ledgr_sweep() evaluates a ledgr_param_grid against a ledgr_experiment
without writing candidate runs to the experiment store. Sweep is an
exploratory surface: it returns candidate summaries, does not rank candidates
automatically, and does not create committed run artifacts.
Usage
ledgr_sweep(
exp,
param_grid,
precomputed_features = NULL,
seed = NULL,
stop_on_error = FALSE
)Arguments
- exp
A
ledgr_experiment.- param_grid
A
ledgr_param_grid.- precomputed_features
Optional
ledgr_precomputed_featuresobject.- seed
Optional integer-like master seed. When supplied, each candidate receives a deterministic derived execution seed.
- stop_on_error
Logical. When
FALSE, candidate-level execution errors are captured as failed rows; whenTRUE, they are rethrown.
Details
For larger grids, precompute shared feature payloads with
ledgr_precompute_features(). When a grid has more than 20 combinations and
precomputed_features = NULL, ledgr warns because feature computation may be
repeated per candidate.
The result carries row-level execution_seed and provenance. Provenance
records what ran; it does not prove that parameter selection was
out-of-sample. The normal discipline is to sweep on a train snapshot, select
a candidate with ledgr_candidate(), and evaluate the locked params on a
held-out test snapshot with ledgr_promote() or ledgr_run(). Same-snapshot
promotion is useful for audit and replay, but remains in-sample.
Failed candidates are retained as rows when stop_on_error = FALSE. Contract
errors such as invalid grids, invalid precomputed feature payloads, and Tier 3
strategy preflight failures still abort.
v0.1.8 does not ship automatic ranking, ledgr_tune(), parallel sweep,
walk-forward/PBO/CSCV helpers, risk-layer insertion, public cost-model
factories, paper/live adapters, intraday-specific support, or full sweep
artifact persistence.
Articles
Exploratory sweeps and promotion:
vignette("sweeps", package = "ledgr")
system.file("doc", "sweeps.html", package = "ledgr")