detarma(phi, theta, n [,masign:Masign] [,arsign:Arsign] [,nfreq:Nfreq]), non-MISSING REAL vectors phi and theta, positive integer vector n, Masign and Arsign +1 or -1, positive integer scalar Nfreq |

Help is not complete but is taken from the comments associated with detarma(). detarma() computes the determinant of the Toeplitz covariance associated with ARMA process. At present it has no provision for seasonal processes. Usage: detarma(phi, theta, n [,masign:Masign] [,arsign:Arsign] [,nfreq:Nfreq]) phi non-MISSING REAL vector defining a causal AR operator theta non-MISSING REAL vector defining a MA operator n vector of positive integers Masign and/or Arsign must be +1 or -1; they determine sign conventions for interpreting phi and theta. Type arimahelp(MASIGN) for information. Nfreq Number of frequencies used in DFT to compute ACFV. Default is the smallest integer >= 2*max(max(n), 200) that has no prime factors > 29 The result is a REAL vector d, the same length as n, with d[i] = det(Sigma(n[i])), where Sigma(n[i]) = Cov[(X(1),...,X(n[i]))] when X(t) is ARMA with AR coefficients phi and MA coefficients theta detarma() computes the autocovariance function as the inverse Fourier transform of the spectrum computed at nfreq frequencies. From this it computes that partial autocorrelations which are used, together with the variance to compute the determinant.

Gary Oehlert 2003-01-15