11
$\begingroup$

Is there a continuous function constructed by elementary functions, or by integral formula involved only elementary functions (like Gamma function) that grows faster than any $e^{e^{e...^x}}$ ($e$ appears $n$ times)?

I ask for the answer with a single formula. Gluing continuous function together is too trivial.

The function need not to be defined on whole $\mathbb{R}$, the domain $(a, \infty)$ is acceptable.

$\endgroup$
18
  • 1
    $\begingroup$ Do you accept iterative procedures? If so The Ackermann Function generalises the series addition, multiplication, exponentiation. Likewise Knuth's Arrow Notation. $\endgroup$ Commented Aug 15, 2016 at 11:54
  • 1
    $\begingroup$ I suppose that applying any of the allowed operations (composition with elementasry function or integration) turns a function that is bounded by some e-tower into a function bounded by some (possibly higher) e-tower. $\endgroup$ Commented Aug 15, 2016 at 11:59
  • 3
    $\begingroup$ Perhaps an annoying example, but for any $\;a>e\;,\;\;a^{a^{a\ldots^x}}\;$ grows faster $\endgroup$
    – DonAntonio
    Commented Aug 15, 2016 at 12:01
  • 3
    $\begingroup$ I suppose you mean to find a function $f$ that eventually grows faster than each function in the set $\{exp^n(x)\mid n\in\Bbb N\}$, where $n$ in the power denotes iteration of function. $\endgroup$
    – edm
    Commented Aug 15, 2016 at 12:31
  • 2
    $\begingroup$ I mean define $f(x) =$ $C_k$ + exp(exp(..(x)) (k iterations) on $[k, k+1]$, $C_k$ is chosen so that $f$ is continuous. $\endgroup$ Commented Aug 15, 2016 at 13:32

2 Answers 2

6
$\begingroup$

a negative answer
No, there is no such function constructed in a "natural" way, if we construe that word propertly.

Perhaps consult the literature on "transseries" ...

For example, transseries in the sense here: G. A. Edgar, "Transseries for Beginners". Real Analysis Exchange 35 (2010) 253-310 .

The set of transseries includes real elementary functions, is closed under indefinite integration, composition, and many other operations.

But no transseries has growth rate beyond all $e^{\dots e^x}$. There is an integer "exponentiality" associated with each (large, positive) transseries; for example Exercise 4.10 in: J. van der Hoeven, Transseries and Real Differential Algebra (LNM 1888) (Springer 2006)

A (large, positive) transseries with exponentiality $n+1$ grows faster than any transseries with exponentiality ${} \le n$. And $e^{\dots e^x}$ with $n$ exponentiations has exponentiality $n$.

$\endgroup$
1
  • 2
    $\begingroup$ If $f$ is a power series that converges on $\mathbb{R}$, do we still get a negative answer? $\endgroup$ Commented Aug 15, 2016 at 14:51
6
$\begingroup$

Here is a proof that there is no such function und elementary function.

Let $A$ be a collection of functions which grow slower than the collection $e^x, e^{e^x} , \dots $.

We know that the sum, product composition and integral of functions which grow slower than $e^x, e^{e^x} , \dots $ also grows slower that $e^x, e^{e^x} , \dots $.

Then the set of functions which are finite iterate sums, $\dots$ of functions in $A$ also grows slower that $e^x, e^{e^x} , \dots $

Now, depending on or definition of elementary function, if you take $A= \{1, x, \exp, \log, x^a \}$ you should be done.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .