Closed agila5 closed 4 years ago
becarioprecario
commented
4 years ago Hi,
Thanks for checking. The code should run fine on R 3.6.3 and the latest version of INLA. The error message that you provide is not very informative... Can you provide your sessionInfo()?
Best,
Virgilio
agila5
commented
4 years ago This is a copy-paste of the error message I obtain in interactive version of Rstudio:
Error in inla.inlaprogram.has.crashed() :
The inla-program exited with an error. Unless you interupted it yourself, please rerun with verbose=TRUE and check the output carefully.
If this does not help, please contact the developers at <help@r-inla.org>.
In addition: Warning message:
In inla.model.properties.generic(inla.trim.family(model), mm[names(mm) == :
Model 'rgeneric' in section 'latent' is marked as 'experimental'; changes may appear at any time.
Use this model with extra care!!! Further warnings are disabled.and this is the sessionInfo:
becarioprecario
commented
4 years ago I’ll try to reproduce the error and go back to you.
Best,
Virgilio
El 6 may 2020, a las 21:07, Andrea Gilardi notifications@github.com<mailto:notifications@github.com> escribió:
This is a copy-paste of the error message I obtain in interactive version of Rstudio:
Error in inla.inlaprogram.has.crashed() : The inla-program exited with an error. Unless you interupted it yourself, please rerun with verbose=TRUE and check the output carefully. If this does not help, please contact the developers at help@r-inla.org<mailto:help@r-inla.org>. In addition: Warning message: In inla.model.properties.generic(inla.trim.family(model), mm[names(mm) == : Model 'rgeneric' in section 'latent' is marked as 'experimental'; changes may appear at any time. Use this model with extra care!!! Further warnings are disabled.
and this is the sessionInfo:
Session info
devtools::session_info()
— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/becarioprecario/INLAMSM/issues/1#issuecomment-624834145, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABYD6WVSP6RORD7CZWPKNSDRQGYO3ANCNFSM4M2SNFGQ.
agila5
commented
4 years ago I had an epiphany a few minutes ago and I think I get it know. The problem is that the second argument to the inla.rgeneric.define function is an unnamed argument, i.e. list(k = k, W = W), which implies that it's passed to the debug parameter:
# packages
options(width = 120)
library(INLA)
library(INLAMSM)
# The k and W parameters are never passed to the inla.rgeneric.indep.IMCAR.model
# but the are assigned to the debug parameter.
k <- 1
W <- diag(2)
inla.rgeneric.define(
inla.rgeneric.indep.IMCAR.model,
list(k = k, W = W)
)
#> $f
#> $f$model
#> [1] "rgeneric"
#>
#> $f$n
#> [1] 2
#>
#> $f$rgeneric
#> $definition
#> function (cmd = c("graph", "Q", "mu", "initial", "log.norm.const",
#> "log.prior", "quit"), theta = NULL)
#> {
#> interpret.theta = function() {
#> param <- sapply(theta[as.integer(1:k)], function(x) {
#> exp(x)
#> })
#> PREC <- diag(param, k)
#> return(list(param = param, PREC = PREC))
#> }
#> graph = function() {
#> PREC <- diag(1, k)
#> G <- kronecker(PREC, Matrix::Diagonal(nrow(W), 1) + W)
#> return(G)
#> }
#> Q = function() {
#> param <- interpret.theta()
#> Q <- kronecker(param$PREC, Matrix::Diagonal(nrow(W),
#> apply(W, 1, sum)) - W)
#> return(Q)
#> }
#> mu = function() {
#> return(numeric(0))
#> }
#> log.norm.const = function() {
#> val <- numeric(0)
#> return(val)
#> }
#> log.prior = function() {
#> param = interpret.theta()
#> val <- -sum(theta)/2 - k * log(2)
#> return(val)
#> }
#> initial = function() {
#> return(c(rep(log(1), k)))
#> }
#> quit = function() {
#> return(invisible())
#> }
#> val = do.call(match.arg(cmd), args = list())
#> return(val)
#> }
#> <environment: 0x000000001cbfa888>
#>
#> $debug
#> $debug$k
#> [1] 1
#>
#> $debug$W
#> [,1] [,2]
#> [1,] 1 0
#> [2,] 0 1
#>
#>
#> attr(,"class")
#> [1] "inla.rgeneric"
#>
#>
#> attr(,"class")
#> [1] "inla.rgeneric"
# debug = FALSE as it should be
inla.rgeneric.define(
inla.rgeneric.indep.IMCAR.model,
k = k,
W = W
)
#> $f
#> $f$model
#> [1] "rgeneric"
#>
#> $f$n
#> [1] 2
#>
#> $f$rgeneric
#> $definition
#> function (cmd = c("graph", "Q", "mu", "initial", "log.norm.const",
#> "log.prior", "quit"), theta = NULL)
#> {
#> interpret.theta = function() {
#> param <- sapply(theta[as.integer(1:k)], function(x) {
#> exp(x)
#> })
#> PREC <- diag(param, k)
#> return(list(param = param, PREC = PREC))
#> }
#> graph = function() {
#> PREC <- diag(1, k)
#> G <- kronecker(PREC, Matrix::Diagonal(nrow(W), 1) + W)
#> return(G)
#> }
#> Q = function() {
#> param <- interpret.theta()
#> Q <- kronecker(param$PREC, Matrix::Diagonal(nrow(W),
#> apply(W, 1, sum)) - W)
#> return(Q)
#> }
#> mu = function() {
#> return(numeric(0))
#> }
#> log.norm.const = function() {
#> val <- numeric(0)
#> return(val)
#> }
#> log.prior = function() {
#> param = interpret.theta()
#> val <- -sum(theta)/2 - k * log(2)
#> return(val)
#> }
#> initial = function() {
#> return(c(rep(log(1), k)))
#> }
#> quit = function() {
#> return(invisible())
#> }
#> val = do.call(match.arg(cmd), args = list())
#> return(val)
#> }
#> <environment: 0x000000001d8b2610>
#>
#> $debug
#> [1] FALSE
#>
#> attr(,"class")
#> [1] "inla.rgeneric"
#>
#>
#> attr(,"class")
#> [1] "inla.rgeneric"Created on 2020-05-06 by the reprex package (v0.3.0)
becarioprecario
commented
4 years ago Hi,
I had an epiphany a few minutes ago and I think I get it know. The problem is that the second argument to the inla.rgeneric.define function is an unnamed argument, i.e. list(k = k, W = W), which implies that it's passed to the debug parameter:
Thanks for this. I believe that we changed this because one of the reviewers of the paper suggested that the arguments needed to be in a list. You can simply pass the values of k and W without the list and it should work.
Best,
Virgilio
agila5
commented
4 years ago Thanks for this. I believe that we changed this because one of the reviewers of the paper suggested that the arguments needed to be in a list. You can simply pass the values of k and W without the list and it should work.
Yes, it works. It was really a minor problem.
Kind regards Andrea
becarioprecario
commented
4 years ago Hi,
Yes, it works. It was really a minor problem.
A minor problem that makes the R code not to run on some platforms. We’ll like into this to make sure that it can also run smoothly on all platforms.
Best,
Virgilio
becarioprecario
commented
4 years ago We have changed the example in the paper so that arguments passed to define the structure of the latent GMRF are not in a list anymore. This should work. Thanks for reporting the error!
Virgilio
agila5
commented
4 years ago You're welcome!
Dear Prof. Gómez-Rubio, I found a minor error running the code in the JSS paper. I installed the INLAMSM package from github while the R code is the same as in the document that you sent me.
Created on 2020-05-06 by the reprex package (v0.3.0)
The same code runs without any issue if I specify
model.indimcar <- inla.INDIMCAR.model(k = k, W = W)instead of theinla.rgeneric.definecode.