Taylor series, Maclaurin series

-

Suppose the function \(f(x)\) is of class \(C^\infty\) in the neighbor of \(x = a\). Then, we can define the following power series:

\[\begin{eqnarray*} \sum_{n=0}^{\infty}\frac{f^{(n)}(a)}{n!}t^n &=& \sum_{n=0}^{\infty}\frac{f^{(n)}(a)}{n!}(x-a)^n\\ &=&f(a) + f'(a)(x-a) + \frac{f''(a)}{2!}(x-a)^2 + \frac{f'''(a)}{3!}(x-a)^3 + \cdots \end{eqnarray*}\]

where \(t = x - a\). If this power series has a positive radius of convergence, and the function defined by it matches \(f(x)\) in the neighbor of \(x = a\), we say the function \(f(x)\) is analytic. \(f(x) = \sum_{n=0}^{\infty}\frac{f^{(n)}(a)}{n!}(x-a)^n\) is called the Taylor series of \(f(x)\) at \(x=a\). A Taylor series at \(x = 0\) is called a Maclaurin series.



Example. Let us define the function \(f(x)\) on the open interval \((a-r, a+r)\) by the following power series

\[f(x) = \sum_{n=0}^{\infty}a_n(x-a)^n\tag{Eq:eg1}\]

where \(r > 0\) is the radius of convergence of the power series. By Corollary 2 in Calculus of power series, \(a_n = \frac{f^{(n)}(a)}{n!}\) \((n = 0, 1, 2, \cdots)\). Therefore, \(f(x)\) is analytic and (Eq:eg1) gives the Taylor series. □

See also: Calculus of power series

Exponential and trigonometric functions

Lemma

Let \(f(x)\) be a function of class \(C^\infty\) in a neighbor of \(x = 0\) and \(r > 0\). Suppose the following condition is satisfied:

  • (\(\dagger\)) There exists an \(M > 0\) such that, for all \(n \in \mathbb{N}_0\) and for all \(x\in \mathbb{R}\), if \(|x| < r\), then \(|f^{(n)}(x)| \leq M\).
Then, \(f(x)\) is analytic at \(x = 0\), and the radius of convergence of its Maclaurin series is at least \(r\).
Proof. Choose an \(x\) such that \(|x| < r\) and consider the finite Maclaurin expansion of \(f(x)\):
\[\begin{eqnarray*} f(x) &=& \sum_{k=0}^{n-1}\frac{f^{(k)}(0)}{k!}x^k + R_n,\\ R_n &=& \frac{f^{(n)}(\theta x)}{n!}x^n ~ (0 < \theta < 1). \end{eqnarray*}\]
For the remainder \(R_n\), we have \(|R_n| \leq M\frac{|x|^n}{n!} \to 0\) (\(n \to \infty\)). Therefore, the above finite Maclaurin expansion converges as \(n\to \infty\) and the limit is equal to \(f(x)\). Thus, \(f(x)\) is analytic at \(x=0\) and its radius of convergence is at least \(r\). ■

Example. Let us show that the exponential function \(e^x\) is analytic and its Maclaurin series is given as
\[e^x = \sum_{n=0}^{\infty}\frac{1}{n!}x^n = 1 + x + \frac{x^2}{2!} + \frac{x^3}{3!} + \cdots. \tag{Eq:Exp}\]
Let \(f(x) = e^x\). For any \(r > 0\), let \(M = e^r\). For any \(x\) such that \(|x| < r\) and for any \(n \in \mathbb{N}_0\), \(|f^{(n)}(x)| = e^x < M\) so that \(f(x)\) is analytic at \(x=0\). Since \(r > 0\) is arbitrary, the radius of convergence is \(+\infty\). For all \(n \in \mathbb{N}_0\), \(f^{(n)}(0) = e^0 = 1\) so that the Maclaurin series is given as in (Eq:Exp). □

Example. It is an exercise to show that the Maclaurin series of \(\sin x\) and \(\cos x\) are given as the following:
\[\begin{eqnarray} \sin x &=& \sum_{n=0}^{\infty}\frac{(-1)^nx^{2n+1}}{(2n+1)!} = x - \frac{x^3}{3!} + \frac{x^5}{5!} - \frac{x^7}{7!} + \cdots,\\ \cos x &=& \sum_{n=0}^{\infty}\frac{(-1)^nx^{2n}}{(2n)!} = 1 - \frac{x^2}{2!} + \frac{x^4}{4!} - \frac{x^6}{6!} + \cdots. \end{eqnarray}\]
Also, show that the radii of convergence of these series are both \(+\infty\). □

Power functions and the binomial theorem

Given \(\alpha \in \mathbb{R}\) and \(n \in \mathbb{N}_0\), we define the binomial coefficient by
\[\binom{\alpha}{n} = \left\{ \begin{array}{cc} 1 & (n = 0),\\ \frac{\alpha(\alpha - 1)(\alpha - 2)\cdots (\alpha - n + 1)}{n!} & (n \geq 1). \end{array} \right.\]
The numerator for the case when \(n \geq 1\) is the product of \(n\) consecutive numbers from \(\alpha\) to \(\{\alpha - (n-1)\}\). For example,
\[\begin{eqnarray*} \binom{\alpha}{0} &=& 1,\\ \binom{\alpha}{1} &=& \alpha,\\ \binom{\alpha}{2} &=& \frac{\alpha(\alpha - 1)}{2},\\ \binom{\alpha}{3} &=& \frac{\alpha(\alpha - 1)(\alpha - 2)}{3!},\\ &\vdots& \end{eqnarray*}\]
Note that the binomial coefficients are defined for all \(\alpha \in \mathbb{R}\).
If \(\alpha\) is a non-negative integer, then \(\binom{\alpha}{n}\) is the number of combinations when we choose \(n\) elements out of \(\alpha\) ("\(\alpha\) choose \(n\)") for \(n = 0, 1, 2, \cdots, \alpha\), or \(\binom{\alpha}{n} = 0\) for \(n > \alpha\).

Lemma 

\[\binom{\alpha - 1}{n-1} + \binom{\alpha - 1}{n} = \binom{\alpha}{n}.\tag{Eq:binsum}\]
Proof. If \(n = 1\), we have
\[\binom{\alpha - 1}{0} + \binom{\alpha - 1}{1} = 1 + (\alpha - 1) = \alpha = \binom{\alpha}{1}.\]
Thus, (Eq:binsum) holds.

If \(n > 1\), 
\[\begin{eqnarray*} \binom{\alpha - 1}{n-1} + \binom{\alpha - 1}{n} &=& \frac{(\alpha-1)(\alpha - 2)(\alpha - 3)\cdots (\alpha -n + 1)}{(n-1)!}\\ && + \frac{(\alpha-1)(\alpha - 2)(\alpha - 3)\cdots (\alpha - n)}{n!} \\ &=&\frac{\{n + (\alpha - n)\}(\alpha-1)(\alpha - 2)(\alpha - 3)\cdots (\alpha - n + 1)}{n!}\\ &=&\frac{\alpha(\alpha-1)(\alpha - 2)(\alpha - 3)\cdots (\alpha - n + 1)}{n!}\\ &=& \binom{\alpha}{n}. \end{eqnarray*}\]
■

For \(\alpha \in \mathbb{R}\), consider the power series
\[\sum_{n=0}^{\infty}\binom{\alpha}{n}x^n.\tag{Eq:binpow}\]
If \(\alpha \in \mathbb{N}_0\), then this is a polynomial of degree \(\alpha\). Otherwise, \(\binom{\alpha}{n} \neq 0\) for all \(n\in\mathbb{N}_0\), and as \(n \to \infty\),
\[\left|\frac{\binom{\alpha}{n+1}}{\binom{\alpha}{n}}\right| = \left|\frac{\alpha - n}{n+1}\right| \to 1\]
so that the radius of convergence of the power series (Eq:binpow) is 1. 
See also: Power series (Theorem (Radius of convergence)).

Theorem (Binomial Theorem)

For \(|x| < 1\) and \(\alpha \in \mathbb{R}\),
\[(1 + x)^\alpha = \sum_{n=0}^{\infty}\binom{\alpha}{n}x^n.\tag{eq:binser}\]
Proof. Let \(f(x)\) be the function on \((-1, 1)\) defined by the right-hand side of (Eq:binser). By term-wise differentiation, we have
\[f'(x) = \sum_{n=1}^{\infty}n\binom{\alpha}{n}x^{n-1} = \sum_{n=1}^{\infty}\alpha\binom{\alpha-1}{n-1}x^{n-1}.\]
Using the above Lemma,
\[\begin{eqnarray*} (1+x)f'(x) &=& \alpha\sum_{n=1}^{\infty}\left\{\binom{\alpha-1}{n-1}x^{n-1} + \binom{\alpha-1}{n-1}x^{n}\right\}\\ &=&\alpha\left[1 + \sum_{n=1}^{\infty}\left\{\binom{\alpha-1}{n-1} + \binom{\alpha-1}{n}\right\}x^{n}\right]\\ &=& \alpha\left\{1 + \sum_{n=1}^{\infty}\binom{\alpha}{n}x^n\right\}\\ &=& \alpha f(x). \end{eqnarray*}\]
Let \(g(x) = (1+x)^{\alpha}\). We have
\[(1+x)g'(x) = (1+x)\alpha (1 + x)^{\alpha - 1} = \alpha g(x).\]
Thus, both \(y = f(x)\) and \(y = g(x)\) satisfies the same differential equation 
\[(1+x)y' = \alpha y.\]
Let us define the function \(h(x)\) on \((-1, 1)\) by
\[h(x) = \frac{f(x)}{g(x)}.\]
Then,
\[h'(x) = \frac{f'(x)g(x) - f(x)g'(x)}{\{g(x)\}^2} = \alpha\cdot\frac{f(x)g(x) - f(x)g(x)}{(1+x)\{g(x)\}^2} = 0\]
so that \(h(x) = c\) (constant) on \((-1, 1)\). But \(c = h(0) = \frac{f(0)}{g(0)} = 1\). Thus, \(f(x) = g(x) = (1+x)^{\alpha}\). ■

Remark. When \(\alpha \in \mathbb{N}_0\), \(\binom{\alpha}{n} = 0\) for \(n > \alpha\). Therefore, in this case, we have
\[(1 + x)^{\alpha} = \sum_{n=0}^{\alpha}\binom{\alpha}{n}x^n.\]
This is the binomial theorem you might have learned in high school. □

Exercise. Show that \(\binom{-1}{n} = (-1)^n\). □

Example. With \(\alpha = -1\), we have
\[\frac{1}{1 - x} = \sum_{n=0}^{\infty}x^n = 1 + x + x^2 + x^3 + \cdots\]
for \(|x| < 1\). This is the formula of geometric series. □

Example. Let us show that \(\arctan x\) is analytic and find its Maclaurin series and radius of convergence. Applying the binomial theorem to \((\arctan x)' = \frac{1}{1 + x^2}\), we have
\[\frac{1}{1 + x^2} = \sum_{n=0}^{\infty}(-1)^{n}x^{2n}. \tag{Eq:atanps}\]
By term-wise integration,
\[\arctan x = \sum_{n=0}^{\infty}\frac{(-1)^n}{2n + 1}x^{2n+1} + C\]
where \(C\) is a constant. But \(\arctan(0) = 0\) so \(C = 0\). Therefore
\[\arctan x = \sum_{n=0}^{\infty}\frac{(-1)^n}{2n + 1}x^{2n+1}.\tag{Eq:atans}\]
The radius of convergence of the right-hand side of (Eq:atans) is equal to that of (Eq:atanps), which is 1 (verify!). Thus, \(\arctan x\) is analytic, and its radius of convergence is 1. □

List of frequently used Maclaurin series

It comes in handy if you memorize the following Maclaurin series. Make sure you can derive them.

\[\frac{1}{1 - x} = \sum_{n=0}^{\infty}x^n ~~~ (|x| < 1).\]
\[(1 + x)^{\alpha} = \sum_{n=0}^{\infty}\binom{\alpha}{n}x^n ~~~  (|x|< 1)\]
where
\[\binom{\alpha}{n} = \left\{ \begin{array}{cc} 1 & (n = 0),\\ \frac{\alpha(\alpha - 1)(\alpha - 2)\cdots (\alpha - n + 1)}{n!} & (n \geq 1). \end{array} \right.\]

\[\log(1 + x) = \sum_{n=1}^{\infty}\frac{(-1)^{n+1}}{n}x^n ~~~ (|x| < 1).\]

\[\sqrt{1 + x} = 1 + \sum_{n=1}^{\infty}\frac{(-1)^{n-1}(2n-3)!!}{(2n)!!}x^n ~~~ (|x| < 1)\]
where the double factorials are defined as
\[n!! = \left\{ \begin{array}{cc} n(n-2)(n-4)\cdots 2 & (\text{$n$ is even}),\\ n(n-2)(n-4)\cdots 1 & (\text{$n$ is odd}),\\ \end{array} \right.\]
and \(0!! = (-1)!! = 1\).

\[\frac{1}{\sqrt{1 - x}} = \sum_{n=0}^{\infty}\frac{(2n-1)!!}{(2n)!!}x^{n} ~~~ (|x| < 1).\]

\[\frac{1}{\sqrt{1 - x^2}} = \sum_{n=0}^{\infty}\frac{(2n-1)!!}{(2n)!!}x^{2n} ~~~ (|x| < 1).\]

\[\arctan x = \sum_{n=0}^{\infty}\frac{(-1)^n}{2n+1}x^{2n+1} ~~ (|x| < 1).\]

\[\arcsin x = \sum_{n=0}^{\infty}\frac{(2n-1)!!}{(2n)!!}\frac{x^{2n+1}}{2n+1} ~~ (|x| < 1).\]

\[\begin{eqnarray*} e^x &=& \sum_{n=0}^{\infty}\frac{x^n}{n!} ~~~ (x\in \mathbb{R}),\\ \sin x &=& \sum_{n=0}^{\infty}\frac{(-1)^n}{(2n+1)!}x^{2n+1} ~~~(x\in \mathbb{R}),\\ \cos x &=& \sum_{n=0}^{\infty}\frac{(-1)^n}{(2n)!}x^{2n} ~~~(x\in \mathbb{R}). \end{eqnarray*}\]

Comments

Post a Comment

Popular posts from this blog

Birth process

-
Image
  Defining the birth process  Consider a colony of bacteria that never dies. We study the following process known as  the birth process , also known as  the Yule process . The colony starts with \(n_0\) cells at time \(t = 0\). Assume that the probability that any individual cell divides in the time interval \((t, t + \delta t)\) is proportional to \(\delta t\) for small \(\delta t\). Further assume that each cell division is independent of others. Let \(\lambda\) be the  birth rate.  The probability of a cell division for a population of \(n\) cells during \(\delta t\) is \(\lambda n \delta t\). We assume that the probability that two or more births take place in the time interval \(\delta t\) is \(o(\delta t)\). That is, it can be ignored. Consequently, the probability that no cell divides during \(\delta t\) is \(1 - \lambda n \delta t - o(\delta t)\). Note that this process is an example of the Markov chain with states \({n_0}, {n_0 + 1}, {n_0 + 2}...
Read more

Branching processes: Mean and variance

-
Image
Generational growth Consider the following scenario (see the figure below): A single individual (cell, organism, etc.) produces \(j (= 0, 1, 2, \cdots)\) descendants with probability \(p_j\), independently of other individuals. The probability of this reproduction, \(\{p_j\}\), is known.  That individual produces no further descendants after the first (if any) reproduction. These descendants each produce further descendants at the next subsequent time with the same probabilities.  This process carries on, creating successive generations. Figure 1. An example of the branching process. Let \(X_n\) be the random variable representing the population size (number of individuals) of generation \(n\). In the above figure, we have \(X_0 = 1\), \(X_1=4\), \(X_2 = 7\), \(X_3=12\), \(X_4 = 9.\) We shall assume \(X_0 = 1\) as the initial condition. Ideally, our goal would be to find how the population size grows through generations, that is, to find the probability \(\Pr(X_n = k)\) for e...
Read more

Combining birth and death processes

-
Image
The birth-death process Combining birth and death processes with birth and death rates \(\lambda\) and \(\mu\), respectively, we expect to have the following differential-difference equations for the birth-death process : \[\begin{eqnarray}\frac{{d}p_0(t)}{{d}t} &=& \mu p_1(t),\\\frac{{d}p_n(t)}{{d}t} &=& \lambda(n-1)p_{n-1}(t) - (\lambda + \mu)np_n(t) + \mu(n+1)p_{n+1}(t),~~(n \geq 1).\end{eqnarray}\] You should derive the above equations based on the following assumptions: Given a population with \(n\) individuals, the probability that an individual is born in the population during a short period \(\delta t\) is \(\lambda n \delta t + o(\delta t)\). Given a population with \(n\) individuals, the probability that an individual dies in the population is \(\mu n \delta t + o(\delta t)\). The probability that multiple individuals are born or die during \(\delta t\) is negligible. (The probability of one birth and one death during \(\delta t\) is also negligible.) Consequ...
Read more