approximating algebraic numbers with linear recurrences
Linear recurrences can be used to obtain rational approximations for realalgebraic numbers. Suppose that is the root of a polynomial
with rational coefficients andfurther assume that the roots of are distinct and that all theother roots of are strictly smaller in absolute value
than .
Consider the recursion
As discussed in the parent entry, the solution of this recurrenceis
where are the roots of — let us agree that — and the ’s are determined by the initial conditionsof the recurrence. If , we have
Because when , we have , and hence
To illustrate this method, we will compute the square root of two. Now,we cannot use the equation because it has two roots, and , which are equal in absolute value. So whatwe shall do instead is to use the equation , or , whose roots are and . Since, we can use our method to approximatethe larger of these roots, namely ; to approximate ,we subtract from our answer. The recursion we should use is
or, moving terms around,
If we choose , then we have
Starting with these values, we obtain the following sequence:
Therefore, as our approximations to , we have
By making suitable transformations, one can compute all the rootsof a polynomial using this technique. A way to do this is to startwith a rational number which is closer to the desired root thanto the other roots, then make a change of variable .
As an example, we shall examine the roots of .Approximating the polynomial by leaving off the constant term, weguess that the roots are close to , , and . Since thetwo complex roots are conjugates, it suffices to find one of them.
To look for the root near , we make the change of variable to obtain , which gives the recursion. Picking some initial values, thisrecursion gives us the sequence
whence we obtain the approximations
To look for the root near , we make the change of variable to obtain , which gives the recursion
Picking some initial values, thisrecursion gives us the sequence