A complete optimization model as plain data

The tables keyed by index tuples are lists of entries rather than named lists, because an index tuple is not a string: params maps a table name to a list of list(key = <index tuple>, value = <number>) entries, indexed_sets maps a name to list(key = ..., value = <element list>) fibres, and fix is a list of list(var = , index = , value = ) pins. Entry order does not matter; encoding sorts them canonically.

Usage

program(
  sets = list(),
  indexed_sets = list(),
  params = list(),
  vars = list(),
  objective = NULL,
  sense = "min",
  constraints = list(),
  fix = list(),
  scenarios = 1,
  scenario_seed = 1,
  sources = list()
)

Arguments

sets: A list of index_set()s.
indexed_sets: Dependent sets carried as data (see above).
params: Named parameter tables (see above).
vars: A list of var_decl()s.
objective: The objective expression node.
sense: "min" or "max".
constraints: A list of constraint()s.
fix: Per-index variable pins (see above).
scenarios: How many scenarios are drawn (at least 1).
scenario_seed: The seed they are drawn from (at least 1).
sources: Named parametric() / empirical() declarations.

Details

A model under uncertainty adds three more: the random variables it draws (sources, a named list of parametric() / empirical() declarations), how many scenarios are drawn and the seed they are drawn from. The last two are model data — they pin the sampled instance, so the same program always sees the same draws. Left at their defaults they say nothing, and the encoded bytes are those of a deterministic model.