Write a JuMP model file from a multimod model

Usage

write_jump(
  model,
  file = NULL,
  model_dir = NULL,
  solver_dir = "jump",
  output_name = "model.jl",
  model_name = "model",
  optimizer = "HiGHS.Optimizer",
  optimizer_attributes = NULL,
  use_folded = TRUE,
  cleanup = TRUE,
  export_data = FALSE,
  export_vars = FALSE,
  export_lp = FALSE,
  export_mps = FALSE,
  data_mode = c("external", "embedded"),
  ...
)

Arguments

model: A multimod model object
file: Output file path (optional, overrides model_dir/solver_dir)
model_dir: Directory where model was saved with save_model() (contains model.rds)
solver_dir: Solver subdirectory name (default: "jump", e.g., "jump-highs")
output_name: Output filename (default: "model.jl")
model_name: Name for the JuMP model variable (default: "model")
optimizer: Julia optimizer to use (e.g., "HiGHS.Optimizer", "GLPK.Optimizer")
optimizer_attributes: Named list of optimizer attributes to set (e.g., list(solver = "ipm", presolve = "on", parallel = "on")). For HiGHS: solver can be "simplex", "ipm" (interior point/barrier), or "choose" (default). "ipm" is usually faster than simplex on large models. If NULL, uses solver defaults with automatic barrier method selection for large models (>1M constraints).
use_folded: logical; whether to use folded equations if available (default: TRUE)
cleanup: logical; whether to remove existing solver_dir before generating new files (default: TRUE)
export_data: logical; whether to export parameter data to CSV files in data_export/ directory (default: FALSE)
export_vars: logical; whether to export variable values to CSV files in vars_export/ directory after solve (default: FALSE). Note: Solution is always saved to solution/ directory regardless of this setting (required for load_results).
export_lp: logical or character; if TRUE, exports model to LP format; if character, specifies filename (default: FALSE)
export_mps: logical or character; if TRUE, exports model to MPS format; if character, specifies filename (default: FALSE)
data_mode: character; "external" (default) generates a data.jl that reads parameter data from files at run time, "embedded" writes the data into the Julia source
...: Additional arguments (not used)

Value

Character vector of Julia code, or writes file if path is determined

Examples

m <- example_models$energyRt$multimod

# Explicit file path
write_jump(m, file = tempfile(fileext = ".jl"))

# Into a saved model workspace: <model_dir>/solvers/<solver_dir>/model.jl
ws <- file.path(tempdir(), "energyRt_ws")
save_model(m, ws, verbose = FALSE)
write_jump(m, model_dir = ws)
write_jump(m, model_dir = ws, solver_dir = "jump-highs")

# Barrier method for large models
write_jump(m, model_dir = ws,
           optimizer_attributes = list(solver = "ipm", parallel = "on"))
#> Removing existing solver directory: C:\Users\admin\AppData\Local\Temp\RtmpgRRJWR/energyRt_ws/solvers/jump