Hacker News new | ask | show | jobs
by machiavelli1024 2873 days ago
Notice how the rejection method becomes useless for higher dimensions.

https://en.m.wikipedia.org/wiki/Curse_of_dimensionality

1 comments

While that's interesting, I'm having a complete lack of imagination fail on why one would be trying to solve this problem for higher dimensions, other than academic pursuits. Could you give a practical example?
“Sampling from the uniform distribution on the N-dimensional Euclidean ball (the N-ball) and its surface (the N-sphere) is a tool useful in many diverse research fields, for instance Monte Carlo integration in physics, generating random directions for MCMC sampling on bounded regions, generating random correlation matrices, Monte Carlo analysis of random packing on the sphere, generating random rotations in cryptography as well as various simulation studies in statistics”

https://www.sciencedirect.com/science/article/pii/S0047259X1...