1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
use crate::*;
use rand::Rng;
#[cfg(feature = "serde_support")]
use serde::{Serialize, Deserialize};
#[derive(Debug, Clone, Copy, Eq, PartialEq)]
#[cfg_attr(feature = "serde_support", derive(Serialize, Deserialize))]
pub enum CoinFlip {
Head,
Tail
}
impl CoinFlip
{
pub fn turn(&mut self) {
*self = match self {
CoinFlip::Head => CoinFlip::Tail,
CoinFlip::Tail => CoinFlip::Head
};
}
}
#[derive(Clone, Debug)]
#[cfg_attr(feature = "serde_support", derive(Serialize, Deserialize))]
pub struct CoinFlipMove{
previouse: CoinFlip,
index: usize,
}
#[derive(Clone, Debug)]
#[cfg_attr(feature = "serde_support", derive(Serialize, Deserialize))]
pub struct CoinFlipSequence<R> {
rng: R,
seq: Vec<CoinFlip>,
steps: usize,
rejected: usize,
accepted: usize,
undo_count: usize
}
impl<R> CoinFlipSequence<R>
where R: Rng,
{
pub fn new(n: usize, mut rng: R) -> Self
{
let mut seq = Vec::with_capacity(n);
seq.extend(
(0..n).map(|_| {
if rng.gen::<bool>() {
CoinFlip::Tail
} else {
CoinFlip::Head
}
})
);
Self{
rng,
seq,
steps:0,
rejected: 0,
accepted: 0,
undo_count: 0
}
}
}
impl<R> CoinFlipSequence<R>
{
pub fn head_count(&self) -> usize
{
self.seq.iter()
.filter(|&item| *item == CoinFlip::Head)
.count()
}
pub fn update_head_count(&self, step: &CoinFlipMove, head_count: &mut usize)
{
match step.previouse {
CoinFlip::Head => {
*head_count -= 1;
},
CoinFlip::Tail => {
*head_count += 1;
}
}
}
pub fn max_heads_in_a_row(&self) -> usize
{
let mut current_heads = 0;
let mut max_heads = 0;
for flip in self.seq.iter()
{
match flip {
CoinFlip::Head => current_heads += 1,
CoinFlip::Tail => {
max_heads = max_heads.max(current_heads);
current_heads = 0;
}
}
}
max_heads.max(current_heads)
}
}
impl<R> MarkovChain<CoinFlipMove, ()> for CoinFlipSequence<R>
where R: Rng
{
fn m_step(&mut self) -> CoinFlipMove {
let pos = self.rng.gen_range(0..self.seq.len());
let previouse = self.seq[pos];
self.seq[pos].turn();
CoinFlipMove{
previouse,
index: pos
}
}
#[inline]
fn m_steps(&mut self, count: usize, steps: &mut Vec<CoinFlipMove>) {
self.steps += 1;
steps.clear();
steps.extend((0..count)
.map(|_| self.m_step())
);
}
#[inline]
fn m_steps_acc<Acc, AccFn>
(
&mut self,
count: usize,
steps: &mut Vec<CoinFlipMove>,
acc: &mut Acc,
mut acc_fn: AccFn
)
where AccFn: FnMut(&Self, &CoinFlipMove, &mut Acc)
{
self.steps += 1;
steps.clear();
steps.extend(
(0..count)
.map(|_| self.m_step_acc(acc, &mut acc_fn))
);
}
fn undo_step(&mut self, step: &CoinFlipMove) {
self.seq[step.index] = step.previouse;
}
#[inline]
fn undo_step_quiet(&mut self, step: &CoinFlipMove) {
self.undo_step(step);
}
fn undo_steps(&mut self, steps: &[CoinFlipMove], res: &mut Vec<()>) {
self.undo_count += 1;
res.clear();
res.extend(
steps.iter()
.rev()
.map(|step| self.undo_step(step))
);
assert_eq!(self.rejected, self.undo_count);
}
fn undo_steps_quiet(&mut self, steps: &[CoinFlipMove]) {
self.undo_count += 1;
steps.iter()
.rev()
.for_each( |step| self.undo_step_quiet(step));
assert_eq!(self.rejected, self.undo_count);
}
fn steps_accepted(&mut self, _steps: &[CoinFlipMove])
{
self.accepted += 1;
if self.accepted + self.rejected != self.steps{
panic!("{} {} {}", self.steps, self.rejected, self.accepted)
}
}
fn steps_rejected(&mut self, _steps: &[CoinFlipMove])
{
self.rejected += 1;
if self.accepted + self.rejected != self.steps{
panic!("{} {} {}", self.steps, self.rejected, self.accepted)
}
}
}
impl<R> HasRng<R> for CoinFlipSequence<R>
where R: Rng
{
fn rng(&mut self) -> &mut R {
&mut self.rng
}
fn swap_rng(&mut self, rng: &mut R) {
std::mem::swap(&mut self.rng, rng);
}
}
#[cfg(test)]
#[cfg(features="replica_exchange")]
mod tests{
use super::*;
use rand::SeedableRng;
use rayon::iter::{IntoParallelRefMutIterator, ParallelIterator};
use statrs::distribution::{Binomial, Discrete};
use rand_pcg::{Pcg64Mcg, Pcg64};
use std::num::*;
#[test]
fn combine()
{
let n = 30;
let intervals = 3;
let hist1 = HistUsizeFast::new_inclusive(0, 2 * n / 3)
.unwrap();
let hist_list1 = hist1.overlapping_partition(intervals, 1).unwrap();
let mut rng = Pcg64Mcg::seed_from_u64(384789);
let ensemble1 = CoinFlipSequence::new
(
n,
Pcg64::from_rng(&mut rng).unwrap()
);
let rewl_builder1 = RewlBuilder::from_ensemble(
ensemble1,
hist_list1,
1,
NonZeroUsize::new(1999).unwrap(),
NonZeroUsize::new(2).unwrap(),
0.000003
).unwrap();
let rewl1 = rewl_builder1.greedy_build(|e| Some(e.head_count()));
let ensemble2 = CoinFlipSequence::new(
n,
Pcg64::from_rng(rng).unwrap()
);
let hist2 = HistUsizeFast::new_inclusive(n / 3, n)
.unwrap();
let hist_list2 = hist2.overlapping_partition(intervals, 1).unwrap();
let rewl_builder2 = RewlBuilder::from_ensemble(
ensemble2,
hist_list2,
1,
NonZeroUsize::new(1999).unwrap(),
NonZeroUsize::new(2).unwrap(),
0.000003
).unwrap();
let mut rewl2 = rewl_builder2.greedy_build(|e| Some(e.head_count()));
rewl2.simulate_until_convergence(|e| Some(e.head_count()));
let mut rewl_slice = vec![rewl1, rewl2];
rewl_slice.par_iter_mut()
.for_each(|rewl| rewl.simulate_until_convergence(|e| Some(e.head_count())));
rewl_slice.iter()
.for_each(
|r|
{
r.walkers().iter()
.for_each(|w| println!("rewl replica_frac {}", w.replica_exchange_frac()));
}
);
let steps: usize = rewl_slice
.iter()
.flat_map(|r|
r.walkers()
.iter()
.map(|w| w.step_count())
).sum();
println!("Ges steps rewl {}", steps);
let binomial = Binomial::new(0.5, n as u64).unwrap();
let prob = rewl::merged_log_prob(&rewl_slice)
.unwrap();
let ln_prob_true: Vec<_> = (0..=n)
.map(|k| binomial.ln_pmf(k as u64))
.collect();
let mut rees_slice: Vec<_> = rewl_slice.into_iter()
.map(|r| r.into_rees())
.collect();
rees_slice.par_iter_mut()
.for_each(|rees| rees.simulate_until_convergence(|e| Some(e.head_count()), |_,_, _|{}));
let steps: usize = rees_slice
.iter()
.flat_map(|r|
r.walkers()
.iter()
.map(|w| w.step_count())
).sum();
println!("Ges steps rees {}", steps);
let prob_rees = rees::merged_log_prob(&rees_slice).unwrap();
let mut max_ln_difference_rewl = f64::NEG_INFINITY;
let mut max_difference_rewl = f64::NEG_INFINITY;
let mut frac_difference_max_rewl = f64::NEG_INFINITY;
let mut frac_difference_min_rewl = f64::INFINITY;
let mut average_p_dif_rewl = 0.0;
let mut max_ln_difference_rees = f64::NEG_INFINITY;
let mut max_difference_rees = f64::NEG_INFINITY;
let mut frac_difference_max_rees = f64::NEG_INFINITY;
let mut frac_difference_min_rees = f64::INFINITY;
for (index, ((val_sim1, val_sim2), val_true)) in prob.0.into_iter().zip(prob_rees.0).zip(ln_prob_true).enumerate()
{
println!("{} {} {} {}", index, val_sim1, val_sim2, val_true);
let val_real = val_true.exp();
let val_simulation1 = val_sim1.exp();
let val_simulation2 = val_sim2.exp();
max_difference_rewl = f64::max((val_simulation1 - val_real).abs(), max_difference_rewl);
max_ln_difference_rewl = f64::max(max_ln_difference_rewl, (val_sim1-val_true).abs());
let frac = val_simulation1 / val_real;
average_p_dif_rewl += (frac - 1.0).abs();
frac_difference_max_rewl = frac_difference_max_rewl.max(frac);
frac_difference_min_rewl = frac_difference_min_rewl.min(frac);
max_difference_rees = f64::max((val_simulation2 - val_real).abs(), max_difference_rees);
max_ln_difference_rees = f64::max(max_ln_difference_rees, (val_sim2-val_true).abs());
let frac = val_simulation2 / val_real;
frac_difference_max_rees = frac_difference_max_rees.max(frac);
frac_difference_min_rees = frac_difference_min_rees.min(frac);
}
average_p_dif_rewl /= (n+1) as f64;
println!("Average_p_dif {}", average_p_dif_rewl);
println!("max_ln_difference rewl: {}", max_ln_difference_rewl);
println!("max absolute difference rewl: {}", max_difference_rewl);
println!("max frac rewl: {}", frac_difference_max_rewl);
println!("min frac rewl: {}", frac_difference_min_rewl);
println!("max_ln_difference rees: {}", max_ln_difference_rees);
println!("max absolute difference rees: {}", max_difference_rees);
println!("max frac rees: {}", frac_difference_max_rees);
println!("min frac rees: {}", frac_difference_min_rees);
assert!(max_difference_rewl < 0.0006);
assert!(max_difference_rees < 0.0005);
assert!(frac_difference_max_rewl - 1.0 < 0.02);
assert!(frac_difference_max_rees - 1.0 < 0.02);
assert!((frac_difference_min_rewl - 1.0).abs() < 0.018);
assert!((frac_difference_min_rees - 1.0).abs() < 0.012);
}
}