The principle of switch-error correction algorithm

[charliechen912ilovbash]

Hi,
I am dealing with somatic SV phasing, and I am very interested in the switch-error correction algorithm in phase.py. Could you please briefly summarize the method of this algorithm? It looks like first searching the SNV with copy number >2 or <0.5, record it, then switch the phase at these points, to keep the same phase within the same copy number region?
Thank you very much.