simulated_annealing_SL: Simulated Annealing Optimization with Categorical Variable...

In covalchemy: Constructing Joint Distributions with Control Over Statistical Properties

Simulated Annealing Optimization with Categorical Variable and R^2 Differences

Description

This function implements the Simulated Annealing algorithm to optimize a solution based on the total variation distance, changes in regression coefficients, R-squared differences, and inter-cluster distance, with respect to a set of categorical and continuous variables.

Usage

simulated_annealing_SL(
  X,
  Y,
  Z,
  X_st,
  Y_st,
  p,
  sd_x = 0.05,
  sd_y = 0.05,
  lambda1 = 1,
  lambda2 = 1,
  lambda3 = 1,
  lambda4 = 1,
  max_iter = 1000,
  initial_temp = 1,
  cooling_rate = 0.99
)

Arguments

X	A numeric vector or matrix of input data (independent variable).
Y	A numeric vector of the dependent variable (target).
Z	A categorical variable (vector), used for grouping data in the analysis.
X_st	A numeric vector of starting values for the composition method of X.
Y_st	A numeric vector of starting values for the composition method of Y.
p	A numeric vector representing the target R^2 values for each category in Z.
sd_x	Standard deviation for the noise added to X during the perturbation (default is 0.05).
sd_y	Standard deviation for the noise added to Y during the perturbation (default is 0.05).
lambda1	Regularization parameter for the total variation distance term (default is 1).
lambda2	Regularization parameter for the coefficient difference term (default is 1).
lambda3	Regularization parameter for the R^2 difference term (default is 1).
lambda4	Regularization parameter for the inter-cluster distance term (default is 1).
max_iter	Maximum number of iterations for the annealing process (default is 1000).
initial_temp	Initial temperature for the annealing process (default is 1.0).
cooling_rate	The rate at which the temperature cools down during annealing (default is 0.99).

Value

A list with the optimized values of X_prime and Y_prime.

covalchemy documentation built on April 12, 2025, 2:15 a.m.