Doubly Recursive Integer Factorizations

As you know, every integer greater than \(1\) can be uniquely factored into a product of powers of prime numbers: \(24 = 8 \times 3 = 2^3 \times 3^1\), that kind of thing. No news there. Knowing that each prime number can be indexed, as the first prime \(p_1 = 2\), the second prime \(p_2 = 3\), the third prime \(p_3 = 5\), the fourth prime \(p_4 = 7\), and so on, each integer greater than \(1\) can then be expressed as

$$\prod p_i^k$$

where \(i\) is the index of the prime and \(k\) is the exponent of the prime in the integer factorization. For example, \(24 = 8 \times 3 = 2^3 \times 3^1 = p_1^3 p_2^1\). No news there either.

What you might not have tried before is to then do the same thing to the indexes and exponents: \(24 = 8 \times 3 = 2^3 \times 3^1 = p_1^3 p_2^1 = p_1^{p_2^1} p_{p_1^1}^1\).

If you keep going, eventually all of the indexes and exponents will be \(1\)’s: \(24 = 8 \times 3 = 2^3 \times 3^1 = p_1^3 p_2^1 = p_1^{p_2^1} p_{p_1^1}^1 = p_1^{p_{p_1^1}^1} p_{p_1^1}^1\).

Once you get that far, you might as well remove all of the 1’s because they don’t add any information, leaving you with \(p_1^{p_{p_1^1}^1} p_{p_1^1}^1 = p^{p_p} p_p\). Using this procedure, any integer greater than 1 can be expressed using this kind of expression.

\(2\)\(= p_1 = p\)
\(3\)\(= p_2 = p_p\)
\(4\)\(= p_1^2 = p_1^{p_1} = p^p\)
\(5\)\(= p_3 = p_{p_2} = p_{p_p}\)
\(6\)\(= p_1 p_2 = p p_p\)
\(7\)\(= p_4 = p_{p^p}\)
\(8\)\(= p_1^3 = p^{p_2} = p^{p_p}\)
\(9\)\(= p_2^2 = p_p^p\)
\(10\)\(= p_1 p_3 = p p_{p_p}\)
\(11\)\(= p_5 = p_{p_{p_p}}\)

Focusing on the shape of the expressions, towers of exponents look like this (just towers of powers of 2):

\(p\)\(= 2\)
\(p^p\)\(= 2^2 = 4\)
\(p^{p^p}\)\(= 2^{2^2} = 16\)
\(p^{p^{p^p}}\)\(= 2^{2^{2^2}} = 65,536\)
\(p^{p^{p^{p^p}}}\)\(= 2^{2^{2^{2^2}}} \approx 2.00352993 \times 10^{19,728}\)

And towers of indexes/subscripts look like this:

\(p\)\(= 2\)
\(p_p\)\(= 3\)
\(p_{p_p}\)\(= 5\)
\(p_{p_{p_p}}\)\(= 11\)
\(p_{p_{p_{p_p}}}\)\(= 31\)
\(p_{p_{p_{p_{p_p}}}}\)\(= 127\)
\(p_{p_{p_{p_{p_{p_p}}}}}\)\(= 709\)
\(p_{p_{p_{p_{p_{p_{p_p}}}}}}\)\(= 5,381\)
\(p_{p_{p_{p_{p_{p_{p_{p_p}}}}}}}\)\(= 52,711\)
\(p_{p_{p_{p_{p_{p_{p_{p_{p_p}}}}}}}}\)\(= 648,391\)
\(p_{p_{p_{p_{p_{p_{p_{p_{p_{p_p}}}}}}}}}\)\(= 9,737,333\)

This is OEIS sequence A007097, \(a(n+1)\) is the \(a(n)\)th prime number.

Combined indexes and exponents:

\(p_p^p\)\(= p_2^2 = 3^2 = 9\)
\(p_{p_p}^{p^p}\)\(= p_3^{2^2} = 5^4 = 625\)
\(p_{p_{p_p}}^{p^{p^p}}\)\(= \ldots\)

Some of these expressions look like typeset versions of the Sierpinski gasket:

\(p\)\(= 2\)
\(p_p^p\)\(= 9\)
\(p_{p_p^p}^{p_p^p}\)\(= 1,801,152,661,463\)
\(p_{p_{p_p^p}^{p_p^p}}^{p_{p_p^p}^{p_p^p}}\) \(= {55,125,235,480,573}^{1,801,152,661,463} \approx 10^{10^{13.39\ldots}}\)

Things might have got a little bit out of hand.


I didn’t really add anything to the work of J. Awbrey, Riffs and Rotes, but still.

Designed and rendered using Wolfram Mathematica 11, with first timer’s thanks to:

Powered by MathJax .
First draft August 2018 by Robert Dickau.

[ home ] || [ 2018-08-14 ]


www.robertdickau.com/drif.html