Below we suppose that M is first-order equivalent to
< N, <,  0,  1,  +, {[/n] | n=1,2,…}, f >
where f is an arbitrary unary function.

Let (M,I) be divided. Let x be a sequence of elements of M. Let Y0 be empty and X0 be the set of all the elements of x. Let Yi+1 be the set of all the elements of b∈I such that there exist term t(x), unbounded terms t1(y) and t2(y), and a∈ (X0∪Yi) such that
t1(b)<t(a)<t2(b).          (4)
For t and a, there is no b∈I or there is the only b∈I such that (4) holds for some unbounded t1 and t2. So Yi is countable for any i. Let Y be the union of all Yi. We call Y the linked part of I for x and denote Y by [I, x].

Let us consider any divided (M,I). Fix two variable sequences x and y. A quantifier-free formula is called divided iff it is a Boolean of atomic divided formulas. An atomic formula is called divided iff it has the form
t1(x)+t2(y)< t3(x)+t4(y).
The theory of (M,I) is divided iff for any first-order formula Φ(x,y), there exist a quantifier-free divided formula Ψ(x,y) such that for any b, a∈(I−[I,b]), and c∈([I,b]∪b), Φ(a,c), is equivalent to Ψ(a,c). A function f is divided iff there exists a divided (M,I) such that the theory of (M,I) is divided and M is equivalent to
< N, <,  0,  1,  +, {[/n] | n=1,2,…}, f >.
The next theorem is a trivial corollary of [7], Theorems 3.1 and 3.2.

Previous page
Next page