Skip to content

Backward compatibility broken

Hi,

When reading older HepMC::Version 2.06.09 files, one get different event structures back.

I attach a number of files to illustrate this

  • testHepMC.cc is a small program that either uses HepMC3 compatibility layer or HepMC2 directly. This should be compiled with

    g++ tests/testHepMC.cc \
 `HepMC3-config --cflags --libs` -lHepMC \
 -Wl,-rpath,`HepMC3-config --libdir`  -o testHepMC

    or equivalent. The program accepts a few options:

    $ ./testHepMC -h
  -d ARG	Debug level (only HepMC3)               	0
  -h    	This help                               	1
  -i ARG	Input file                              	event1.hepmc
  -n ARG	Number of events (<0 means all)         	-1
  -v ARG	HepMC version (2 or 3)                  	2

  • event1.hepmc A single EPOS-LHC event generated via CRMC.

  • event2.hepmc A single Pythia(8) event

  • event1.out3, event1.out2, event2.out2, and event2.out3 are outputs of the above testHepMC program (see below).

Both data files have version HepMC::Version 2.06.09.

Reading event1.hepmc with HepMC2 produces a single event with 120 particles.

$ ./testHepMC -v 2 -i event1.hepmc | tee event1.out2
________________________________________________________________________
GenEvent: #1
 Momentum units: GEV Position units: CM
 Entries in this event: 1 vertices, 120 particles, 0 weights.
...

However, reading the same file with the HepMC3 compatibility layer yields and empty event

$ ./testHepMC -v 3 -i event1.hepmc | tee event1.out3
________________________________________________________________________
GenEvent: #1
 Momentum units: GEV Position units: CM
 Entries in this event: 0 vertices, 0 particles, 0 weights.

The reason being, that the compatibility layer prunes the event for particles that are not attached to a production vertex. Since CRMC didn't set this information, we get back an empty event.

Consider the Pythia event in event2.hepmc

$ ./testHepMC -v 2 -i event2.hepmc | tee event2.out2
________________________________________________________________________
GenEvent: #1
 Momentum units: GEV Position units: CM
 Entries in this event: 90 vertices, 246 particles, 0 weights.

$ ./testHepMC -v 3 -i event2.hepmc  | tee event2.out3
________________________________________________________________________
GenEvent: #1
 Momentum units: GEV Position units: CM
 Entries in this event: 89 vertices, 246 particles, 0 weights.

As is evident from the above, HepMC2 gives 90 vertices, while the compatibility layer of HepMC3 gives only 89. A closer inspection of the output likewise reveals more differences. For example, the first three verteces read by HepMC2 are

Vtx:     -1 stat:   0 (X,cT): -0.0071 0.0024 0 0
 O:       1     2212 +0.00e+00,+0.00e+00,+2.51e+03,+2.51e+03   4    -1
          2     2212 +0.00e+00,+0.00e+00,-2.51e+03,+2.51e+03   4    -1
          7       21 -1.30e+00,-4.89e-01,-2.37e+00,+2.74e+00  21    -1
          8        1 +1.14e+00,+2.44e+00,-7.69e+01,+7.70e+01  21    -1
         16        1 -2.63e-01,+5.46e-02,-1.67e+01,+1.67e+01  12    -1
         17       -1 +2.63e-01,-5.46e-02,+8.68e+00,+8.69e+00  11    -1
         24       21 -1.76e-01,-1.52e-01,-1.70e+02,+1.70e+02  12    -1
         25       21 +1.76e-01,+1.52e-01,+3.87e+00,+3.87e+00  11    -1
Vtx:     -2 stat:   0 (X,cT): -0.0071 0.0024 0 0
 I:       1     2212 +0.00e+00,+0.00e+00,+2.51e+03,+2.51e+03   4    -1
 O:       3       21 +2.67e-01,+9.02e-01,+7.94e+02,+7.94e+02  21    -2
         15     2101 -3.82e-01,-3.01e-01,+1.34e+03,+1.34e+03  11    -2
         18        2 +1.15e-01,-6.01e-01,+3.69e+02,+3.69e+02  12    -2
Vtx:     -3 stat:   0 (X,cT): -0.0071 0.0024 0 0
 I:       2     2212 +0.00e+00,+0.00e+00,-2.51e+03,+2.51e+03   4    -1
 O:       4        1 -2.42e-01,+7.68e-01,-8.83e+01,+8.83e+01  21    -3
         23     2203 +2.42e-01,-7.68e-01,-2.24e+03,+2.24e+03  11    -3

while the equivalent from the HepMC3 reader are the 2 first vertices

Vtx:     -1 stat:   0 (X,cT): -0.0071 0.0024 0 0
 I:       1     2212 +0.00e+00,+0.00e+00,+2.51e+03,+2.51e+03   4     0
 O:       2       21 +2.67e-01,+9.02e-01,+7.94e+02,+7.94e+02  21    -1
          3     2101 -3.82e-01,-3.01e-01,+1.34e+03,+1.34e+03  11    -1
          4        2 +1.15e-01,-6.01e-01,+3.69e+02,+3.69e+02  12    -1
Vtx:     -2 stat:   0 (X,cT): -0.0071 0.0024 0 0
 I:       5     2212 +0.00e+00,+0.00e+00,-2.51e+03,+2.51e+03   4     0
 O:       6        1 -2.42e-01,+7.68e-01,-8.83e+01,+8.83e+01  21    -2
          7     2203 +2.42e-01,-7.68e-01,-2.24e+03,+2.24e+03  11    -2

In both cases (event1.hepmc and event2.hepmc) the output of the compatibility layer are not consistent with the old HepMC2 output. In the latter case (event2.hepmc) the output of HepMC3 seems more correct, while in the first case (event1.hepmc) it seems that HepMC2 is correct.

Thanks.

Yours,

Christian