[R-SIG-Finance] bilinear and non linear time series?

[Spencer Graves]

Hello: 

      Are there any facilities in R for estimating bilinear time series, 
i.e., ARMA models with cross products between AR and MA terms?  I ask, 
because Tsay (2005, sec. 4.1.1) discusses this.  RSiteSearch("bilinear 
time series") produced a post from May 2004 on this issue.  Is the 
answer today the same as in 2004:  Use 'nls' or write a likelihood 
function and use 'optim'? 

      Thanks,
      Spencer