B.3. TRANSCENDENTAL NUMBERS 403

where α (k) = γk− γ0. These are still distinct algebraic numbers. Therefore, α (k) is a rootof a polynomial

Qk (x) = vkxmk + · · ·+uk (2.14)

having integer coefficients, vk,uk ̸= 0. Recall algebraic numbers were defined as roots ofpolynomial equations having rational coefficients. Just multiply by the denominators to getone with integer coefficients. Let the roots of this polynomial equation be{

α (k)1 , · · · ,α (k)mk

}and suppose they are listed in such a way that α (k)1 = α (k). Thus, by Theorem B.2.7every symmetric polynomial in these roots is rational.

Letting ik be an integer in {1, · · · ,mk} it follows from the assumption 2.12 that

∏(i1,··· ,in)

ik∈{1,··· ,mk}

(K +b1eα(1)i1 +b2eα(2)i2 + · · ·+bneα(n)in

)= 0 (2.15)

This is because one of the factors is the one occurring in 2.13 when ik = 1 for every k. Theproduct is taken over all distinct ordered lists (i1, · · · , in) where ik is as indicated. Expandthis possibly huge product. This will yield something like the following.

K′+ c1

(eβ (1)1 + · · ·+ eβ (1)µ(1)

)+c2

(eβ (2)1 + · · ·+ eβ (2)µ(2)

)+ · · ·+

cN

(eβ (N)1 + · · ·+ eβ (N)µ(N)

)= 0 (2.16)

These integers c j come from products of the bi and K. You group these exponentials ac-cording to which ci they multiply. The β (i) j are the distinct exponents which result, eachbeing a sum of some of the α (r)ir . Since the product included all roots for each Qk (x),interchanging their order does not change the distinct exponents β (i) j which result. Theymight occur in a different order however, but you would still have the same distinct ex-ponents associated with each cs as shown in the sum. Thus any symmetric polynomialin the β (s)1 ,β (s)2 , · · · ,β (s)

µ(s) is also a symmetric polynomial in the roots of Qk (x) ,α (k)1 ,α (k)2 , · · · ,α (k)mk

for each k.Doesn’t this contradict Corollary B.3.6? This is not yet clear because we don’t know

that the β (i)1 , ...,β (i)µ(i) are roots of a polynomial having rational coefficients. For a

given r,β (r)1 , · · · ,β (r)µ(r) are roots of the polynomial

(x−β (r)1)(x−β (r)2) · · ·(

x−β (r)µ(r)

)(2.17)

the coefficients of which are elementary symmetric polynomials in the β (r)i , i ≤ µ (r).Thus the coefficients are symmetric polynomials in the

α (k)1 ,α (k)2 , · · · ,α (k)mk

for each k. Say the polynomial is of the form

µ(r)

∑l=0

xn−lBl (A(1) , · · · ,A(n))

B.3. TRANSCENDENTAL NUMBERS 403where & (k) = Yj — Yo. These are still distinct algebraic numbers. Therefore, @ (k) is a rootof a polynomialOx (x) = vygx'"® + +++ + Ug (2.14)having integer coefficients, vz, uz, 40. Recall algebraic numbers were defined as roots ofpolynomial equations having rational coefficients. Just multiply by the denominators to getone with integer coefficients. Let the roots of this polynomial equation be{a (14-0 (by, $and suppose they are listed in such a way that w@(k), = @(k). Thus, by Theorem B.2.7every symmetric polynomial in these roots is rational.Letting i, be an integer in {1,--- ,m,} it follows from the assumption 2.12 thatTL (bie 4 dpe 0 4-4 bpe"n) = 0 (2.15)(i++ in)iE {1,-++ uy}This is because one of the factors is the one occurring in 2.13 when i, = 1 for every k. Theproduct is taken over all distinct ordered lists (i,,--- ,i,) where i, is as indicated. Expandthis possibly huge product. This will yield something like the following.K'+c Ga $e eFu)+e) (eh +108 )u0)) eeecy (eo qe + Mua) ~( (2.16)These integers c; come from products of the b; and K. You group these exponentials ac-cording to which c; they multiply. The (i) ; are the distinct exponents which result, eachbeing a sum of some of the a (r);. Since the product included all roots for each Q, (x),interchanging their order does not change the distinct exponents B (i) , which result. Theymight occur in a different order however, but you would still have the same distinct ex-ponents associated with each cs; as shown in the sum. Thus any symmetric polynomialin the B(s),,B(s),--+,B (s)y(s) is also a symmetric polynomial in the roots of Qx (x) ,a (k)) ,A(k)o,+++, (Kk), for each k.Doesn’t this contradict Corollary B.3.6? This is not yet clear because we don’t knowthat the B (7), ,..-,8 (yj are roots of a polynomial having rational coefficients. For agiven r,B (r),,---,B(r) u(r) are roots of the polynomial(«= B ()1) @-B (a): (xB On) (2.17)the coefficients of which are elementary symmetric polynomials in the B (r);,i < u(r).Thus the coefficients are symmetric polynomials in thea(k), ,A(k)o,+°° ,a(k)mkfor each k. Say the polynomial is of the formH(r)YB, (A(1),-+ ,A(n))i=0