vars and nvars for MPolyElem{T} about abstractalgebra.jl HOT 8 CLOSED

Should it return the number of variable of the ring, or the actual number of variables used in that particular polynomial?

from abstractalgebra.jl.

Sorry I mean, should it return the variables for the ring or the poly.

from abstractalgebra.jl.

I was thinking the number of variables in the ring, but I hadn't thought about this ambiguity. So maybe it's better to only implement these functions for the ring?

from abstractalgebra.jl.

from abstractalgebra.jl.

Sounds good, then I think the method nvars should be defined for the ring instead of the elements.

It came up here BTW, and I think using the vars of the parent object is the only reasonable thing here right?

function AbstractAlgebra.subst(p::MPolyElem, v::Vector)
    @assert length(vars(parent(p))) == length(v) 
    r = zero(p)
    for i = 1:length(p)
        r += p.coeffs[i] * prod(v[k]^p.exps[k, i] for k = 1:length(vars(parent(p))))
    end
    r
end

from abstractalgebra.jl.

Yes, I think that's the only reasonable thing here. We'll have to add some more methods for substitution of only certain variables, since that is also a reasonable use case.

from abstractalgebra.jl.

from abstractalgebra.jl.

I agree. vars and nvars are too ambiguous for elements. They should be defined for the parent.

from abstractalgebra.jl.