With a moderately large data and unspecified twindow, tmpInd will have countless rows, and apply on it is literally time-consuming. I adapt it for a vectorized computation and the efficiency is much improved.
data[indSp[,1],1] extracts only one column. For a data.frame, the default is to drop if only one column is left, and the result reduces to a vector. However, for a tibble, it's not the default. data[indSp[,1],1] is still a tibble, and hence passing a tibble into mean returns NA. I add drop=TRUE to coerce tibble to drop.
https://github.com/r-spatial/gstat/blob/8335d8e2fb1a16b02d256148d502a7edcff9142a/R/variogramST.R#L234
With a moderately large data and unspecified
twindow,tmpIndwill have countless rows, andapplyon it is literally time-consuming. I adapt it for a vectorized computation and the efficiency is much improved.https://github.com/r-spatial/gstat/blob/8335d8e2fb1a16b02d256148d502a7edcff9142a/R/variogramST.R#L255-L256
data[indSp[,1],1]extracts only one column. For adata.frame, the default is to drop if only one column is left, and the result reduces to a vector. However, for atibble, it's not the default.data[indSp[,1],1]is still atibble, and hence passing atibbleintomeanreturns NA. I adddrop=TRUEto coercetibbleto drop.