the chambers around the forward beam
hole. Environmental conditions in the hall (beam related backgrounds,
RF noise, etc.) can often be time dependent. Think of not only having
a "calibration database" but also a "MC tuning constants" database.
The problem is that unlike calibration, sometimes MC tuning is not a
deterministic process.
I haven't looked at David's mcsmear -- it may fit the bill exactly,
and we just need to have it read from the existing response factories
that have been developed for the calorimeter. I'm also happy to help
with development, but having some consensus as to whether this is a
significant issue and how we want to handle it would be good.
-Matt