Performance issue when indexing the data set of 'double' key type

[baotonglu]

Hi Xinjing,

When I tested the DPTree using the data sets of the 'double' type, I observe that the performance of DPTree is extremely low and even has the segmentation fault. Probably it is because you assume that key-type is 'uint64_t' when writing the code.

The data sets I tested are longlat and longitudes. The method to read the keys from the file is here.

Could you take a look how to fix it?

Best, Baotong