10

1) Is there a reason why the binomial expansion of $(a+x)^n$ is the same as a Taylor series approximation of $(a+x)^n$ centered at zero?

2) The binomial expansion of $(a+x)^n$ is

$a^n + na^{n-1}x + \frac{n(n-1)}{2!}a^{n-2}x^2 +$....

If the expansion is written this way, then $n$ can be an integer (positive or negative) or a fraction? If the binomial expansion is written in summation notation using nCr, then n can only be positive because nCr cannot have a negative $n$?

3) For the expansion of $(a+x)^n$ I gave in question 2, does $a$ have to be $a = 1$ with $-1 < x < 1$? What are these restrictions?

Update : An infinite geometric series converges when the common ratio, $x$ in this case, is between -1 and 1. The infinite binomial expansion I wrote in question 2 is a valid expansion of $(a+x)^n$ when $-1 < x < 1$. So if I put $x = 0.5$ into $(a+x)^n$ for a given $n$ and $a$, $(a+0.5)^n$ and the infinite expansion for $x = 0.5$ will give the same answer. If I use $x = 40$, the expansion will diverge and not give the same answer as the original function $(a+40)^n$. Does this mean that the binomial expansion is actually a power series (a geometric series is a special case of a power series)? And does $a$ need to be 1 for the $-1 < x < 1$ to be required? Or is it required regardless of what $a$ is?

DWade64
  • 1,308
  • If $n$ is a positive integer, if $i\ge n+1$ the $i$-th derivative of $(a+x)^n$ is $0$. So after a while the Taylor series for $(a+x)^n$ terminates, all the terms are after a while of shape $0\cdot x^i$. Thus the series converges for all $x$. – André Nicolas Aug 21 '14 at 18:27
  • The link is a nice source, so do not be afraid of using it due to the downvotes. The downvotes come from technical inner workings of the review system on the site, not from fundamental errors in the answer. – Vladhagen Aug 21 '14 at 21:06
  • This is the link to @MhenniBenghorbal's answer. http://en.wikipedia.org/wiki/Binomial_theorem#Generalisations – Vladhagen Aug 21 '14 at 22:38
  • Note that we are not dealing with an infinite geometric series here, as $x$ may be whatever we wish it to be. You may note that the more terms you write out, the more they converge on $0$ because of either a $n-m$ term, where $m=n$, thus making the term equal to $0$, or because of the large factorial in the denominator. – Simply Beautiful Art Dec 26 '15 at 17:04

1 Answers1

5

1) They are the same function, so they have the same power series.

2) In this answer, it is shown that for the generalized binomial theorem, we have for negative exponents, $$ \binom{-n}{k}=(-1)^k\binom{n+k-1}{k} $$ Thus, we have $$ \begin{align} (a+x)^{-3} &=a^{-3}\left(1+\frac xa\right)^{-3}\\ &=a^{-3}\sum_{k=0}^\infty\binom{-3}{k}\left(\frac xa\right)^k\\ &=a^{-3}\sum_{k=0}^\infty\binom{k+2}{k}\left(\frac xa\right)^k\\ &=\sum_{k=0}^\infty\binom{k+2}{2}\frac{x^k}{a^{k+3}}\\ \end{align} $$ The same can be done for fractional exponents, but the formulas for the coefficients are more complicated.

3) In the answer to 2), we factored out the $a^{-3}$ so that one term of the sum was $1$. This allows us to use the binomial theorem in an open-ended way; that is, we don't need to worry about what the exponent of $n-k$ needs to be. In particular, the generalized binomial theorem reads $$ (1+x)^n=\sum_{k=0}^\infty\binom{n}{k}x^k $$ where $$ \binom{n}{k}=\frac{n(n-1)(n-2)\dots(n-k+1)}{k!} $$ Furthermore, if $n$ is not a non-negative integer, the binomial expansion does not terminate. In that case, the series for $$ (a+x)^n=a^n\left(1+\frac xa\right)^n $$ converges for $|x|\lt|a|$.


Extension for $\boldsymbol{|x|\gt|a|}$

We can extend the convergence of a series for $(a+x)^n$ for $|x|\gt|a|$ if we allow Laurent expansions and write $$ (a+x)^n=x^n\left(1+\frac ax\right)^n $$ Using the same example as above, $$ \begin{align} (a+x)^{-3} &=x^{-3}\left(1+\frac ax\right)^{-3}\\ &=x^{-3}\sum_{k=0}^\infty\binom{-3}{k}\left(\frac ax\right)^k\\ &=x^{-3}\sum_{k=0}^\infty\binom{k+2}{k}\left(\frac ax\right)^k\\ &=\sum_{k=0}^\infty\binom{k+2}{2}\frac{a^k}{x^{k+3}}\\ \end{align} $$ which converges for $|x|\gt|a|$.

robjohn
  • 345,667