ncatlab.org

Dedekind-Hasse norm in nLab

The notion of a Dedekind-Hasse norm is a generalization of the notion of the degree function in a Euclidean domain. It makes clear the links between a principal ideal domain and an Euclidean domain: a pid is like an Euclidean domain but with a weaker notion of Euclidean division.

Proof

Suppose that RR possesses a Dedekind-Hasse norm. Let II be a non-zero ideal. Let bb be a non-zero element of II of minimal norm. We know that (b)⊆I(b) \subseteq I. Let aa be an element of II. Suppose that bb doesn’t divide aa. Then, there exists p,q,r∈Rp,q,r \in R such that pa=bq+rpa = bq+r and 0<v(r)<v(b)0 \lt v(r) \lt v(b). Thus, r=pa−bq∈Ir=pa-bq \in I, r≠0r \neq 0 and v(r)<v(b)v(r) \lt v(b), absurd! Therefore b|ab|a and a∈(b)a \in (b). Thus, I⊆(b)I \subseteq (b) and I=(b)I = (b). We have proved that RR is a pid.

Suppose that RR is a pid. Thus, it is a UFD. Put v(s)=0v(s)=0 if s=0s=0, and v(s)v(s) equal to 2 n2^{n} where nn is the number of irreducible elements in the factorization of nn, if s≠0s \neq 0. Let a∈Ra \in R and (b≠0)∈R(b \neq 0) \in R. Suppose that bb doesn’t divide aa. We know that (a,b)=(r)(a,b) = (r) and thus there exists p,q,r∈Rp,q,r \in R such that pa+bq=rpa+bq = r. rr divides bb but bb doesn’t divide rr because it would imply that bb divides aa. Thus, there is strictly less irreducible elements in the factorization of rr than in the one of bb and v(r)<v(b)v(r) \lt v(b). Moreover r≠0r \neq 0 because (a,b)=(r)(a,b) = (r) and b≠0b \neq 0. Thus 0<v(r)<v(b)0 \lt v(r) \lt v(b). We have proved that vv is a Dedekind-Hasse norm.

In constructive mathematics, there are different types of integral domains, yielding different types of Dedekind-Hasse norms. For example, if the integral domain has a tight apartness relation a#ba # b, such as in a Heyting integral domain, then one can use the tight apartness relation instead of denial inequality in the second condition