GSI Forum: Bugs, Fixes, Releases » [FIXED] Real vs. Ideal Tracking

Home » PANDA » PandaRoot » Bugs, Fixes, Releases » [FIXED] Real vs. Ideal Tracking - Ghosts?

Show: Today's Messages :: Polls :: Message Navigator

[FIXED] Real vs. Ideal Tracking - Ghosts? [message #14587]

Mon, 08 April 2013 10:15

Ralf Kliemt
Messages: 507
Registered: May 2007
Location: GSI, Darmstadt

first-grade participant

From: *gsi.de

Hello everybody.

While going through the analysis tools I simulated quickly 1000 events of
Psi(3770) --> D+D- --> K-pi+pi+ K+pi-pi-.

Now at the D reconstruction I find much "dirt" in the mass spectrum. More unsettling is the spiky behavior. To see whats going on I did a MC truth match and tried the ideal Pattaern recognition as well.

The plots are attached below. My ideas so far are that the MC indexing is done by the first hit or something, which would be error prone. Also I think the Pattrern Recognition adds LOTS of ghosts and fakes mostly many in one event.

Cheers Ralf

Attachment: Dplus-raw.pdf
(Size: 5.15KB, Downloaded 220 times)
Attachment: Dplus-raw-simple.pdf
(Size: 4.94KB, Downloaded 233 times)
Attachment: Dplus-truth.pdf
(Size: 4.98KB, Downloaded 263 times)
Attachment: Dplus-truth-simple.pdf
(Size: 4.86KB, Downloaded 239 times)

[Updated on: Fri, 19 April 2013 12:30] by Moderator

Report message to a moderator

Re: Real vs. Ideal Tracking - Ghosts? [message #14588 is a reply to message #14587]

Mon, 08 April 2013 12:43

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *vpn.unito.it

About the MC indexing, this comes from the most common MC id present in the track. If more than one mc tracks are present in the same reco track, only the "majority" part is used as id.
As far as it concerns the doubles, which trunk are you using? The last message I got from Gianluigi and Klaus is that doubles should be less than 1%.

Report message to a moderator

Re: Real vs. Ideal Tracking - Ghosts? [message #14589 is a reply to message #14587]

Mon, 08 April 2013 13:05

Ralf Kliemt
Messages: 507
Registered: May 2007
Location: GSI, Darmstadt

first-grade participant

From: *gsi.de

Hi Stefano,

The simulations were performed with a trunk from March. The MC tracking issue from Klaus was closed in the beginning of February.
So it's "new".

Cheers
Ralf

Report message to a moderator

Re: Real vs. Ideal Tracking - Ghosts? [message #14590 is a reply to message #14589]

Mon, 08 April 2013 15:06

Gianluigi Boca
Messages: 177
Registered: March 2004

first-grade participant

From: *pv.infn.it

Hi Ralf,
please update your trunk version of the
tracking
directory and tell me if you still see the problem.
I have been working a lot on that directory changing things back and forth, maybe by chance you have a bad version. The latest version should contain at most 1% ghosts.

Please try and tell me if that is NOT the case, thanks in advance
Gianluigi

ps
in the future the ghosts will be definitely removed at the pattern recognition level, I have that in my to do list.

Ralf Kliemt wrote on Mon, 08 April 2013 13:05

Hi Stefano,

The simulations were performed with a trunk from March. The MC tracking issue from Klaus was closed in the beginning of February.
So it's "new".

Cheers
Ralf

Report message to a moderator

Re: Real vs. Ideal Tracking - Ghosts? [message #14592 is a reply to message #14590]

Mon, 08 April 2013 15:26

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *vpn.unito.it

(and if you could post here your dec sim digi reco pid and analysis macro I could do some tests this evening).

Report message to a moderator

Re: Real vs. Ideal Tracking - Ghosts? [message #14594 is a reply to message #14592]

Mon, 08 April 2013 23:01

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *vpn.unito.it

Hi Ralf,
I have done tests with the last trunk, and everything seems fine:

D+ invariant mass for all the combinations, ideal (blue) and real (red) tracking:

index.php?t=getfile&id=7330&private=0

and with pid selection using montecarlo truth of kaons and pions:

index.php?t=getfile&id=7331&private=0

Everything seems fine to me. I attach here the analysis macro I have used.

I have few comments to add.

I have used the following PndAnalysis constructor:

PndAnalysis* theAnalysis = new PndAnalysis("SttMvdGemGenTrack","FtsIdealGenTrack");

The system is not able to load them automatically, I think it should be stressed that one has to write the names of the TCAs. Not clear to me what is doing the code with the track objects.

I have tried to do the "mother" mc identification:

mcm.SetType(dplus,"D+");

It is not working, while it was worlking for the simple particles. Maybe I have done something wrong (could you please correct me?) or maybe there is some bug in the PndMcMatch for composite candidates.

Attachment: hdplus_nopid.gif
(Size: 9.14KB, Downloaded 225 times)
Attachment: hdplus_pid.gif
(Size: 7.02KB, Downloaded 307 times)
Attachment: anatut_psi2s.C
(Size: 5.63KB, Downloaded 181 times)

Report message to a moderator

Re: Real vs. Ideal Tracking - Ghosts? [message #14596 is a reply to message #14594]

Tue, 09 April 2013 07:55

Klaus Götzen
Messages: 293
Registered: June 2006
Location: GSI

first-grade participant

From: *gsi.de

Hi Stefano,

did you use evtgen->SetStoreTree() in simulation? If not, the mc matcher cannot do composites correctly.

You could also try with

mcm.SetType(dplus, 411)

to be independent of any naming scheme. Just to mention, charged conjugates are not automatically set, so you need to set the type for D+ and D- separately.

Best,
Klaus

Report message to a moderator

Re: Real vs. Ideal Tracking - Ghosts? [message #14597 is a reply to message #14596]

Tue, 09 April 2013 21:02

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *vpn.unito.it

Hi,
you are right, I had a SetStoreTree(kFALSE). Setting it to kTRUE now the D+ id works.
Thanks

[Updated on: Tue, 09 April 2013 21:03]

Report message to a moderator

Previous Topic:	[FIXED] Detector geometry is not visualized inside the FairBaseParSet Container
Next Topic:	[FIXED] eventdisplay does not load on Fedora 18

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Apr 25 21:18:42 CEST 2024

Total time taken to generate the page: 0.00898 seconds