gcd of rational numbers

[smaasz]

MultivariatePolynomials.jl/src/gcd.jl

Line 845 in c0fd0f0

return gcd(a.num, b.num) // gcd(a.den, b.den)

shouldn't it be the lcm of the denominators?

[blegat]

Could you provide an example of gcd computation that is incorrect?

[smaasz]

using DynamicPolynomials
@polyvar x y z
gcd(1//2*x + 1//4*y + 1//3*z, 1)

outputs:

1//2

[blegat]

You expect it to be 1 or 1//12 ?

[smaasz]

I'd expect it to be 1//12.

Note that the issue only occurs when there're more than 2 terms (as only then _simplifier is called):

using DynamicPolynomials
@polyvar x y
gcd(1//3*x + 1//4*y, 1)

outputs:

1//12