Helps solve closest pair problems with generic interparticle and particle-beam distances. More...

#include <fastjet/NNBase.hh>

Inheritance diagram for fastjet::NNBase< I >:

[legend]

Collaboration diagram for fastjet::NNBase< I >:

Public Member Functions
	NNBase ()
	Default constructor.

	NNBase (I *info)
	Constuctor with additional Info.

virtual void	start (const std::vector< PseudoJet > &jets)=0
	initialisation from a given list of particles

virtual double	dij_min (int &iA, int &iB)=0
	returns the dij_min and indices iA, iB, for the corresponding jets. More...

virtual void	remove_jet (int iA)=0
	removes the jet pointed to by index iA

virtual void	merge_jets (int iA, int iB, const PseudoJet &jet, int jet_index)=0
	merges the jets pointed to by indices A and B and replaces them with jet, assigning it an index jet_index. More...

Public Member Functions inherited from fastjet::NNInfo< I >
	NNInfo (I *info)

template<class BJ >
void	init_jet (BJ *briefjet, const fastjet::PseudoJet &jet, int index)

Detailed Description

template<class I = _NoInfo>
class fastjet::NNBase< I >

Helps solve closest pair problems with generic interparticle and particle-beam distances.

Description and derived classes:

This is an abstract base class which defines the interface for several classes that help carry out nearest-neighbour clustering:

NNH provides an implementation for generic measures,
NNFJN2Plain provides an implementation for distances satisfying the FastJet lemma i.e. distances for which the minimum dij has the property that i is the geometrical nearest neighbour of j, or vice versa. I.e. the distance can be factorised in a momentum factor and a geometric piece. This is based on the fastjet N2Plain clustering strategy
NNFJN2Tiled is a tiled version of NNFJN2Plain (based on the N2Tiled FastJet clustering strategy). Like NNPlain2 it applies to distance measures that satisfy the FastJet lemma, with the additional restriction that: (a) the underlying geometry should be cylindrical (e.g. rapidity–azimuth) and (b) the search for the geometric nearest neighbour of each particle can be limited to that particle's tile and its neighbouring tiles.

If you can use NNFJN2Plain it will usually be faster than NNH. NNFJN2Tiled, where it can be used, will be faster for multiplicities above a few tens of particles.

NOTE: IN ALL CASES, THE DISTANCE MUST BE SYMMETRIC (dij=dji)!!!

Underlying BriefJet (BJ) class:

All derived classes must be templated with a BriefJet (BJ) class — BJ should basically cache the minimal amount of information that is needed to efficiently calculate interparticle distances and particle-beam distances.

This class can be used with or without an extra "Information" template, i.e. NN*<BJ> or NN*<BJ,I>. Accordingly BJ must provide one of the two following init functions:

void BJ::init(const PseudoJet & jet); // initialise with a PseudoJet

void BJ::init(const PseudoJet & jet, I * info); // initialise with a PseudoJet + info

where info might be a pointer to a class that contains, e.g., information about R, or other parameters of the jet algorithm

The BJ then provides information about interparticle and particle-beam distances. The exact requirements depend on whether you use NNH, NNFJN2Plain or NNFJN2Tiled. (See the corresponding classes for details).

Workflow:

In all cases, the usage of NNBase classes works as follows:

First, from the list of particles, create an NN*<BJ> object of the appropriate type with the appropriate BJ class (and optional extra info).

Then, cluster using a loop like this (assuming a FastJet plugin)

while (njets > 0) {
  int i, j, k;
  // get the i and j that minimize the distance
  double dij = nn.dij_min(i, j);  
  // do the appropriate recombination and update the nn
  if (j >= 0) {    // interparticle recombination
    cs.plugin_record_ij_recombination(i, j, dij, k);
    nn.merge_jets(i, j, cs.jets()[k], k); 
  } else {         // bbeam recombination
    double diB = cs.jets()[i].E()*cs.jets()[i].E(); // get new diB
    cs.plugin_record_iB_recombination(i, diB);
    nn.remove_jet(i);
  }
  njets--;
}

For an example of how the NNH<BJ> class is used, see the JadePlugin or EECambridgePlugin.

Definition at line 164 of file NNBase.hh.

Member Function Documentation

◆ dij_min()

template<class I = _NoInfo>

virtual double fastjet::NNBase< I >::dij_min	(	int &	iA,
		int &	iB
	)

pure virtual

returns the dij_min and indices iA, iB, for the corresponding jets.

If iB < 0 then iA recombines with the beam

Implemented in fastjet::NNFJN2Tiled< BJ, I >, fastjet::NNFJN2Plain< BJ, I >, and fastjet::NNH< BJ, I >.

◆ merge_jets()