Fast Deviance: Numerical Equality Checks

We validate that the fast deviance evaluator matches the generic gamlss.dist density route (up to floating-point tolerance) across common families.

What you’ll learn

  • How to call check_fast_vs_generic() on fitted gamlss models and read the returned diagnostics.
  • How to iterate over multiple families/generators to create a validation table.
  • How to attempt the wide sweep helper for dozens of families while skipping unsupported ones gracefully.
library(gamlss)
library(SelectBoost.gamlss)

set.seed(2025)
n <- 4000
families <- list(
  NO = gamlss.dist::NO(),
  PO = gamlss.dist::PO(),
  LOGNO = gamlss.dist::LOGNO(),
  GA = gamlss.dist::GA(),
  IG = gamlss.dist::IG(),
  LO = gamlss.dist::LO(),
  LOGITNO = gamlss.dist::LOGITNO(),
  GEOM = gamlss.dist::GEOM(),
  BE = gamlss.dist::BE()
)
gen_fun <- list(
  NO = function(n) gamlss.dist::rNO(n, mu = 0, sigma = 1),
  PO = function(n) gamlss.dist::rPO(n, mu = 2.5),
  LOGNO = function(n) gamlss.dist::rLOGNO(n, mu = 0, sigma = 0.6),
  GA = function(n) gamlss.dist::rGA(n, mu = 2, sigma = 0.5),
  IG = function(n) gamlss.dist::rIG(n, mu = 2, sigma = 0.5),
  LO = function(n) gamlss.dist::rLO(n, mu = 0, sigma = 1),
  LOGITNO = function(n) gamlss.dist::rLOGITNO(n, mu = 0, sigma = 1),
  GEOM = function(n) gamlss.dist::rGEOM(n, mu = 3),
  BE = function(n) gamlss.dist::rBE(n, mu = 0.4, sigma = 0.2)
)

summ <- lapply(names(families), function(fname) {
  fam <- families[[fname]]; gen <- gen_fun[[fname]]
  y <- gen(n); dat <- data.frame(y = y)
  fit <- try(gamlss::gamlss(y ~ 1, data = dat, family = fam), silent = TRUE)
  if (inherits(fit, "try-error")) return(NULL)
  chk <- check_fast_vs_generic(fit, dat, tol = 1e-6)
  data.frame(family = fname, ll_fast = chk$ll_fast, ll_generic = chk$ll_generic,
             abs_diff = chk$abs_diff, pass = chk$pass)
})
summ <- do.call(rbind, Filter(Negate(is.null), summ))
summ

Wide family sweep (auto)

Below we attempt an automatic sweep across many families; those without installed densities or generators will be skipped gracefully.

fams <- c("NO","PO","LOGNO","GA","IG","LO","LOGITNO","GEOM","BE",
          "NBI","NBII","LOGLOG","DEL","ZAGA","ZIP","ZINBI","DPO","GPO",
          "ZAIG","ZALG","BCT","BCPE","ZIPF","ZIPFmu","SICHEL","GLG",
          "BETA4","RS","WEI","GIG")
n <- 3000
res <- lapply(fams, function(fam) {
  y <- try(.gen_family(fam, n), silent = TRUE)
  if (inherits(y, "try-error") || is.null(y)) return(NULL)
  dat <- data.frame(y = y)
  fam_obj <- try(getFromNamespace(fam, "gamlss.dist")(), silent = TRUE)
  if (inherits(fam_obj, "try-error")) return(NULL)
  fit <- try(gamlss::gamlss(y ~ 1, data = dat, family = fam_obj), silent = TRUE)
  if (inherits(fit, "try-error")) return(NULL)
  chk <- check_fast_vs_generic(fit, dat, tol = 1e-6)
  data.frame(family = fam, abs_diff = chk$abs_diff, pass = chk$pass)
})
res <- do.call(rbind, Filter(Negate(is.null), res))
res

Wide family sweep with skip reasons

fams <- c("NO","PO","LOGNO","GA","IG","LO","LOGITNO","GEOM","BE",
          "NBI","NBII","LOGLOG","DEL","ZAGA","ZIP","ZINBI","DPO","GPO",
          "ZAIG","ZALG","BCT","BCPE","ZIPF","ZIPFmu","SICHEL","GLG",
          "BETA4","RS","WEI","GIG")
n <- 3000
tols <- SelectBoost.gamlss:::.family_tolerance()
tol_default <- tols[['.default']]
rows <- lapply(fams, function(fam) {
  # try generator
  y <- try(.gen_family(fam, n), silent = TRUE)
  if (inherits(y, "try-error") || is.null(y)) {
    return(data.frame(family=fam, status="skip", reason="generator missing/failed", abs_diff=NA_real_, pass=NA))
  }
  dat <- data.frame(y = y)
  # try family object
  fam_obj <- try(getFromNamespace(fam, "gamlss.dist")(), silent = TRUE)
  if (inherits(fam_obj, "try-error")) {
    return(data.frame(family=fam, status="skip", reason="family constructor missing", abs_diff=NA_real_, pass=NA))
  }
  # try fit
  fit <- try(gamlss::gamlss(y ~ 1, data = dat, family = fam_obj), silent = TRUE)
  if (inherits(fit, "try-error")) {
    return(data.frame(family=fam, status="skip", reason="fit failed", abs_diff=NA_real_, pass=NA))
  }
  # evaluate
  tol_fam <- tols[[fam]] %||% tol_default
  chk <- check_fast_vs_generic(fit, dat, tol = tol_fam)
  data.frame(family=fam, status="ok", reason="", abs_diff=chk$abs_diff, pass=chk$pass)
})
res <- do.call(rbind, rows)
res