Problem Description

AtlfastTemp has the concept of cells (rectangles in $\eta -\phi$ space representing calorimeter cells), and clusters (the struck cells lying in a cone centered on a high cell).

The isolation of a particle is currently calculated using both clusters and cells.

The AtlfastTemp algorithm is as follows (abbreviated description, from Elzbieta.)

• find kine-electron in the list of generated particles
• check kinematic thresholds (in fact I also apply smearing before so you might find some residuals from here)
• choose the closest cluster to the kine-axis
• if the closest cluster within given cone it is marked as associated
• check isolation criteria from all OTHER clusters
• check isolation criteria from all cells within given cone after subtracting energy of kine-particle (in fact it subtract already smeared energy so you mind find some residuals from here)
• if isolation passed remove marked cluster from cluster list

Clusters are used in the isolation calculation if they are not associated to a particle (motivation: you do not want an electron to be marked as non-isolated by its own cluster, or the cluster due to another electron or gamma).

Clusters are assigned to the particles as the particles are being formed. They are then removed from the list of clusters available for the isolation calculation.

The clusters participating in the calculation of the isolation of a given particle is thus dependent on the order in which the particles are processed.