Difference between revisions of "Chebyshev polynomials of the first kind"
(→Rational roots) |
|||
Line 57: | Line 57: | ||
===Rational roots=== | ===Rational roots=== | ||
− | + | The rational roots of <math>T_n(x) - 1</math> for any <math>n</math> must be elements of the set <math>S = \{ -1, -\frac{1}{2}, 0, \frac{1}{2}, 1 \}</math>. | |
− | |||
− | Because any value <math>\cos \frac{a\pi}{b}</math> for integers <math>a</math> and <math>b</math> is a root of <math> | + | Any root other than <math>1</math> of <math>T_{2n+1}(x) - 1</math> must be a root of <math>W_n(x)</math>. The polynomials <math>w_n(x) = W_n \left( \frac{x}{2} \right)</math> satisfy the recursive formulas <cmath>\begin{align*} w_0(x) &= 1 \\ w_1(x) &= x + 1 \\ w_{n+1}(x) &= xw_n(x) - w_{n-1}(x), \end{align*}</cmath> ensuring that they have integer coefficients and are monic for all <math>n</math>. Furthermore, their constant terms satisfy <cmath>w_{n+1}(0) = -w_{n-1}(0),</cmath> and so are always either <math>-1</math> or <math>1</math>. Thus, by the [[Rational Root Theorem]], any rational roots of <math>w_n(x)</math> must be <math>-1</math> or <math>1</math>, so the only possible rational roots of <math>W_n(x)</math> are <math>-\frac{1}{2}</math> and <math>\frac{1}{2}</math>. Thus, the only possible rational roots of <math>T_{2n+1}(x) - 1</math> are <math>-\frac{1}{2}</math>, <math>\frac{1}{2}</math>, and <math>1</math>, each an element of <math>S</math>. |
+ | |||
+ | By the composition identity, <math>T_{2n}(x) - 1 = T_{n}(T_{2}(x)) - 1</math>. If <math>x</math> is rational, then <math>T_2(x) = 2x^2 - 1</math> is rational also. Therefore, if all rational roots of <math>T_n(x) - 1</math> lie in <math>S</math>, then any rational root of <math>T_{2n}(x) - 1</math> must be a solution of <math>2x^2 - 1 = s</math> for some <math>s \in S</math>. The following five cases show that the rational roots of <math>T_{2n}(x) - 1</math> must lie in <math>S</math> as well: | ||
+ | <cmath>\begin{align*} 2x^2 = 0 &\Longrightarrow x = 0, \\ 2x^2 - \frac{1}{2} = 0 &\Longrightarrow x = \pm \frac{1}{2}, \\ 2x^2 - 1 = 0 &\Longrightarrow x = \pm \frac{\sqrt 2}{2}, \\ 2x^2 - \frac{3}{2} = 0 &\Longrightarrow x = \pm \frac{\sqrt 3}{2}, \\ 2x^2 - 2 = 0 &\Longrightarrow x = \pm 1. \end{align*}</cmath> | ||
+ | |||
+ | Because any value <math>\cos \frac{a\pi}{b}</math> for integers <math>a</math> and <math>b</math> is a root of <math>T_{2b}(x) - 1</math>, the five elements of <math>S</math> are the only rational values of the cosine of a rational multiple of <math>\pi</math>. This result is known as <b>Niven's Theorem</b>. |
Revision as of 11:05, 3 March 2022
The Chebyshev polynomials of the first kind are defined recursively by or equivalently by
Contents
Proof of equivalence of the two definitions
In the proof below, will refer to the recursive definition.
For the base case, for the base case,
Now for the inductive step, let , so that . We then assume that and , and we wish to prove that .
From the cosine sum and difference identities we have and The sum of these equations is rearranging, Substituting our assumptions yields as desired.
Composition identity
For nonnegative integers and , the identity holds.
First proof
By the trigonometric definition, .
As before, let . We have for some integer . Multiplying by and distributing gives ; taking the cosine gives .
For now this proof only applies where the trigonometric definition is defined; that is, for . However, is a degree- polynomial, and so is , so the fact that for some distinct is sufficient to guarantee that the two polynomials are equal over all real numbers.
Second proof (Induction)
First we prove a lemma: that for all . To prove this lemma, we fix and induct on .
For all , we have and for all , proving the lemma for and respectively.
Suppose the lemma holds for and ; that is, and . Then multiplying the first equation by and subtracting the second equation gives which simplifies to using the original recursive definition, as long as . Thus, the lemma holds for (as long as ), completing the inductive step.
To prove the claim, we now fix and induct on .
For all , we have and proving the claim for and respectively.
Suppose the claim holds for and ; that is, and . We may also assume , since the smaller cases have already been proven, in order to ensure that . Then by the lemma (with ) and the original recursive definition, completing the inductive step.
Roots
Since , and the values of for which are for integers , the roots of are of the form
These roots are also called Chebyshev nodes.
Connection to roots of unity
Because cosine is only at integer multiples of , the roots of the polynomial follow the simpler formula . The th roots of unity have arguments of and magnitude , so the roots of are the real parts of the th roots of unity. This lends intuition to several patterns.
All roots of are also roots of , since all th roots of unity are also th roots of unity. This can also be shown algebraically as follows: Suppose . Then using the composition identity and the fact that for all .
Particular cases include that , being a root of , is a root of for all positive , and , being a root of , is a root of for all positive even . All other roots of correspond to roots of unity which fall into conjugate pairs with the same real part. One might therefore suspect that all remaining roots of are double roots. In fact, and for polynomials and satisfying the same recurrence relation as , but with the different base cases and . The polynomials are known as Chebyshev polynomials of the second kind.
Rational roots
The rational roots of for any must be elements of the set .
Any root other than of must be a root of . The polynomials satisfy the recursive formulas ensuring that they have integer coefficients and are monic for all . Furthermore, their constant terms satisfy and so are always either or . Thus, by the Rational Root Theorem, any rational roots of must be or , so the only possible rational roots of are and . Thus, the only possible rational roots of are , , and , each an element of .
By the composition identity, . If is rational, then is rational also. Therefore, if all rational roots of lie in , then any rational root of must be a solution of for some . The following five cases show that the rational roots of must lie in as well:
Because any value for integers and is a root of , the five elements of are the only rational values of the cosine of a rational multiple of . This result is known as Niven's Theorem.