Skip to contents

Create and Verify Optimization Parameters

Source: R/optim_params.R
optim_params.Rd

Create and verify optimization parameters. These parameters are used to control the behaviour of simulated annealing algorithm.

Usage

optim_params(
  swap_count = 1,
  swap_all_blocks = FALSE,
  adaptive_swaps = FALSE,
  start_temp = 100,
  cooling_rate = 0.99,
  random_initialisation = 0,
  adj_weight = 1,
  bal_weight = 1,
  stop_at_optimal = TRUE
)

Arguments

swap_count

Number of treatment swaps per iteration (default: 1).

swap_all_blocks

Logical; if TRUE, performs swaps in all blocks at each iteration (default: FALSE).

adaptive_swaps

Logical; if TRUE, adjusts swap parameters based on temperature (default: FALSE).

start_temp

Starting temperature for simulated annealing (default: 100). A higher start temperature allows the algorithm to accept worse solutions early on, encouraging exploration of the solution space and helping to avoid local optima. Lower values make the algorithm greedier from the start, which can speed up convergence but increases the risk of getting stuck in a poor solution. A good starting temperature allows moderately worse solutions to be accepted with a probability of 70-90% at the beginning of the optimisation.

cooling_rate

Rate at which temperature decreases for simulated annealing (default: 0.99). This controls how quickly the algorithm shifts from exploration to exploitation. The temperature is updated at each iteration by multiplying it by this rate: T_i = start_temp * cooling_rate^i. A higher cooling rate (e.g. 0.995-0.999) results in slower cooling and a longer exploration phase, which is generally better for complex or noisy optimisation landscapes. Lower values (e.g. 0.95-0.98) cool quickly, leading to faster convergence but greater risk of premature convergence to a suboptimal design.

random_initialisation

Number of times to randomly shuffle items within swap_within; the design with the best score is used as an initial design (default: 0).

adj_weight

Weight for adjacency score (default: 1).

bal_weight

Weight for balance score (default: 1).

stop_at_optimal

Logical; if TRUE, stops the level as soon as the best score reaches the lowest score its layout allows (default: TRUE). The bound is only derivable for the default objective_function() with non-negative weights and no relationship matrix; otherwise the level runs to its usual stopping rules regardless of this setting.

Value

A named list of optimization parameters.

See also

speed() for examples