I'm currently doing some work which involves Thom Encodings. However I'm having trouble about their representation. This is the algorithm which uses them:
INPUT: Real algebraic numbers α and β given by their Thom encodings $σ_{P_1}(α)$ and $σ_{P_2}(β)$ with respect to polynomials $P_1, P_2 ∈ Z[X]$ such that $deg(P_1), deg(P_2) ≤ d (d ≥ 2)$ and $H(P_1), H(P_2) ≤ H$. Where H is the height of a polynomial.
OUTPUT: The sign $s := sign(e^β − α)$.
- Let $c := (2^{d+1}(d + 1)H)^{−2^{41}d^6(5d+4⌈log(H)⌉)}$.
- Compute $w ∈Q$ such that $|e^β −w|
(a) Compute $w_1 ∈Q$ such that $|β−w_1|< (\frac{c}{(2.3^{H+2}})$
(b) Compute $w ∈Q$ such that $|e^{w_1} −w|< (\frac{c}{2})$- Compute $s = sign(w − α)$.
What I'm wondering is what is α and β? I believe that a Thom Encoding is a tuple of the polynomial and the signs when calculated at their derivatives. If this is true, how could α and β, a tuple, be used in the algorithm? If possible, could you provide an example please?
Thanks, help appreciated.