Calibrate a global covariance scale to a target MixSim BarOmega — scale_covariances_to

Binary-searches \(s>0\) so that compute_average_overlap() of \(\{s\Sigma_j\}\) matches target. Larger \(s\) inflates the Gaussians and increases average overlap while leaving graph zeros intact. This is the graph-constrained analogue of MixSim / FSDA's average-overlap generator, which cannot take a precision support as input.

Usage

scale_covariances_to_overlap(
  mu,
  sigma,
  p,
  target,
  n_mc = 4000L,
  s_range = c(0.001, 1000),
  tol = 0.005,
  seed = NULL,
  verbose = FALSE
)

Arguments

mu: \(G\times J\) mean signature.
sigma: Base \(G\times G\times J\) covariances.
p: Simplex weights (length \(J\)).
target: Target average pairwise overlap in \((0,1)\).
n_mc: Monte Carlo size forwarded to overlap_gaussian_mc() when \(G\ge 4\).
s_range: Length-2 search interval for \(s\).
tol: Absolute overlap tolerance.
seed: Optional seed for overlap evaluations.
verbose: If TRUE, print the MixSim-to-MC switch.

Value

A list with sigma, Theta, scale, baromega, and target.