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
//! Core copula modeling for validated pseudo-observations.
//!
//! `rscopulas` is the main Rust surface for fitting, evaluating, and
//! sampling:
//!
//! - single-family copulas such as Gaussian, Student t, Clayton, Frank, and
//! Gumbel-Hougaard,
//! - low-level pair-copula kernels with h-functions and inverse h-functions,
//! including **Khoudraji** asymmetric pair copulas,
//! - C-vine, D-vine, and R-vine copulas.
//!
//! The crate assumes your data is already in pseudo-observation form: finite
//! values strictly inside `(0, 1)`. If you start from raw observations, estimate
//! or transform the marginals first, then build a [`PseudoObs`] matrix.
//!
//! # Quick start
//!
//! ```no_run
//! use ndarray::array;
//! use rand::{rngs::StdRng, SeedableRng};
//! use rscopulas::{CopulaModel, FitOptions, GaussianCopula, PseudoObs};
//!
//! # fn main() -> Result<(), Box<dyn std::error::Error>> {
//! let data = PseudoObs::new(array![
//! [0.12, 0.18],
//! [0.21, 0.25],
//! [0.27, 0.22],
//! [0.35, 0.42],
//! [0.48, 0.51],
//! [0.56, 0.49],
//! [0.68, 0.73],
//! [0.82, 0.79],
//! ])?;
//!
//! let fit = GaussianCopula::fit(&data, &FitOptions::default())?;
//! println!("AIC: {}", fit.diagnostics.aic);
//!
//! let log_pdf = fit.model.log_pdf(&data, &Default::default())?;
//! println!("first log density = {}", log_pdf[0]);
//!
//! let mut rng = StdRng::seed_from_u64(7);
//! let sample = fit.model.sample(4, &mut rng, &Default::default())?;
//! println!("sample = {:?}", sample);
//! # Ok(())
//! # }
//! ```
//!
//! # Fit a vine copula
//!
//! ```no_run
//! use ndarray::array;
//! use rscopulas::{
//! PairCopulaFamily, PseudoObs, SelectionCriterion, VineCopula, VineFitOptions,
//! };
//!
//! # fn main() -> Result<(), Box<dyn std::error::Error>> {
//! let data = PseudoObs::new(array![
//! [0.12, 0.18, 0.21],
//! [0.21, 0.25, 0.29],
//! [0.27, 0.22, 0.31],
//! [0.35, 0.42, 0.39],
//! [0.48, 0.51, 0.46],
//! [0.56, 0.49, 0.58],
//! [0.68, 0.73, 0.69],
//! [0.82, 0.79, 0.76],
//! ])?;
//!
//! let options = VineFitOptions {
//! family_set: vec![
//! PairCopulaFamily::Independence,
//! PairCopulaFamily::Gaussian,
//! PairCopulaFamily::Clayton,
//! PairCopulaFamily::Frank,
//! PairCopulaFamily::Gumbel,
//! PairCopulaFamily::Khoudraji,
//! ],
//! include_rotations: true,
//! criterion: SelectionCriterion::Aic,
//! truncation_level: Some(1),
//! ..VineFitOptions::default()
//! };
//!
//! let fit = VineCopula::fit_r_vine(&data, &options)?;
//! println!("structure = {:?}", fit.model.structure());
//! println!("order = {:?}", fit.model.order());
//! # Ok(())
//! # }
//! ```
//!
//! # Conditional sampling from a vine
//!
//! A fitted vine natively supports conditional simulation through the
//! Rosenblatt transform. Pin the conditioning column at the Rosenblatt
//! anchor position `variable_order[0]` by placing it at the end of `order`,
//! then feed its uniforms into `inverse_rosenblatt` alongside fresh random
//! uniforms for the free columns:
//!
//! ```no_run
//! use ndarray::{Array2, Axis, s};
//! use rand::{rngs::StdRng, Rng, SeedableRng};
//! use rscopulas::{PseudoObs, SampleOptions, VineCopula, VineFitOptions};
//!
//! # fn main() -> Result<(), Box<dyn std::error::Error>> {
//! # let data = PseudoObs::new(ndarray::array![
//! # [0.12, 0.18, 0.21], [0.21, 0.25, 0.29], [0.27, 0.22, 0.31],
//! # [0.35, 0.42, 0.39], [0.48, 0.51, 0.46], [0.56, 0.49, 0.58],
//! # [0.68, 0.73, 0.69], [0.82, 0.79, 0.76],
//! # ])?;
//! let target = 2usize; // column to condition on
//! let order = vec![0, 1, target];
//! let vine = VineCopula::fit_c_vine_with_order(
//! &data, &order, &VineFitOptions::default(),
//! )?.model;
//! assert_eq!(vine.variable_order()[0], target);
//!
//! // Build U: known column at target, fresh uniforms elsewhere.
//! let n = 1_000;
//! let d = vine.variable_order().len();
//! let known = Array2::<f64>::from_elem((n, 1), 0.9);
//! let mut rng = StdRng::seed_from_u64(0);
//! let mut u = Array2::<f64>::zeros((n, d));
//! u.column_mut(target).assign(&known.index_axis(Axis(1), 0));
//! for obs in 0..n {
//! for var in 0..d {
//! if var != target {
//! u[(obs, var)] = rng.random();
//! }
//! }
//! }
//! let v = vine.inverse_rosenblatt(u.view(), &SampleOptions::default())?;
//! assert_eq!(v.shape(), &[n, d]);
//! # Ok(())
//! # }
//! ```
//!
//! The Python API ships a `VineCopula.sample_conditional(known, n, seed)`
//! convenience that does this plumbing for you and also supports `k >= 2`
//! known columns via a partial forward Rosenblatt pass.
//!
//! # Backend expectations
//!
//! The crate exposes explicit execution policy controls through [`ExecPolicy`]
//! and [`Device`]. Today, `Auto` is conservative and does not promise that
//! every numerically heavy path uses CUDA or Metal. If you need a deterministic
//! backend choice, prefer `ExecPolicy::Force(...)`.
//!
pub use PseudoObs;
pub use ;
pub use ;
pub use ;