I have a series of points in lat-long space, ranging from having their nearest neighbour very close to 1000s of km away. Say I do a geo-hash of these to a high precision, my search of nearest neighbouring squares must be of a lower or equal precision.
How can I find the closest point/s efficiently when I have no guarantee if there will be any neighbouring hashes given the precision of my search grid?
If I set the grid too high I include too many, If I set it too low I don't find any. My set can contain both these scenarios simultaneously depending on which point you look at.
Is there a preferred way to find the closest point/s when we don't know if the closest point is right next to the other or 1000s of km away?