Phases switches in local ancestry inference (LAMP-LD vs RFMix)

Question

[Mr Locuace]

Hello. I am working on the revision of a manuscript were we used LAMP-LD software to infer local ancestry of a SNP dataset. One reviewer made the following comment:

"The local ancestry inference program LAMP-LD does not correct for phases switches, which can substantially lower confidence in local ancestry calls for heterozygous sites. They should consider carefully phasing the data and/or using the local ancestry inference program RF-Mix, which does correct for phasing switches"

I am not sure what does the reviewer mean by "phasing switches". Any explanation will be very welcome. Thanks a lot.