19.07.2014 Views

Contents - Student subdomain for University of Bath

Contents - Student subdomain for University of Bath

Contents - Student subdomain for University of Bath

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

3.1. EQUATIONS IN ONE VARIABLE 59<br />

In fact, 2 if we choose such a polynomial “at random”, the probability <strong>of</strong> its<br />

having a solution that can be expressed in terms <strong>of</strong> radicals is zero. Of course,<br />

any particular quintic, or higher degree equation, may have solutions expressible<br />

in radicals, such as x 5 − 2, whose solutions are 5√ 2, but this is the exception<br />

rather than the rule.<br />

Hence algebra systems, if they handle such concepts, can only regard the<br />

roots <strong>of</strong> such equations as being defined by the polynomial <strong>of</strong> which they are<br />

a root. A Maple example 3 is given in figure 1.5.1, where the Maple operator<br />

RootOf is generated. It is normal to insist that the argument to RootOf (or its<br />

equivalent) is square-free: then different rooots are genuinely different. Then α,<br />

the first root <strong>of</strong> f(x), satisfies f(α) = 0, the second root β satisfies f(x)/(x−α) =<br />

0, and so on. Even if f is irreducible, these later polynomials may not be, but<br />

determining the factorisations if they exist is a piece <strong>of</strong> Galois theory which<br />

would take us too far out <strong>of</strong> our way [FM89]. It is, however, comparatively easy<br />

to determine the Monte Carlo question: “such factorisations definitely do not<br />

exist”/“they probably do exist” [DS00].<br />

3.1.5 Reducible defining polynomials<br />

It should be noted that handling such constructs when the defining polynomial<br />

is not irreducible can give rise to unexpected results. For example, in Maple,<br />

1<br />

if α is RootOf(x^2-1,x), then<br />

α−1<br />

returns that, but attempting to evaluate<br />

this numerically gives infinity, which is right if α = 1, but wrong if α = −1,<br />

the other, equally valid, root <strong>of</strong> x 2 − 1. In this case, the mathematical answer<br />

to “is α − 1 zero?” is neither ‘yes’ nor ‘no’, but rather ‘it depends which α<br />

you mean’, and Maple is choosing the 1 value (as we can see from 1<br />

α+1 , which<br />

evaluates to 0.5). However, the ability to use polynomials not guaranteed to<br />

be irreducible can be useful in some cases — see section 3.3.7. In particular,<br />

algorithm 10 asks if certain expressions are invertible, and a ‘no’ answer here<br />

entrains a splitting into cases, just as asking “is α−1 zero?” entrains a splitting<br />

<strong>of</strong> RootOf(x^2-1,x).<br />

In general, suppose we are asking if g(α) is invertible, where α = RootOf(f(x), x),<br />

i.e. we are asking <strong>for</strong> d(α) such that d(α)g(α) = 1 after taking account <strong>of</strong> the<br />

fact that α = RootOf(f(x), x). This is tantamount to asking <strong>for</strong> d(x) such that<br />

d(x)g(x) = 1 modulo f(x) = 0, i.e. d(x)g(x) + c(x)f(x) = 1 <strong>for</strong> some c(x). But<br />

applying the Extended Euclidean Algorithm (Algorithm 4) to f and g gives us<br />

c and d such that cf + dg = gcd(f, g). Hence if the gcd is in fact 1, g(α) is<br />

invertible, and we have found the inverse.<br />

If in fact the gcd is not 1, say some h(x) ≠ 1, then we have split f as f = hĥ,<br />

2 The precise statement is as follows. For all n ≥ 5, the fraction <strong>of</strong> polynomials <strong>of</strong> degree n<br />

and coefficients at most H which have a root expressible in radicals tends to zero as H tends<br />

to infinity.<br />

3 By default, Maple will also use this <strong>for</strong>mulation <strong>for</strong> roots <strong>of</strong> most quartics, and the expression<br />

in figure 3.2 is obtained by convert(%,radical) and then locating the sub-expressions by<br />

hand. This can be seen as an application <strong>of</strong> Carette’s view <strong>of</strong> simplification (page 21), though<br />

historically Carette’s paper is a retrospective justification.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!