Solve a Pyomo model exported by multimod — solve

Runs the generated model.py (or solve.py, written by write_pyomo(split = TRUE)) in solvers/pyomo/ with the configured Python interpreter, and returns basic status information together with the status the script writes to solution/status.json. With load_results = TRUE the variable values the script exports are read back into model.

Usage

solve_pyomo(
  model = NULL,
  model_dir,
  solver_dir = "pyomo",
  env_name = "multimod",
  ensure_env = FALSE,
  packages = NULL,
  python_version = NULL,
  ensure_miniconda = TRUE,
  reinstall_env = FALSE,
  python = NULL,
  solver = NULL,
  env = NULL,
  load_results = FALSE,
  verbose = c("buffer", "stream", "quiet"),
  timeout = NULL,
  ...
)

Arguments

model: Optional multimod model. Required when load_results = TRUE: the solution of each variable is stored in model$variables[[name]]$solution.
model_dir: Path to the saved model directory created by save_model(). Must contain solvers/<solver_dir>/model.py (generated via write_pyomo()).
solver_dir: Name of the solver sub-directory inside model_dir/solvers. Defaults to "pyomo" to match write_pyomo().
env_name: Name of the Conda environment handled by mm_setup_python(). Defaults to "multimod".
ensure_env: Logical; when TRUE the environment is checked with mm_check_python() and mm_setup_python() runs if it is not usable. Default FALSE: the configured interpreter is used as it is, and nothing is installed.
packages: Character vector of Python packages to request when the helper runs. NULL (default) uses mm_setup_python()'s own list, so the environment contents have a single owner.
python_version: Optional Python version passed through to mm_setup_python().
ensure_miniconda: Forwarded to mm_setup_python().
reinstall_env: When TRUE, asks the helper to recreate the Conda environment before solving.
python: Optional path to a Python executable. When omitted the Python binary reported by reticulate::py_config() for the activated environment is used.
solver: Optional solver name that is exported to the Python process via the PYOMO_SOLVER environment variable.
env: Named character vector of additional environment variables to pass to the Python process.
load_results: Logical; when TRUE, read solution/variables.json into model and return the model as result$model. The file is written by the generated script only when the scaffold was produced with write_pyomo(export_vars = TRUE).
verbose: Controls solver logging capture. Logical (TRUE = buffer output, FALSE = quiet) or one of "buffer", "stream", "quiet".
timeout: Optional timeout (in seconds). Requires the processx package.
...: Additional arguments forwarded to mm_setup_python().

Value

A list describing the solver run (exit code, success flag, elapsed time, raw output, parsed status metadata when available, and environment details). With load_results = TRUE, also model, the input model with a solution data frame on every variable the solver reported. objective is the objective value the script records in status.json.

Details

The Python environment is created once with mm_setup_python(). With ensure_env = TRUE, solve_pyomo() checks the environment and creates it when it is not usable, which may install Miniconda and Python packages.