Ooops sorry... I had just a look without checking the degree with coeff :-(...
Anyway it should be really better if linsolve could check for linearity.
In data luned? 11 maggio 2009 22:31:07, reyssat ha scritto:
: > Stefano Ferri a ?crit :
> > I had a look at your first problem, and it is linear in y.
>
> No, it is not linear but cubic ; for instance
> (%i70) coeff(expand(det1),y,3);
> (%o70)
> x3^3-9*x2*x3^2+9*x1*x3^2-3*x0*x3^2+27*x2^2*x3-54*x1*x2*x3+18*x0*x2*x3
> +27*x1^2*x3-18*x0*x1*x3+3*x0^2*x3-27*x2^3+81*x1*x2^2-27*x0*x2^2
>
> -81*x1^2*x2+54*x0*x1*x2-9*x0^2*x2+27*x1^3-27*x0*x1^2+9*x0^2*x1-x0^3
>
> and linsolve is dangerous since it doesn't check for linearity (and its
> documentation says
> "The expressions must each be polynomials in the variables and may be
> equations. ", I don't understand why the word polynomial is used here).
> For instance :
> (%i1) linsolve(y^3+y+1,y);
> (%o1) [y = -
> 1] looks like a bug.
>
> Eric Reyssat
>
> > Instead of solve,
> > try linsolve. It solves your problem in less than a second.
> > But look how complicated is the solution, and simplifier cannot help you
> > here.