tulpa_grid_log_quad: Per-cell log quadrature weight of an outer grid

In tulpa: Templated Unified Library for Posterior Approximation in Bayesian Hierarchical Models

Per-cell log quadrature weight of an outer grid

Description

The per-cell log quadrature weight (prior mass) of a nested-Laplace outer grid: every path that turns a fit's log_marginal into posterior cell weights goes through this one rule, so the prior mass a cell carries is decided in one place. Rebuilds axis specs from the grid's own columns when specs is not supplied. Intended for reconstructing a fit's outer-grid posterior weights (with tulpa_theta_matrix() and tulpa_normalise_weights_safe()) when the fit doesn't already carry fit$log_quad.

Usage

tulpa_grid_log_quad(
  theta_grid,
  specs = NULL,
  copy_slab = "exponential",
  close_domain = TRUE,
  folded_axes = NULL,
  refining = NULL
)

Arguments

theta_grid	A named ⁠[n_cells x n_axes]⁠ matrix (see tulpa_theta_matrix()).
specs	Optional pre-built per-axis spec list; NULL rebuilds it from theta_grid's columns via tulpa_joint_axis_specs_from_grid().
copy_slab	"exponential" or "flat"; the copy-scale axis's continuum measure (see ?tulpa_joint_axis_specs_from_grid).
close_domain	Whether an unbounded axis's outer cells are closed at the grid's own edge rather than left open to infinity.
folded_axes	Optional names of axes folded onto ⁠[0, Inf)⁠ (e.g. a correlation axis reflected at 0).
refining	Optional refinement tag vector (see tulpa_hyper_slice_home()); when supplied, specs are rebuilt from the grid's declared cells only.

Value

Numeric vector of per-cell log quadrature weights, length nrow(theta_grid), or NULL if theta_grid has no axis names.

See Also

tulpa_theta_matrix(), tulpa_normalise_weights_safe(), tulpa_joint_axis_specs_from_grid()

Examples

set.seed(1)
S <- 30L                                   # spatial units in a chain
nb <- lapply(seq_len(S), function(s) setdiff(c(s - 1L, s + 1L), c(0L, S + 1L)))
nn <- lengths(nb)
field <- as.numeric(scale(cumsum(rnorm(S, 0, 0.4))))
idx <- rep(seq_len(S), each = 6L); n <- length(idx); x <- rnorm(n)
y <- rbinom(n, 1L, plogis(-0.2 + 0.6 * x + field[idx]))
prior <- list(type = "icar", n_spatial_units = S, spatial_idx = idx,
              adj_row_ptr = c(0L, cumsum(nn)), adj_col_idx = unlist(nb) - 1L,
              n_neighbors = nn, tau_grid = c(0.5, 1, 2, 4, 8))
fit <- tulpa_nested_laplace(y, rep(1L, n), cbind(1, x), prior = prior,
                            family = "binomial",
                            control = list(progress = FALSE))
tg <- tulpa_theta_matrix(fit)
lq <- tulpa_grid_log_quad(tg)
# The fit's own cell weights, rebuilt from its log marginals:
tulpa_normalise_weights_safe(fit$log_marginal, log_quad = lq)

tulpa documentation built on Oct. 19, 2026, 5:09 p.m.