Last updated on 2020-02-19 10:48:46 CET.
| Flavor | Version | Tinstall | Tcheck | Ttotal | Status | Flags |
|---|---|---|---|---|---|---|
| r-devel-linux-x86_64-debian-clang | 2.0 | 8.25 | 26.20 | 34.45 | ERROR | |
| r-devel-linux-x86_64-debian-gcc | 2.0 | 3.38 | 20.36 | 23.74 | ERROR | |
| r-devel-linux-x86_64-fedora-clang | 2.0 | 42.48 | ERROR | |||
| r-devel-linux-x86_64-fedora-gcc | 2.0 | 37.72 | ERROR | |||
| r-devel-windows-ix86+x86_64 | 2.0 | 11.00 | 61.00 | 72.00 | NOTE | |
| r-devel-windows-ix86+x86_64-gcc8 | 2.0 | 20.00 | 77.00 | 97.00 | NOTE | |
| r-patched-linux-x86_64 | 2.0 | 4.12 | 28.91 | 33.03 | NOTE | |
| r-patched-solaris-x86 | 2.0 | 64.70 | NOTE | |||
| r-release-linux-x86_64 | 2.0 | 3.92 | 28.82 | 32.74 | NOTE | |
| r-release-windows-ix86+x86_64 | 2.0 | 13.00 | 61.00 | 74.00 | NOTE | |
| r-release-osx-x86_64 | 2.0 | NOTE | ||||
| r-oldrel-windows-ix86+x86_64 | 2.0 | 10.00 | 61.00 | 71.00 | NOTE | |
| r-oldrel-osx-x86_64 | 2.0 | NOTE |
Version: 2.0
Check: DESCRIPTION meta-information
Result: NOTE
BugReports field is not a suitable URL but contains an email address:
use the Contact field instead
Flavors: r-devel-linux-x86_64-debian-clang, r-devel-linux-x86_64-debian-gcc, r-devel-linux-x86_64-fedora-clang, r-devel-linux-x86_64-fedora-gcc, r-devel-windows-ix86+x86_64, r-devel-windows-ix86+x86_64-gcc8, r-patched-linux-x86_64, r-patched-solaris-x86, r-release-linux-x86_64, r-release-windows-ix86+x86_64
Version: 2.0
Check: examples
Result: ERROR
Running examples in 'bayesDccGarch-Ex.R' failed
The error most likely occurred in:
> base::assign(".ptime", proc.time(), pos = "CheckExEnv")
> ### Name: bayesDccGarch-package
> ### Title: bayesDccGARCH: Methods and tools for Bayesian analysis of
> ### DCC-GARCH(1,1) Model.
> ### Aliases: bayesDccGarch-package
> ### Keywords: package multivariate DCC-GARCH GARCH Bayes
>
> ### ** Examples
>
>
> data(DaxCacNik)
> Dax = DaxCacNik[,1]
>
> ### DCC-GARCH(1,1) ###
> out = bayesDccGarch(DaxCacNik[1:10,], nSim=1000) # more data is necessary
Maximizing the log-posterior density function.
Done.
One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.
Calibrating the standard deviations for simulation:
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.29 0.91 0.71 0.86 0.91 0.91 0.65 0.90 0.92 0.92 0.74
phi_12 phi_13 phi_14 phi_15
0.88 0.93 0.73 0.66
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.25 0.86 0.67 0.79 0.87 0.88 0.54 0.84 0.87 0.86 0.61
phi_12 phi_13 phi_14 phi_15
0.84 0.87 0.64 0.54
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.87 0.20 0.66 0.80 0.88 0.23 0.57 0.81 0.87 0.27 0.53
phi_12 phi_13 phi_14 phi_15
0.75 0.84 0.56 0.48
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.82 0.21 0.52 0.70 0.82 0.22 0.49 0.73 0.81 0.24 0.43
phi_12 phi_13 phi_14 phi_15
0.69 0.80 0.45 0.45
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.80 0.20 0.44 0.63 0.78 0.24 0.48 0.59 0.73 0.27 0.45
phi_12 phi_13 phi_14 phi_15
0.62 0.74 0.43 0.51
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.12 0.81 0.41 0.47 0.68 0.86 0.41 0.55 0.65 0.81 0.44
phi_12 phi_13 phi_14 phi_15
0.50 0.63 0.41 0.34
Computing the covariance matrix of pilot sample.
----------- FAILURE REPORT --------------
--- failure: the condition has length > 1 ---
--- srcref ---
:
--- package (from environment) ---
bayesDccGarch
--- call from context ---
bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- call from argument ---
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
} else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
--- R stacktrace ---
where 1: bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- value of length: 2 type: logical ---
[1] TRUE TRUE
--- function from context ---
function (mY, nSim = 10000, tail_ini = 8, omega_ini = rep(0.03,
ncol(mY)), alpha_ini = rep(0.03, ncol(mY)), beta_ini = rep(0.8,
ncol(mY)), a_ini = 0.03, b_ini = 0.8, gamma_ini = rep(1,
ncol(mY)), errorDist = 2, control = list())
{
ptm <- proc.time()
if (missing(mY))
stop("'mY' is missing")
if (any(is.na(mY)))
stop("'mY' contains 'NA' values")
if (errorDist == 3 && tail_ini == 8) {
tail_ini = 2
}
mY = as.matrix(mY)
n = nrow(mY)
k = ncol(mY)
if (k == 1) {
npar = 5
}
else {
npar = 4 * k + 3
}
Lout = try(logLikDccGarch(mY, omega_ini, alpha_ini, beta_ini,
a_ini, b_ini, gamma_ini, tail_ini, errorDist))
if (class(Lout) == "try-error")
stop("The likelihood function can not be computed for the initial values.")
if (!is.list(control))
stop("control must be an object of list class.")
if (is.null(control$data)) {
control$data = mY
}
if (is.null(control$errorDist)) {
control$errorDist = errorDist
}
if (is.null(control$mu_a)) {
control$mu_a = 0
}
else {
if (length(control$mu_a) != 1) {
stop("control$mu_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_b)) {
control$mu_b = 0
}
else {
if (length(control$mu_b) != 1) {
stop("control$mu_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_omega)) {
control$mu_omega = rep(0, k)
}
else {
if (length(control$mu_omega) != k) {
stop("control$mu_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_alpha)) {
control$mu_alpha = rep(0, k)
}
else {
if (length(control$mu_alpha) != k) {
stop("control$mu_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_beta)) {
control$mu_beta = rep(0, k)
}
else {
if (length(control$mu_beta) != k) {
stop("control$mu_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_gamma)) {
control$mu_gamma = rep(0, k)
}
else {
if (length(control$mu_gamma) != k) {
stop("control$mu_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_tail)) {
control$mu_tail = 0
}
else {
if (length(control$mu_tail) != 1) {
stop("control$mu_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_a)) {
control$sigma_a = 10
}
else {
if (length(control$sigma_a) != 1) {
stop("control$sigma_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_b)) {
control$sigma_b = 10
}
else {
if (length(control$sigma_b) != 1) {
stop("control$sigma_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_omega)) {
control$sigma_omega = rep(10, k)
}
else {
if (length(control$sigma_omega) != k) {
stop("control$sigma_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_alpha)) {
control$sigma_alpha = rep(10, k)
}
else {
if (length(control$sigma_alpha) != k) {
stop("control$sigma_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_beta)) {
control$sigma_beta = rep(10, k)
}
else {
if (length(control$sigma_beta) != k) {
stop("control$sigma_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_gamma)) {
control$sigma_gamma = rep(10, k)
}
else {
if (length(control$sigma_gamma) != k) {
stop("control$sigma_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_tail)) {
control$sigma_tail = 1.25
}
else {
if (length(control$sigma_tail) != 1) {
stop("control$sigma_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$print)) {
control$print = 1
}
else {
if (!any(control$print == 0:1))
control$print = 1
}
if (is.null(control$simAlg)) {
control$simAlg = 3
}
else {
if (!any(control$simAlg == 1:3))
control$simAlg = 3
}
if (control$simAlg == 1) {
option = 1
if (is.null(control$cholCov)) {
stop("control$cholCov was not found.")
}
else {
if (!is.matrix(control$cholCov)) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
else {
if (ncol(control$cholCov) != npar || nrow(control$cholCov) !=
npar) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
}
}
}
if (control$simAlg == 2 && !is.null(control$sdSim)) {
option = 2
if (length(control$sdSim) != npar)
stop("control$sdSim must be a numeric positive vector with length 5 if k=1 or 4k+3 if k>1, where 'k=ncol(mY)'")
}
if (control$simAlg == 3) {
if (k == 1) {
control$sdSim = c(0.2, 0.2, 0.4, 0.4, 0.2)
}
else {
control$sdSim = c(0.2, rep(c(0.2, 0.4, 0.4, 0.2),
k), 1, 1.5)
}
}
if (is.null(control$lambda)) {
control$lambda = 0.5
}
if (is.null(control$nPilotSim)) {
control$nPilotSim = 1000
}
phi_ini = .C("original_to_real_scale", phi = numeric(4 *
k + 3), omega = as.double(omega_ini), alpha = as.double(alpha_ini),
beta = as.double(beta_ini), a = as.double(a_ini), b = as.double(b_ini),
gamma = as.double(gamma_ini), tail = as.double(tail_ini),
k = as.integer(k), errorDist = as.integer(errorDist))$phi
memoryAllocation(mY, errorDist, control$mu_omega, control$sigma_omega,
control$mu_alpha, control$sigma_alpha, control$mu_beta,
control$sigma_beta, control$mu_a, control$sigma_a, control$mu_b,
control$sigma_b, control$mu_gamma, control$sigma_gamma,
control$mu_tail, control$sigma_tail, control$print)
if (control$simAlg == 3) {
writeLines("Maximizing the log-posterior density function.")
if (k == 1)
phi_ini = phi_ini[1:5]
opt = optim(par = phi_ini, fn = logPosterior_phi, hessian = TRUE,
control = list(fnscale = -abs(Lout$value)))
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
if (opt$convergence == 0 || opt$convergence == 1) {
writeLines("Done.")
phi_ini = opt$par
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
control$cholCov = try(t(chol(solve(-opt$hessian))),
silent = TRUE)
if (class(control$cholCov) != "try-error") {
option = 1
}
else {
writeLines("One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.")
option = 2
}
}
else {
writeLines("The optim function convergence was not obtained.")
option = 2
}
}
repeat {
if (option == 1 && control$simAlg != 2) {
if (control$simAlg == 3) {
writeLines("Calibrating the Lambda coefficient:")
cont = 1
proceed = TRUE
while (proceed) {
writeLines(paste("lambda:", control$lambda))
MC_phi = MH_oneBlock(phi_ini, k, n_sim = control$nPilotSim,
control$lambda * control$cholCov)
acceptRate = 1 - rejectionRate(MC_phi)[5]
writeLines(paste("Accept Rate:", round(acceptRate,
2)))
if (acceptRate > 0.2 && acceptRate < 0.5 ||
cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
if (acceptRate < 0.2) {
control$lambda = control$lambda * 0.8
}
if (acceptRate > 0.5) {
control$lambda = control$lambda * 1.2
}
}
}
writeLines("Done.")
}
writeLines("Starting the simulation by one-block random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneBlock(phi_ini, k, n_sim = nSim, control$lambda *
control$cholCov)
writeLines("Done.")
break
}
if (option == 2 && control$simAlg != 1) {
if (control$simAlg == 3) {
writeLines("Calibrating the standard deviations for simulation:")
proceed = TRUE
cont = 1
while (proceed) {
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = control$nPilotSim,
control$sdSim)
acceptRate = 1 - rejectionRate(MC_phi)
writeLines("Accept Rate:")
print(round(acceptRate, 2))
if (errorDist == 1) {
acceptRate[1] = 0.3
}
lowerRate = which(acceptRate < 0.15)
upperRate = which(acceptRate > 0.5)
if (length(lowerRate) == 0 && length(upperRate) ==
0 || cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
control$sdSim[lowerRate] = control$sdSim[lowerRate] *
0.6
control$sdSim[upperRate] = control$sdSim[upperRate] *
1.4
}
}
if (errorDist == 1) {
MC_phi[, 1] = rnorm(control$nPilotSim)
}
writeLines("Computing the covariance matrix of pilot sample.")
control$cholCov = try(t(chol(cov(MC_phi))), silent = TRUE)
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
}
else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
}
if (option == 2) {
writeLines("Starting the simulation by one-dimensional random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = nSim,
sd_phi_sim = control$sdSim)
writeLines("Done.")
break
}
}
}
vMMeanH = .C("getMeanH", vMMeanH = numeric(n * k * k))$vMMeanH
mMeanH = matrix(vMMeanH, nrow = n, ncol = k * k, byrow = TRUE)
name = rep("NA", k * k)
for (i in 1:k) {
for (j in 1:k) {
name[(i - 1) * k + j] = paste("H_", i, ",", j, sep = "")
}
}
colnames(mMeanH) = name
logLike_mean = .C("getLogLikelihood_mean", value = numeric(1))$value
memoryDeallocation()
name = rep("NA", npar)
MC = matrix(NA, nrow = nSim, ncol = npar)
if (errorDist == 1 || errorDist == 3) {
MC[, 1] = exp(MC_phi[, 1])
name[1] = "delta"
}
if (errorDist == 2) {
MC[, 1] = exp(MC_phi[, 1]) + 2
name[1] = "nu"
}
for (i in 1:k) {
MC[, 4 * (i - 1) + 2:5] = cbind(exp(MC_phi[, 4 * (i -
1) + 2:3]), exp(MC_phi[, 4 * (i - 1) + 4:5])/(1 +
exp(MC_phi[, 4 * (i - 1) + 4:5])))
name[4 * (i - 1) + 2:5] = paste(c("gamma", "omega", "alpha",
"beta"), "_", i, sep = "")
}
if (k > 1) {
MC[, 4 * k + 2:3] = exp(MC_phi[, 4 * k + 2:3])/(1 + exp(MC_phi[,
4 * k + 2:3]))
name[4 * k + 2:3] = c("a", "b")
}
colnames(MC) = name
D_mean = -2 * logLike_mean
par_est = apply(X = MC, MARGIN = 2, FUN = mean)
tail_est = par_est[1]
gamma_est = par_est[2 + 4 * (0:(k - 1))]
omega_est = par_est[3 + 4 * (0:(k - 1))]
alpha_est = par_est[4 + 4 * (0:(k - 1))]
beta_est = par_est[5 + 4 * (0:(k - 1))]
if (k > 1) {
a_est = par_est[2 + 4 * k]
b_est = par_est[3 + 4 * k]
}
else {
a_est = 0.1
b_est = 0.1
}
D_at_mean = -2 * logLikDccGarch(mY, omega_est, alpha_est,
beta_est, a_est, b_est, gamma_est, tail_est, errorDist)$value
if (errorDist == 1) {
EAIC = D_mean + 2 * (npar - 1)
EBIC = D_mean + (npar - 1) * log(n)
}
else {
EAIC = D_mean + 2 * npar
EBIC = D_mean + npar * log(n)
}
DIC = D_at_mean + 2 * (D_mean - D_at_mean)
IC = matrix(c(EAIC, EBIC, DIC), nrow = 3, ncol = 1)
rownames(IC) = c("EAIC", "EBIC", "DIC")
colnames(IC) = c("Estimative")
out = list()
out$control = control
out$MC = as.mcmc(MC)
out$H = mMeanH
out$IC = IC
out$elapsedTime = proc.time() - ptm
return(structure(out, class = "bayesDccGarch"))
}
<bytecode: 0x376add0>
<environment: namespace:bayesDccGarch>
--- function search by body ---
Function bayesDccGarch in namespace bayesDccGarch has this body.
----------- END OF FAILURE REPORT --------------
Error in if (class(control$cholCov) != "try-error") { :
the condition has length > 1
Calls: bayesDccGarch
Execution halted
Flavor: r-devel-linux-x86_64-debian-clang
Version: 2.0
Check: examples
Result: ERROR
Running examples in ‘bayesDccGarch-Ex.R’ failed
The error most likely occurred in:
> base::assign(".ptime", proc.time(), pos = "CheckExEnv")
> ### Name: bayesDccGarch-package
> ### Title: bayesDccGARCH: Methods and tools for Bayesian analysis of
> ### DCC-GARCH(1,1) Model.
> ### Aliases: bayesDccGarch-package
> ### Keywords: package multivariate DCC-GARCH GARCH Bayes
>
> ### ** Examples
>
>
> data(DaxCacNik)
> Dax = DaxCacNik[,1]
>
> ### DCC-GARCH(1,1) ###
> out = bayesDccGarch(DaxCacNik[1:10,], nSim=1000) # more data is necessary
Maximizing the log-posterior density function.
Done.
One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.
Calibrating the standard deviations for simulation:
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.29 0.91 0.71 0.86 0.91 0.91 0.65 0.90 0.92 0.92 0.74
phi_12 phi_13 phi_14 phi_15
0.88 0.93 0.73 0.66
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.25 0.86 0.67 0.79 0.87 0.88 0.54 0.84 0.87 0.86 0.61
phi_12 phi_13 phi_14 phi_15
0.84 0.87 0.64 0.54
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.87 0.20 0.66 0.80 0.88 0.23 0.57 0.81 0.87 0.27 0.53
phi_12 phi_13 phi_14 phi_15
0.75 0.84 0.56 0.48
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.82 0.21 0.52 0.70 0.82 0.22 0.49 0.73 0.81 0.24 0.43
phi_12 phi_13 phi_14 phi_15
0.69 0.80 0.45 0.45
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.80 0.20 0.44 0.63 0.78 0.24 0.48 0.59 0.73 0.27 0.45
phi_12 phi_13 phi_14 phi_15
0.62 0.74 0.43 0.51
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.12 0.81 0.41 0.47 0.68 0.86 0.41 0.55 0.65 0.81 0.44
phi_12 phi_13 phi_14 phi_15
0.50 0.63 0.41 0.34
Computing the covariance matrix of pilot sample.
----------- FAILURE REPORT --------------
--- failure: the condition has length > 1 ---
--- srcref ---
:
--- package (from environment) ---
bayesDccGarch
--- call from context ---
bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- call from argument ---
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
} else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
--- R stacktrace ---
where 1: bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- value of length: 2 type: logical ---
[1] TRUE TRUE
--- function from context ---
function (mY, nSim = 10000, tail_ini = 8, omega_ini = rep(0.03,
ncol(mY)), alpha_ini = rep(0.03, ncol(mY)), beta_ini = rep(0.8,
ncol(mY)), a_ini = 0.03, b_ini = 0.8, gamma_ini = rep(1,
ncol(mY)), errorDist = 2, control = list())
{
ptm <- proc.time()
if (missing(mY))
stop("'mY' is missing")
if (any(is.na(mY)))
stop("'mY' contains 'NA' values")
if (errorDist == 3 && tail_ini == 8) {
tail_ini = 2
}
mY = as.matrix(mY)
n = nrow(mY)
k = ncol(mY)
if (k == 1) {
npar = 5
}
else {
npar = 4 * k + 3
}
Lout = try(logLikDccGarch(mY, omega_ini, alpha_ini, beta_ini,
a_ini, b_ini, gamma_ini, tail_ini, errorDist))
if (class(Lout) == "try-error")
stop("The likelihood function can not be computed for the initial values.")
if (!is.list(control))
stop("control must be an object of list class.")
if (is.null(control$data)) {
control$data = mY
}
if (is.null(control$errorDist)) {
control$errorDist = errorDist
}
if (is.null(control$mu_a)) {
control$mu_a = 0
}
else {
if (length(control$mu_a) != 1) {
stop("control$mu_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_b)) {
control$mu_b = 0
}
else {
if (length(control$mu_b) != 1) {
stop("control$mu_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_omega)) {
control$mu_omega = rep(0, k)
}
else {
if (length(control$mu_omega) != k) {
stop("control$mu_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_alpha)) {
control$mu_alpha = rep(0, k)
}
else {
if (length(control$mu_alpha) != k) {
stop("control$mu_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_beta)) {
control$mu_beta = rep(0, k)
}
else {
if (length(control$mu_beta) != k) {
stop("control$mu_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_gamma)) {
control$mu_gamma = rep(0, k)
}
else {
if (length(control$mu_gamma) != k) {
stop("control$mu_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_tail)) {
control$mu_tail = 0
}
else {
if (length(control$mu_tail) != 1) {
stop("control$mu_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_a)) {
control$sigma_a = 10
}
else {
if (length(control$sigma_a) != 1) {
stop("control$sigma_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_b)) {
control$sigma_b = 10
}
else {
if (length(control$sigma_b) != 1) {
stop("control$sigma_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_omega)) {
control$sigma_omega = rep(10, k)
}
else {
if (length(control$sigma_omega) != k) {
stop("control$sigma_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_alpha)) {
control$sigma_alpha = rep(10, k)
}
else {
if (length(control$sigma_alpha) != k) {
stop("control$sigma_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_beta)) {
control$sigma_beta = rep(10, k)
}
else {
if (length(control$sigma_beta) != k) {
stop("control$sigma_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_gamma)) {
control$sigma_gamma = rep(10, k)
}
else {
if (length(control$sigma_gamma) != k) {
stop("control$sigma_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_tail)) {
control$sigma_tail = 1.25
}
else {
if (length(control$sigma_tail) != 1) {
stop("control$sigma_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$print)) {
control$print = 1
}
else {
if (!any(control$print == 0:1))
control$print = 1
}
if (is.null(control$simAlg)) {
control$simAlg = 3
}
else {
if (!any(control$simAlg == 1:3))
control$simAlg = 3
}
if (control$simAlg == 1) {
option = 1
if (is.null(control$cholCov)) {
stop("control$cholCov was not found.")
}
else {
if (!is.matrix(control$cholCov)) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
else {
if (ncol(control$cholCov) != npar || nrow(control$cholCov) !=
npar) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
}
}
}
if (control$simAlg == 2 && !is.null(control$sdSim)) {
option = 2
if (length(control$sdSim) != npar)
stop("control$sdSim must be a numeric positive vector with length 5 if k=1 or 4k+3 if k>1, where 'k=ncol(mY)'")
}
if (control$simAlg == 3) {
if (k == 1) {
control$sdSim = c(0.2, 0.2, 0.4, 0.4, 0.2)
}
else {
control$sdSim = c(0.2, rep(c(0.2, 0.4, 0.4, 0.2),
k), 1, 1.5)
}
}
if (is.null(control$lambda)) {
control$lambda = 0.5
}
if (is.null(control$nPilotSim)) {
control$nPilotSim = 1000
}
phi_ini = .C("original_to_real_scale", phi = numeric(4 *
k + 3), omega = as.double(omega_ini), alpha = as.double(alpha_ini),
beta = as.double(beta_ini), a = as.double(a_ini), b = as.double(b_ini),
gamma = as.double(gamma_ini), tail = as.double(tail_ini),
k = as.integer(k), errorDist = as.integer(errorDist))$phi
memoryAllocation(mY, errorDist, control$mu_omega, control$sigma_omega,
control$mu_alpha, control$sigma_alpha, control$mu_beta,
control$sigma_beta, control$mu_a, control$sigma_a, control$mu_b,
control$sigma_b, control$mu_gamma, control$sigma_gamma,
control$mu_tail, control$sigma_tail, control$print)
if (control$simAlg == 3) {
writeLines("Maximizing the log-posterior density function.")
if (k == 1)
phi_ini = phi_ini[1:5]
opt = optim(par = phi_ini, fn = logPosterior_phi, hessian = TRUE,
control = list(fnscale = -abs(Lout$value)))
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
if (opt$convergence == 0 || opt$convergence == 1) {
writeLines("Done.")
phi_ini = opt$par
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
control$cholCov = try(t(chol(solve(-opt$hessian))),
silent = TRUE)
if (class(control$cholCov) != "try-error") {
option = 1
}
else {
writeLines("One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.")
option = 2
}
}
else {
writeLines("The optim function convergence was not obtained.")
option = 2
}
}
repeat {
if (option == 1 && control$simAlg != 2) {
if (control$simAlg == 3) {
writeLines("Calibrating the Lambda coefficient:")
cont = 1
proceed = TRUE
while (proceed) {
writeLines(paste("lambda:", control$lambda))
MC_phi = MH_oneBlock(phi_ini, k, n_sim = control$nPilotSim,
control$lambda * control$cholCov)
acceptRate = 1 - rejectionRate(MC_phi)[5]
writeLines(paste("Accept Rate:", round(acceptRate,
2)))
if (acceptRate > 0.2 && acceptRate < 0.5 ||
cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
if (acceptRate < 0.2) {
control$lambda = control$lambda * 0.8
}
if (acceptRate > 0.5) {
control$lambda = control$lambda * 1.2
}
}
}
writeLines("Done.")
}
writeLines("Starting the simulation by one-block random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneBlock(phi_ini, k, n_sim = nSim, control$lambda *
control$cholCov)
writeLines("Done.")
break
}
if (option == 2 && control$simAlg != 1) {
if (control$simAlg == 3) {
writeLines("Calibrating the standard deviations for simulation:")
proceed = TRUE
cont = 1
while (proceed) {
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = control$nPilotSim,
control$sdSim)
acceptRate = 1 - rejectionRate(MC_phi)
writeLines("Accept Rate:")
print(round(acceptRate, 2))
if (errorDist == 1) {
acceptRate[1] = 0.3
}
lowerRate = which(acceptRate < 0.15)
upperRate = which(acceptRate > 0.5)
if (length(lowerRate) == 0 && length(upperRate) ==
0 || cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
control$sdSim[lowerRate] = control$sdSim[lowerRate] *
0.6
control$sdSim[upperRate] = control$sdSim[upperRate] *
1.4
}
}
if (errorDist == 1) {
MC_phi[, 1] = rnorm(control$nPilotSim)
}
writeLines("Computing the covariance matrix of pilot sample.")
control$cholCov = try(t(chol(cov(MC_phi))), silent = TRUE)
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
}
else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
}
if (option == 2) {
writeLines("Starting the simulation by one-dimensional random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = nSim,
sd_phi_sim = control$sdSim)
writeLines("Done.")
break
}
}
}
vMMeanH = .C("getMeanH", vMMeanH = numeric(n * k * k))$vMMeanH
mMeanH = matrix(vMMeanH, nrow = n, ncol = k * k, byrow = TRUE)
name = rep("NA", k * k)
for (i in 1:k) {
for (j in 1:k) {
name[(i - 1) * k + j] = paste("H_", i, ",", j, sep = "")
}
}
colnames(mMeanH) = name
logLike_mean = .C("getLogLikelihood_mean", value = numeric(1))$value
memoryDeallocation()
name = rep("NA", npar)
MC = matrix(NA, nrow = nSim, ncol = npar)
if (errorDist == 1 || errorDist == 3) {
MC[, 1] = exp(MC_phi[, 1])
name[1] = "delta"
}
if (errorDist == 2) {
MC[, 1] = exp(MC_phi[, 1]) + 2
name[1] = "nu"
}
for (i in 1:k) {
MC[, 4 * (i - 1) + 2:5] = cbind(exp(MC_phi[, 4 * (i -
1) + 2:3]), exp(MC_phi[, 4 * (i - 1) + 4:5])/(1 +
exp(MC_phi[, 4 * (i - 1) + 4:5])))
name[4 * (i - 1) + 2:5] = paste(c("gamma", "omega", "alpha",
"beta"), "_", i, sep = "")
}
if (k > 1) {
MC[, 4 * k + 2:3] = exp(MC_phi[, 4 * k + 2:3])/(1 + exp(MC_phi[,
4 * k + 2:3]))
name[4 * k + 2:3] = c("a", "b")
}
colnames(MC) = name
D_mean = -2 * logLike_mean
par_est = apply(X = MC, MARGIN = 2, FUN = mean)
tail_est = par_est[1]
gamma_est = par_est[2 + 4 * (0:(k - 1))]
omega_est = par_est[3 + 4 * (0:(k - 1))]
alpha_est = par_est[4 + 4 * (0:(k - 1))]
beta_est = par_est[5 + 4 * (0:(k - 1))]
if (k > 1) {
a_est = par_est[2 + 4 * k]
b_est = par_est[3 + 4 * k]
}
else {
a_est = 0.1
b_est = 0.1
}
D_at_mean = -2 * logLikDccGarch(mY, omega_est, alpha_est,
beta_est, a_est, b_est, gamma_est, tail_est, errorDist)$value
if (errorDist == 1) {
EAIC = D_mean + 2 * (npar - 1)
EBIC = D_mean + (npar - 1) * log(n)
}
else {
EAIC = D_mean + 2 * npar
EBIC = D_mean + npar * log(n)
}
DIC = D_at_mean + 2 * (D_mean - D_at_mean)
IC = matrix(c(EAIC, EBIC, DIC), nrow = 3, ncol = 1)
rownames(IC) = c("EAIC", "EBIC", "DIC")
colnames(IC) = c("Estimative")
out = list()
out$control = control
out$MC = as.mcmc(MC)
out$H = mMeanH
out$IC = IC
out$elapsedTime = proc.time() - ptm
return(structure(out, class = "bayesDccGarch"))
}
<bytecode: 0x558a69d3f620>
<environment: namespace:bayesDccGarch>
--- function search by body ---
Function bayesDccGarch in namespace bayesDccGarch has this body.
----------- END OF FAILURE REPORT --------------
Error in if (class(control$cholCov) != "try-error") { :
the condition has length > 1
Calls: bayesDccGarch
Execution halted
Flavor: r-devel-linux-x86_64-debian-gcc
Version: 2.0
Check: compiled code
Result: NOTE
File ‘bayesDccGarch/libs/bayesDccGarch.so’:
Found no calls to: ‘R_registerRoutines’, ‘R_useDynamicSymbols’
It is good practice to register native routines and to disable symbol
search.
See ‘Writing portable packages’ in the ‘Writing R Extensions’ manual.
Flavors: r-devel-linux-x86_64-fedora-clang, r-devel-linux-x86_64-fedora-gcc
Version: 2.0
Check: examples
Result: ERROR
Running examples in ‘bayesDccGarch-Ex.R’ failed
The error most likely occurred in:
> ### Name: bayesDccGarch-package
> ### Title: bayesDccGARCH: Methods and tools for Bayesian analysis of
> ### DCC-GARCH(1,1) Model.
> ### Aliases: bayesDccGarch-package
> ### Keywords: package multivariate DCC-GARCH GARCH Bayes
>
> ### ** Examples
>
>
> data(DaxCacNik)
> Dax = DaxCacNik[,1]
>
> ### DCC-GARCH(1,1) ###
> out = bayesDccGarch(DaxCacNik[1:10,], nSim=1000) # more data is necessary
Maximizing the log-posterior density function.
Done.
One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.
Calibrating the standard deviations for simulation:
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.29 0.91 0.71 0.86 0.91 0.91 0.65 0.90 0.92 0.92 0.74
phi_12 phi_13 phi_14 phi_15
0.88 0.93 0.73 0.66
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.25 0.86 0.67 0.79 0.87 0.88 0.54 0.84 0.87 0.86 0.61
phi_12 phi_13 phi_14 phi_15
0.84 0.87 0.64 0.54
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.87 0.20 0.66 0.80 0.88 0.23 0.57 0.81 0.87 0.27 0.53
phi_12 phi_13 phi_14 phi_15
0.75 0.84 0.56 0.48
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.82 0.21 0.52 0.70 0.82 0.22 0.49 0.73 0.81 0.24 0.43
phi_12 phi_13 phi_14 phi_15
0.69 0.80 0.45 0.45
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.80 0.20 0.44 0.63 0.78 0.24 0.48 0.59 0.73 0.27 0.45
phi_12 phi_13 phi_14 phi_15
0.62 0.74 0.43 0.51
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.12 0.81 0.41 0.47 0.68 0.86 0.41 0.55 0.65 0.81 0.44
phi_12 phi_13 phi_14 phi_15
0.50 0.63 0.41 0.34
Computing the covariance matrix of pilot sample.
----------- FAILURE REPORT --------------
--- failure: the condition has length > 1 ---
--- srcref ---
:
--- package (from environment) ---
bayesDccGarch
--- call from context ---
bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- call from argument ---
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
} else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
--- R stacktrace ---
where 1: bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- value of length: 2 type: logical ---
[1] TRUE TRUE
--- function from context ---
function (mY, nSim = 10000, tail_ini = 8, omega_ini = rep(0.03,
ncol(mY)), alpha_ini = rep(0.03, ncol(mY)), beta_ini = rep(0.8,
ncol(mY)), a_ini = 0.03, b_ini = 0.8, gamma_ini = rep(1,
ncol(mY)), errorDist = 2, control = list())
{
ptm <- proc.time()
if (missing(mY))
stop("'mY' is missing")
if (any(is.na(mY)))
stop("'mY' contains 'NA' values")
if (errorDist == 3 && tail_ini == 8) {
tail_ini = 2
}
mY = as.matrix(mY)
n = nrow(mY)
k = ncol(mY)
if (k == 1) {
npar = 5
}
else {
npar = 4 * k + 3
}
Lout = try(logLikDccGarch(mY, omega_ini, alpha_ini, beta_ini,
a_ini, b_ini, gamma_ini, tail_ini, errorDist))
if (class(Lout) == "try-error")
stop("The likelihood function can not be computed for the initial values.")
if (!is.list(control))
stop("control must be an object of list class.")
if (is.null(control$data)) {
control$data = mY
}
if (is.null(control$errorDist)) {
control$errorDist = errorDist
}
if (is.null(control$mu_a)) {
control$mu_a = 0
}
else {
if (length(control$mu_a) != 1) {
stop("control$mu_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_b)) {
control$mu_b = 0
}
else {
if (length(control$mu_b) != 1) {
stop("control$mu_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_omega)) {
control$mu_omega = rep(0, k)
}
else {
if (length(control$mu_omega) != k) {
stop("control$mu_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_alpha)) {
control$mu_alpha = rep(0, k)
}
else {
if (length(control$mu_alpha) != k) {
stop("control$mu_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_beta)) {
control$mu_beta = rep(0, k)
}
else {
if (length(control$mu_beta) != k) {
stop("control$mu_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_gamma)) {
control$mu_gamma = rep(0, k)
}
else {
if (length(control$mu_gamma) != k) {
stop("control$mu_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_tail)) {
control$mu_tail = 0
}
else {
if (length(control$mu_tail) != 1) {
stop("control$mu_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_a)) {
control$sigma_a = 10
}
else {
if (length(control$sigma_a) != 1) {
stop("control$sigma_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_b)) {
control$sigma_b = 10
}
else {
if (length(control$sigma_b) != 1) {
stop("control$sigma_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_omega)) {
control$sigma_omega = rep(10, k)
}
else {
if (length(control$sigma_omega) != k) {
stop("control$sigma_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_alpha)) {
control$sigma_alpha = rep(10, k)
}
else {
if (length(control$sigma_alpha) != k) {
stop("control$sigma_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_beta)) {
control$sigma_beta = rep(10, k)
}
else {
if (length(control$sigma_beta) != k) {
stop("control$sigma_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_gamma)) {
control$sigma_gamma = rep(10, k)
}
else {
if (length(control$sigma_gamma) != k) {
stop("control$sigma_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_tail)) {
control$sigma_tail = 1.25
}
else {
if (length(control$sigma_tail) != 1) {
stop("control$sigma_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$print)) {
control$print = 1
}
else {
if (!any(control$print == 0:1))
control$print = 1
}
if (is.null(control$simAlg)) {
control$simAlg = 3
}
else {
if (!any(control$simAlg == 1:3))
control$simAlg = 3
}
if (control$simAlg == 1) {
option = 1
if (is.null(control$cholCov)) {
stop("control$cholCov was not found.")
}
else {
if (!is.matrix(control$cholCov)) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
else {
if (ncol(control$cholCov) != npar || nrow(control$cholCov) !=
npar) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
}
}
}
if (control$simAlg == 2 && !is.null(control$sdSim)) {
option = 2
if (length(control$sdSim) != npar)
stop("control$sdSim must be a numeric positive vector with length 5 if k=1 or 4k+3 if k>1, where 'k=ncol(mY)'")
}
if (control$simAlg == 3) {
if (k == 1) {
control$sdSim = c(0.2, 0.2, 0.4, 0.4, 0.2)
}
else {
control$sdSim = c(0.2, rep(c(0.2, 0.4, 0.4, 0.2),
k), 1, 1.5)
}
}
if (is.null(control$lambda)) {
control$lambda = 0.5
}
if (is.null(control$nPilotSim)) {
control$nPilotSim = 1000
}
phi_ini = .C("original_to_real_scale", phi = numeric(4 *
k + 3), omega = as.double(omega_ini), alpha = as.double(alpha_ini),
beta = as.double(beta_ini), a = as.double(a_ini), b = as.double(b_ini),
gamma = as.double(gamma_ini), tail = as.double(tail_ini),
k = as.integer(k), errorDist = as.integer(errorDist))$phi
memoryAllocation(mY, errorDist, control$mu_omega, control$sigma_omega,
control$mu_alpha, control$sigma_alpha, control$mu_beta,
control$sigma_beta, control$mu_a, control$sigma_a, control$mu_b,
control$sigma_b, control$mu_gamma, control$sigma_gamma,
control$mu_tail, control$sigma_tail, control$print)
if (control$simAlg == 3) {
writeLines("Maximizing the log-posterior density function.")
if (k == 1)
phi_ini = phi_ini[1:5]
opt = optim(par = phi_ini, fn = logPosterior_phi, hessian = TRUE,
control = list(fnscale = -abs(Lout$value)))
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
if (opt$convergence == 0 || opt$convergence == 1) {
writeLines("Done.")
phi_ini = opt$par
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
control$cholCov = try(t(chol(solve(-opt$hessian))),
silent = TRUE)
if (class(control$cholCov) != "try-error") {
option = 1
}
else {
writeLines("One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.")
option = 2
}
}
else {
writeLines("The optim function convergence was not obtained.")
option = 2
}
}
repeat {
if (option == 1 && control$simAlg != 2) {
if (control$simAlg == 3) {
writeLines("Calibrating the Lambda coefficient:")
cont = 1
proceed = TRUE
while (proceed) {
writeLines(paste("lambda:", control$lambda))
MC_phi = MH_oneBlock(phi_ini, k, n_sim = control$nPilotSim,
control$lambda * control$cholCov)
acceptRate = 1 - rejectionRate(MC_phi)[5]
writeLines(paste("Accept Rate:", round(acceptRate,
2)))
if (acceptRate > 0.2 && acceptRate < 0.5 ||
cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
if (acceptRate < 0.2) {
control$lambda = control$lambda * 0.8
}
if (acceptRate > 0.5) {
control$lambda = control$lambda * 1.2
}
}
}
writeLines("Done.")
}
writeLines("Starting the simulation by one-block random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneBlock(phi_ini, k, n_sim = nSim, control$lambda *
control$cholCov)
writeLines("Done.")
break
}
if (option == 2 && control$simAlg != 1) {
if (control$simAlg == 3) {
writeLines("Calibrating the standard deviations for simulation:")
proceed = TRUE
cont = 1
while (proceed) {
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = control$nPilotSim,
control$sdSim)
acceptRate = 1 - rejectionRate(MC_phi)
writeLines("Accept Rate:")
print(round(acceptRate, 2))
if (errorDist == 1) {
acceptRate[1] = 0.3
}
lowerRate = which(acceptRate < 0.15)
upperRate = which(acceptRate > 0.5)
if (length(lowerRate) == 0 && length(upperRate) ==
0 || cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
control$sdSim[lowerRate] = control$sdSim[lowerRate] *
0.6
control$sdSim[upperRate] = control$sdSim[upperRate] *
1.4
}
}
if (errorDist == 1) {
MC_phi[, 1] = rnorm(control$nPilotSim)
}
writeLines("Computing the covariance matrix of pilot sample.")
control$cholCov = try(t(chol(cov(MC_phi))), silent = TRUE)
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
}
else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
}
if (option == 2) {
writeLines("Starting the simulation by one-dimensional random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = nSim,
sd_phi_sim = control$sdSim)
writeLines("Done.")
break
}
}
}
vMMeanH = .C("getMeanH", vMMeanH = numeric(n * k * k))$vMMeanH
mMeanH = matrix(vMMeanH, nrow = n, ncol = k * k, byrow = TRUE)
name = rep("NA", k * k)
for (i in 1:k) {
for (j in 1:k) {
name[(i - 1) * k + j] = paste("H_", i, ",", j, sep = "")
}
}
colnames(mMeanH) = name
logLike_mean = .C("getLogLikelihood_mean", value = numeric(1))$value
memoryDeallocation()
name = rep("NA", npar)
MC = matrix(NA, nrow = nSim, ncol = npar)
if (errorDist == 1 || errorDist == 3) {
MC[, 1] = exp(MC_phi[, 1])
name[1] = "delta"
}
if (errorDist == 2) {
MC[, 1] = exp(MC_phi[, 1]) + 2
name[1] = "nu"
}
for (i in 1:k) {
MC[, 4 * (i - 1) + 2:5] = cbind(exp(MC_phi[, 4 * (i -
1) + 2:3]), exp(MC_phi[, 4 * (i - 1) + 4:5])/(1 +
exp(MC_phi[, 4 * (i - 1) + 4:5])))
name[4 * (i - 1) + 2:5] = paste(c("gamma", "omega", "alpha",
"beta"), "_", i, sep = "")
}
if (k > 1) {
MC[, 4 * k + 2:3] = exp(MC_phi[, 4 * k + 2:3])/(1 + exp(MC_phi[,
4 * k + 2:3]))
name[4 * k + 2:3] = c("a", "b")
}
colnames(MC) = name
D_mean = -2 * logLike_mean
par_est = apply(X = MC, MARGIN = 2, FUN = mean)
tail_est = par_est[1]
gamma_est = par_est[2 + 4 * (0:(k - 1))]
omega_est = par_est[3 + 4 * (0:(k - 1))]
alpha_est = par_est[4 + 4 * (0:(k - 1))]
beta_est = par_est[5 + 4 * (0:(k - 1))]
if (k > 1) {
a_est = par_est[2 + 4 * k]
b_est = par_est[3 + 4 * k]
}
else {
a_est = 0.1
b_est = 0.1
}
D_at_mean = -2 * logLikDccGarch(mY, omega_est, alpha_est,
beta_est, a_est, b_est, gamma_est, tail_est, errorDist)$value
if (errorDist == 1) {
EAIC = D_mean + 2 * (npar - 1)
EBIC = D_mean + (npar - 1) * log(n)
}
else {
EAIC = D_mean + 2 * npar
EBIC = D_mean + npar * log(n)
}
DIC = D_at_mean + 2 * (D_mean - D_at_mean)
IC = matrix(c(EAIC, EBIC, DIC), nrow = 3, ncol = 1)
rownames(IC) = c("EAIC", "EBIC", "DIC")
colnames(IC) = c("Estimative")
out = list()
out$control = control
out$MC = as.mcmc(MC)
out$H = mMeanH
out$IC = IC
out$elapsedTime = proc.time() - ptm
return(structure(out, class = "bayesDccGarch"))
}
<bytecode: 0x2e71ae8>
<environment: namespace:bayesDccGarch>
--- function search by body ---
Function bayesDccGarch in namespace bayesDccGarch has this body.
----------- END OF FAILURE REPORT --------------
Error in if (class(control$cholCov) != "try-error") { :
the condition has length > 1
Calls: bayesDccGarch
Execution halted
Flavor: r-devel-linux-x86_64-fedora-clang
Version: 2.0
Check: examples
Result: ERROR
Running examples in ‘bayesDccGarch-Ex.R’ failed
The error most likely occurred in:
> ### Name: bayesDccGarch-package
> ### Title: bayesDccGARCH: Methods and tools for Bayesian analysis of
> ### DCC-GARCH(1,1) Model.
> ### Aliases: bayesDccGarch-package
> ### Keywords: package multivariate DCC-GARCH GARCH Bayes
>
> ### ** Examples
>
>
> data(DaxCacNik)
> Dax = DaxCacNik[,1]
>
> ### DCC-GARCH(1,1) ###
> out = bayesDccGarch(DaxCacNik[1:10,], nSim=1000) # more data is necessary
Maximizing the log-posterior density function.
Done.
One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.
Calibrating the standard deviations for simulation:
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.29 0.91 0.71 0.86 0.91 0.91 0.65 0.90 0.92 0.92 0.74
phi_12 phi_13 phi_14 phi_15
0.88 0.93 0.73 0.66
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.25 0.86 0.67 0.79 0.87 0.88 0.54 0.84 0.87 0.86 0.61
phi_12 phi_13 phi_14 phi_15
0.84 0.87 0.64 0.54
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.87 0.20 0.66 0.80 0.88 0.23 0.57 0.81 0.87 0.27 0.53
phi_12 phi_13 phi_14 phi_15
0.75 0.84 0.56 0.48
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.82 0.21 0.52 0.70 0.82 0.22 0.49 0.73 0.81 0.24 0.43
phi_12 phi_13 phi_14 phi_15
0.69 0.80 0.45 0.45
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.80 0.20 0.44 0.63 0.78 0.24 0.48 0.59 0.73 0.27 0.45
phi_12 phi_13 phi_14 phi_15
0.62 0.74 0.43 0.51
Simulation number = 100
Simulation number = 200
Simulation number = 300
Simulation number = 400
Simulation number = 500
Simulation number = 600
Simulation number = 700
Simulation number = 800
Simulation number = 900
Simulation number = 1000
Accept Rate:
phi_1 phi_2 phi_3 phi_4 phi_5 phi_6 phi_7 phi_8 phi_9 phi_10 phi_11
0.12 0.81 0.41 0.47 0.68 0.86 0.41 0.55 0.65 0.81 0.44
phi_12 phi_13 phi_14 phi_15
0.50 0.63 0.41 0.34
Computing the covariance matrix of pilot sample.
----------- FAILURE REPORT --------------
--- failure: the condition has length > 1 ---
--- srcref ---
:
--- package (from environment) ---
bayesDccGarch
--- call from context ---
bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- call from argument ---
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
} else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
--- R stacktrace ---
where 1: bayesDccGarch(DaxCacNik[1:10, ], nSim = 1000)
--- value of length: 2 type: logical ---
[1] TRUE TRUE
--- function from context ---
function (mY, nSim = 10000, tail_ini = 8, omega_ini = rep(0.03,
ncol(mY)), alpha_ini = rep(0.03, ncol(mY)), beta_ini = rep(0.8,
ncol(mY)), a_ini = 0.03, b_ini = 0.8, gamma_ini = rep(1,
ncol(mY)), errorDist = 2, control = list())
{
ptm <- proc.time()
if (missing(mY))
stop("'mY' is missing")
if (any(is.na(mY)))
stop("'mY' contains 'NA' values")
if (errorDist == 3 && tail_ini == 8) {
tail_ini = 2
}
mY = as.matrix(mY)
n = nrow(mY)
k = ncol(mY)
if (k == 1) {
npar = 5
}
else {
npar = 4 * k + 3
}
Lout = try(logLikDccGarch(mY, omega_ini, alpha_ini, beta_ini,
a_ini, b_ini, gamma_ini, tail_ini, errorDist))
if (class(Lout) == "try-error")
stop("The likelihood function can not be computed for the initial values.")
if (!is.list(control))
stop("control must be an object of list class.")
if (is.null(control$data)) {
control$data = mY
}
if (is.null(control$errorDist)) {
control$errorDist = errorDist
}
if (is.null(control$mu_a)) {
control$mu_a = 0
}
else {
if (length(control$mu_a) != 1) {
stop("control$mu_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_b)) {
control$mu_b = 0
}
else {
if (length(control$mu_b) != 1) {
stop("control$mu_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$mu_omega)) {
control$mu_omega = rep(0, k)
}
else {
if (length(control$mu_omega) != k) {
stop("control$mu_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_alpha)) {
control$mu_alpha = rep(0, k)
}
else {
if (length(control$mu_alpha) != k) {
stop("control$mu_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_beta)) {
control$mu_beta = rep(0, k)
}
else {
if (length(control$mu_beta) != k) {
stop("control$mu_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_gamma)) {
control$mu_gamma = rep(0, k)
}
else {
if (length(control$mu_gamma) != k) {
stop("control$mu_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$mu_tail)) {
control$mu_tail = 0
}
else {
if (length(control$mu_tail) != 1) {
stop("control$mu_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_a)) {
control$sigma_a = 10
}
else {
if (length(control$sigma_a) != 1) {
stop("control$sigma_a must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_b)) {
control$sigma_b = 10
}
else {
if (length(control$sigma_b) != 1) {
stop("control$sigma_b must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$sigma_omega)) {
control$sigma_omega = rep(10, k)
}
else {
if (length(control$sigma_omega) != k) {
stop("control$sigma_omega must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_alpha)) {
control$sigma_alpha = rep(10, k)
}
else {
if (length(control$sigma_alpha) != k) {
stop("control$sigma_alpha must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_beta)) {
control$sigma_beta = rep(10, k)
}
else {
if (length(control$sigma_beta) != k) {
stop("control$sigma_beta must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_gamma)) {
control$sigma_gamma = rep(10, k)
}
else {
if (length(control$sigma_gamma) != k) {
stop("control$sigma_gamma must be a numeric vector with length equal to 'ncol(mY)'.")
}
}
if (is.null(control$sigma_tail)) {
control$sigma_tail = 1.25
}
else {
if (length(control$sigma_tail) != 1) {
stop("control$sigma_tail must be a numeric vector with length equal to 1.")
}
}
if (is.null(control$print)) {
control$print = 1
}
else {
if (!any(control$print == 0:1))
control$print = 1
}
if (is.null(control$simAlg)) {
control$simAlg = 3
}
else {
if (!any(control$simAlg == 1:3))
control$simAlg = 3
}
if (control$simAlg == 1) {
option = 1
if (is.null(control$cholCov)) {
stop("control$cholCov was not found.")
}
else {
if (!is.matrix(control$cholCov)) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
else {
if (ncol(control$cholCov) != npar || nrow(control$cholCov) !=
npar) {
stop("control$cholCov must be a positive definite matrix of dimension 5x5 if k=1 or (4k+3)x(4k+3) if k>1, where 'k=ncol(mY)'")
}
}
}
}
if (control$simAlg == 2 && !is.null(control$sdSim)) {
option = 2
if (length(control$sdSim) != npar)
stop("control$sdSim must be a numeric positive vector with length 5 if k=1 or 4k+3 if k>1, where 'k=ncol(mY)'")
}
if (control$simAlg == 3) {
if (k == 1) {
control$sdSim = c(0.2, 0.2, 0.4, 0.4, 0.2)
}
else {
control$sdSim = c(0.2, rep(c(0.2, 0.4, 0.4, 0.2),
k), 1, 1.5)
}
}
if (is.null(control$lambda)) {
control$lambda = 0.5
}
if (is.null(control$nPilotSim)) {
control$nPilotSim = 1000
}
phi_ini = .C("original_to_real_scale", phi = numeric(4 *
k + 3), omega = as.double(omega_ini), alpha = as.double(alpha_ini),
beta = as.double(beta_ini), a = as.double(a_ini), b = as.double(b_ini),
gamma = as.double(gamma_ini), tail = as.double(tail_ini),
k = as.integer(k), errorDist = as.integer(errorDist))$phi
memoryAllocation(mY, errorDist, control$mu_omega, control$sigma_omega,
control$mu_alpha, control$sigma_alpha, control$mu_beta,
control$sigma_beta, control$mu_a, control$sigma_a, control$mu_b,
control$sigma_b, control$mu_gamma, control$sigma_gamma,
control$mu_tail, control$sigma_tail, control$print)
if (control$simAlg == 3) {
writeLines("Maximizing the log-posterior density function.")
if (k == 1)
phi_ini = phi_ini[1:5]
opt = optim(par = phi_ini, fn = logPosterior_phi, hessian = TRUE,
control = list(fnscale = -abs(Lout$value)))
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
if (opt$convergence == 0 || opt$convergence == 1) {
writeLines("Done.")
phi_ini = opt$par
if (k == 1)
phi_ini = c(phi_ini, -2, 1)
control$cholCov = try(t(chol(solve(-opt$hessian))),
silent = TRUE)
if (class(control$cholCov) != "try-error") {
option = 1
}
else {
writeLines("One approximation for covariance matrix of parameters cannot be directly computed through the hessian matrix.")
option = 2
}
}
else {
writeLines("The optim function convergence was not obtained.")
option = 2
}
}
repeat {
if (option == 1 && control$simAlg != 2) {
if (control$simAlg == 3) {
writeLines("Calibrating the Lambda coefficient:")
cont = 1
proceed = TRUE
while (proceed) {
writeLines(paste("lambda:", control$lambda))
MC_phi = MH_oneBlock(phi_ini, k, n_sim = control$nPilotSim,
control$lambda * control$cholCov)
acceptRate = 1 - rejectionRate(MC_phi)[5]
writeLines(paste("Accept Rate:", round(acceptRate,
2)))
if (acceptRate > 0.2 && acceptRate < 0.5 ||
cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
if (acceptRate < 0.2) {
control$lambda = control$lambda * 0.8
}
if (acceptRate > 0.5) {
control$lambda = control$lambda * 1.2
}
}
}
writeLines("Done.")
}
writeLines("Starting the simulation by one-block random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneBlock(phi_ini, k, n_sim = nSim, control$lambda *
control$cholCov)
writeLines("Done.")
break
}
if (option == 2 && control$simAlg != 1) {
if (control$simAlg == 3) {
writeLines("Calibrating the standard deviations for simulation:")
proceed = TRUE
cont = 1
while (proceed) {
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = control$nPilotSim,
control$sdSim)
acceptRate = 1 - rejectionRate(MC_phi)
writeLines("Accept Rate:")
print(round(acceptRate, 2))
if (errorDist == 1) {
acceptRate[1] = 0.3
}
lowerRate = which(acceptRate < 0.15)
upperRate = which(acceptRate > 0.5)
if (length(lowerRate) == 0 && length(upperRate) ==
0 || cont > 5) {
proceed = FALSE
}
else {
cont = cont + 1
control$sdSim[lowerRate] = control$sdSim[lowerRate] *
0.6
control$sdSim[upperRate] = control$sdSim[upperRate] *
1.4
}
}
if (errorDist == 1) {
MC_phi[, 1] = rnorm(control$nPilotSim)
}
writeLines("Computing the covariance matrix of pilot sample.")
control$cholCov = try(t(chol(cov(MC_phi))), silent = TRUE)
if (class(control$cholCov) != "try-error") {
writeLines("Done.")
option = 1
}
else {
control$cholCov = NULL
writeLines("The approximately covariance matrix is not positive definitely.")
option = 2
}
}
if (option == 2) {
writeLines("Starting the simulation by one-dimensional random walk Metropolis-Hasting algorithm.")
MC_phi = MH_oneDimension(phi_ini, k = k, n_sim = nSim,
sd_phi_sim = control$sdSim)
writeLines("Done.")
break
}
}
}
vMMeanH = .C("getMeanH", vMMeanH = numeric(n * k * k))$vMMeanH
mMeanH = matrix(vMMeanH, nrow = n, ncol = k * k, byrow = TRUE)
name = rep("NA", k * k)
for (i in 1:k) {
for (j in 1:k) {
name[(i - 1) * k + j] = paste("H_", i, ",", j, sep = "")
}
}
colnames(mMeanH) = name
logLike_mean = .C("getLogLikelihood_mean", value = numeric(1))$value
memoryDeallocation()
name = rep("NA", npar)
MC = matrix(NA, nrow = nSim, ncol = npar)
if (errorDist == 1 || errorDist == 3) {
MC[, 1] = exp(MC_phi[, 1])
name[1] = "delta"
}
if (errorDist == 2) {
MC[, 1] = exp(MC_phi[, 1]) + 2
name[1] = "nu"
}
for (i in 1:k) {
MC[, 4 * (i - 1) + 2:5] = cbind(exp(MC_phi[, 4 * (i -
1) + 2:3]), exp(MC_phi[, 4 * (i - 1) + 4:5])/(1 +
exp(MC_phi[, 4 * (i - 1) + 4:5])))
name[4 * (i - 1) + 2:5] = paste(c("gamma", "omega", "alpha",
"beta"), "_", i, sep = "")
}
if (k > 1) {
MC[, 4 * k + 2:3] = exp(MC_phi[, 4 * k + 2:3])/(1 + exp(MC_phi[,
4 * k + 2:3]))
name[4 * k + 2:3] = c("a", "b")
}
colnames(MC) = name
D_mean = -2 * logLike_mean
par_est = apply(X = MC, MARGIN = 2, FUN = mean)
tail_est = par_est[1]
gamma_est = par_est[2 + 4 * (0:(k - 1))]
omega_est = par_est[3 + 4 * (0:(k - 1))]
alpha_est = par_est[4 + 4 * (0:(k - 1))]
beta_est = par_est[5 + 4 * (0:(k - 1))]
if (k > 1) {
a_est = par_est[2 + 4 * k]
b_est = par_est[3 + 4 * k]
}
else {
a_est = 0.1
b_est = 0.1
}
D_at_mean = -2 * logLikDccGarch(mY, omega_est, alpha_est,
beta_est, a_est, b_est, gamma_est, tail_est, errorDist)$value
if (errorDist == 1) {
EAIC = D_mean + 2 * (npar - 1)
EBIC = D_mean + (npar - 1) * log(n)
}
else {
EAIC = D_mean + 2 * npar
EBIC = D_mean + npar * log(n)
}
DIC = D_at_mean + 2 * (D_mean - D_at_mean)
IC = matrix(c(EAIC, EBIC, DIC), nrow = 3, ncol = 1)
rownames(IC) = c("EAIC", "EBIC", "DIC")
colnames(IC) = c("Estimative")
out = list()
out$control = control
out$MC = as.mcmc(MC)
out$H = mMeanH
out$IC = IC
out$elapsedTime = proc.time() - ptm
return(structure(out, class = "bayesDccGarch"))
}
<bytecode: 0x29e5b48>
<environment: namespace:bayesDccGarch>
--- function search by body ---
Function bayesDccGarch in namespace bayesDccGarch has this body.
----------- END OF FAILURE REPORT --------------
Error in if (class(control$cholCov) != "try-error") { :
the condition has length > 1
Calls: bayesDccGarch
Execution halted
Flavor: r-devel-linux-x86_64-fedora-gcc
Version: 2.0
Check: DESCRIPTION meta-information
Result: NOTE
BugReports field is not a suitable URL but contains an email address
which will be used as from R 3.4.0
Flavors: r-release-osx-x86_64, r-oldrel-windows-ix86+x86_64, r-oldrel-osx-x86_64