Particle container broken in certain cases

[marek]

There is an issue when using the custom particle containers. The following run card

(run){
  SCALES VAR{sqr(91.2)}
  MI_HANDLER None; FRAGMENTATION Off;

  MASSIVE[5] 1;
  PARTICLE_CONTAINER 95 bquarks 5 -5;
}(run)

(beam){
  BEAM_1 2212; BEAM_ENERGY_1 3500;
  BEAM_2 2212; BEAM_ENERGY_2 3500;
}(beam)

(processes){
  Process 93 93 -> 11 -11 95 95
  Order_EW 2;
  Print_Graphs Graphs;
  End process;
}(processes)

(selector){
  Mass 11 -11 60 120
}(selector)

produces plenty of nan's for Comix and a wrong result for Amegic (wrong but converging central value, constant error estimate). Working with 93 93 -> 11 -11 5 -5 explicitly works just fine and gives consistent results. The problem is identical in rel-2-0-beta2 and trunk. The list of processes built up appears to be fine. The particle container works fine for 93 93 -> 95 95, 93 93 -> 23 95 95, 11 -11 -> 95 95, 11 -11 -> 95 95 93, 11 -11 -> 95 95 95 95. Any ideas?

[anonymous]

The issue was quickly detected with Frank's help: The orderings of the particle container and the underlying processes differed, mostly because the properties of the container all default to zero. Possible solutions are: a) use PRIORITY[5]=1 to manually interfere with the ordering b) manually set the relevant properties of the container "PARTICLE_CONTAINER 95[Q:1] bquarks 5 -5"

The latter should be preferable. Generally set particle container properties on the basis of their content should prove difficult for general cases.