bauer/WHAM
1
0
Fork 0
mirror of https://github.com/dnlbauer/WHAM.git synced 2026-09-11 06:35:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

seedable rng

Browse Source
This commit is contained in:
Daniel Bauer
2020-03-08 12:29:52 +01:00
parent 936cbe1c58
commit 08e7c93327
4 changed files with 28 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/error_analysis.rs
View File

@@ -1,4 +1,4 @@
use rand;
use rand::{SeedableRng, StdRng, Rng};
use super::histogram::{Dataset};
use super::perform_wham;
use super::Config;
@@ -7,9 +7,9 @@ use rgsl::statistics;
// returns a set of num_windows continious weights by
// a) generate num_windows-1 random variables and sort them
// b) each weight n is the difference between n+1 and n, where n0=0 and nN+1=1
fn generate_random_weights(num_windows: usize) -> Vec<f64> {
fn generate_random_weights(num_windows: usize, rng: &mut StdRng) -> Vec<f64> {
// create a list of num_windows - 1 sorted random numbers and append/prepend 0 and 1
let mut tmp = (0..num_windows-1).map(|_| rand::random::<f64>()).collect::<Vec<f64>>();
let mut tmp = (0..num_windows-1).map(|_| rng.gen::<f64>()).collect::<Vec<f64>>();
tmp.sort_by(|a,b| { a.partial_cmp(b).unwrap() });
let mut rnds = vec![0.0];
rnds.append(&mut tmp);
@@ -24,8 +24,8 @@ fn generate_random_weights(num_windows: usize) -> Vec<f64> {
}
// Generate a random weighted dataset from the given dataset by changing the weights
fn generate_random_weighted_dataset(ds: Dataset) -> Dataset {
let weights = generate_random_weights(ds.num_windows);
fn generate_random_weighted_dataset(ds: Dataset, rng: &mut StdRng) -> Dataset {
let weights = generate_random_weights(ds.num_windows, rng);
Dataset::new_weighted(ds, weights)
}
@@ -33,10 +33,13 @@ fn generate_random_weighted_dataset(ds: Dataset) -> Dataset {
// datasets. The standard deviation is calculated on the bootstrapped probabilities of each bin. The
// standard deviation of the free eneergy is then deduced by error propagation (A_std = kT*1/P*P_std)
pub fn run_bootstrap(cfg: &Config, ds: Dataset, P: &[f64], num_runs: usize) -> (Vec<f64>,Vec<f64>) {
// seed the rng
let mut rng: StdRng = SeedableRng::seed_from_u64(cfg.bootstrap_seed);
// Calculate bootstrapped probabilities
let bootstrapped_Ps: Vec<Vec<f64>> = (0..num_runs).map(|x| {
println!("Bootstrap run {}/{}", x, num_runs);
let rnd_weighted_dataset = generate_random_weighted_dataset(ds.clone());
let rnd_weighted_dataset = generate_random_weighted_dataset(ds.clone(), &mut rng);
perform_wham(cfg, &rnd_weighted_dataset).unwrap().0
}).collect();