Separation of two overlapped electromagnetic or electromagnetic-hadronic showers in CALICE highly granular physics calorimeter prototypes using Pandora, Garlic and Arbor Particle Flow Algorithms

CALICE collaboration is developing highly granular calorimetry for the future ILD detector. Usage of Particle Flow Algorithm (PFA) for individual reconstruction of particles in jets provides the best jet energy resolution. For high jet energies (typically above 70 - 100 GeV), particles start overlapping in the calorimeter, causing confusion, and recognition of individual particles becomes complicated. This is the natural limit for PFA at high energies. In this work we are presenting the results on the separation of - 2 overlapped electromagnetic showers in the CALICE silicon-tungsten electromagnetic calorimeter (SiW ECAL) physics prototype; - electromagnetic-hadron showers in the physics prototypes of SiW ECAL and the analogue scintillator-steel hadronic calorimeter (AHCAL); - photon-photon and photon-hadron showers in ILD Monte-Carlo (MC) simulation with $5 \times 5$ mm$^2$ and $2.5 \times 2.5$ mm$^2$ ECAL cell size for ILD baseline models with AHCAL or semi-digital hadronic calorimeter (SDHCAL). Physics prototype data were collected during beam tests (TB) in CERN (2007) and FermiLab (2011). Results are compared with Monte-Carlo simulations. Three different reconstruction algorithms are used:Pandora, Garlic and Arbor.