Intermediate Boson missing from events in LHE output file

Typical question:

I'm using MadGraph on the web to generate the process e+ e- > a > mu+ mu- (The same occur for e+ e- > Z > mu+ mu-).
The problem is that the events in the output LHE file do not contain statistics for the photon. How can I get the intermediate photon to show up in the event information as well?

A sample events looks like this:

<event>
 4 0 0.4975400E-04 0.1183216E+03 0.7957747E-01 0.1139134E+00
      -11 -1 0 0 0 0 0.00000000000E+00 0.00000000000E+00 0.70000000000E+04 0.70000000000E+04 0.00000000000E+00 0. 1.
       11 -1 0 0 0 0 0.00000000000E+00 0.00000000000E+00 -0.70000000000E+04 0.70000000000E+04 0.00000000000E+00 0. -1.
      -13 1 1 2 0 0 -0.67882574179E+04 -0.42589375412E+03 -0.16547434057E+04 0.70000000000E+04 0.00000000000E+00 0. 1.
       13 1 1 2 0 0 0.67882574179E+04 0.42589375412E+03 0.16547434057E+04 0.70000000000E+04 0.00000000000E+00 0. -1.
</event>

I feel like what should have happened is that the photon shows up as an intermediary in the LHE file.

*******************************************************************************************
** ANSWER **
*******************************************************************************************

In madgraph, we wrote in the lhe file only particles which are ONSHELL.
The reason is that for process like e+ e- > mu+ mu-, they are no physical meaning in the intermediate particles, if the
intermediate particle is off-shell. Indeed the interference between the two diagrams are not negligible outside the Z peak.

The definition of being on-shell (and therefore being written in the lhe file) depend of the "bw_cut" parameter of the run_card.
All particles with invariant mass between M - bw_cut * Width and M + bw_cut * Width are consider as on-shell and therefore written in the lhe file.