Quick gradient ascent with sensible defaults

Convenience wrapper for mle_gradient_ascent with simplified interface. Automatically enables line search for better convergence.

Usage

mle_grad(loglike, score, theta0, use_linesearch = TRUE, ...)

Arguments

loglike

Log-likelihood function

score

Score function (gradient)

theta0

Initial parameters

use_linesearch

Use backtracking line search (default: TRUE)

...

Additional config parameters passed to mle_config_linesearch or mle_config_gradient

Value

mle_gradient_ascent object

Examples

if (FALSE) { # \dontrun{
# Quick usage with defaults
result <- mle_grad(loglike, score, theta0 = c(0, 1))

# Override config parameters
result <- mle_grad(
  loglike, score, theta0 = c(0, 1),
  max_iter = 200,
  rel_tol = 1e-6
)

# Without line search
result <- mle_grad(
  loglike, score, theta0 = c(0, 1),
  use_linesearch = FALSE,
  eta = 0.1
)
} # }