corcounts (version 1.4)

pc2c: Calculate a corresponding correlation matrix to partial correlations

Description

'c2pc' is used to calculate the corresponding correlation matrix of dimension T times T out of partial correlations.

Usage

pc2c(Theta)

Arguments

Theta

A T times T matrix with partial correlations. See details.

Value

A symmetric positive definite correlation matrix of dimension T times T.

Details

The partial correlations in Theta have to be specified as

Theta =

12........13........14..........15...........16

..........23|1......24|1........25|1.........26|1 ....................34|12.......35|12........36|12 ... ................................45|123.......46|123 .............................................56|1234

...

and may be NA elsewhere. Theta has to be of dimension T times T.

This routine only calculates partial correlations conditional on 1, 12, 123, 1234, etc.. Partial correlations conditional on other margins can be obtained by a permutation of margins.

Examples

# create random uniform(0,1) partial correlations in dimension 8
dimension <- 8
Theta <- matrix(NA,dimension,dimension)
for (i in 2:dimension) {
  for (j in 1:(i-1)) {
    Theta[j,i] <- runif(1,-1,1)
  }
}
Theta

# calculate corresponding correlation matrix
C <- pc2c(Theta)
C

# transform back to partial correlations
c2pc(C)

# equivalence with original Theta
Theta - c2pc(C)