34404 Chebyshev多項式 和Morgan-Voyce多項式、Fibonacci數、Pell數、Lucas數等的關係探討
Chebyshev多項式 和Morgan-Voyce多項式、Fibonacci數、Pell數、Lucas數等的關係探討

1. 前言

Chebyshev 多項式在數學方面都有很多之應用 (參見), 譬如在組合數學、數論方面等等。

1959 年, A. M. Morgan-Voyce 研究 electric ladder network of resistors , 藉由計算 ladder network 的等效電阻, 他發現了兩個多項式家族: $b_n(x)$ 和 $B_n(x)$, 其中

$b_n(x)$ 和 $B_n(x)$ 的遞迴關係式如下: \begin{eqnarray} b_n(x) &=& xB_{n-1}(x) + b_{n-1}(x) \\ %(1.1) B_n(x) &=& (x+1)B_{n-1}(x) + b_{n-1}(x) %(1.2) \end{eqnarray} 其中 $b_0(x)=B_0(x)=1$ 。

而 M. N. S. Swamy, S. L. Basin of Sylnania Electronic System, V. E. Hoggatt, Jr., 和 M. Bicknell 等人更在 1967 和 1968 深入的研究這些多項式的特性

本篇文章的主要目的有三:

  1. 簡化 Morgan-Voyce 多項式 $b_n(x)$、 $B_n(x)$ 一般式的計算過程, 並得到 $b_n(x)$、 $B_n(x)$ 與 Chebyshev 多項式的關係式。 (見第 2 節)
  2. 利用簡單的計算得到一般性的數列 $\{a_n\}$ 滿足遞迴關係式: \begin{equation} a_n = a\cdot a_{n-1} + b\cdot a_{n-2}, ~~a_0 = c, ~~a_1 = d, \quad \hbox{其中 $a$, $b$, $c$, $d$ 為任意實數} \end{equation} 與 Chebyshev 多項式的關係式。 進而得到 Fibonacci 數、 Pell 數、 Jacobsthal 數、 Mersenne 數、 Lucas 數、 Pell-Lucas 數、 Jacobsthal-Lucas 數、 Fermat 數和 Chebyshev 多項式的關係式。 (見第 3 節)
  3. 我們也將 (1.3) 式之數列 $\{a_n\}$ 以 $e^\theta$ 之形式表示出來。 (見第 4 節)

2. Chebyshev 多項式與 Morgan-Voyce 多項式

首先, 讓我們來看什麼是Chebyshev多項式。

Chebyshev 多項式是與棣美弗定理 (de Moivre identity) 有關, 以遞迴方式定義的一系列正交多項式序列。 Chebyshev 多項式在逼近理論中有重要的應用, 可用於多項式插值, 並且提供多項式在連續函數的最佳一致逼近

我們知道 \begin{eqnarray*} && \cos 0\theta = 1 \\ && \cos 1\theta = \cos\theta \\ && \cos 2\theta = 2\cos^2\theta - 1 \\ && \cos 3\theta = 4\cos^3\theta - 3\cos\theta \\ && \cos 4\theta = 8\cos^4\theta - 8\cos^2\theta + 1 \\ && \cos 5\theta = 16\cos^5\theta - 20\cos^3\theta + 5\cos\theta \\ && \cos 6\theta = 32\cos^6\theta - 48\cos^4\theta + 18\cos^2\theta - 1 \\ \hbox{如 }~ \cos 2\theta &=& 2\cos^2\theta - 1 = 2\cos\theta(\cos 1\theta) - \cos 0\theta \\ \cos 4\theta &=& 8\cos^4\theta - 8\cos^2\theta + 1 = 2\cos\theta(\cos 3\theta) - \cos 2\theta \hskip 3cm~ \end{eqnarray*} 由於上述的三角恆等式, Chebyshev 多項式的定義如下: $$ P_n(x) = 2xP_{n-1}(x) - P_{n-2}(x), \quad P_0(x) = 1 $$ 隨著 $P_1(x)$ 不同的值, 共可分成四種形式的 Chebyshev 多項式 (表一)

舉例: 若令 $x=\cos\theta$, 我們有 $T_n(\cos\theta)=\cos n\theta$ \begin{eqnarray*} && T_0(x) = 1 \\ && T_1(x) = x \\ && T_2(x) = 2x^2 - 1 \\ && T_3(x) = 4x^3 - 3x \\ && T_4(x) = 8x^4 - 8x^2 + 1 \\ && T_5(x) = 16x^5 - 20x^3 + 5x \\ && T_6(x) = 32x^6 - 48x^4 + 18x^2 -1\hskip 3cm~ \end{eqnarray*} 所以 Chebyshev 多項式第一類是由 $\cos n\theta$ 的展開式所定義出來的。

(表一)

第一類
$T_n(x)$
第二類
$U_n(x)$
第三類
$V_n(x)$
第四類
$W_n(x)$
三角定義
$x=\cos\theta$
$\cos n\theta$ $\displaystyle\frac{\sin(n+1)\theta}{\sin\theta}$ $\displaystyle\frac{\cos(n+1/2)\theta}{\cos(\theta/2)}$ $\displaystyle\frac{\sin(n+1/2)\theta}{\sin(\theta/2)}$
$n=1$ $x$ $2x$ $2x-1$ $2x-1$
和 $U_n(x)$ 的關係 $\displaystyle\frac{U_n(x)\!-\!U_{n-2}(x)}{2}$ $-$ $U_n(x)-U_{n-1}(x)$ $U_n(x)+U_{n-1}(x)$
一般式
$\alpha_i=x\!+\!\sqrt{x^2\!-\!1}$
$\beta_i=x\!-\!\sqrt{x^2\!-\!1}$
$i=1, 2, 3, 4$
$\displaystyle\frac{\alpha_1^n+\beta_1^n}{2}$ $\displaystyle\frac{\alpha_2^{n+1}\!-\!\beta_2^{n+1}}{\alpha_2\!-\!\beta_2}$ $\displaystyle\frac{\alpha_3^n(\alpha_3\!-\!1)\!-\!\beta_3^n(\beta_3\!-\!1)}{\alpha_3\!-\!\beta_3}$ $\displaystyle\frac{\alpha_4^n(\alpha_4\!+\!1)\!-\!\beta_4^n(\beta_4\!+\!1)}{\alpha_4\!-\!\beta_4}$

由 (1.1) 及 (1.2) 可觀察出 $b_n(x)$ 和 $B_n(x)$ 的前幾項: \begin{eqnarray*} && b_0(x) = 1, ~~b_1(x) = x+1, ~~b_2(x) = x^2+3x+1 ~~\cdots\cdots \\ && B_0(x) = 1, ~~B_1(x) = x+2, ~~B_2(x) = x^2+4x+3 ~~\cdots\cdots \end{eqnarray*} 將 (1.1) 及 (1.2) 經由整理, 可以得到下列式子: \begin{eqnarray} b_n(x) &=& (x+2)b_{n-1}(x) - b_{n-2}(x) \\ %(2.1) B_n(x) &=& (x+2)B_{n-1}(x) - B_{n-2}(x) %(2.2) \end{eqnarray}

2.1. Chebyshev 多項式 $V_n(y)$ 和 Morgan-Voyce 多項式 $b_n(x)$ 的關係

Morgan-Voyce 多項式 $b_n(x)$ \begin{equation} %(2.1) b_n(x) = (x+2)b_{n-1}(x) - b_{n-2}(x), ~~n \ge 2, ~~b_0(x) = 1, ~~b_1(x) = x+1 \end{equation}

Chebshev 多項式第三類 $V_n(y)$ $$ V_n(y) = 2yV_{n-1}(y) - V_{n-2}(y), ~~n \ge 2, ~~V_0(y) = 1, ~~V_1(y) = 2y-1 $$ 將 (2.1) 和 Chebyshev 多項式第三類 $V_n(y)$ 做比較, 發現若令 $x+2=2y$, 則 $b_n$ 多項式和 $V_n$ 多項式有同樣的遞迴關係式以及同樣的初始條件: $$ b_n(2y-2) = 2yb_{n-1}(2y-2) - b_{n-2}(2y-2), ~~b_0(2y-2) = 1, ~~b_1(2y-2) = 2y-1 $$ 因此 $b_n(x)$ 就是 Chebyshev 多項式第三類。

而 $V_n(y)$ 的一般式如下: \begin{equation} %(2.3) V_n(y) = \frac{(\alpha_3^{n+1}-\alpha_3^n)-(\beta_3^{n+1}-\beta_3^n)}{\alpha_3-\beta_3}, ~\alpha_3 = y+\sqrt{y^2-1}, ~~\beta_3 = y - \sqrt{y^2-1} \end{equation} 將 $y=\displaystyle\frac{x+2}{2}$ 代入 $\alpha_3$ 和 $\beta_3$ 中, 得到: $$ \alpha_3 = \frac{(x+2)+\sqrt{x^2+4x}}{2} \qquad \beta_3 = \frac{(x+2)-\sqrt{x^2+4x}}{2} $$ 所以 $b_n(x)$ 的一般式為: $$ \frac{\displaystyle\bigg [\Big (\frac{x\!+\!2\!+\!\sqrt{x^2\!+\!4x}}{2}\Big )^{n+1} - \Big (\frac{x\!+\!2\!+\!\sqrt{x^2\!+\!4x}}{2}\Big )^n\bigg ] - \bigg [\Big (\frac{x\!+\!2\!-\!\sqrt{x^2\!+\!4x}}{2}\Big )^{n+1} - \Big (\frac{x\!+\!2\!-\!\sqrt{x^2\!+\!4x}}{2}\Big )^n\bigg ]}{\sqrt{x^2+4x}} $$

將 $y=\displaystyle\frac{x+2}{2}$ 代入 $V_n(y)$ 的三角表示式 $\displaystyle\frac{\cos(n+1/2)\theta}{\cos(\theta/2)}$ 中, 其中 $y=\cos\theta$。

得到 $b_n(x)$ 的三角表示式 $\displaystyle\frac{\cos(n+1/2)\theta}{\cos(\theta/2)}$, 其中 $\displaystyle\frac{x+2}{2}=\cos\theta$。

2.2. Chebyshev 多項式 $U_n(y)$ 和 Morgan-Voyce 多項式 $B_n(x)$ 的關係

Morgan-Voyce 多項式 $B_n(x)$ \begin{equation} %(2.2) B_n(x) = (x+2)B_{n-1}(x) - B_{n-2}(x), ~~B_0(x) = 1, ~~B_1(x) = x+2 \end{equation}

Chebyshev 多項式第二類 $U_n(y)$ $$ U_n(y) = 2yU_{n-1}(y) - U_{n-2}(y), ~~U_0(y) = 1, ~~U_1(y) = 2y $$ 將 (2.2) 和Chebyshev 多項式第二類 $U_n(y)$ 做比較, 發現若令 $x+2=2y$, 則 $B_n$ 多項式和 $U_n$ 多項式有同樣的遞迴關係式以及同樣的初始條件: $$ B_n(2y-2) = 2yB_{n-1}(2y-2) - B_{n-2}(2y-2), ~~B_0(2y-2) = 1, ~~B_1(2y-2) = 2y, $$ 因此 $B_n(x)$ 多項式就是 Chebyshev 多項式第二類。

而 $U_n(y)$ 的一般式如下: \begin{equation} %(2.4) U_n(y) = \frac{\alpha_2^{n+1}-\beta_2^{n+1}}{\alpha_2-\beta_2}, ~~\alpha_2 = y+\sqrt{y^2-1}, ~~\beta_2 = y-\sqrt{y^2-1} \end{equation} 將 $y=\displaystyle\frac{x+2}{2}$ 代入中 $\alpha_2$ 和 $\beta_2$ 中, 得到: $$ \alpha_2 = \frac{(x+2)+\sqrt{x^2+rx}}{2}, \qquad \beta_2 = \frac{(x+2)-\sqrt{x^2+4x}}{2} $$ 所以 $B_n(x)$ 的一般式為 $B_n(x)=\displaystyle\frac{(\frac{x+2+\sqrt{x^2+4x}}{2})^{n+1}-(\frac{x+2-\sqrt{x^2+4x}}{2})^{n+1}}{\sqrt{x^2+4x}}$。

將 $y=\displaystyle\frac{x+2}{2}$ 代入 $U_n(y)$ 的三角表示式 $\displaystyle\frac{\sin(n+1)\theta}{\sin\theta}$ 中, 其中 $y=\cos\theta$。

得到 $B_n(x)$ 的三角表示式 $\displaystyle\frac{\sin(n+1)\theta}{\sin\theta}$, 其中 $\displaystyle\frac{x+2}{2}=\cos\theta$。

[註] Morgan-Voyce 在他的文章中, 並未發現上述與 Chebyshev 多項式關係之事實, 因此他們的結果雖與本文相同, 但他們的計算很長。

3. Fibonacci數、Pell數、Lucas數等和 Chebyshev多項式的關係

讓我們先從較一般性的遞迴數列看起: \begin{equation} %(3.1) a_n = a\cdot a_{n-1} + b\cdot a_{n-2} ~(n\ge 2), ~~a_0 = c, ~~a_1 = d, ~\hbox{其中 $a$, $b$, $c$, $d$ 為任意實數。} \end{equation}

利用生成函數求得 $a_n$ 的一般式: \begin{equation} %(3.2) a_n \!=\! \frac{c(\alpha^{n+1}\!-\!\beta^{n+1})\!+\!(d\!-\!ac)(\alpha^n-\!\beta^n)}{\alpha-\beta}, ~~\alpha \!=\! \frac{a\!+\!\sqrt{a^2\!+\!4b}}{2}, ~\beta \!=\! \frac{a\!-\!\sqrt{a^2\!+\!4b}}{2} \end{equation}

已知 Chebyshev 多項式 $P_n(x)$ 一般式的兩根分別為 $r(x)$ 和 $s(x)$, 其中 \begin{eqnarray*} r(x) &=& x + \sqrt{x^2-1} \\ s(x) &=& x - \sqrt{x^2-1} \end{eqnarray*}

[註] 在本文中, 不考慮等根的情況。

由於想求得 Chebyshev 多項式和一般遞迴數列之間的關係, 不妨令 $r(x)=k\alpha$, $s(x)=k\beta$, $k$ 不限為實數。

若令 $r(x)=k\alpha$, $s(x)=k\beta$, $k$ 為實數, 得到: \begin{eqnarray} x + \sqrt{x^2-1} &=& k\alpha \\ %(3.3) x - \sqrt{x^2-1} &=& k\beta %(3.4) \end{eqnarray} 將 (3.3) (3.4) 兩式相加, 得到: \begin{equation} %(3.5) 2x = k(\alpha+\beta) = ka \end{equation} 將 (3.3) (3.4) 兩式相減, 得到: \begin{equation} %(3.6) 2\sqrt{x^2-1} = k(\alpha-\beta) = k\sqrt{a^2+4b} \end{equation} 再將 (3.5) 代入 (3.6), 並將等式左右兩邊平方, 得到: \begin{eqnarray} && \hskip -13pt k^2a^2 - 4 = k^2a^2 + 4k^2b \nonumber \\ && \rightarrow -4 = 4k^2b \nonumber \\ && \rightarrow k^2 = \frac{1}{-b} \\ %(3.7) && \rightarrow k = \pm \Big (-\frac{1}{b}\Big )^{1/2} %(3.8) \end{eqnarray}

[註] 在本文只考慮 $k=\displaystyle\Big (\!\!-\frac{1}{b}\Big )^{1/2}$ 的情況, 而 $k=-\displaystyle\Big (\frac{-1}{b}\Big )^{1/2}$ 的情況可用相同方法處理。

利用已知: $U_n(x)=\displaystyle\frac{r^{n+1}-s^{n+1}}{r-s}$ 及 $U_{n-1}(x)=\displaystyle\frac{r^n-s^n}{r-s}$。

並將 $r(x)=k\alpha$、 $s(x)=k\beta$ 及 (3.5)、 (3.8) 代回 (3.2) 中, 得到: \begin{equation} %(3.9) a_n = \Big (\!\!-\frac{1}{b}\Big )^{-n/2} cU_n \Big (\frac{a}{2} \Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big ) + \Big (\!\!-\frac{1}{b}\Big )^{(1-n)/2} (d-ac) U_{n-1} \Big (\frac{a}{2} \Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big ) \end{equation} 若想要將 $a_n$ 單純用 $U_n(x)$ 表示, 則在 (3.9) 中, 令 $d-ac=0$, 得到 $d=ac$。故 \begin{equation} %(3.10) a_n = \Big (\!\!-\frac{1}{b}\Big )^{-n/2} cU_n \Big (\frac{a}{2} \Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big ) \quad \hbox{其中 $a$, $c$ 為任意實數, $d=ac$。} \end{equation} 若想要將 $a_n$ 單純用 $U_{n-1}(x)$ 表示, 則在 (3.9) 中, 令 $c=0$, 得到 $d-ac=d$。故 \begin{equation} %(3.11) a_n = \Big (\!\!-\frac{1}{b}\Big )^{(1-n)/2} dU_{n-1} \Big (\frac{a}{2} \Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big ) ~\hbox{ 其中 $a$, $d$ 為任意實數, $c=0$。} \end{equation} 由 (3.10) (3.11) 可得定理 3.1:

定理 3.1. $\{a_n\}$ 為一數列, 此數列滿足二次遞迴式如下: $$ a_n = a\cdot a_{n-1} + b\cdot a_{n-2} ~(n\ge 2), ~~a_0=c, ~~a_1=d $$

若 $(a, b, c, d)\!\to\!(a, b, c, ac)$, $a_n\!=\!\displaystyle\Big (\!\!-\frac{1}{b}\Big )^{-n/2} cU_n\Big (\frac{a}{2}\Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big )$, $a$, $c$ 為任意實數。

若 $(a, b, c, d)\!\to\!(a, b, 0, d)$, $a_n\!=\!\displaystyle\Big (\!\!-\frac{1}{b}\Big )^{(1-n)/2} dU_{n-1}\Big (\frac{a}{2}\Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big )$, $a,d$ 為任意實數。

推論 3.1. $\{u_n\}$ 為一數列, 此數列滿足二次遞迴式: $u_n=a\cdot u_{n-1}+b\cdot u_{n-2}$ $(n\ge 2)$,

$u_0=0$, $u_1=1$, 則 $u_n=\displaystyle\Big (\!\!-\frac{1}{b}\Big )^{(1-n)/2} U_{n-1} \Big (\frac{a}{2}\Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big )$。

證明: $\{u_n\}$ 所滿足的遞迴式就是 (3.1) 取 $(a, b, c, d)=(a, b, 0, 1)$, 因此根據定理 3.1, 得到 $$u_n=\displaystyle\Big (\!\!-\frac{1}{b}\Big )^{(1-n)/2} U_{n-1} \Big (\frac{a}{2}\Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big )\hbox{。}$$

推論 3.2. $$ F_n = (-1)^{n-1} (i)^{n-1} U_{n-1} \Big (\frac{i}{2}\Big ) $$ 其中 $\{F_n\}$ 為 Fibonacci 數列, $F_0=0$, $F_1=1$, $F_n=F_{n-1}+F_{n-2}$ $(n\ge 2)$。

證明: Fibonacci 數的遞迴關係式為 $F_n=F_{n-1}+F_{n-2}$, $F_0=0$, $F_1=1$。 即在 (3.1) 取 $(a, b, c, d)=(1, 1, 0, 1)$。根據定理 3.1, 取 $(a, b, c, d)=(1, 1, 0, 1)$, 得到 $F_n=(-1)^{n-1}$ $(i)^{n-1} U_{n-1} \Big (\displaystyle\frac{i}{2}\Big )$。

推論 3.3. $$ p_n = (-1)^{n-1} (i)^{n-1} U_{n-1}(i) $$ 其中 $p_n$ 為 Pell 數, $p_0=0$, $p_1=1$, $p_n=2p_{n-1}+p_{n-2}$ $(n\ge 2)$

證明: Pell 數的遞迴關係式為 $p_n=2p_{n-1}+p_{n-2}$ $(n\ge 2)$, $p_0=0$, $p_1=1$。 即在 (3.1) 中, 取 $(a, b, c, d)=(2, 1, 0, 1)$。 根據定理 3.1, 得到 $p_n=(-1)^{n-1} (i)^{n-1} U_{n-1}(i)$。

推論 3.4. $$ J_n = (-\sqrt{2}i)^{n-1} U_{n-1} \Big (\frac{i}{2\sqrt{2}}\Big ) $$ 其中 $J_n$ 為 Jacobsthal 數, $J_0=0$, $J_1=1$, $J_n=J_{n-1}+2J_{n-2}$ $(n\ge 2)$

證明: Jacobsthal 數的遞迴關係式為 $J_n=J_{n-1}+2J_{n-2}$ $(n\ge 2)$, $J_0=0$, $J_1=1$。 即

在 (3.1) 中, 取 $(a, b, c, d)=(1, 2, 0, 1)$, 根據定理 3.1, 得到 $J_n=(-\sqrt{2}i)^{n-1} U_{n-1} \Big (\displaystyle\frac{i}{2\sqrt{2}}\Big )$。

推論 3.5. $$ M_n = (\sqrt{2})^{n-1} U_{n-1} \Big (\frac{3}{2\sqrt{2}}\Big ) $$ 其中 $M_n$ 為 Mersenne 數, $M_0=0$, $M_1=1$, $M_n=3M_{n-1}-2M_{n-2}$ $(n\ge 2)$

證明: Mersenne 數的遞迴關係式為 $M_n=3M_{n-1}-2M_{n-2}$ $(n\ge 2)$, $M_0=0$, $M_1=1$。 即在 (3.1) 中, 取 $(a, b, c, d)=(3, -2, 0, 1)$, 根據定理 3.1, 得到 $M_n=(\sqrt{2})^{n-1} U_{n-1}$

$\Big (\displaystyle\frac{3}{2\sqrt{2}}\Big)$。

定理3.2. $\{v_n\}$ 為一數列, 此數列滿足二次遞迴式如下: \begin{equation} %(3.12) v_n = a\cdot v_{n-1} + b\cdot v_{n-2} ~(n\ge 2), ~~v_0 = 2, ~~v_1 = a \end{equation} 則 $v_n=\displaystyle 2\Big (\!\!-\frac{1}{b}\Big )^{-n/2} T_n \Big (\frac{a}{2} \Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big )$。

證明: 與定理 3.1的作法一樣, 令 $r(x)=k\alpha$, $s(x)=k\beta$, $k$ 為實數, 得到 (3.3)$\sim$(3.8)。 而 $v_n$ 的一般式為 $c=2$, $d=a$ 代入 (3.2), 得到 \begin{equation} %(3.13) v_n = \frac{2(\alpha^{n+1}-\beta^{n+1})-a(\alpha^n-\beta^n)}{\alpha-\beta} = \frac{\alpha^n(2\alpha-a)+\beta^n(a-2\beta)}{\alpha-\beta} \end{equation} 再利用 $\alpha+\beta=a$, 得到 $2\alpha-a=\alpha-\beta$、 $a-2\beta=\alpha-\beta$ 代入 (3.13), 經由整理得到: \begin{equation} %(3.14) v_n = \alpha^n + \beta^n \end{equation} 而 \begin{equation} %(3.15) T_n \Big (\frac{a}{2} \Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big ) = \frac{\alpha^n+\beta^n}{2} \Big (\!\!-\frac{1}{b}\Big )^{n/2} \end{equation} 將 (3.15) 代入 (3.14), 即得到 \begin{equation} %(3.16) v_n = 2 \Big (\!\!-\frac{1}{b}\Big )^{-n/2} T_n \Big (\frac{a}{2} \Big (\!\!-\frac{1}{b}\Big )^{1/2}\Big ) \hbox{。} \end{equation}

推論 3.6. $$ L_n = 2(-1)^n (i)^n T_n \Big (\frac{i}{2}\Big ) $$ 其中 $\{L_n\}$ 為 Lucas 數列, $L_0=2$, $L_1=1$, $L_n=L_{n-1}+L_{n-2}$ ($n\ge 2$)。

證明: Lucas 數的遞迴關係式為 $L_n=L_{n-1}\!+\!L_{n-2}\,(n\!\ge\! 2)$, $L_0=2$, $L_1=1$。 根據定理

3.2, 取 $(a, b)=(1, 1)$, 得到 $L_n=2(-1)^n(i)^nT_n\Big (\displaystyle\frac{i}{2}\Big )$。

推論 3.7. $$ D_n = 2(-i)^n T_n(i) $$ 其中 $\{D_n\}$ 為 Pell-Lucas 數列, $D_0=2$, $D_1=2$, $D_n=2D_{n-1}+D_{n-2}$ $(n\ge 2)$

證明: Pell-Lucas 數的遞迴關係式為 $D_n=2D_{n-1}+D_{n-2}$ $(n\ge 2)$, $D_0=2$, $D_1=2$。 根據定理 3.2, 取 $(a, b)=(2, 1)$, 得到 $D_n=2(-i)^n T_n(i)$。

推論 3.8. $$ j_n = 2(-\sqrt{2}i)^n T_n \Big (\frac{i}{2\sqrt{2}}\Big ) $$ 其中 $\{j_n\}$ 為 Jacobsthal-Lucas 數列, $j_0=2$, $j_1=1$, $j_n=j_{n-1}+2j_{n-2}$ $(n\ge 2)$

證明: Jacobsthal-Lucas 數的遞迴關係式為$j_n=j_{n-1}+2j_{n-2}$ $(n\ge 2)$,$j_0=2$,

$j_1=1$。 根據定理 3.2, 取 $(a, b)=(1, 2)$, 得到 $j_n=2(-\sqrt{2}i)^n T_n\Big (\displaystyle\frac{i}{2\sqrt{2}}\Big )$。

推論 3.9. $$ f_n = 2(\sqrt{2})^n T_n \Big (\frac{3}{2\sqrt{2}}\Big ) $$ 其中 $\{f_n\}$ 為 Fermat 數列, $f_0=2$, $f_1=3$, $f_n=3f_{n-1}-2f_{n-2}$ $(n\ge 2)$

證明: Fermat 數的遞迴關係式為 $f_n=3f_{n-1}-2f_{n-2}$ $(n\ge 2)$, $f_0=2$, $f_1=3$, 根據

定理 3.2, 取 $(a, b)=(3, -2)$, 得到 $f_n=2(\sqrt{2})^n T_n \Big (\displaystyle\frac{3}{2\sqrt{2}}\Big )$。

4. $a_n$ 的另外一種表示方式

在 Askey 的文章中, 將 $F_n$ 和 $L_n$ 用 $e^\theta$ 表示, 而在 Ismail 的文章中, 是透過和 Chebyshev 多項式的關係, 將 $F_n$ 和 $L_n$ 用 $e^\theta$ 表示成一般式。 而在本文中, 則是透過比上述兩篇文章更直接的方法將 $a_n$ 用 $e^\theta$ 之形式表示。

4.1. $a_n$ 滿足遞迴關係式 $a_n=a\cdot a_{n-1}+b\cdot a_{n-2}$ $(n\ge 2)$, $a_0=c$, $a_1=d$

$a_n$ 滿足遞迴關係式 (3.1), 且 $a_n$ 的一般式如下: \begin{equation} a_n \!=\! \frac{c(\alpha^{n+1}\!-\!\beta^{n+1})\!+\!(d\!-\!ac)(\alpha^n\!-\!\beta^n)}{\alpha-\beta}, ~~a \!=\! \frac{a\!+\!\sqrt{a^2\!+\!4b}}{2}, ~\beta \!=\! \frac{a\!-\!\sqrt{a^2\!+\!4b}}{2} \end{equation} 利用 $\alpha+\beta=a$ 和 $\alpha\beta=-b$, 並考慮 $b\gt 0$ 和 $b\lt 0$ 的情形。

(在這裡不失一般性令 $\alpha\gt \beta$)

4.1.1. $b\gt 0$ 的情形

若 $b\gt 0$, 則兩根一正一負, 即 $\alpha\gt 0$ 且 $\beta\lt 0$

由 $e^\theta\cdot e^{-\theta}=1$ 和 $\alpha\beta=-b$, 令 $\alpha=\sqrt{b}e^\theta$, $\beta=-\sqrt{b}e^{-\theta}$, 並將 $\alpha\cdot\beta$ 代入 $a_n$ 的一般式 (3.2) , 得到: \begin{equation} a_n(\theta) = (\sqrt{b})^{n-1} \frac{c\sqrt{b}(e^{(n+1)\theta} - (-1)^{n+1}e^{-(n+1)\theta}) +(d-ac)(e^{n\theta}-(-1)^ne^{-n\theta})}{e^\theta+e^{-\theta}} \end{equation} 將 $n=0$ 和 $n=1$ 代回 (4.1), 得到 $a_0=c$, $a_1=2c\sqrt{b}\sinh\theta+(d-ac)$。

由以上結果, 可以得到定理 4.1:

定理4.1. $a_n$ 滿足遞迴關係式

$a_n=a\cdot a_{n-1}\!+\!b\cdot a_{n-2}\,(n\!\ge\! 2)$, $b\!\gt \!0$, $a_0\!=\!c$, $a_1=2c\sqrt{b}\sinh\theta\!+\!(d\!-\!ac)$, 其中 $a, c, d$ 為任意實數。 則 $a_n(\theta)=(\sqrt{b})^{n-1} \displaystyle\frac{c\sqrt{b}(e^{(n+1)\theta}\!-\!(-1)^{n+1}e^{-(n+1)\theta}) +(d\!-\!ac)(e^{n\theta}\!-\!(-1)^ne^{-n\theta})}{e^\theta+e^{-\theta}}$

推論 4.1. $$ F_n(\theta) = \frac{e^{n\theta}-(-1)^ne^{-n\theta}}{e^\theta+e^{-\theta}} $$ 其中 $\{F_n(\theta)\}$ 為 Fibonacci 數列, $F_0(\theta)=0$, $F_1(\theta)=1$, $F_n(\theta)=F_{n-1}(\theta)+F_{n-2}(\theta)$ $(n\ge 2)$。

證明: Fibonacci 數的遞迴關係式為 $F_n(\theta)=F_{n-1}(\theta)+F_{n-2}(\theta)$ $(n\ge 2)$, $F_0(\theta)=0$,

$F_1(\theta)=1$。根據定理 4.1, 取 $(a, b, c, d)=(1, 1, 0, 1)$, 得到 $F_n(\theta)=\displaystyle\frac{e^{n\theta}-(-1)^n e^{-n\theta}}{e^\theta+e^{-\theta}}$, 和 Ismail 的結果相同。

4.1.2. $b\lt 0$ 的情形

若 $b\lt 0$, 則兩根同號, 由 $\alpha+\beta=a$ 辨別, 若 $a\gt 0$, 則兩根皆為正根; 若 $a\lt 0$, 則兩根皆為負根。

  1. 若 $a\gt 0$, 令 $\alpha=\sqrt{-b}e^\theta$, $\beta=\sqrt{-b}e^{-\theta}$ 將 $\alpha\cdot\beta$ 代入 $a_n$ 的一般式 (3.2), 得到: \begin{equation} %(4.2) a_n(\theta) = (\sqrt{-b})^{n-1} \frac{c(\sqrt{-b}(e^{(n+1)\theta}-e^{-(n+1)\theta})+(d-ac)(e^{n\theta}-e^{-n\theta})} {e^\theta-e^{-\theta}} \end{equation} 將 $n=0$ 和 $n=1$ 代回 (4.2), 得到 $a_0=c$, $a_1=2c\sqrt{-b}\cosh\theta+(d-ac)$。
  2. 若 $a\lt 0$, 令 $\alpha=-\sqrt{-b}e^\theta$, $\beta=-\sqrt{-b}e^{-\theta}$ 將 $\alpha\cdot\beta$ 代入 $a_n$ 的一般式 (3.2), 得到: \begin{equation} %(4.3) a_n(\theta) = (-\sqrt{-b})^{n-1} \frac{c\cdot(-\sqrt{-b})(e^{(n+1)\theta}-e^{-(n+1)\theta}) +(d-ac)(e^{n\theta}-e^{-n\theta})}{e^\theta-e^{-\theta}} \end{equation} 將 $n=0$ 和 $n=1$ 代回 (4.3), 得到 $a_0=c$, $a_1=-2c\sqrt{-b}\cosh\theta+(d-ac)$。

由以上結果, 可以得到定理 4.2:

定理 4.2. $a_n$ 滿足遞迴關係式 $a_n=a\cdot a_{n-1}+b\cdot a_{n-2}$ $(n\ge 2)$, $b\lt 0$, $a_0=c$, 其中 $a$, $c$, $d$ 為任意實數。則若

  1. $a\gt 0$, $a_1=2c\sqrt{-b}\cosh\theta+(d-ac)$ $a_n(\theta)=(\sqrt{-b})^{n-1} \displaystyle\frac{c\sqrt{-b}(e^{(n+1)\theta}-e^{-(n+1)\theta}) +(d-ac)(e^{n\theta}-e^{-n\theta})}{e^\theta-e^{-\theta}}$
  2. $a\lt 0$, $a_1=-2c\sqrt{-b}\cosh\theta+(d-ac)$ $a_n(\theta)=(-\sqrt{-b})^{n-1} \displaystyle\frac{c\cdot(-\sqrt{-b})(e^{(n+1)\theta}-e^{-(n+1)\theta}) +(d-ac)(e^{n\theta}-e^{-n\theta})}{e^\theta-e^{-\theta}}$

4.2. $a_n$ 滿足遞迴關係式 $a_n=a\cdot a_{n-1}+b\cdot a_{n-2}$ $(n\ge 2)$, $a_0=2$, $a_1=a$

由 (3.14) 知 $a_n$ 的一般式為: \begin{equation} %(4.4) a_n = \alpha^n + \beta^n, ~~\alpha = \frac{a+\sqrt{a^2+4b}}{2}, ~~\beta = \frac{a-\sqrt{a^2+4b}}{2} \end{equation} 考慮 $b\gt 0$ 和 $b\lt 0$ 的情形:

  1. 當 $b\gt 0$ 時, 令 $\alpha=\sqrt{b}e^\theta$, $\beta=-\sqrt{b}e^{-\theta}$, 並將 $\alpha\cdot\beta$ 代入 $a_n$ 的一般式 (4.4), 得到: \begin{equation} %(4.5) a_n(\theta) = (\sqrt{b})^n e^{n\theta} + (-1)^n (\sqrt{b})^n e^{-n\theta} \end{equation} 將 $n=0$ 和 $n=1$ 代回 (4.5), 得到 $a_0=2$, $a_1=2\sqrt{b}\sinh\theta$。
  2. $b\lt 0$, $a\gt 0$ 時, 令 $\alpha=\sqrt{-b}e^\theta$, $\beta=\sqrt{-b}e^{-\theta}$, 並將 $\alpha\cdot\beta$ 代入 $a_n$ 的一般式 (4.4), 得到: \begin{equation} %(4.6) a_n(\theta) = (\sqrt{-b})^n (e^{n\theta}+e^{-n\theta}) \end{equation} 將 $n=0$ 和 $n=1$ 代回 (4.6), 得到 $a_0=2$, $a_1=2\sqrt{-b}\cosh\theta$。
  3. $b\lt 0$, $a\lt 0$ 時, 令 $\alpha=-\sqrt{-b}e^{\theta}$, $\beta=-\sqrt{-b}e^{-\theta}$, 並將 $\alpha\cdot\beta$ 代入 $a_n$ 的一般式 (4.4), 得到: \begin{equation} %(4.7) a_n(\theta) = (-\sqrt{-b})^n (e^{n\theta}+e^{-n\theta}) \end{equation} 將 $n=0$ 和 $n=1$ 代回 (4.7), 得到 $a_0=2$, $a_1=-2\sqrt{-b}\cosh\theta$。

由以上結果, 可以得到定理 4.3:

定理 4.3. $a_n$ 滿足遞迴關係式 $a_n=a\cdot a_{n-1}+b\cdot a_{n-2}$ $(n\ge 2)$, $a_0=2$, 其中 $a$, $b$, $c$, $d$ 為任意實數。則

  1. 當 $b\gt 0$ 時, $a_1=2\sqrt{b}\sinh\theta$, $a_n(\theta)=(\sqrt{b})^ne^{n\theta}+(-1)^n(\sqrt{b})^ne^{-n\theta}$
  2. 當 $b\lt 0$, $a\gt 0$ 時, $a_1=2\sqrt{-b}\cosh\theta$, $a_n(\theta)=(\sqrt{-b})^n (e^{n\theta}+e^{-n\theta})$
  3. 當 $b\lt 0$, $a\lt 0$ 時, $a_1=-2\sqrt{-b}\cosh\theta$, $a_n(\theta)=(-\sqrt{-b})^n (e^{n\theta}+e^{-n\theta})$

推論 4.2. $$ L_n(\theta) = e^{n\theta} + (-1)^n e^{-n\theta} $$ 其中 $\{L_n(\theta)\}$ 為 Lucas 數, $L_0(\theta)=2$, $L_1(\theta)=2\sinh\theta$, $L_n(\theta)=L_{n-1}(\theta)+L_{n-2}(\theta)$ $(n\ge 2)$。

證明: Lucas 數的遞迴關係式為 $L_n=L_{n-1}+L_{n-2}$ $(n\ge 2)$, $L_0=2$, $L_1=1$。 根據定理 4.3, 取 $(a, b)=(1, 1)$, 得到 $L_n(\theta)=e^{n\theta}+(-1)^ne^{-n\theta}$, 和 Ismail 的結果相同。

5.致謝

感謝中研院數學所的暑期組合數學專題研究計畫提供機會與資源, 讓作者有機會在暑假跟著美國內華達大學數學系薛昭雄教授做暑期研究, 也由衷感謝薛昭雄教授的指導, 由於薛教授適當地鼓勵與建議, 本文才能完稿。

參考文獻

M. Abramowitz and I. A. Stegun, eds. Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, Chapter 22. New York: Dover, 1972. R. Askey, Fibonacci and Lucas Numbers, Mathematics Teacher, p.610-614, 2005. William Y. C. Chen and James D. Louck, The Combinatorial Power of the Companion Matrix, Linear Algebra and its Applications, 232, p.261-278, 1996. Haci Civciv and Ramazan Türkmen, Notes on the $(s, t)$-Lucas and Lucas Matrix Sequences, Ars Conbinatoria 89, p.271-285, 2008. Mourad E. H. Ismail, One parameter generalizations of the Fibonacci and lucas numbers, The Fibonacci Quarterly, Volume 46/47, Number 2, p.167-179, 2006. J. Lahr., Fibonacci and Lucas Numbers and the Morgan-Voyce Polynomials in Ladder Networks and in Electrical Line Theory, In Fibonacci Numbers and Their Applications, ed. G. E. Bergum, A. N. Philippou, & A. F. Horadam, I:141-61. Dordrecht: Kluwer, 1986. A. M. Morgan-Voyce, Ladder Network Analysis using Fibonacci Numbers, IRE. Trans. On Circuit Theory, CT-6(Sept.), p.321-322, 1959. T. J. Rivlin, Chebyshev Polynomials: From Approximation Theory to Algebra and Number Theory, John Wiley, New York, 1990.

---本文作者為台灣大學電機系學生---