Internal AR(1) evaluation, for the gates

Internal AR(1) evaluation, for the gates

Usage

eval_internal_ar1(
  formula,
  data,
  pedigree = NULL,
  subject,
  time,
  theta,
  missing_code = NULL,
  with_dense = NULL,
  metafounders = NULL,
  gamma = NULL
)

Arguments

formula

the same one as in model_ar1()

data

data.frame

pedigree

data.frame or NULL

subject

subject column

time

time column

theta

components at which to evaluate (groups, s2e, rho)

missing_code

missing-value code or NULL

with_dense

the dense V form, the reference of the small gates, O(n^3) in the n observed cells: NULL (the default) computes it up to 2000 cells and returns neg2logl_V = NA above; TRUE computes it up to 6000 cells and above stops before allocating, with the size; FALSE skips it. The sparse evaluation (neg2logl, the score) is always computed, on the time axis of model_ar1() (time_step). This default differs from eval_internal() and eval_internal_mt(), where with_dense = TRUE computes the dense form with no size limit. Above 3000 fixed-effect columns a message gives the size first, as in model_ar1(): the rank of the fixed design is found on a dense triangular factor, O(p^3)

metafounders

as in model()

gamma

as in model()