An atennna-based jet clustering algorithm is under development. So far, only a set of elementary 3→2 clustering maps are provided.
The VinClu clustering map is a 3->2 clustering
map rather similar to that of the ARCLUS jet clustering algorithm.
Its principal advantage is that it
contains the exact inverses of the Vincia kinematics mappings, and
hence it can be used to "undo" branchings and/or reconstruct the exact
intermediate shower states from a given momentum configuration. The
clustering is invoked by creating an instance of the class
VinClu and using either of the two methods:
iMap is the
index number of the kinematics map
to be used (the numbering
follows kineMapType), pIn is a vector of
arbitrary size containing the input momentum configuration (in
PYTHIA 8's Vec4 format),
and the indices a, r, b
represent the 3 particles to be clustered in the backwards
reconstruction of AB->arb. For the variant with massive
momenta, mA and mB should be used to
specify the on-shell masses of A and B.
Technically, the clustering is done as follows.
pClu is initially
set equal to pIn.
The reconstructed momenta of the
parents, A
and B, are then stored in pClu[a]
and pClu[b], respectively, and
finally pClu[r] is erased. (Note that this last step
may cause the indices of A and/or B
in pClu to be different from
those of a and b in pIn. For the most
common case, a
colour-ordered clustering with r=a+1 and b=a+2,
the indices of A and B after the clustering will
be A=a and B=A+1.)