Package index • numerical.mle

Configuration

Create configuration objects for optimization algorithms

mle_config(): Create optimization configuration
mle_config_gradient(): Create gradient-based optimization configuration
mle_config_linesearch(): Create line search configuration
mle_constraint(): Create domain constraint specification
is_mle_config(): Check if object is an mle_config
is_mle_constraint(): Check if object is an mle_constraint

Core Solvers

Main optimization algorithms

mle_gradient_ascent(): Maximum likelihood estimation via gradient ascent
mle_newton_raphson(): Maximum likelihood estimation via Newton-Raphson

Meta-Solvers

Global optimization strategies

mle_grid_search(): MLE via grid search
mle_random_restart(): MLE via random restarts

Convenience Wrappers

Quick access to solvers with sensible defaults

mle_grad(): Quick gradient ascent with sensible defaults
mle_nr(): Quick Newton-Raphson with sensible defaults
with_constraint(): Quick constrained optimization

Function Transformers

Transform log-likelihood functions

with_subsampling(): Create stochastic log-likelihood with subsampling
with_penalty(): Add penalty term to log-likelihood
penalty_l1(): L1 penalty function (LASSO)
penalty_l2(): L2 penalty function (Ridge)
penalty_elastic_net(): Elastic net penalty (combination of L1 and L2)
compose(): Compose multiple function transformations

Generic Methods

Methods for mle_numerical objects

is_converged(): is_converged
is_mle_numerical(): is_mle_numerical
num_iterations(): num_iterations
mle_numerical(): mle_numerical

Package

Package documentation

numerical.mle-package numerical.mle: `numerical.mle`: A package for numerically solving maximum likelihood estimators from log-likelihood functions.