Bootstrap Configuration (Auto-Convergence)
Tip
Use Default Parameters: It is strongly recommended to use the provided default parameters to ensure methodological consistency. If different parameters are used, they should be scientifically justified and clearly documented in your research methodology.
The bootstrap_* structs (e.g., bootstrap_thresholds) support the following
nested parameters to control the convergence algorithm:
| Field | Type | Default (Thr/CI/Rank) | Description |
|---|---|---|---|
B_start |
int | 100 / 100 / 50 |
Initial number of bootstrap iterations. |
B_step |
int | 100 / 200 / 25 |
Iterations to add in each step. |
B_end |
int | 10000 / 20000 / 2500 |
Maximum number of iterations. |
n_trials |
int | 25 / 30 / 15 |
Number of independent trials per step to check stability. |
convergence_tolerance |
double | 0.01 / 0.03 / 0.005 |
Max allowed variation (e.g., 0.005 = 0.5%). |
smoothing_window |
int | 3 / 3 / 3 |
Window size for moving average smoothing. |
convergence_streak_needed |
int | 3 / 3 / 3 |
Consecutive steps required to pass tolerance. |
min_steps_for_convergence_check |
int | 1 |
Minimum steps before checking convergence. |