Hi, Omar.
I'm just going to CC this thread to the software list from now.
I can run the reconstruction now over all the events in this file after your fix:
http://www.slac.stanford.edu/~meeg/hps2/meeg/mock_data/tritrig-beam-tri_1-10_readout.slcio
I am using the steering file HPS2014OfflineRecon.lcsim which you recommended.
But I am still seeing a lot of problems with the ReconstructedParticle output...
1) Most of the events have no ReconstructedParticle collections in them at all.
Every event should have these collections in them, even if they are empty:
BeamspotConstrainedV0Candidates
FinalStateParticles
UnconstrainedV0Candidates
TargetConstrainedV0Candidates
If the same set of collections are not present in every event, it becomes clunky to access the data, requiring a lot of unnecessary checks in the Driver code to see if the collection exists. (EventHeader throws fatal exceptions if you try to access a collection that is not there!)
2) For the events that do have ReconstructedParticle collections in them, mostly they are empty, from what I can tell. I just stepped through maybe the first 100 events in the sample, so I would need to confirm that this is true for the 12k events. I only found one collection that had objects in it, and....
3) For the few events that I did find with ReconstructedParticles in them, the data is still all zeros.
Can you confirm these problems on your end? Or did I do something wrong here in my configuration?
BTW, when running the recon, you might put this Driver into your job along with the AidaSaveDriver to save the plots:
/hps-users/src/main/java/org/hps/users/jeremym/MockDataChallengeDiagnosticDriver.java
It plots basically all of the data in all the collections, so it can be used for diagnostics.
Or run its TestCase on your output data (just hand hack the TestCase for now):
/hps-users/src/test/java/org/hps/users/jeremym/MockDataChallengeDiagnosticDriverTest.java
Though right now it doesn't plot the ReconstructedParticle collections at all due to issue #1 (I set everything up automatically based on the collections in the first event but for this data set those collections aren't present in the first event!).
We can cover these issues in more detail at the software meeting.
--Jeremy
-----Original Message-----
From: Omar Moreno [mailto:[log in to unmask]]
Sent: Wednesday, May 07, 2014 2:59 PM
To: Uemura, Sho
Cc: Graham, Mathew Thomas; McCormick, Jeremy I.; Moreno, Omar
Subject: Re: problem running recon on MDC data
Ah, OK. Thanks Sho.
On Wed, May 7, 2014 at 2:58 PM, Sho Uemura <[log in to unmask]> wrote:
It's slow because the time cuts are very loose, i.e. right now there is a trade-off between speed and recon efficiency, and we've chosen to maximize efficiency. It's still fast enough for the MDC.
On Wed, 7 May 2014, Omar Moreno wrote:
My recon is running unusually slow. I thought this issue was fixed but
maybe I'm mistaken. Is anyone else having this problem??
On Wed, May 7, 2014 at 2:50 PM, Omar Moreno <[log in to unmask]> wrote:
The recon works now.
On Tue, May 6, 2014 at 4:04 PM, Mathew Graham <[log in to unmask]>wrote:
Right?that?s pretty unlikely.
On May 6, 2014, at 3:10 PM, McCormick, Jeremy I. <
[log in to unmask]> wrote:
But I'm not sure that explanation makes sense, anyways, because it does
get through 173 events before crashing, unless all those events had no
tracks and Billoir was not activated until the crash...
--Jeremy
-----Original Message-----
From: Graham, Mathew Thomas
Sent: Tuesday, May 06, 2014 2:44 PM
To: McCormick, Jeremy I.; Moreno, Omar
Cc: Uemura, Sho
Subject: Re: problem running recon on MDC data
Yes...so I have no answers.
On 5/6/2014 2:43 PM, McCormick, Jeremy I. wrote:
Okay, I see what you mean. But I'm running the entire reconstruction
chain from HPS2014OfflineRecon.lcsim on the simulated readout data. So
shouldn't those classes be available?
-----Original Message-----
From: Graham, Mathew Thomas
Sent: Tuesday, May 06, 2014 2:41 PM
To: McCormick, Jeremy I.; Moreno, Omar
Cc: Uemura, Sho
Subject: Re: problem running recon on MDC data
That line is casting the track as a SeedTrack and then getting a
SeedCandidate...however, you can only do this if you run reconstructions
reconstruction since SeedCandidate isn't persisted and isn't even known in
lcio.
On 5/6/2014 2:01 PM, McCormick, Jeremy I. wrote:
Hi,
I'm attempting to run the current version of the recon on this data
from Sho.
http://www.slac.stanford.edu/~meeg/hps2/meeg/mock_data/tritrig-beam-tr <http://www.slac.stanford.edu/~meeg/hps2/meeg/mock_data/tritrig-beam-tr>
i_1-10_readout.slcio
However, I get a fatal exception some events into the job.
Event 174
java.lang.NullPointerException
at
org.hps.recon.particle.HpsReconParticleDriver.toBilliorTrack(HpsReconParticleDriver.java:100)
at
org.hps.recon.particle.HpsReconParticleDriver.vertexParticles(HpsReconParticleDriver.java:62)
at
org.hps.recon.particle.ReconParticleDriver.process(ReconParticleDriver.java:142)
at org.lcsim.util.Driver.doProcess(Driver.java:273)
at org.lcsim.util.Driver.processChildren(Driver.java:284)
at org.lcsim.util.Driver.process(Driver.java:198)
at
org.lcsim.util.DriverAdapter.recordSupplied(DriverAdapter.java:74)
at
org.freehep.record.loop.DefaultRecordLoop.consumeRecord(DefaultRecordLoop.java:832)
at
org.freehep.record.loop.DefaultRecordLoop.loop(DefaultRecordLoop.java:668)
at
org.freehep.record.loop.DefaultRecordLoop.execute(DefaultRecordLoop.java:566)
at org.lcsim.util.loop.LCSimLoop.loop(LCSimLoop.java:152)
at
org.lcsim.job.JobControlManager.run(JobControlManager.java:416)
Any idea on what the problem could be and if it can be easily fixed?
--Jeremy
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the HPS-SOFTWARE list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1
|
