3.4 Chebyshev Polynomials of the First Kind

Definition 3.4.1. Chebyshev’s polynomials of the first kind denoted by \(T_n(x)\), are defined by \[T_n(x) = \cos \left (n\, \cos ^{-1}x\right ),\] \(n= 0,\, 1,\, 2,\, \cdots \) and \(0\leq \cos ^{-1}(x) \leq \pi \).

We immediately set that \(\, -1 \leq T_n(x) \leq 1\).

Theorem 3.4.2. \[T_n(x) = \frac {1}{2}\left [\left \{x + i\sqrt {1 - x^2}\right \}^n + \left \{x - i\sqrt {1 - x^2}\right \}^n\right ]\]

Proof. Let \(x = \cos \theta \). Then \begin {align*} T_n(x) = \cos \left (n\, \cos ^{-1}x\right ) & = \cos n\theta \\ & = \frac {1}{2}\left [e^{in\theta } + e^{-in\theta }\right ]\\ & = \frac {1}{2}\left [\left (e^{i\theta }\right )^n + \left (e^{-i\theta }\right )^n\right ]\\ & = \frac {1}{2}\left [\left \{\cos \theta + i\sin \theta \right \}^n + \left \{\cos \theta - i\sin \theta \right \}^n\right ]\\ & = \frac {1}{2}\left [\left \{x + i \sqrt {1 - x^2}\right \}^n + \left \{x - i\sqrt {1 - x^2}\right \}^n \right ]. \end {align*} □

Theorem 3.4.3. \[T_n(x) = \sum _{k =0 }^{\left [\frac {n}{2}\right ]}\frac {(-1)^n\, n!}{(2k)!\, (n-2k)!}\, (1 - x^2)^k\, x^{n-2k}.\]

Proof. From 4.4.2 \begin {align*} T_n(x) & = \frac {1}{2}\left [\left \{x + i\sqrt {1 - x^2}\right \}^n + \left \{x - i\sqrt {1 - x^2}\right \}^n\right ]\\ & = \frac {1}{2}\left [\sum ^{n}_{k = 0} \binom {n}{k}\, x^{n-k}\, \left \{i\sqrt {1 - x^2}\right \}^k \, + \, \sum _{k=0}^n\binom {n}{k}\, x^{n-k}\, \left \{-i\sqrt {1 - x^2}\right \}^k\right ]\\ & = \frac {1}{2}\sum ^n_{k=0}\binom {n}{k}\, x^{n-k}\, (1 - x^2)^{k/2}\, \left [1 + (-1)^k\right ] \, i^k\\ & = \frac {1}{2}\sum ^n_{k=0}\binom {n}{k}x^{n-k}(1-x^2)^{k/2}\left [1 + (-1)^k\right ]\, i^k. \end {align*}

Now when \(k = \) odd then \(1 + (-1)^k = 0\), and when \(k = \) even then \(1 + (-1)^k = 2\). Hence, \[T_n(x) = \frac {1}{2}\sum _{k = \, \text {even}\, \leq n} \binom {n}{k} x^{n-k}(1 - x^2)^{k/2}\, i^k \cdot 2\] But when \(k = \) even, we have \(k = 2s\) for some integers \(s\). \(k \leq n\) means that \(s \leq \frac {n}{2}\). Since \(s\) is an integer, \(s \leq \frac {n}{2}\) means \(s \leq \left [\frac {n}{2}\right ]\). Thus \begin {align*} T_n(x) & = \sum ^{\left [\frac {n}{2}\right ]}_{s = 0} \binom {n}{2s} x^{n-2s} (1 - x^2)^{s}\, i^{2s}\\ & = \sum ^{\left [\frac {n}{2}\right ]}_{s =0} \frac {n!}{(2s)!\, (n-2s)!}x^{n-2s}(1 - x^2)^s(-1)^s. \end {align*} □

Using theorem 4.4.3 we have \[ T_0(x) = 1, \hspace {0.5cm} T_1(x) = x, \hspace {0.5cm} T_2(x) = 2x^2 - 1, \hspace {0.5cm} T_3(x) = 4x^3 - 3x, \hspace {0.5cm} T_4(x) = 8x^4 - 8x^2 + 1,\] \[ T_5(x) = 16x^5 - 20x^3 + 5x.\]

Theorem 3.4.4.

(a).
\(T_n(1) = 1\)
(b).
\(T_n(-1) = (-1)^n\)
(c).
\(T_n(-x) = (-1)^n\, T_n(x)\)
(d).
\(T_{2n}(0) = (-1)^n\)
(e).
\( T_{2n+1}(0) = 0.\)

Proof. We use the definitions of \(T_n(x)\)

(a).
\(T_n(1) = \cos \left (n\, \cos ^{-1} 1\right ) = \cos (n\, (0)) = 1.\)
(b).
\(T_n(-1) = \cos \left (n\, \cos ^{-1}\, (-1)\right ) = \cos (n\pi ) = (-1)^n.\)
(c).
\(T_n(-x)\) \begin {align*} T_n(-x) & = \frac {1}{2}\left [\left \{-x + i\sqrt {1 - x^2}\right \}^n + \left \{-x - i\sqrt {1 - x^2}\right \}^n\right ]\\ & = \frac {1}{2}\left [(-1)^n\left \{x - i\sqrt {1 - x^2}\right \}^n + (-1)^n\left \{x + i\sqrt {1 - x^2}\right \}^n\right ]\\ & = (-1)^n\, T_n(x). \end {align*}
(d).
\(T_{2n}(0) = \frac {1}{2}\left [(i)^{2n} + (-i)^{2n}\right ] = \frac {1}{2}\left [(-1)^n + (-1)^n\right ] = (-1)^n\,\) by using 4.4.2
(e).
\(T_{n+1}(0) = \frac {1}{2}\left [(i)^{2n+1} + (-i)^{2n+1}\right ] = \frac {1}{2}\left [(-1)^n\, i - (-1)^n\, i\right ] = 0.\)

Exercise 3.4.5. Show that \(\, T_n(x)\,\) is a solution of Chebyshev’s equation \[(1 - x^2)\frac {d^2y}{dx^2} - x\frac {dy}{dx} + n^2 y = 0.\] Hint: let \(y = T_n(x)\).

Theorem 3.4.6 (Orthogonality property.). \[\int _{-1}^1\frac {T_m(x)\, T_n(x)}{\sqrt {1 - x^2}}\, dx = \begin {cases} 0, & m\neq n\\ \frac {\pi }{2}, & m = n \neq 0\\ \pi , & m = n = 0. \end {cases}\]

Proof. Let \(\, x = \cos \theta , \hspace {0.5cm} dx = -\sin \theta \, d\theta \,\,\) and \(\sqrt {1 - x^2} = \sin \theta \), so \begin {align*} \int _{-1}^1\frac {T_m(x)\, T_n(x)}{\sqrt {1 - x^2}}\, dx & = \int _{\pi }^0\frac {T_m(\cos \theta )\, T_n(\cos \theta )\, (-\sin \theta \, d\theta )}{\sin \theta }\\ & = \int ^{\pi }_0\cos n\theta \, \cos m\theta \, d\theta , \end {align*}

\begin {align*} T_n(\cos \theta ) = \cos \left (n\, \cos ^{-1}(\cos \theta )\right ) & = \cos n\theta \\ & = \frac {1}{2}\int _0^{\pi }\left [\cos (m+n)\theta + \cos (n-m)\theta \right ]\, d\theta \\ & = \frac {1}{2}\left [\frac {\sin (m+n)\theta }{m+n} + \frac {\sin (n-m)\theta }{n-m}\right ]^{\pi }_0\\ & = 0, \hspace {0.6cm} m\neq n. \end {align*}

If \(m = n \neq 0\), we have \begin {align*} \int _{-1}^1T_n(x)\, T_n(x)\, dx & = \int _0^{\pi }\cos ^2 n\theta \, d\theta \\ & = \int _0^{\pi }\frac {1}{2}(1 + \cos 2n\theta )\, d\theta \\ & = \frac {\pi }{2}. \end {align*}

If \(m = n = 0,\) \[\int ^1_{-1}\frac {T_0(x)\, T_0(x)}{\sqrt {1 - x^2}}\, dx = \int ^{\pi }_0 d\theta = \pi .\] □

Theorem 3.4.7 (Recurrence Relations).

(a).
\(T_{n+1}(x) - 2xT_n(x) + T_{n-1}(x) = 0\)
(b).
\((1 - x^2)T_n'(x) = -nxT_n(x) + nT_{n-1}(x).\)

Notice that \[x = \cos \theta \, \implies \, \begin {cases} T_{n+1}(x) & = \cos (n+1)\theta \\ T_{n-1}(x) & = \cos (n-1)\theta . \end {cases}\]

Exercise 3.4.8.

(a).
Show that \[\, T_{m+n}(x) + T_{n-m}(x) = 2\, T_m(x)\, T_n(x).\]
(b).
Show that \[2\left \{T_n(x)\right \}^2 = 1 + T_{2n}(x).\]
(c).
Show that \[\left \{T_n(x)\right \}^2 - T_{n+1}(x)\, T_{n-1}(x) = 1 - x^2.\]
(d).
Show that \[T_m(T_n(x)) = T_n(T_m(x)) = T_{m+n}(x).\]
(e).
Let \(\{T_n(x)\}^{\infty }_{n=0}\) be a sequence of the Chebyshev polynomials of the first kind. Let \[\xi _i = \cos \left [(2i - 1)\frac {\pi }{2n}\right ]\, , \hspace {0.4cm} i = 1,\, 2,\, \cdots \, n.\]
(i).
Verify that \(\xi _1, \, \xi _2, \, \cdots \, , \, \xi _n\) are zeros of \(T_n(x)\) that is \(\,T_n(\xi _i) = 0\,\) for \(i = 1, \, 2, \, \cdots \, , \, n\).
(ii).
Show that \(\xi _1, \, \xi _2, \, \cdots \, , \, \xi _n\) are simple zeros of \(\,T_n(x)\). i.e \(\,T_n'(\xi _i) \neq 0, \hspace {0.3cm} i = 1, \, 2,\, \cdots \, , \, n.\)

x−1−1TTTT 1 11234

Figure 3.2: The Chebyshev polynomials \(T_1\) to \(T_4\) on \([-1,1]\). Each oscillates between \(-1\) and \(+1\) and attains both bounds \(n+1\) times: the equal-ripple property. Setting \(x=\cos \theta \) turns \(T_n\) into \(\cos n\theta \), which is where the oscillation comes from, and it is this uniform bound that makes these the polynomials of choice for approximation.

Questions on this section

Stuck on something here? Ask below and it stays attached to this topic.