// dgp_plr_CCDDHNR.mbt
//
// Pure-MoonBit port of upstream
// `doubleml.plm.datasets.dgp_plr_CCDDHNR2018.make_plr_CCDDHNR2018`.
//
// DGP from Chernozhukov, Chetverikov, Demirer, Duflo, Hansen,
// Newey, Robins (2018) "Double/debiased machine learning for
// treatment and structural parameters", Econometrics Journal 21:
// C1-C68, doi:10.1111/ecoj.12097, Figure 1 (a partially linear
// regression model with confounded continuous `d` and `y`).
//
// The data generating process (DGP) is
//
// d_i = a_0 * x_{i,1} + a_1 * sigmoid(x_{i,3})
// + s_1 * v_i, & v_i ~ N(0, 1),
//
// y_i = theta * d_i + b_0 * sigmoid(x_{i,1})
// + b_1 * x_{i,3} + s_2 * zeta_i,
// & zeta_i ~ N(0, 1),
//
// with covariates x_i ~ N(0, Sigma),
// where Sigma_{kj} = 0.7^|j-k|.
//
// Defaults: a_0=1.0, a_1=0.25, b_0=1.0, b_1=0.25, s_1=1.0, s_2=1.0.
//
// The MoonBit port omits the upstream `return_type` parameter
// (`DataFrame` / `DoubleMLData` / tuple); callers wrap the result in
// `DoubleMLData::new(...)` as needed.
///|
/// Result of the PLR CCDDHNR 2018 DGP.
struct PlrCcddhnr2018 {
/// Causal parameter used in `y = theta * d + g(X) + noise`.
theta : Double
/// `Matrix` of shape `(n_obs, dim_x)`.
x : Matrix
/// Outcome vector of length `n_obs`.
y : Array[Double]
/// Treatment vector of length `n_obs`.
d : Array[Double]
}
///|
/// Generate the PLR DGP from Chernozhukov et al. (2018) Figure 1.
pub fn make_plr_CCDDHNR2018(
n_obs : Int,
dim_x : Int,
theta : Double,
seed : Int,
) -> PlrCcddhnr2018 {
let a_0 = 1.0
let a_1 = 0.25
let b_0 = 1.0
let b_1 = 0.25
let s_1 = 1.0
let s_2 = 1.0
let rng = chacha8_rng(seed)
// 1. Pre-draw n_obs * dim_x standard normals via Box-Muller.
let n_normals_x = n_obs * dim_x
let z_flat : Array[Double] = Array::make(n_normals_x, 0.0)
let half_z = (n_normals_x + 1) / 2
for i = 0; i < half_z; i = i + 1 {
let (z1, z2) = box_muller_pair(rng)
let idx_a = 2 * i
let idx_b = 2 * i + 1
if idx_a < n_normals_x {
z_flat[idx_a] = z1
}
if idx_b < n_normals_x {
z_flat[idx_b] = z2
}
}
// 2. Build X ~ N(0, Sigma) with Sigma_{kj} = 0.7^|j-k| via
// the closed-form lower-triangular L (L_{kj} = 0.7^(k-j)
// for k >= j). Store X in row-major order.
let x_flat : Array[Double] = Array::make(n_normals_x, 0.0)
for i = 0; i < n_obs; i = i + 1 {
for k = 0; k < dim_x; k = k + 1 {
let mut s = 0.0
let mut acc = 1.0
let mut j = k
while j >= 0 {
s = s + acc * z_flat[i * dim_x + j]
acc = acc * 0.7
if j == 0 {
break
}
j = j - 1
}
x_flat[i * dim_x + k] = s
}
}
// 3. Pre-draw v, zeta ~ N(0, 1) independently via Box-Muller.
let v_flat : Array[Double] = Array::make(n_obs, 0.0)
let zeta_flat : Array[Double] = Array::make(n_obs, 0.0)
let half_n = (n_obs + 1) / 2
for i = 0; i < half_n; i = i + 1 {
let (z1, z2) = box_muller_pair(rng)
let idx_a = 2 * i
let idx_b = 2 * i + 1
if idx_a < n_obs {
v_flat[idx_a] = z1
}
if idx_b < n_obs {
v_flat[idx_b] = z2
}
}
for i = 0; i < half_n; i = i + 1 {
let (z1, z2) = box_muller_pair(rng)
let idx_a = 2 * i
let idx_b = 2 * i + 1
if idx_a < n_obs {
zeta_flat[idx_a] = z1
}
if idx_b < n_obs {
zeta_flat[idx_b] = z2
}
}
// 4. Build d and y per the DGP formula.
let d : Array[Double] = Array::make(n_obs, 0.0)
let y : Array[Double] = Array::make(n_obs, 0.0)
for i = 0; i < n_obs; i = i + 1 {
let x1 = x_flat[i * dim_x + 1]
let x3 = x_flat[i * dim_x + 3]
let sig3 = 1.0 / (1.0 + @math.exp(-x3))
let sig1 = 1.0 / (1.0 + @math.exp(-x1))
d[i] = a_0 * x1 + a_1 * sig3 + s_1 * v_flat[i]
y[i] = theta * d[i] + b_0 * sig1 + b_1 * x3 + s_2 * zeta_flat[i]
}
let x_mat = Matrix::from_array(x_flat, n_obs, dim_x)
{ theta, x: x_mat, y, d }
}
///|
/// Get the causal parameter (theta).
pub fn PlrCcddhnr2018::theta_get(self : PlrCcddhnr2018) -> Double {
self.theta
}
///|
/// Get the design matrix.
pub fn PlrCcddhnr2018::x_get(self : PlrCcddhnr2018) -> Matrix {
self.x
}
///|
/// Get the outcome vector.
pub fn PlrCcddhnr2018::y_get(self : PlrCcddhnr2018) -> Array[Double] {
self.y
}
///|
/// Get the treatment vector.
pub fn PlrCcddhnr2018::d_get(self : PlrCcddhnr2018) -> Array[Double] {
self.d
}