Assemble a model into LP arrays

Usage

model_to_lp(
  model,
  col_index = NULL,
  row_index = NULL,
  on_duplicate = c("error", "last", "first"),
  verbose = FALSE,
  index_names = FALSE,
  on_empty_row = c("user_constraints", "all", "warn", "ignore"),
  chunk_rows = 1000000L
)

Arguments

model: A multimod model with data attached.
col_index, row_index: Optional index tables; built when not supplied.
on_duplicate: Passed to build_triplets(); how to resolve a parameter that stores one index tuple with several distinct values.
verbose: Logical; report progress.
index_names: Logical; carry the readable name column in the index tables. Default FALSE: nothing on the solve/MPS/decode path reads the names, and on a large model they are the biggest single allocation of index building (one interned string per LP row and column).
on_empty_row: What to do about equation rows that carry no coefficients. Such a row reads 0 <op> rhs: it constrains nothing, yet the model still assembles and the solver still reports Optimal, for a problem other than the one intended. "user_constraints" (the default) refuses when a user constraint (an equation named eqCns*, the convention of energyRt models) is empty, since a relaxed user constraint changes the problem, and warns about any other empty rows. "all" refuses on any empty row, "warn" only warns, "ignore" says nothing.
chunk_rows: Passed to build_triplets(): evaluate each equation over at most this many LP rows at a time. It bounds the pre-aggregation intermediate of a sum() (outer rows x inner tuples), which is the transient that dominates peak memory on a large full-year build. Slicing by row is exact – each LP row's terms are independent, and the per-chunk triplets concatenate without renumbering.

Value

A list with A (a Matrix::dgCMatrix), obj, row_lo, row_up, col_lo, col_up, maximum, and the row_index / col_index used. Row bounds encode the constraint sense: == gives equal bounds, <= an upper bound only, >= a lower bound only.

Examples

lp <- model_to_lp(example_models$toy)
dim(lp$A)
#> [1]  9 13
head(lp$row_index)
#>        i   symbol     i1     i2  sense   rhs
#>    <int>   <char> <char> <char> <char> <num>
#> 1:     1 eqCapLim   coal  north     <=    NA
#> 2:     2 eqCapLim    gas  north     <=    NA
#> 3:     3 eqCapLim   wind  north     <=    NA
#> 4:     4 eqCapLim   coal  south     <=    NA
#> 5:     5 eqCapLim    gas  south     <=    NA
#> 6:     6 eqCapLim   wind  south     <=    NA