1import numpy as np
2
3# =====================================================================
4# EDITABLE: implement run_causal_discovery below
5# =====================================================================
6def run_causal_discovery(X: np.ndarray) -> np.ndarray:
7    """
8    Input:  X of shape (n_samples, n_variables)
9    Output: adjacency matrix B of shape (n_variables, n_variables)
10            B[i, j] != 0  means j -> i  (follows causal-learn convention)
11    """
12    n = X.shape[1]
13    return np.zeros((n, n))
14# =====================================================================
15

1"""Evaluation harness for the causal-observational-nonlinear task."""
2import argparse
3import os
4import sys
5
6sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
7
8from data_gen import simulate_nonlinear_anm
9from metrics import compute_metrics
10from custom_algorithm import run_causal_discovery
11
12
13def main():
14    parser = argparse.ArgumentParser(
15        description="Evaluate a causal discovery algorithm on synthetic nonlinear ANM data."
... 49 more lines

1"""Synthetic nonlinear additive noise model (ANM) data generator.
2
3Structural equation:  x_j = f_j(pa(j)) + e_j
4where f_j is a nonlinear function and e_j is additive non-Gaussian noise.
5"""
6import numpy as np
7import networkx as nx
8
9
10def simulate_dag(n_nodes, graph_type, seed, er_prob=0.3, sf_m=2):
11    """Return a binary adjacency matrix for a random DAG.
12
13    Convention: adj[i, j] = 1 means i -> j  (i is a parent of j).
14    The DAG is enforced by keeping only edges i -> j with i < j, imposing
15    a topological ordering by node index.
... 145 more lines

1"""Evaluation metrics for directed causal graph recovery."""
2import numpy as np
3
4
5def compute_metrics(B_est, B_true, threshold=0.01):
6    """Compute SHD, F1, precision, and recall for directed edge recovery.
7
8    Convention: B[i, j] != 0 means j -> i.
9
10    SHD definition (each type counts as exactly 1 error):
11        - Reversed edge : correct skeleton edge but wrong direction
12        - Extra edge    : present in estimate but absent in truth (non-reversal)
13        - Missing edge  : present in truth but absent in estimate (non-reversal)
14
15    F1 / precision / recall are computed on the directed edge set
... 38 more lines