I got some R code that I don't understand.
Question as comment in code
//where is t comming from, what is phi inverse
rAC <- function(name, n, d, theta){
#generic function for Archimedean copula simulation
illegalpar <- switch(name,
clayton = (theta < 0),
gumbel = (theta < 1),
frank = (theta < 0),
BB9 = ((theta[1] < 1) | (theta[2] < 0)),
GIG = ((theta[2] < 0) | (theta[3] < 0) | ((theta[1]>0) &
(theta[3]==0)) |
((theta[1]<0) & (theta[2]==0))))
if(illegalpar)
stop("Illegal parameter value")
independence <- switch(name,
clayton = (theta == 0),
gumbel = (theta == 1),
frank = (theta == 0),
BB9 = (theta[1] == 1),
GIG=FALSE)
U <- runif(n * d)
U <- matrix(U, nrow = n, ncol = d)
if(independence)
return(U)
Y <- switch(name,
clayton = rgamma(n, 1/theta),
gumbel = rstable(n, 1/theta) * (cos(pi/(2 * theta)))^theta,
frank = rFrankMix(n, theta),
BB9 = rBB9Mix(n, theta),
GIG = rGIG(n,theta[1],theta[2],theta[3]))
Y <- matrix(Y, nrow = n, ncol = d)
phi.inverse <- switch(name,
clayton = function(t, theta)
//where is t comming from, what is phi inverse
{
(1 + t)^(-1/theta)
}
,
gumbel = function(t, theta)
{
exp( - t^(1/theta))
}
,
frank = function(t, theta)
{
(-1/theta) * log(1 - (1 - exp( - theta)) * exp( - t))
}
,
BB9 = function(t, theta)
{
exp( - (theta[2]^theta[1] + t)^(1/theta[1]) + theta[2])
}
,
GIG = function(t, theta)
{
lambda <- theta[1]
chi <- theta[2]
psi <- theta[3]
if (chi==0)
out <- (1+2*t/psi)^(-lambda)
else if (psi==0)
out <- 2^(lambda+1)*exp(besselM3(lambda,sqrt(2*chi*t),log
value=TRUE)-lambda*log(2*chi*t)/2)/gamma(-lambda)
else
out <- exp(besselM3(lambda,sqrt(chi*(psi+2*t)),logvalue=T
RUE)+lambda*log(chi*psi)/2-besselM3(lambda,sqrt(chi*psi),logvalue=TRUE)-lambda*log(chi*(psi+2*t))/2)
out
}
)
phi.inverse( - log(U)/Y, theta)
}
--
View this message in context:
http://www.nabble.com/Newbie-that-don%27t-understand-R-code-tp25009034p25009034.html
Sent from the R help mailing list archive at Nabble.com.
kfcnhl wrote:> > I got some R code that I don't understand. > > Question as comment in code > //where is t comming from, what is phi inverse > > rAC <- function(name, n, d, theta){ > ................. > phi.inverse <- switch(name, > clayton = function(t, theta) > //where is t comming from, what is phi inverse > { > (1 + t)^(-1/theta) > } > , > gumbel = function(t, theta) > { > exp( - t^(1/theta)) > } > , > frank = function(t, theta) > { > (-1/theta) * log(1 - (1 - exp( - theta)) * exp( - t)) > } > , > BB9 = function(t, theta) > { > exp( - (theta[2]^theta[1] + t)^(1/theta[1]) + theta[2]) > } > , > GIG = function(t, theta) > { > lambda <- theta[1] > chi <- theta[2] > psi <- theta[3] > if (chi==0) > out <- (1+2*t/psi)^(-lambda) > else if (psi==0) > out <- 2^(lambda+1)*exp(besselM3(lambda,sqrt(2*chi*t),log > value=TRUE)-lambda*log(2*chi*t)/2)/gamma(-lambda) > else > out <- exp(besselM3(lambda,sqrt(chi*(psi+2*t)),logvalue=T > RUE)+lambda*log(chi*psi)/2-besselM3(lambda,sqrt(chi*psi),logvalue=TRUE)-lambda*log(chi*(psi+2*t))/2) > out > } > ) > phi.inverse( - log(U)/Y, theta) > } >First of all: R comments start with # and not // 1. t comes from the argument of the function object returned when <name> has the value clayton clayton = function(t, theta) ####where is t comming from, what is phi inverse { (1 + t)^(-1/theta) } 2. In phi.inverse <- switch(....) the value returned by the switch expression is assigned to phi.inverse. The switch expression appears to return a function object depending on the value of the <name> argument to rAC. 3. the code at the end of the rAC function contains two errors (spaces): log value should be logvalue T RUE should be TRUE After corrections it appears to work. Berend -- View this message in context: http://www.nabble.com/Newbie-that-don%27t-understand-R-code-tp25009034p25011959.html Sent from the R help mailing list archive at Nabble.com.
Hi, Comments inline and at end: On Aug 17, 2009, at 11:36 AM, kfcnhl wrote:> I got some R code that I don't understand. > > Question as comment in code > //where is t comming from, what is phi inverse > > rAC <- function(name, n, d, theta){ > #generic function for Archimedean copula simulation > illegalpar <- switch(name, > clayton = (theta < 0), > gumbel = (theta < 1), > frank = (theta < 0), > BB9 = ((theta[1] < 1) | (theta[2] < 0)), > GIG = ((theta[2] < 0) | (theta[3] < 0) | ((theta[1]>0) & > (theta[3]==0)) | > ((theta[1]<0) & (theta[2]==0)))) > if(illegalpar) > stop("Illegal parameter value") > independence <- switch(name, > clayton = (theta == 0), > gumbel = (theta == 1), > frank = (theta == 0), > BB9 = (theta[1] == 1), > GIG=FALSE) > U <- runif(n * d) > U <- matrix(U, nrow = n, ncol = d) > if(independence) > return(U) > Y <- switch(name, > clayton = rgamma(n, 1/theta), > gumbel = rstable(n, 1/theta) * (cos(pi/(2 * theta)))^theta, > frank = rFrankMix(n, theta), > BB9 = rBB9Mix(n, theta), > GIG = rGIG(n,theta[1],theta[2],theta[3])) > Y <- matrix(Y, nrow = n, ncol = d) > phi.inverse <- switch(name, > clayton = function(t, theta) > //where is t comming from, what is phi inverse > { > (1 + t)^(-1/theta) > }t isn't coming from anywhere, it's just a parameter to the function definition here. phi.inverse will be THE FUNCTION returned by the switch statement here, depending on the value of ``name``.> , > gumbel = function(t, theta) > { > exp( - t^(1/theta)) > } > , > frank = function(t, theta) > { > (-1/theta) * log(1 - (1 - exp( - theta)) * exp( - t)) > } > , > BB9 = function(t, theta) > { > exp( - (theta[2]^theta[1] + t)^(1/theta[1]) + theta[2]) > } > , > GIG = function(t, theta) > { > lambda <- theta[1] > chi <- theta[2] > psi <- theta[3] > if (chi==0) > out <- (1+2*t/psi)^(-lambda) > else if (psi==0) > out <- 2^(lambda+1)*exp(besselM3(lambda,sqrt(2*chi*t),log > value=TRUE)-lambda*log(2*chi*t)/2)/gamma(-lambda) > else > out <- exp(besselM3(lambda,sqrt(chi*(psi+2*t)),logvalue=T > RUE)+lambda*log(chi*psi)/2- > besselM3(lambda,sqrt(chi*psi),logvalue=TRUE)-lambda*log(chi*(psi > +2*t))/2) > out > } > ) > phi.inverse( - log(U)/Y, theta)phi.inverse was defined as the function returned by the switch statement. ``- log(U)/Y`` is passed in to the function's ``t`` argument. Does that help? -steve -- Steve Lianoglou Graduate Student: Computational Systems Biology | Memorial Sloan-Kettering Cancer Center | Weill Medical College of Cornell University Contact Info: http://cbio.mskcc.org/~lianos/contact