Sweeps a grid of (n, p, K, signal) configurations and records recovery
angle and timing for each combination.
Usage
control_ladder(
ns = c(20L, 40L),
ps = c(500L, 5000L),
Ks = c(2L, 3L),
signals = c(10, 20, 30, 50),
strategy_name = "hogsvd_averaged",
seed = 42L
)
Arguments
- ns
integer vector of sample sizes
- ps
integer vector of feature counts (genes per omic layer)
- Ks
integer vector of omic-layer counts
- signals
numeric vector of signal singular values to test
- strategy_name
character Decomposer strategy to benchmark
- seed
integer base seed (incremented per config)
Value
A ControlLadderResult data frame with requested configuration,
per-cell seed, status, typed failure class and reason, BBP and
recovery fields, and repeated requested/completed/failed cell totals. The
same totals are available in the accounting attribute.