radix_id: Mixed-radix codes and group indices for row combinations

In xfun: Supporting Functions for Packages Maintained by 'Yihui Xie'

Mixed-radix codes and group indices for row combinations

Description

radix_id() encodes each row as a numeric code via a mixed-radix key: each vector contributes one digit in base nlevels + 1, with NA mapped to 0 (its own group). Equal combinations share a code; use !duplicated(radix_id(...)) to flag the first occurrence of each. Faster than paste() — no string allocation.

Usage

radix_id(...)

radix_group(...)

Arguments

...	Equal-length vectors (factors, or coercible via factor()).

Details

radix_group() returns match(radix_id(...), unique(radix_id(...))): a 1-based group index in first-appearance order. Use radix_id() when only !duplicated() is needed; radix_group() when the index itself matters.

Value

radix_id(): numeric vector; only equality comparisons meaningful. radix_group(): integer vector in 1:k (k = distinct combinations).

Examples

# radix_id(): equal combinations share a code; !duplicated() flags first occurrences
k = radix_id(c('b', 'a', 'b', 'a', 'c'))
!duplicated(k)  # TRUE FALSE TRUE FALSE TRUE

# two vectors: combinations are matched row-wise
k2 = radix_id(c(1, 1, 2), c('x', 'x', 'y'))
k2[1] == k2[2]  # TRUE  (same combination)
k2[1] == k2[3]  # FALSE (different combination)

# radix_group(): 1-based group index in first-appearance order
radix_group(c("b", "a", "b", "a", "c"))  # 1 2 1 2 3

# expand a per-group result back to the full row order
x = c(3, 1, 3, 2, 1)
gid = radix_group(x)
u = !duplicated(gid)
per_group = x[u]^2  # 9 1 4 (one value per distinct group)
per_group[gid]      # 9 1 9 4 1 (expanded back)

xfun documentation built on Oct. 16, 2026, 9:06 a.m.