Given generators a,b,c,... of a finite algebraic extension of the
rationals, would it not be possible for Gap 3.4+epsilon to calculate a
minimal polynomial for a single element t such that
Q(t)=Q(a,b,c,...), by applying an algorithmic version of the primitive
element theorem? It seems like most of what is necessary is already
in Gap 3.4.

Leonard Soicher.