Greatest common divisor

[ahumenberger]

The gcd behaviour seems weird:

@polyvar x

gcd(2*x^2 + 4, 2*x) # Result: 4.0

gcd(2*x^2 + 2, x) # Result: 2.0

gcd(4*x^2 + 8*x + 4, 4 * x^2) # Result: 1.0

gcd(2*x^2 + 4*x, 6*x) # Result: 6.0x

[saschatimme]

The gcd is only defined up to an invertible element of the field. If a polynomial with Integer coefficients is considered as a polynomial in ℝ[x] then this is correct. Also note the promotion to Float64 which seems to happen here:

MultivariatePolynomials.jl/src/division.jl

Line 27 in 95dc00a

(coefficient(t1) / coefficient(t2)) * _div(monomial(t1), monomial(t2))

The questions is whether we want to treat a polynomial with Integer coefficients as a polynomial in ℤ[x].

[ahumenberger]

The questions is whether we want to treat a polynomial with Integer coefficients as a polynomial in ℤ[x].

If I'm explicitly constructing a Polynomial{true, Int} I would expect it to be treated as a polynomial in ℤ[x]. What's your take on this?

[saschatimme]

I think this would make sense.