Description
Derivative-Based Optimization with User-Defined Convergence Criteria.
Description
Provides a derivative-based optimization framework that allows users to combine eight convergence criteria. Unlike standard optimization functions, this package includes a built-in mechanism to verify the positive definiteness of the Hessian matrix at the point of convergence. This additional check helps prevent the solver from falsely identifying non-optimal solutions, such as saddle points, as valid minima.
README.md
optimflex
optimflex provides a highly flexible suite of derivative-based non-linear optimization algorithms. It is specifically designed for researchers who require rigorous convergence control, particularly in complex models like SEM.
Why Use optimflex?
Standard optimization functions often rely on a single, fixed stopping rule. optimflex offers:
- Strict Convergence Control: Choose from 8 distinct criteria.
- The “AND” Rule: All selected criteria must be met simultaneously.
- Hessian Verification: Verify local minima by checking positive definiteness at the final point.
Installation
# install.packages("devtools")
# devtools::install_github("yourusername/optimflex")
Basic Usage
library(optimflex)
rosenbrock <- function(x) {
100 * (x[2] - x[1]^2)^2 + (1 - x[1])^2
}
res <- double_dogleg(
start = c(-1.2, 1.0),
objective = rosenbrock,
control = list(
use_grad = TRUE,
use_rel_x = TRUE,
use_posdef = TRUE
)
)
print(res$par)
#> [1] 0.9999955 0.9999910
Convergence Criteria Overview
| Flag | Description |
|---|---|
use_abs_f | Absolute function change |
use_rel_f | Relative function change |
use_abs_x | Absolute parameter change |
use_rel_x | Relative parameter change |
use_grad | Gradient infinity norm |
use_posdef | Hessian Verification |
use_pred_f | Predicted Decrease. |