Problems with linsolve: quotient by polynomial of higher degree



On 5/26/2012 11:51 AM, Robert Dodier wrote:
> I don't know much about linsolve or the code for GCD and so on. I tried
> the code (linsolve_error.mac) and made a little progress.
>
> (1) with algebraic = false, there is an error ("Quotient by a polynomial
> of higher degree") in PQUOTIENT. (It is the first one of the two such
> errors in that function.) The call which triggers the error is:
>
> (PQUOTIENT '4 '(#:|3^(1/2)15991| 2 1))
this is a call to compute the quotient of
4
divided by the polynomial

1*r^2   where r is an indeterminate representing sqrt(3).

I assume there is some computation which computed the GCD of two 
polynomials A and B and decided the gcd was  r^2.
Then somewhere else the program realized, hey,  A is 4.

For what it is worth, the commercial Macsyma does the same thing.

RJF




>
> Question: what is the second argument supposed to represent?
>
>
> (2) with algebraic = true, it bumps into the same error by a different
> path. This time the call which triggers it is:
>
> (PQUOTIENT '-2985984 '(#:_BK16206 1 288))

this is dividing -2986984  by   288* bk ^ 1.