Solve a restoration optimization problem

Solve a restoration optimization problem to generate a solution.

Usage

# S3 method for class 'RestoptProblem'
solve(a, b, ...)

Arguments

a

restopt_problem() Restoration problem object.

b

Argument not used.

...

Additional arguments: verbose (logical) if TRUE, output solver logs. (FALSE by default) search_strategy (character) specify the solver's search strategy, among: "" -> default "RANDOM", "DOM_OVER_W_DEG", "DOM_OVER_W_DEG_REF", "MIN_DOM_LB", "MIN_DOM_UB", "ACTIVITY_BASED", "CONFLICT_HISTORY", "FAILURE_RATE", "FAILURE_LENGTH" lns (logical) if TRUE, activate Large Neighborhood Search (LNS). LNS can boost the optimization efficiency for large problems, with the price of less guarantees. Warning: it is highly recommended to set a time limit when using LNS because optimality proof may become unreachable (it depends on the characteristics of the problem), causing the solver to run forever without a time limit. Also, while LNS may boost the solver, it is very dependent on the problem's setting and size. It is recommended to first try solving a problem without LNS, and then try it if the results are not satisfactory.

Value

A restopt_solution() object.

Details

This function relies on the Choco-solver (https://choco-solver.org/) to solve a restoration optimization problem. If the solver finds a solution, it outputs a raster with 5 possible values: NA : NA (or NO_DATA) areas from the input habitat raster. 0 : non-habitat areas that were locked out. 1 : non-habitat areas that were available for selection. 2 : habitat areas. 3 : selected planning units for restoration. If the solve function return a no-solution error, it is either because the solver could not find a solution within the time limit that was set (see add_settings), or because the solver has detected that this is not possible to satisfy the constraints (the constraints are contradictory). In the first case, you can try to increase the time limit. In the second case, you should modify your targets.

Examples

if (FALSE) { # \dontrun{
# load data
habitat_data <- rast(
  system.file("extdata", "habitat_hi_res.tif", package = "restoptr")
)

available <- vect(
  system.file("extdata", "accessible_areas.gpkg", package = "restoptr")
)

# create problem with locked out constraints
p <- restopt_problem(
    existing_habitat = habitat_data,
    aggregation_factor = 16,
    habitat_threshold = 0.7
  ) %>%
  set_max_mesh_objective() %>%
  add_restorable_constraint(
    min_restore = 5,
    max_restore = 5,
  ) %>%
  add_available_areas_constraint(available) %>%
  add_settings(time_limit = 1)

# print problem
print(p)

# solve problem
s <- solve(p)

# plot solution
plot(s)
} # }