The proper way to do this (which is how most other experiments do it) is loading an “ideal” detector model and then applying alignment constants to that model, then recomputing the global transformations of the detector elements from the resulting (ideal+delta) transformations. I would probably implement this in our framework by having a class within the conditions manager which applied the alignment constants to the detector based on the run number. If no alignment constants were found it would simply use the detector “as is”.
This would require that the detector model being used could accept a set of SvtAlignmentConstant objects or some data structure representing them and then apply them to the detector rather than only read them from definitions in the compact.xml file. So I believe the building of the detector should be split into two separate steps (building the ideal model and then applying alignment constants). I’d need to talk with Pelle to find out what the new model supports for this. Perhaps we already support some/most of this. Otherwise, I think we will be stuck with supplying a unique compact.xml file for each set of unique alignment parameters if we don’t do it this way.
The MC simulation is another issue given that SLIC only accepts LCDD files. So we would still need some way to generate those for each configuration, and probably having separate compact.xml files for different alignment configurations is the only real solution here that could be implemented easily.
On Apr 29, 2015, at 11:33 PM, Sho Uemura <[log in to unmask]> wrote:
> I put a run script and an Auger template in /u/group/hps/production/data/auger/onlinedq.
>
> "./runjob.sh recon_10k.xml 5153 0 2mm" does what you'd expect - runs 10k events (enough for most of the things I want to look at), does recon and DQM. All output gets dumped in /volatile/hallb/hps/engrun/onlinedq.
>
> I also added a script (/u/group/hps/hps_soft/hps-java/get_nexus_jar.sh) that downloads the most recent Nexus jar and updates a symlink. I think it's useful.
>
> On Thu, 30 Apr 2015, Graham, Mathew Thomas wrote:
>
>>
>> Nobody is implying otherwise...
>>
>>> On Apr 29, 2015, at 6:13 PM, Nathan Baltzell <[log in to unmask]> wrote:
>>>
>>> Make no mistake, we are all "hugely busy".
>>>
>>>
>>>
>>>
>>> On Apr 29, 2015, at 8:56 PM, "Graham, Mathew Thomas" <[log in to unmask]> wrote:
>>>
>>>>
>>>>> On Apr 29, 2015, at 5:51 PM, Nathan Baltzell <[log in to unmask]> wrote:
>>>>>
>>>>> If 4-6 is not suitable, then addressing this now is worthwhile.
>>>>> We have 2 weeks of beam upcoming.
>>>>>
>>>>> I highly doubt we need a new conditions backend just
>>>>> to choose SVT position based on run number.
>>>>
>>>> We don?t need an entire new backend, just the SVT table?and the java to get them and put them somewhere. There are some details about what happens first (db reading or detector making) as well. It?s not a huge job but people are hugely busy and were still at the point where ?about 2mm? or ?about 3mm" is good enough.
>>>>
>>>>> Make
>>>>> detectors according to positions and prescribe run
>>>>> numbers accordingly in the db.
>>>>>
>>>>> Ok, my 2 cents.
>>>>>
>>>>> -Nathan
>>>>>
>>>>>
>>>>> On Apr 29, 2015, at 8:42 PM, Nathan Baltzell <[log in to unmask]> wrote:
>>>>>
>>>>>> It seems SVT position is a pretty fundamental thing to have in the db,
>>>>>> and well before the end of this May run.
>>>>>>
>>>>>>
>>>>>> On Apr 29, 2015, at 8:34 PM, "Graham, Mathew Thomas" <[log in to unmask]> wrote:
>>>>>>
>>>>>>> That would be awesome if you have time?
>>>>>>>
>>>>>>>> On Apr 29, 2015, at 5:28 PM, Sho Uemura <[log in to unmask]> wrote:
>>>>>>>>
>>>>>>>> For now I would be happy to have a script that I can run by hand to fire off an Auger job.
>>>>>>>>
>>>>>>>> e.g. "runCurrentRecon XXXX Y 2mm" to do the recon on file hps_00XXXX.evio.Y and put the output in volatile.
>>>>>>>>
>>>>>>>> I might do it tonight if I have time.
>>>>>>>>
>>>>>>>> On Wed, 29 Apr 2015, Nathan Baltzell wrote:
>>>>>>>>
>>>>>>>>> If we want to process data with all 1-6 layers tracking, we will need it in the db
>>>>>>>>> such that hps-java knows which detector to use based on the run number.
>>>>>>>>> Else someone has to tune the batch farm scripts by run number according to
>>>>>>>>> SVT position. Another option is to do only 4-6 recon.
>>>>>>>>>
>>>>>>>>> Let me know what you think.
>>>>>>>>>
>>>>>>>>> -Nathan
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On Apr 28, 2015, at 5:34 PM, Nathan Baltzell <[log in to unmask]> wrote:
>>>>>>>>>
>>>>>>>>>> If "detector" name needs to change by run, we need it in database.
>>>>>>>>>>
>>>>>>>>>> On Apr 28, 2015, at 4:29 PM, "Graham, Mathew Thomas" <[log in to unmask]> wrote:
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> I?m going go ahead and say it?s good to go:
>>>>>>>>>>>
>>>>>>>>>>> org.hps.steering.recon.EngineeringRun2015FullRecon.lcsim
>>>>>>>>>>> HPS-EngRun2015-XXX-v1 detectors
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>> On Apr 28, 2015, at 11:46 AM, Graham, Mathew Thomas <[log in to unmask]> wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Sho has been putting one together?Sho, do you think it?s ready for prime-time?
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>> On Apr 28, 2015, at 11:44 AM, Nathan Baltzell <[log in to unmask]> wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>> Hi Matt,
>>>>>>>>>>>>>
>>>>>>>>>>>>> It is probably time to process a bunch of new data.
>>>>>>>>>>>>> Do we have a steering and jar file with SVT ready to go
>>>>>>>>>>>>> that I should use?
>>>>>>>>>>>>>
>>>>>>>>>>>>> -Nathan
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>>
>
> ########################################################################
> 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
########################################################################
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
|
