[Scilab-users] pfss question...

[Samuel Gougeon]

Hello,

Le 22/06/2014 06:59, tanthiamhuat a écrit :
>   
> -->pfss((x^2)/(x^3-3*x+2))
>   ans  =
>         ans(1)
>   
>      0.4444444
>      ---------
>        2 + s
>   
>         ans(2)
>   
>    - 0.2222222 + 0.5555556s
>      ----------------------
>                      2
>            1 - 2s + s
>
> But I expect this answer:
> 5/(9*(x - 1)) + 1/(3*(x - 1)^2) + 4/(9*(x + 2))
>
> how can Scilab return above answer?
Apart the x=>s bug, the uncomplete decomposition is due to the 
uncomplete factorization/simplification. Such a limitation has been 
reported here:
http://bugzilla.scilab.org/show_bug.cgi?id=4191
This bug has been tagged FIXED instead of WONT (be fixed), while IMO the 
algorithm could post-process the initial set of raw roots, agregating 
subsets of roots near enough from each others to be considered as the 
same (average over the subset). In this way, i think that roots of high 
multiplicity would be better handled.

Regards
Samuel

PS : a similar effect is reported there for complex coefficients: 
http://bugzilla.scilab.org/show_bug.cgi?id=8099