Diff for /imach096d/doc/imach.htm between versions 1.8 and 1.10

version 1.8, 2002/03/11 14:18:06 version 1.10, 2002/03/11 22:26:00
Line 425  warranty!): </p> Line 425  warranty!): </p>
 23 0.0 0.0</pre>  23 0.0 0.0</pre>
 </blockquote>  </blockquote>
   
   <p> In order to speed up the convergence you can make a first run with
   a large stepm i.e stepm=12 or 24 and then decrease the stepm until
   stepm=1 month. If newstepm is the new shorter stepm and stepm can be
   expressed as a multiple of newstepm, like newstepm=n stepm, then the
   following approximation holds:
   <pre>aij(stepm) = aij(n . stepm) - ln(n)
   </pre> and
   <pre>bij(stepm) = bij(n . stepm) .</pre>
   
   <p> For example if you already ran for a 6 months interval and
   got:<br>
    <pre># Parameters
   12 -13.390179  0.126133
   13  -7.493460  0.048069
   21   0.575975 -0.041322
   23  -4.748678  0.030626
   </pre>
   If you now want to get the monthly estimates, you can guess the aij by
   substracting ln(6)= 1,7917<br> and running<br>
   <pre>12 -15.18193847  0.126133
   13 -9.285219469  0.048069
   21 -1.215784469 -0.041322
   23 -6.540437469  0.030626
   </pre>
   and get<br>
   <pre>12 -15.029768 0.124347
   13 -8.472981 0.036599
   21 -1.472527 -0.038394
   23 -6.553602 0.029856
   </br>
   which is closer to the results. The approximation is probably useful
   only for very small intervals and we don't have enough experience to
   know if you will speed up the convergence or not.
   <pre>         -ln(12)= -2.484
    -ln(6/1)=-ln(6)= -1.791
    -ln(3/1)=-ln(3)= -1.0986
   -ln(12/6)=-ln(2)= -0.693
   </pre>
   
 <h4><font color="#FF0000">Guess values for computing variances</font></h4>  <h4><font color="#FF0000">Guess values for computing variances</font></h4>
   
 <p>This is an output if <a href="#mle">mle</a>=1. But it can be  <p>This is an output if <a href="#mle">mle</a>=1. But it can be

Removed from v.1.8  
changed lines
  Added in v.1.10


FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>