3.5 Constructible numbers revisited
Earlier (1.36) we showed that a real number is constructible if and only if it is contained in a subfield of of the form with each a positive element of . In particular
Now we can prove a partial converse to this last statement.
If is contained in a subfield of that is Galois of degree over , then it is constructible.
Suppose where is Galois of degree over , and let . Because finite -groups are solvable (GT, 6.7), there exists a sequence of groups
with of order . Correspondingly, there will be a sequence of fields,
with of degree over . The next lemma shows that for some , and because otherwise would not be real. This proves the theorem. □
Let be a quadratic extension of fields of characteristic . Then for some .
Let , , and let be the minimal polynomial of . Then , and so . □
If is a prime of the form , then is constructible.
Thus a regular -gon, prime, is constructible if and only if is a Fermat prime, i.e., of the form . For example, we have proved that the regular -polygon is constructible, without (happily) having to exhibit an explicit formula for .
The converse to (7) is false; in particular, there are nonconstructible algebraic numbers of degree over . The polynomial is irreducible, and we’ll show below (LABEL:cg8a) that the Galois group of a splitting field for is . Each root of lies in an extension of degree of . If the four roots of were constructible, then all the elements of would be constructible (1.36a), but if denotes a Sylow -subgroup of , then has odd degree over , and so no element of is constructible.33 3 It is possible to prove this without appealing to the Sylow theorems. If a root of were constructible, then there would exist a tower of quadratic extensions . By Galois theory, the groups have orders and respectively. As , would be . But has no subgroup of order , a contradiction. Thus no root of is constructible. (Actually , but that does not matter here.)