Dinamica EGO Allocation Methods — alloc

Methods for running Dinamica EGO allocation and evaluating allocation parameters, with two entry points for allocation:

db$alloc_dinamica(id_periods, ...) runs a whole sequence of periods. After each period it commits the allocated map to lulc_data_t and recomputes the neighbour predictors the next period is predicted from. Use it for plain simulations.
alloc_dinamica_one_period(db, id_period_post, ...) allocates one period for the active id_run and returns the map without committing it. Use it to write your own loop: edit trans_pot_t between prediction and allocation (e.g. interventions), edit the allocated map afterwards, or draw an ensemble of single-period realisations. The caller then commits the map with db$commit(x, "lulc_data_t", method = "upsert") and, if a later period follows, runs db$upsert_new_neighbors(id_period_post), in that order.

Both reuse transition potentials already in trans_pot_t for the run and period, and only predict them when none exist (or when force_predict_trans_pot = TRUE). They mirror alloc_clumpy_one_period() and db$alloc_clumpy().

eval_alloc_params_t() runs alloc_dinamica() for every run in alloc_params_t over the observed periods and scores each against the observation with fuzzy similarity.

Usage

alloc_dinamica_one_period(
  db,
  id_period_post,
  select_score,
  select_maximize,
  work_dir = NULL,
  use_parent_trans_pot = FALSE,
  force_predict_trans_pot = FALSE
)

alloc_dinamica(
  self,
  id_periods,
  select_score,
  select_maximize,
  work_dir = NULL,
  keep_intermediate = FALSE,
  use_parent_trans_pot = FALSE,
  force_predict_trans_pot = FALSE,
  update_neighbors = TRUE
)

eval_alloc_params_t(
  self,
  select_score,
  select_maximize,
  work_dir = NULL,
  keep_intermediate = FALSE
)

Arguments

db: An evoland_db instance; uses its active id_run.
id_period_post: Integer posterior period ID.
select_score: Character; mlr3 measure ID for model selection.
select_maximize: Logical; whether to maximise select_score.
work_dir: Character or NULL; directory for Dinamica's input and output files. If NULL (default), a temporary directory is used and removed afterwards; a directory passed explicitly is kept.
use_parent_trans_pot: Logical; if TRUE, predict (or reuse) transition potentials under the parent run, so sibling runs share one set, e.g. for Monte-Carlo ensembles.
force_predict_trans_pot: Logical; if TRUE, recompute transition potentials even if trans_pot_t already holds them for this run and period.
self: An evoland_db instance.
id_periods: Integer vector of contiguous posterior period IDs to simulate; data from the period before the first is used as the anterior state.
keep_intermediate: Logical, whether to keep intermediate files from simulations
update_neighbors: Logical; whether to recompute neighbour predictors after the last requested period (default TRUE). Intermediate periods are always updated, since the next period's prediction reads them. Set FALSE when nothing is allocated onwards from the last period, e.g. for single-period ensembles.

Value

alloc_dinamica_one_period(): an lulc_data_t with the simulated posterior LULC. alloc_dinamica(): called for its side effects on lulc_data_t and pred_data_t. eval_alloc_params_t(): the alloc_params_t rows, with similarity filled in.

Functions

alloc_dinamica_one_period(): Allocate a single period and return the map without committing it; see Description for when to use which.
alloc_dinamica(): Allocate a contiguous sequence of periods, committing each one and recomputing neighbour predictors in between; the method behind db$alloc_dinamica().
eval_alloc_params_t(): Evaluate allocation parameters using fuzzy similarity over different runs.