Question

I am making a tree which will eventually turn into a set of arms with probabilities attached to going down each arm, with a value at the end of each arm.

This is the final arm of me tree (I am building it backwards).

require(mc2d)
ndvar(1000)

sderror_dfail<-(0.30-0.86)/1.96
alpha_dfail<-0.62*{{0.62*{1-0.62}/{sderror_dfail*sderror_dfail}}-1}
beta_dfail<-alpha_dfail*{{1-0.62}/0.62}

probs_d_if_fail<-mcstoc(rbetagen,type="V",shape1=alpha_dfail, shape2=beta_dfail, 
min=0, max=1)

lots1<-mcdata(1)
test<-lots1-probs_d_if_fail

probs1 <- mcstoc(rempiricalD, values=1:2, type="V", prob= c(probs_d_if_fail,test))

The next step is to build a branch of a tree as follows:

node1 <- mcprobtree(probs1, list("1"=mcdata(0), "2"=mcdata(1), type="V")

However the line probs1<- mcstoc(….) when I run it gives me the following error

Error in function (n, values, prob = NULL)  : 
Prob and values should be of same length or have the same number of columns.

I get the feeling it doesn’t like me using mcstoc object as the prob=c(probs_d_if_fail,test) part of the equation, but am not sure how to work around this.

Any help would be greatly appreciated.

Timothy

Answer 1

rempiricalD requires a a multivariate mcnode as prob argument. This is tricky.

Try:

require(mc2d)
ndvar(10000)

sderror_dfail<-(0.30-0.86)/1.96
alpha_dfail<-0.62*{{0.62*{1-0.62}/{sderror_dfail*sderror_dfail}}-1}
beta_dfail<-alpha_dfail*{{1-0.62}/0.62}

probs_d_if_fail<-mcstoc(rbetagen,type="V",shape1=alpha_dfail, shape2=beta_dfail, 
min=0, max=1)

lots1<-mcdata(1)
test<-lots1-probs_d_if_fail
probs <- mcdata(c(probs_d_if_fail,test),type="V",nvariates=2)
probs1 <- mcstoc(rempiricalD, values=1:2, type="V", prob= probs)

node1 <- mcprobtree(probs1, list("1"=mcdata(0), "2"=mcdata(1)), type="V")
node1

But the following code might be simpler if it fits your needs:

require(mc2d)
ndvar(10000)

sderror_dfail<-(0.30-0.86)/1.96
alpha_dfail<-0.62*{{0.62*{1-0.62}/{sderror_dfail*sderror_dfail}}-1}
beta_dfail<-alpha_dfail*{{1-0.62}/0.62}

probs_d_if_fail<-mcstoc(rbeta,type="V",shape1=alpha_dfail,
shape2=beta_dfail)

node1 <- mcstoc(rbern, type="V",prob = 1-probs_d_if_fail)
node1

Note that rbern will provide a 1 with probability (1-probs_d_if_fail), as you need if I understand well.

RP