Fast Strauss simulator (Rcpp backend) returning an sf point layer — simulate_points_strauss

Fixed-\(n\) Metropolis–Hastings sampler for the Strauss process with interaction radius r and parameter gamma (\(0<\gamma\le 1\)). Uses a cell grid, cached neighbor counts, squared-distance checks, and light auto-tuning for the proposal radius. This is a fast, self-contained alternative to spatstat for baseline point generation.

Usage

simulate_points_strauss_fast(
  domain,
  n_target,
  r,
  gamma,
  sweeps = 2000,
  burnin = 200,
  thin = 1,
  step0 = NA_real_,
  target_acc = 0.35,
  tune_every = 200
)

Arguments

domain: An sf polygon/multipolygon defining the window.
n_target: Integer; number of points to return.
r: Interaction radius (map units).
gamma: Interaction parameter (\(\le 1\) for inhibition).
sweeps: Number of MH sweeps (proposals per point). Default 2000.
burnin: Number of initial sweeps reserved for adaptation. Default 200.
thin: Currently unused (kept for API symmetry). Default 1.
step0: Optional initial proposal radius (defaults to 0.5 * r).
target_acc: Target acceptance for auto-tuning. Default 0.35.
tune_every: Sweeps between tuning updates. Default 200.

Value

An sf POINT layer with n_target rows.