I am trying to come up with an algorithm that takes N 32 bit integers and computes a set of N pseudo random floating point values from 0 to 1 by scrambling the bits of the integers.
I am trying to search this up but all I get are hashing algorithms for cryptography or for data structures like hashtables.
The main challenge I am facing atm is that merely casting a large bit sequence into floats is very unpredictable so trying to get a normalized floating point value is proving tircky.
The final use case is to have a consistent way to generate a random vector for a given set of integer coordinates.
i.e. a mathematical map from a tuple of natural numbers N^n to a tuple of floating point coordinates sampled from the unit hyper cube. I am using this to generate "random" vector fields in a grid.
Aucun commentaire:
Enregistrer un commentaire