Flavor sums: phase space insufficient

[kilian]

When doing some real life LHC process like

alias q = d:u:s:c
alias Q = D:U:S:C
alias parton = g:d:u:s:c:D:U:S:C

process ttbar = parton, parton => b, bbar, e1, N1, q, Q

WHIZARD fails to adapt the grids to a useful result, for two reasons at least:

• the default phase space channels seem to be insufficient (I'll check this, probably off_shell has to be increased or similar)
• O'Mega reorders all particles, including the leading b, bbar, so that the assignment of bincodes in the phase space file is screwed up.

I'll try to adapt the WHIZARD setup to this situation, which should be quite common. But comments or suggestions are welcome.

[kilian]

Oops, no, I was mistaken, it's probably just reason #2. The reordering is confusing, but not the poblem.

[ohl]

Replying to kilian:

The reordering is confusing

It's the easiest way to remove duplicate final states.

If necessary, I could invest the extra work to maintain the order given on the command line (I do so whenever $restrictions are in play), but you would still need to consult FLAVOR_STATES to find out which representative of an equivalence class was chosen if there were equivalent final states.

[kilian]

Update: the signal diagrams (t-tbar in this case) are missing from the phase space setup, but only if I include the c and s quark in the flavor summation. Very strange ...

[kilian]

Found the cause: only one channel is actualy used, this is due to an error in hashing the cascades. Simply fixing this is not enough, since then the execution time blows up. I'll try a new approach with doing the channels separately and merging them only at the end.

[kilian]

Revised phase-space setup, in particular for flavor sums [2508]. Flavor sums should work now, plain processes should also benefit from the elimination of redundancies. I hope there are no regressions.