It is the motivation for the subroot labeling in #5 . It is proposed as a progress measure (runtime prediction method), not tree size estimation. In the context of tree size estimation, this progress measure is closely related to the weight sum of the encountered leaf nodes.
The original publication suggests to partition the open node frontier every time a new incumbent solution is found. A periodic repartitioning is discussed, but not implemented.
The subset sum gap method has been introduced in Ozaltin et al. 2011
It is the motivation for the subroot labeling in #5 . It is proposed as a progress measure (runtime prediction method), not tree size estimation. In the context of tree size estimation, this progress measure is closely related to the weight sum of the encountered leaf nodes.
The original publication suggests to partition the open node frontier every time a new incumbent solution is found. A periodic repartitioning is discussed, but not implemented.