On 09/22/2011 05:53 PM, Gero Flucke wrote: > On Tue, 20 Sep 2011, Gyoergy Vesztergombi wrote: > >> >> Dear Gero, >> >> Sorry for the long silence, but I just arrived back to work. >> I shall contact my people immediately. I hope by Thursday we shall be >> ready to move. > > Hi Gyoergy, > I was lucky and caught Joerg whose other meeting was just finished - he > will write you a few lines how to setup millepede for the pixel > alignment, based on what our DESY summer student did. > Many thanks to Joerg (whom I put in CC)! Dear all, here comes now a short summary of what you have to do to perform the pixel alignment. A few weeks ago I have prepared a setup for our summer student here at DESY which is quite similar to the setup you will need for the pixel alignment. You can use that setup as a basis ( directory mp0789 in the common MP area). These things you have to do: - Adjust setup_align.pl. The variable my $mssDir has to be changed in order to be usable with EOS. Change it to something like my $mssDir = "/store/caf/user/jbehr/MPproduction/mp0999"; Additionally you have to comment out #system "nsrm -rf $mssDir"; #system "nsmkdir $mssDir"; and out instead something like my @rmfiles = `cmsLs -l $mssDir`; foreach my $i (@rmfiles) { if($i =~ /\/store\//) { my @k = split " ", $i; system "cmsRm $k[4]"; } } system "cmsRmdir $mssDir"; system "cmsMkdir $mssDir"; in the file. If somebody knows how to recursively remove files or how to delete directories WITH the content in the directory using the NEW commands, PLEASE let me know!!!! cmsRmdir -rf bla seems not to work anymore if files are in the directory bla. - Adjust in setup_align.pl the input data sample: 'data' => "${datasetdir}/SingleMu_Run2011A_TkAlMuonIsolated_v4_169810_to_172255.txt", You have to prepare a text file with a list of root files. Examples can be found in /afs/cern.ch/cms/CAF/CMSALCA/ALCA_TRACKERALIGN/MP/MPproduction/datasetfiles/2011alignment. I usually just DBS to prepare these files. A possible syntax would be dbs search --query='find file where dataset=/DoubleMu/Run2011A-TkAlZMuMu-v1/ALCARECO and file.numevents>0' | grep -i store > bla.txt Certainly you have to adjust the name of the dataset file. The output is stored in bla.txt. The first line of this text file has to be changed to CastorPool=cmscaf. - Change the global tag: 'globaltag' => 'GR_R_42_V14::All', Probably it is 'globaltag' => 'GR_P_V20::All', but this you should verify. - You need a JSON file. Here you find them: file:///afs/cern.ch/cms/CAF/CMSCOMM/COMM_DQM/certification/Collisions11/7TeV/Prompt/ Chose one appropriate file with _JSON.txt in the name. The format has to be converted so that it is usable in the alignment setup. You can find a script in my public folder: /afs/cern.ch/user/j/jbehr/public/convertlumis.pl In the current configuration the script will print out only even run numbers. If you want to change it, adjust this variable: my $selectruns = "even"; #odd even chose "" for all runs Lets say your downloaded JSON file is named bla_json.txt you can then convert this with printJSON.py bla_json.txt | perl -n ~/bin/convertlumis.pl Copy the output to a new text file and edit the file until the syntax is similar to the file /afs/cern.ch/cms/CAF/CMSALCA/ALCA_TRACKERALIGN/MP/MPproduction/mp0789/new_muon_json_evenrunnr.txt Change the line 'json' => "$ENV{'PWD'}/new_muon_json_evenrunnr.txt", in setup_align.pl so that it will point to the new json file you just have created. - You have to change the starting geometry. Open startgeometry.txt and change tag = cms.string('Alignments_3') to tag = cms.string('Alignments_8') - The job in mp0789 is configured to use isolated muons. If you want to use other data sets you might have to use a different .py file (and of course changes have to be applied in setup_align.pl accordingly). - One important issue: You have to configure the IOV boundaries in alignables.txt . Add something like process.AlignmentProducer.RunRangeSelection = [ cms.PSet(RunRanges = cms.vstring('-1:158383', '158384:161406', '161407:162974', '163078:164060', '164162:165408', '165415:165999', '166004:166396', '166399:167029', '167033:-1' ), selector = cms.vstring('TrackerTPBLayer,111111', 'TrackerTPEHalfDisk,111111') ), ] to alignables.txt whereas of course the numbers have to be adjusted. Hope thats all and I have not forgotten anything ... Cheers, Joerg > > But you will also have to run offline validation, as has been done for > the sign-off, so you are probably aware of how to do that - you could > already start to look for the relevant files, i.e. data after beginniung > of July. For datasets see > https://twiki.cern.ch/twiki/bin/viewauth/CMS/TrackerAlignment2011#Lists_of_2011_Collision_Datasets > > > Cheers > > Gero >> >> Cheers >> >> gv >> On Mon, 19 Sep 2011, Gero Flucke wrote: >> >>> Dear Hungarian aligners, >>> you are probably aware that the pixel detector seems to have slightly >>> moved at least twice since end of June (day 180), i.e. the last data >>> included in our currently signed-off alignment, see at lower right >>> plot of http://musich.web.cern.ch/musich/CronPVValidation/ >>> >>> Since reprocessing is not yet scheduled, there is time to correct for >>> that, i.e. to run alignment of the pixel barrel layers and forward >>> half discs keeping the strip detecctor fixed. >>> >>> Would your group be able to take the responsibility of this task, >>> i.e. providing this alignment and its basic validation within the >>> next two weeks or so? >>> >>> Please answer soon so that we can exchange about details. >>> >>> Cheers >>> >>> Gero >>> >>> -- >>> ----------------------------------------------------------------------- >>> Gero Flucke >>> - Analysis Centre, Helmholtz Alliance "Physics at the Terascale" >>> * Statistics Tools >>> - CMS: Tracker Alignment Convenor >>> DESY/CMS, Notkestr. 85, D-22607 Hamburg, Germany >>> Bldg. 1e, Rm. 02.501 >>> phone: +49 (0)40 8998 3525 >>> fax: +49 (0)40 8998 3092 >>> >> >