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
use crate::covertree::node::CoverNode;
use crate::covertree::CoverTreeReader;
use crate::plugins::*;
use rand::distributions::{Distribution, Uniform};
use rand::Rng;
#[derive(Debug, Clone, Default)]
pub struct Categorical {
pub(crate) child_counts: Vec<(NodeAddress, f64)>,
pub(crate) singleton_count: f64,
}
impl Categorical {
pub fn new() -> Categorical {
Categorical {
child_counts: Vec::new(),
singleton_count: 0.0,
}
}
pub fn total(&self) -> f64 {
self.singleton_count
+ self
.child_counts
.iter()
.map(|(_, c)| c)
.fold(0.0, |x, a| x + a)
}
pub fn prob_vector(&self) -> Option<(Vec<(NodeAddress, f64)>, f64)> {
let total = self.total();
if total > 0.0 {
let v: Vec<(NodeAddress, f64)> = self
.child_counts
.iter()
.map(|(na, f)| (*na, f / total))
.collect();
Some((v, self.singleton_count / total))
} else {
None
}
}
pub(crate) fn merge(&mut self, other: &Categorical) {
for (na, c) in &other.child_counts {
self.add_child_pop(Some(*na), *c);
}
self.add_child_pop(None, other.singleton_count);
}
pub(crate) fn add_child_pop(&mut self, loc: Option<NodeAddress>, count: f64) {
match loc {
Some(ca) => match self.child_counts.binary_search_by_key(&ca, |&(a, _)| a) {
Ok(index) => self.child_counts[index].1 += count,
Err(index) => self.child_counts.insert(index, (ca, count)),
},
None => self.singleton_count += count,
}
}
pub(crate) fn remove_child_pop(&mut self, loc: Option<NodeAddress>, count: f64) {
match loc {
Some(ca) => {
if let Ok(index) = self.child_counts.binary_search_by_key(&ca, |&(a, _)| a) {
if self.child_counts[index].1 < count {
self.child_counts[index].1 = 0.0;
} else {
self.child_counts[index].1 -= count;
}
}
}
None => {
if self.singleton_count < count as f64 {
self.singleton_count = 0.0;
} else {
self.singleton_count -= count as f64;
}
}
}
}
pub fn ln_pdf(&self, loc: Option<&NodeAddress>) -> Option<f64> {
let total = self.total();
if total > 0.0 {
let ax = match loc {
Some(ca) => self
.child_counts
.binary_search_by_key(&ca, |(a, _)| a)
.map(|i| self.child_counts[i].1)
.unwrap_or(0.0),
None => self.singleton_count,
};
Some(ax.ln() - total.ln())
} else {
None
}
}
pub fn sample<R: Rng>(&self, rng: &mut R) -> Option<NodeAddress> {
let sum = self.total() as usize;
let uniform = Uniform::from(0..sum);
let sample = uniform.sample(rng) as f64;
let mut count = 0.0;
for (a, c) in &self.child_counts {
count += c;
if sample < count {
return Some(*a);
}
}
None
}
pub fn kl_divergence(&self, other: &Categorical) -> Option<f64> {
let my_total = self.total();
let other_total = other.total();
if my_total == 0.0 || other_total == 0.0 {
None
} else {
let ln_total = my_total.ln() - other_total.ln();
let mut sum: f64 = 0.0;
if self.singleton_count > 0.0 && other.singleton_count > 0.0 {
sum += (self.singleton_count / my_total)
* (self.singleton_count.ln() - other.singleton_count.ln() - ln_total);
}
for ((ca, ca_count), (other_ca, other_ca_count)) in
self.child_counts.iter().zip(other.child_counts.iter())
{
assert_eq!(ca, other_ca);
sum += (ca_count / my_total) * (ca_count.ln() - other_ca_count.ln() - ln_total);
}
Some(sum)
}
}
}
impl<D: PointCloud> NodePlugin<D> for Categorical {}
#[derive(Debug, Clone)]
pub struct GokoCategorical {}
impl<D: PointCloud> GokoPlugin<D> for GokoCategorical {
type NodeComponent = Categorical;
fn node_component(
_parameters: &Self,
my_node: &CoverNode<D>,
my_tree: &CoverTreeReader<D>,
) -> Option<Self::NodeComponent> {
let mut bucket = Categorical::new();
if let Some((nested_scale, child_addresses)) = my_node.children() {
my_tree.get_node_plugin_and::<Self::NodeComponent, _, _>(
(nested_scale, *my_node.center_index()),
|p| {
bucket.add_child_pop(
Some((nested_scale, *my_node.center_index())),
p.total() as f64,
);
},
);
for ca in child_addresses {
my_tree.get_node_plugin_and::<Self::NodeComponent, _, _>(*ca, |p| {
bucket.add_child_pop(Some(*ca), p.total() as f64);
});
}
bucket.add_child_pop(None, my_node.singletons_len() as f64);
} else {
bucket.add_child_pop(None, my_node.singletons_len() as f64 + 1.0);
}
Some(bucket)
}
}
#[cfg(test)]
pub(crate) mod tests {
use super::*;
#[test]
fn empty_bucket_sanity_test() {
let buckets = Categorical::new();
assert_eq!(buckets.ln_pdf(None), None);
assert_eq!(buckets.ln_pdf(Some(&(0, 0))), None);
assert_eq!(buckets.kl_divergence(&buckets), None)
}
#[test]
fn singleton_bucket_sanity_test() {
let mut buckets = Categorical::new();
buckets.add_child_pop(None, 5.0);
assert_approx_eq!(buckets.ln_pdf(None).unwrap(), 0.0);
assert_approx_eq!(buckets.kl_divergence(&buckets).unwrap(), 0.0);
assert_eq!(buckets.ln_pdf(Some(&(0, 0))), Some(std::f64::NEG_INFINITY));
}
#[test]
fn child_bucket_sanity_test() {
let mut buckets = Categorical::new();
buckets.add_child_pop(Some((0, 0)), 5.0);
assert_approx_eq!(buckets.ln_pdf(Some(&(0, 0))).unwrap(), 0.0);
assert_approx_eq!(buckets.kl_divergence(&buckets).unwrap(), 0.0);
assert_eq!(buckets.ln_pdf(None).unwrap(), std::f64::NEG_INFINITY);
}
#[test]
fn mixed_bucket_sanity_test() {
let mut bucket1 = Categorical::new();
bucket1.add_child_pop(None, 6.0);
bucket1.add_child_pop(Some((0, 0)), 6.0);
println!("{:?}", bucket1);
let mut bucket2 = Categorical::new();
bucket2.add_child_pop(None, 4.0);
bucket2.add_child_pop(Some((0, 0)), 8.0);
println!("{:?}", bucket2);
assert_approx_eq!(bucket1.ln_pdf(None).unwrap(), (0.5f64).ln());
assert_approx_eq!(
bucket2.ln_pdf(Some(&(0, 0))).unwrap(),
(0.666666666f64).ln()
);
assert_approx_eq!(bucket1.kl_divergence(&bucket1).unwrap(), 0.0);
assert_approx_eq!(bucket1.kl_divergence(&bucket2).unwrap(), 0.05889151782);
assert_approx_eq!(bucket2.kl_divergence(&bucket1).unwrap(), 0.05663301226);
}
}