partacf(rho), rho a REAL matrix whose columns are autocorrelations partacf(phikk, inverse:T), phikk a REAL matrix whose columns are partical autocorrelations |

partacf(rho), where rho is a REAL vector, computes the partial autocorrelations corresponding to the autocorrelation function in the REAL vector rho. Row k of rho should contain the lag k autocorrelation. The Levinson-Durbin algorithm is used. If rho is a matrix, partacf(rho) is a matrix of the same shape whose j-th column contains partial autocorrelations corresponding to autocorrelations in column j of rho. If any column of rho is not a valid autocorrelation function, that is, it does not define a positive definite Toeplitz matrix, a warning message is printed. partacf(phikk,inverse:T) is the inverse function to partacf. Each column of REAL vector or matrix phikk is considered to be the partial autocorrelation function of a time series. The corresponding column of the result is the corresponding autocorrelation function. All the elements of phikk must be less than 1 in absolute value. See also yulewalker().

Gary Oehlert 2003-01-15