Difference: EmbedParticles (1 vs. 7)

jmohep(i)=mod(isthep(i),100)*4000*4000 + jmohep(1,i)*4000+jmohep(2,i) jdahep(i)=mod(isthep(j)/100,100)*4000*4000+jdahep(1,i)*4000+jdahep(2,i)

Modify (extend) an existing ntuple

TODO: there has to be a more elegant solution than just write out to a second ntuple. Maybe hropen(... 'U' ...) ?

1. Create an ntuple with the HROPEN procedure (say test.ntpl), and define its structure with HBNAME. Note: ntuple's ID must be 100 (unpacked) or 101 (packed; a bit tricky). Additonal note: Ntuple variable names have to be in capital letters. Don't ask why...
2. Do a loop over the events, generate some particles, etc; at the end of the loop call hfnt(ID)

1. Create an ntuple with the HROPEN procedure (say test.ntpl), and define its structure with HBNAME. Note: ntuple's ID must be 100 (unpacked) or 101 (packed; a bit tricky). Additonal note: Ntuple variable names has to be in capital letters. Don't ask why...

The HEPEVT ntuple structure

Block HEPEVT:
   integer  NEVHEP = event number
   integer  NHEP   = number of entries (particles, partons)
   integer ISTHEP(NHEP) = status code
   integer  IDHEP(NHEP)  = PDG identifier
   integer  JMOHEP(2,NHEP) = position of 1st and 2nd  mother
   integer  JDAHEP(2,NHEP) = position of 1st and last daughter
   real PHEP(5,NHEP)   = 4-momentum and mass            (single precision in ntuple file)
   real VHEP(4,NHEP)   = vertex xyz and production time (single precision in ntuple file)

Block MC_PARAM:
   integer    IRNMCP       = run number
   integer    IEVMCP         = event number
   real  WGTMCP         = event weight
   real  XSECN          = cross section equivalent
   integer    IFILTER        = filter pattern
   integer    NVRMCP         = number of additional variables
   real VARMCP(NMXMCP) = list of additional variables

Writing ntuples

1. Define the data blocks as Fortran COMMON's.
2. Create an ntuple with the HROPEN procedure (say test.ntpl), and define its structure with HBNAME. Note: ntuple's ID must be 100 (unpacked) or 101 (packed; a bit tricky)
3. Do a loop over the events, generate some particles, etc; at the end of the loop call hfnt(ID)
4. Close the ntuple.

Compile the program with g77 -c test.f and link with g77 -o test.run test.o `cernlib packlib,mathlib` This requires a running g77 system with cernlib properly installed (on debian: apt-get install cernlib)

Run the executable.

Now the file test.ntpl should contain the particles. You can test it by opening with PAW:

h/file 0 test.ntpl 4096
n/print 100
n/plot 100.idhep

TODO: this example program does not call Pythia to decay instable particles, but who cares anyway? We want now to control everything manually.

Feed into OSCAR

Copy or symlink test.ntpl into the OSCAR working directory (in my case, it's OSCAR_3_9_9/src/Workspace).

Edit oscarrc:

...
OutputDataSet = /System/sim/test
EventNtplReader:NtplFileName = test.ntpl
EventNtplReader:NtplID = 100
...

Do the ritual eval `scram run -sh` (or -csh), and run oscar -c oscarrc

This will produce a lot of sim and xml files (see ProduceSimulatedEvents)

Further readings

CMKIN

HBOOK Reference

-- GergelyPatay - 08 Mar 2006