5.2 ALGEBRAIC MANIPULATION REVISITED 153and we have g(a) = g(e) = 0, yielding the factorization
b-e e-a a-b ( b-a)( b-e)Q(a)
-a -+
-b -+
-e -=
abc 'where Q(x) is a different linear polynomial. We conclude that
b-e e-a a-b ((a-b )(b-e)( e-a))
--+--+--= K ,
abc abcfor some constant K. Plugging in values (for example, a = 1, b = 2, e = 3) establishes
that K = -1. •
Example 5.2.20 (Putnam 19 39) Let x^3 + b;(2 + ex + d = 0 have integral coefficients
and roots r, s, t. Find a polynomial equation with integer coefficients written in terms
of a, b, e, d whose roots are r^3 , s^3 , t^3.
Solution: An incredibly ugly way to do this would be to solve for r, s, t, in terms
of a , b, e and then construct the cubic polynomial (x - r^3 )(x - s^3 )(x - t^3 ). Instead we
define p(x) : = x^3 + b;(2 + ex + d and note that
p(�) =0
is satisfied by x = r^3 , s^3 , t^3. We must thus convert
into an equivalent polynomial equation. Cubing comes to mind, but what should we
cube? Cubing anything but a binomial is too painful. If we put the radicals on one side
and the non-radicals on the other, we have
(1)and now cubing both sides will remove all of the radicals. We shall employ the more
useful form of 5.2.4, which states that
(x+y)^3 =x^3 +y^3 +3xy(x+y),
and cubing both sides of (1) yields
_(x+d)^3 = (b�9)^3 + (e�)^3 + 3b�e�(b� + e�)
= b^3 ;(2 +e^3 x+3bex(b�+ e�).
On the surface, this does not look like much of an improvement, since the right-hand
side still contains radicals. But (1) allows us to substitute -(x+d) for those pesky
radicals! Our equation becomes
_(x+d)^3 = b^3 .x2 + e3x -3bex(x+d),
a cubic with integral coefficients. •