AdaptivMC: Adaptive MC procedure

In Zink-Antoine/Gum.Output: Output following Gum supplement 1 (MCMC) and 2 (multivariate)

Adaptive MC procedure

Description

adapted from GUM suppl.2 7.8 Adaptive Monte Carlo procedure

Usage

AdaptivMC(
  nOutput = 2,
  ndig = 2,
  p = 0.95,
  outliers = TRUE,
  FUN,
  n.iter = 10^4,
  MC = TRUE,
  ...
)

Arguments

nOutput	integer (with default value) number of model output(default nOutput=2)
ndig	integer (with default value) number of significant decimal digits (default ndig=2)
p	integer (with default value) coverage probability (default p= 0.95)
outliers	logical(with default) logical parameters indicating whether or not outliers are discarded. TRUE, the outliers are keeping, FALSE, the outliers are rejected ( default outliers=TRUE)
FUN	function (required) a simulation MC model
n.iter	integer (with default) number of iteration for each simulation (default n.iter= 1e4)
MC	logical (with default value) MC=TRUE Monte-Carlo simulation; MC=FALSE Martkov Chain Monte Carlo
...	further parameters passed to the FUN function

Value

a list including

ndig integer number of significant decimal digits

M integer the number of draws M= h x n.iter (h number of simulation and n.iter number of iteration for each simulation)

y vector an estimate of measurand Y

uy vector the standard uncertainties associated with the estimates,

Ry matrix correlation coefficients rij = r(yi, yj) associated with pairs of the estimates,

kp integer a coverage factor defining a 100p % coverage region for Y,

as stored file

trial_h mcmc list of the simulations

References

JCGM-WG1 (2011) data – Supplement 2 to the “Guide to the expression of uncertainty in measurement” – Extension to any number of output quantities. Guide JCGM 102:2011. Sèvres: BIPM, IEC, IFCC, ILAC, ISO, IUPAC, IUPAP and OIML.

Geyer, C. (2011) Introduction to Markov Chain Monte Carlo, in: Handbook of Markov Chain Monte Carlo. pp. 3–48.

Examples

## example 9.2.2 GUM suppl.2 case1
Model<-
function(n.iter){
X1<-rnorm(n.iter,0,1)
X2<-rnorm(n.iter,0,1)
X3<-rnorm(n.iter,0,1)
Y1<-X1+X3
Y2<-X2+X3
cbind(Y1,Y2)
}

AdaptivMC(nOutput=2,ndig=3,p=0.95,FUN=Model,n.iter=1000)


## Not run: 
AR(1) example
AR1<-
function(n.iter,rho=1,sig=1){
 X<-Xn<-rnorm(1,0,sig)
 Y<-0
 for (i in 2:n.iter){
   Yn<-rnorm(1,0,sig)
   Xn<-rho*Xn+Yn
   X<-rbind(X,Xn)
   Y<-rbind(Y,Yn)
 }
 out<-cbind(X,Y)
}

AdaptivMC(nOutput=2,ndig=2,p=0.95,FUN=AR1,n.iter=1000,rho=0.99)


## End(Not run)


## Not run: 
require(TLpack)

data(TLetru)
table<-Lum(TLetru,Doseb0=360,Dosea=0,alpha=FALSE,supra=FALSE)
B<-table$b
N<-table$n
table.norm<-B
 for (j in 1:3){
   for (i in 1:3)
     {
       table.norm[,i,j]<-(B[,i,j])/sum(N[seq(350,400),1,(j-1)*3+i])
       }
   }
 table.data<-cbind(table.norm[,,1],table.norm[,,2],table.norm[,,3])
 ii<-c(1,4,7,2,5,8,3,6,9)
 table.data<-table.data[seq(1,475),ii]
 Dose<-as.numeric(colnames(table.data))
 df.T<-matrix(rep(seq(26,500),9),475,9)
 df.y<-table.data[,1:9]
AdaptivMC(nOutput=5,ndig=2,outliers=FALSE,FUN=Slice5,Dose=Dose,df.T=df.T,df.y=df.y,n.thin=1)

## End(Not run)

Zink-Antoine/Gum.Output documentation built on Feb. 23, 2025, 6:27 p.m.