cessen/psychopath
1
0
Fork 0
Code Issues 1 Pull Requests Projects Releases Wiki Activity

Removed unsafe code from sobol sampler and improved its documentation.

Browse Source
This commit is contained in:
Nathan Vegdahl 2020-03-11 21:30:14 +09:00
parent b081424ba6
commit db9efc6a55
1 changed files with 16 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
sub_crates/sobol/src/lib.rs
View File

@ -38,7 +38,9 @@ pub fn sample(dimension: u32, index: u32) -> f32 {
/// scramble parameter. /// scramble parameter.
/// ///
/// To get proper random digit scrambling, you need to use a different scramble /// To get proper random digit scrambling, you need to use a different scramble
/// value for each dimension. /// value for each dimension, and those values should be generated more-or-less
/// randomly. For example, using a 32-bit hash of the dimension parameter
/// works well.
#[inline] #[inline]
pub fn sample_rd_scramble(dimension: u32, index: u32, scramble: u32) -> f32 { pub fn sample_rd_scramble(dimension: u32, index: u32, scramble: u32) -> f32 {
u32_to_0_1_f32(sample_u32(dimension, index) ^ scramble) u32_to_0_1_f32(sample_u32(dimension, index) ^ scramble)
@ -47,7 +49,7 @@ pub fn sample_rd_scramble(dimension: u32, index: u32, scramble: u32) -> f32 {
/// Same as `sample()` except applies Owen scrambling using the given seed. /// Same as `sample()` except applies Owen scrambling using the given seed.
/// ///
/// To get proper Owen scrambling, you need to use a different seed for each /// To get proper Owen scrambling, you need to use a different seed for each
/// dimension. /// dimension. For example, reusing the dimension parameter itself works well.
#[inline] #[inline]
pub fn sample_owen_scramble(dimension: u32, index: u32, seed: u32) -> f32 { pub fn sample_owen_scramble(dimension: u32, index: u32, seed: u32) -> f32 {
// Get the sobol point. // Get the sobol point.
@ -59,13 +61,17 @@ pub fn sample_owen_scramble(dimension: u32, index: u32, seed: u32) -> f32 {
// below. Instead, this simply serves to seed the Owen scrambling. // below. Instead, this simply serves to seed the Owen scrambling.
n ^= seed; n ^= seed;
// Do owen scrambling. This uses the technique presented in the paper // Do Owen scrambling.
// "Stratified Sampling for Stochastic Transparency" by Laine and Karras. // This uses the technique presented in the paper "Stratified Sampling for
// The basic idea is that we're running a hash function on the final valuw, // Stochastic Transparency" by Laine and Karras.
// but which only allows avalanche to happen upwards (e.g. a bit is never // The basic idea is that we're running a special kind of hash function
// affected by higher bits). This is acheived by only using multiplies by // that only allows avalanche to happen upwards (i.e. a bit is only
// even numbers. Normally this would be considered a poor hash function, // affected by the bits lower than it). This is achieved by only doing
// but in this case that behavior is exactly what we want. // mixing via multiplication by even numbers. Normally this would be
// considered a poor hash function, because normally you want all bits to
// have an equal chance of affecting all other bits. But in this case that
// only-upwards behavior is exactly what we want, because it ends up being
// equivalent to Owen scrambling.
for _ in 0..4 { for _ in 0..4 {
// The constant here is a large prime * 2. // The constant here is a large prime * 2.
n ^= n * 0xa97774e6; n ^= n * 0xa97774e6;
@ -85,7 +91,7 @@ fn sample_u32(dimension: u32, mut index: u32) -> u32 {
let mut i = (dimension as usize) * SIZE; let mut i = (dimension as usize) * SIZE;
while index != 0 { while index != 0 {
if (index & 1) != 0 { if (index & 1) != 0 {
result ^= unsafe { *MATRICES.get_unchecked(i) }; result ^= MATRICES[i];
} }
index >>= 1; index >>= 1;