The BinarySearch is strange I don't understand what it is doing, see example below, where the output is the same as the max(abs(x)) value
set.seed(45)
x <- rnorm(15702)
RGCCA:::BinarySearch(x, 1)
#> [1] 4.112402
The effect seems to take lower values as the sumabs increase:

Created on 2018-11-07 by the reprex package (v0.2.1)