16. First, we have P(x)=Q(x)R(x) for Q(x)=xm−∣a∣me40 and R(x)= xm−∣a∣me−iθ, where eiφ means of course cosφ+isinφ. It remains to factor both Q and R. Suppose that Q(x)=(x−q1)⋯(x−qm) and R(x)=(x−r1)⋯(x−rm).
Considering Q(x), we see that ∣qkm∣=∣a∣m and also ∣qk∣=∣a∣ for k= 1,…,m. Thus we may put qk=∣a∣eiβk and obtain by de Moivre's formula qkm=∣a∣meimβk. It follows that mβk=θ+2jπ for some j∈Z, and we have exactly m possibilities for βk modulo 2π:βk=mθ+2(k−1)π for k=1,2,…,m.
Thus qk=∣a∣eiβk; analogously we obtain for R(x) that rk=∣a∣e−iβk. Consequently,
xm−∣a∣meiθ=k=1∏m(x−∣a∣eiβk) and xm−∣a∣me−iθ=k=1∏m(x−∣a∣e−iβk).
Finally, grouping the k.th factors of both polynomials, we get
P(x)=k=1∏m(x−∣a∣eiβk)(x−∣a∣e−iβk)=k=1∏m(x2−2∣a∣xcosβk+a2)=k=1∏m(x2−2∣a∣xcosmθ+2(k−1)π+a2).