Evaluate kernel on one subset

For maximum robustness, this prefers the scalar eval() path when available and otherwise falls back to eval_batch() for a single-row batch.

Usage

score_eval(score, idxs, data, ctx = NULL, rng = NULL)

Arguments

score

ustat_score

idxs

integer vector of length s

data

data.frame/matrix

ctx

optional context

rng

optional RNG key/seed

Value

numeric(1)

Examples

sc <- as_score(function(idxs, data) sum(data[idxs]), s = 2L)
score_eval(sc, c(1L, 2L), c(3, 4, 5), NULL)
#> [1] 7