R help - Dec 2011 - try to silence errors

I am trying to use the dmt function in the package {mnormt}.  Throughout my
algorithm, the covariance matrix is sometime calculated to be singular. 
When attempting to calculate the dmt function with a covariance that is not
positive definite, I would like it to return Inf or NaN instead of an error
message.

I have been using the try function, however it is not yeilding the desired
result. (I did not include the values since they are 17 dimensions, all you
need to know is that sigij is singular)

try(out <- dmt(y[,j],new$mu[,,4],sigij,ceiling(new$nu[4])),silent=TRUE)
Error in chol.default(x, pivot = FALSE) : 
  the leading minor of order 16 is not positive definite

I have similarly tried the trycatch function (see below for code) however I
am using this within a function made my be  (which i call "ll") so I
get
errors like
 ll <- function(new,y){	
.......
out <- tryCatch(dmt(y[,j],new$mu[,,3],sigij,
 	ceiling(new$nu[3])),error = function(e) -Inf)
....
}

ll(new,y)
Error in chol.default(x, pivot = FALSE) : 
  the leading minor of order 16 is not positive definite
[1] -Inf

If anyone could shed some light on this problem, it would be greatly
appreciated.


--
View this message in context:
http://r.789695.n4.nabble.com/try-to-silence-errors-tp4225754p4225754.html
Sent from the R help mailing list archive at Nabble.com.

Something like the following may be what you are looking for.

try.out <- try(dmt(y[,j],new$mu[,,4],sigij,ceiling(new$nu[4])),silent=TRUE)

if (class(try.out) == 'try-error'} ) {
  out <- NaN
} else {  out <- try.out
}
-Don


-- 
Don MacQueen

Lawrence Livermore National Laboratory
7000 East Ave., L-627
Livermore, CA 94550
925-423-1062





On 12/22/11 7:59 AM, "statfan" <irene_vrbik at hotmail.com>
wrote:
>I am trying to use the dmt function in the package {mnormt}.  Throughout
>my
>algorithm, the covariance matrix is sometime calculated to be singular.
>When attempting to calculate the dmt function with a covariance that is
>not
>positive definite, I would like it to return Inf or NaN instead of an
>error
>message.
>
>I have been using the try function, however it is not yeilding the desired
>result. (I did not include the values since they are 17 dimensions, all
>you
>need to know is that sigij is singular)
>
>try(out <- dmt(y[,j],new$mu[,,4],sigij,ceiling(new$nu[4])),silent=TRUE)
>Error in chol.default(x, pivot = FALSE) :
>  the leading minor of order 16 is not positive definite
>
>I have similarly tried the trycatch function (see below for code) however
>I
>am using this within a function made my be  (which i call "ll") so
I get
>errors like
> ll <- function(new,y){	
>.......
>out <- tryCatch(dmt(y[,j],new$mu[,,3],sigij,
> 	ceiling(new$nu[3])),error = function(e) -Inf)
>....
>}
>
>ll(new,y)
>Error in chol.default(x, pivot = FALSE) :
>  the leading minor of order 16 is not positive definite
>[1] -Inf
>
>If anyone could shed some light on this problem, it would be greatly
>appreciated.
>
>
>--
>View this message in context:
>http://r.789695.n4.nabble.com/try-to-silence-errors-tp4225754p4225754.html
>Sent from the R help mailing list archive at Nabble.com.
>
>______________________________________________
>R-help at r-project.org mailing list
>https://stat.ethz.ch/mailman/listinfo/r-help
>PLEASE do read the posting guide
>http://www.R-project.org/posting-guide.html
>and provide commented, minimal, self-contained, reproducible code.