I am using an unordered_map>float, unsigned short> to implement a hash table in C++.
I know that using floats as keys to a hash table is a BAD idea under most circumstances because comparing them is error prone. However, under these circumstances I am reading the floats in from large files and their precision is known and constant.
However, I would like to know the details of how unordered_map is hashing my floats in order to estimate collision frequency. I am not overriding the default hash implementation when I create the unordered map. According to the documentation, the default hash function is std::hash>Key>. Which in my case is std::hash>float>. However, when I look at the std::hash documentation, it is only defined for only defined for "template arguments of type char*, const char*, crope, wrope, and the built-in integral types".
Does anyone know what function is being called to hash the values as I add them to the unordered_map?
unordered_map - http://msdn.microsoft.com/en-us/library/bb982522.aspx
std::hash - http://www.sgi.com/tech/stl/hash.html#1