How SQL Server sampling data for constructing statistics such as histograms a histograms?

Question

I see the statement in a paper that "If ∆obs ≤ ∆req we are done, otherwise the required block-level sample size rblk can be derived from Theorem 1 to be

my q