4

Since $e$ is a transcendence number, so it is certain that it is not zero of any polynomial with rational coefficients. However, I wonder can we find a power series with rational coefficient such that it is zero evaluated at $e$. If such series exists, can we explicitly find the coefficients?

Ken.Wong
  • 1,227
  • 2
    Ethan Bolker's answer is great. I just wanted to comment on the last line, about series that are particularly nice. You are perhaps thinking about $\pi$ that is the zero of a particularly nice power series: the Taylor series of the sin function. Now one thing that sets $\pi$ apart from $e$ despite both of them being transcendental is that $\pi$ is a 'period' and $e$ conjecturally is not. I am inclined to believe b/c of this that no 'particularly nice' series for $e$ exists. The answers to my earlier MO question about this topic: https://mathoverflow.net/q/180035/41139 might be helpful here. – Vincent Nov 25 '21 at 16:05
  • This is, of course, all speculation – Vincent Nov 25 '21 at 16:08
  • 2
    Similar to Ethan Bolker's answer you can just use reciprocals so something like $1-\frac{x}{3}-\frac{x^2}{79}-\frac{x^3}{53748}-\ldots$ and even those terms have got you to below $10^{-8}$ and the next to below $10^{-19}$ – Henry Nov 25 '21 at 16:16

1 Answers1

8

Yes. You can build such a power series recursively.

Start with $a_0 = 1$ (or anywhere else you like).

Now find a rational multiple $a_1$ of $e$ such that $$ 0 < 1 + a_1e < 1/2. $$ Then find rational $a_2$ such that $$ 0 < 1 + a_1e + a_2e^2< 1/4. $$ Continue in the obvious way.

Clearly this procedure produces many such series. I don't know whether there's one that's particularly nice.

Ethan Bolker
  • 95,224
  • 7
  • 108
  • 199