 Usage:  y = simvar(u,y0,a)  
 
 Input:

  u                      (K x T)-matrix of 'noise'. Each column represents 
                         the K-dimensional noise, or innovation of a point in time. 
                         
  y0                     (K x p)-matrix of starting ('pre-sample')-values of 
                         time series. 
                         
  a                      (K x K*p) or (K x K*p+1)-matrix of model parameters. 
                         The model can be specified with [(K x K*p+1)] or without 
                         [(K x K*p)] intercept. If an intercept is specified simvar() 
                         regards the first column of 'a' as the intercept. 
                         
 Output:

  y                      (K x p+T)-matrix of autoregressive time series. The first 
                         p columns of 'y' are 'y0', the remaining are the computed 
                         time series. 
                         
--------------------------------------------------------------
(C) MD*TECH Method and Data Technologies, 17.8.2000
