library("mvtnorm") set.seed(29) chk <- function(...) stopifnot(isTRUE(all.equal(..., check.attributes = FALSE))) ### N samples with N different covariance matrices N <- 10 J <- 5 prm <- runif(N * J * (J + 1) / 2) + 1 m <- matrix(rnorm(N * J), nrow = J) Z <- matrix(rnorm(N * J), ncol = N) W <- matrix(runif((J - 1) * 100), nrow = J - 1) thischeck <- expression({ lt <- ltMatrices(matrix(prm[1: (N * J * (J + c(-1, 1)[dg + 1L]) / 2)], ncol = N), diag = dg) lt <- ltMatrices(lt, diag = dg, byrow = br) d <- Mult(lt, m) Y <- solve(lt, Z) + m ll1 <- sum(dnorm(Mult(lt, Y - m), log = TRUE)) + sum(log(diagonals(lt))) S <- as.array(Tcrossprod(solve(lt))) ll2 <- sum(l2 <- sapply(1:N, function(i) mvtnorm:::dmvnorm(x = Y[,i], mean = m[,i], sigma = S[,,i], log = TRUE))) chk(ll1, ll2) l3 <- ldmvnorm(obs = Y, mean = m, invchol = lt, logLik = FALSE) l4 <- ldmvnorm(obs = Y, mean = m, chol = solve(lt), logLik = FALSE) chk(l2, l3) chk(l2, l4) l3d <- ldmvnorm(obs = Y, invcholmean = d, invchol = lt, logLik = FALSE) l4d <- ldmvnorm(obs = Y, invcholmean = d, chol = solve(lt), logLik = FALSE) chk(l3, l3d) chk(l4, l4d) ll1 <- sum(dnorm(Mult(lt[1,], Y - m[,1]), log = TRUE)) + N * sum(log(diagonals(lt[1,]))) S <- as.array(Tcrossprod(solve(lt))) ll2 <- sum(l2 <- sapply(1:N, function(i) mvtnorm:::dmvnorm(x = Y[,i], mean = m[,1], sigma = S[,,1], log = TRUE))) chk(ll1, ll2) l3 <- ldmvnorm(obs = Y, mean = m[,1], invchol = lt[1,], logLik = FALSE) l4 <- ldmvnorm(obs = Y, mean = m[,1], chol = solve(lt[1,]), logLik = FALSE) chk(l2, l3) chk(l2, l4) l3d <- ldmvnorm(obs = Y, invcholmean = d[,1], invchol = lt[1,], logLik = FALSE) l4d <- ldmvnorm(obs = Y, invcholmean = d[,1], chol = solve(lt[1,]), logLik = FALSE) chk(l3, l3d) chk(l4, l4d) ### check scores if (require("numDeriv", quietly = TRUE)) { f <- function(L) { L <- ltMatrices(L, diag = dg, byrow = br) ldmvnorm(obs = Y, mean = m, invchol = L) } s0 <- grad(f, unclass(lt)) s1 <- sldmvnorm(obs = Y, mean = m, invchol = lt) chk(Lower_tri(ltMatrices(matrix(s0, ncol = N), diag = dg, byrow = br), diag = dg), Lower_tri(s1$invchol, diag = dg)) f <- function(L) { L <- ltMatrices(L, diag = dg, byrow = br) ldmvnorm(obs = Y, invcholmean = d, invchol = L) } s0 <- grad(f, unclass(lt)) s1 <- sldmvnorm(obs = Y, invcholmean = d, invchol = lt) chk(Lower_tri(ltMatrices(matrix(s0, ncol = N), diag = dg, byrow = br), diag = dg), Lower_tri(s1$invchol, diag = dg)) f <- function(L) { L <- ltMatrices(L, diag = dg, byrow = br) ldmvnorm(obs = Y, mean = m, chol = L) } s0 <- grad(f, unclass(lt)) s1 <- sldmvnorm(obs = Y, mean = m, chol = lt) chk(Lower_tri(ltMatrices(matrix(s0, ncol = N), diag = dg, byrow = br), diag = dg), Lower_tri(s1$chol, diag = dg)) f <- function(L) { L <- ltMatrices(L, diag = dg, byrow = br) ldmvnorm(obs = Y, invcholmean = d, chol = L) } s0 <- grad(f, unclass(lt)) s1 <- sldmvnorm(obs = Y, invcholmean = d, chol = lt) chk(Lower_tri(ltMatrices(matrix(s0, ncol = N), diag = dg, byrow = br), diag = dg), Lower_tri(s1$chol, diag = dg)) f <- function(x) ldmvnorm(obs = x, mean = m, invchol = lt) s0 <- grad(f, Y) s1 <- sldmvnorm(obs = Y, mean = m, invchol = lt) chk(matrix(s0, ncol = N), s1$obs) f <- function(x) ldmvnorm(obs = x, invcholmean = d, invchol = lt) s0 <- grad(f, Y) s1 <- sldmvnorm(obs = Y, invcholmean = d, invchol = lt) chk(matrix(s0, ncol = N), s1$obs) f <- function(m) ldmvnorm(obs = Y, mean = m, invchol = lt) s0 <- grad(f, m) s1 <- sldmvnorm(obs = Y, mean = m, invchol = lt) chk(matrix(s0, ncol = N), -s1$obs) f <- function(d) ldmvnorm(obs = Y, invcholmean = d, invchol = lt) s0 <- grad(f, d) s1 <- sldmvnorm(obs = Y, invcholmean = d, invchol = lt) chk(matrix(s0, ncol = N), s1$invcholmean) } }) dg <- TRUE br <- FALSE eval(thischeck) dg <- FALSE br <- FALSE eval(thischeck) dg <- FALSE br <- TRUE eval(thischeck) dg <- FALSE br <- FALSE eval(thischeck)