Build Candidate Support Profiles for SSRHE Adaptive-Radius Tuning

Constructs a compact grid of adaptive-radius support profiles for use with support.selection = "cv" in SSRHE fitting functions. Each row gives an adaptive.k.scale value and a requested min.support. The grid is intentionally small by default because support selection nests operator construction inside response cross-validation.

Usage

ssrhe.support.grid(
  n,
  tangent.dim,
  derivative.order = 2L,
  support.buffer = 2L,
  max.candidates = 8L
)

Arguments

n

Number of observations.

tangent.dim

Tangent dimension used by the SSRHE local polynomial design.

derivative.order

SSRHE derivative order, currently 2L or 3L.

support.buffer

Nonnegative integer added to the local design size when constructing the smallest candidate support.

max.candidates

Maximum number of candidate support profiles to return.

Value

A data frame with columns adaptive.k.scale, min.support, and max.support. max.support is NA by default, meaning no truncation.

Examples

ssrhe.support.grid(n = 50L, tangent.dim = 2L, max.candidates = 4L)
#>   adaptive.k.scale min.support max.support
#> 1                2           7          NA
#> 2                4          12          NA
#> 3                5          17          NA
#> 4                9          28          NA