Speeding up spatial analyses by integrating `sf` and `data.table`: a test case

The problem Last week, I replied to this interesting question posted by @Tim_K over stackoverflow. He was seeking efficient solutions to identify all points falling within a maximum distance of xx meters with respect to each single point in a spatial points dataset. If you have a look at the thread, you will see that a simple solution based on creating a “buffered” polygon dataset beforehand and then intersecting it with the original points is quite fast for “reasonably sized” datasets, thanks to sf spatial indexing capabilities which reduce the number of the required comparisons to be done (See http://r-spatial.