2
$\begingroup$

Let $\sigma(n)=\sum_{1\leq d\mid n}d$ the sum of divisors, $\varphi(n)$ the Euler's totient function and we denote the primorial $\prod_{k=1}^n p_k$ as $N_n$, where $p_k$ denotes the $k$-th prime number. For integers $1\leq z<n$ and $m\geq 1$, and $a\geq 0$ I've consider if it is possible to find the solutions of some cases of the following equations of next problems.

Problem 1. Solve for positive integers $1\leq z,n, m$ and $a\geq 0$ $$\sigma\left(\frac{N_n}{N_z}\right)=2^aN_m\tag{1}$$ where also is required the condition $z<n$.

Problem 2. Solve for positive integers $1\leq z,n, m$ and $a\geq 0$ $$\varphi\left(\frac{N_n}{N_z}\right)=2^aN_m\tag{2}$$ where also is required the condition $z<n$.

Question. Is it possible to make a remarkable progress in studying the solutions of Problem 1? Is it possible to make a remarkable progress in studying the solutions of Problem 2? Many thanks.

I denote the solutions as $(n,z,m;a)$ or $(n,z,m)$ for a given integer $a$, and from th context we know if it are solutions of the corresponding problem. Below I add the cases for wich I've calculated some solutions using a Pari/GP program, over the segments of integers $1\leq z,n,m\leq 100$.

Here I add a claim for which I did an easy draft for its proof.

Claim. The equation $\varphi\left(\frac{N_n}{N_z}\right)=2^aN_m$ implies $n-z\leq a+1$ (the identity holds for the case $a=0$). Similarly we've a claim for the other Problem 1. Further we know that the Euler's totient function and the sum of divisors function are multiplicative functions, and each (odd) prime number $p$ of the form $4\lambda+1$ contributes as $\sigma(p)\equiv2\text{ mod }4$ and $\varphi(p)\equiv 0\text{ mod }4$.

Some solutions for Problem 1. This is a summary of the solutions that I know for Problem 1, when $1\leq z,n,m\leq 100$ and $z<n$:

  • For the case $a=0$ the solutions $(n,z,m)=(3,2,2)$ and $(10,9,3)$.
  • For the case $a=1$ the solutions $(n,z,m)=(2,1,1)$,$(5,4,2)$,$(17,16,3)$ and $(81,80,4)$.
  • For the case $a=2$ the solutions $(n,z,m)=(3,1,2)$, $(4,3,1)$ and $(9,8,2)$.
  • For the case $a=3$ the solutions $(n,z,m)=(4,2,2)$, $(15,14,2)$ and $(52,51,3)$.

Some solutions for Problem 2. This is a summary of the solutions that I know for Problem 2, when $1\leq z,n,m\leq 100$ and $z<n$:

  • For the case $a=0$ the solutions $(n,z,m)=(2,1,1)$,$(4,3,2)$,$(11,10,3)$ and $(47,46,4)$.
  • For the case $a=1$ the solutions $(n,z,m)=(3,2,1)$, $(5,3,3)$,$(6,5,2)$,$(18,17,3)$,$(20,18,5)$ and $(82,81,4)$.
  • For the case $a=2$ the solutions $(n,z,m)=(3,1,1)$, $(4,2,2)$,$(6,4,3)$ and $(11,9,4)$.
  • For the case $a=3$ the solutions are $(n,z,m)=(4,1,2)$,$(5,2,3)$,$(7,6,1)$,$(11,8,5)$, $(14,12,4)$ and $(53,52,3)$.
$\endgroup$
2
  • $\begingroup$ For large values of our parameters, our integers in $(n,z,m;a)$, I evoke that maybe my Claim can be improved by counting how many primes there are of the form $4\lambda+1$ in comparison with the size of $a$. $\endgroup$
    – user142929
    Sep 26, 2019 at 9:47
  • $\begingroup$ I was inspired in Florian Luca, Equations Involving Arithmetic Functions of Factorials, Divulgaciones Matemáticas Vol 8. No. 1 (2000). $\endgroup$
    – user142929
    Sep 26, 2019 at 9:59

1 Answer 1

3
+50
$\begingroup$

First we should compute the left side of the first equation: $$\begin{array}{rcl} \displaystyle\sigma\left(\frac{N_n}{N_z}\right) & = & \displaystyle\sigma\left( \prod_{z < k \leq n}p_k \right) \\ \\ & = & \displaystyle\prod_{z < k \leq n}\dfrac{p_k^2-1}{p_k-1} \\ \\ & = & \displaystyle\prod_{z < k \leq n}(p_k+1) \end{array}$$ Then problem 1 is solving: $$\displaystyle\prod_{z < k \leq n}(p_k+1) = 2^a N_m$$ The reason why this equation is hard is the case $p_k+1=2^b$, then $p_k$ is Mersenne prime, and we don't know if the set of Mersenne's primes has infinite cardinality or not.

Your problem2 is solving: $$\varphi\left(\frac{N_n}{N_z}\right)=\displaystyle\prod_{z < k \leq n}(p_k-1)=2^aN_m$$

Same of problem 1:

If $p_k-1=2^b$ , then $p_k$ is Fermat prime number and we don't know if the set of Fermat's prime numbers has infinite cardinality or not.

$\endgroup$
2
  • 1
    $\begingroup$ Many thanks for your great contribution. $\endgroup$
    – user142929
    Sep 26, 2019 at 11:26
  • $\begingroup$ And your discussion for Problem 1 also works for the equation $$\psi\left(\frac{N_n}{N_z}\right)=2^aN_m,$$ that is a different problem, a problem that involves the Dedekind psi function $\psi(l)$. Many thanks for you great answer again. $\endgroup$
    – user142929
    Apr 30, 2020 at 19:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.