If $N$ is deficient-perfect, under what conditions does this inequality hold

divisor-suminequalitynumber theoryperfect numbersupper-lower-bounds

This question is an offshoot of the following answer to a closely related MSE question.

Let $N$ be a deficient-perfect number, i.e. $N$ is a positive integer such that $D(N) \mid N$ where $D(N)=2N-\sigma(N)$ is the deficiency and $\sigma(N)$ is the sum of divisors of $N$, respectively.

Since $N$ is deficient-perfect, it follows that $N$ is deficient. It is then known that $N$ satisfies the inequality
$$\frac{2N}{N + D(N)} \leq I(N) < \frac{2N+D(N)}{N+D(N)}$$
where $I(N)=\sigma(N)/N$ is the abundancy index of $N$. (Equality holds if and only if $N=1$.)

Consequently, since $N$ is deficient-perfect, then $N/D(N)$ is an integer, and since $N/D(N) \mid N$, then we have
$$I\bigg(\frac{N}{D(N)}\bigg) \leq I(N) < \frac{2N+D(N)}{N+D(N)}=\frac{2\bigg(\frac{N}{D(N)}\bigg)+1}{\bigg(\frac{N}{D(N)}\bigg)+1}.$$

Here is my question:

If $N$ is deficient-perfect, under what conditions does
$$\frac{2\bigg(\frac{N}{D(N)}\bigg)}{\bigg(\frac{N}{D(N)}\bigg)+1} \leq I\bigg(\frac{N}{D(N)}\bigg)$$
hold?

It appears that it only holds when $D(N)=1$, i.e. when $N$ is almost perfect. Are there other conditions under which the inequality in the question holds?

Added April 11 2020 (5:49 PM Manila time)

The inequality
$$\frac{2\bigg(\frac{N}{D(N)}\bigg)}{\bigg(\frac{N}{D(N)}\bigg)+1} \leq I\bigg(\frac{N}{D(N)}\bigg) < \frac{2\bigg(\frac{N}{D(N)}\bigg)+1}{\bigg(\frac{N}{D(N)}\bigg)+1}$$
holds if and only if $N/D(N)$ is almost perfect.

So here is my reformulated question:

If $N$ is deficient-perfect, when is $N/D(N)$ almost perfect?

Best Answer

I've got some partial results.

This answer proves the following claims :

Claim 1 : If $N=2^k$, then $\frac{N}{D(N)}$ is an almost perfect number.

Claim 2 : If $N = 2^k(2^{k+1} + 2^t - 1)$ where $2^{k+1} + 2^t - 1$ is an odd prime and $0\lt t\le k$, then $\frac{N}{D(N)}$ is not an almost perfect number.

Claim 3 : If $N$ is not of the form $2^k$ and $N\le 49024$, then $\frac{N}{D(N)}$ is not an almost perfect number.

Claim 4 : If $\frac{N}{D(N)}=2^k$ where $k\ge 1$ with $D(N)\not=1$, then

  • $D(N)$ is even.

  • $D(N)$ has at least three distinct prime factors.

  • $D(N)$ is not of the form $2pq$ where $p,q$ are distinct odd primes.

  • $D(N)$ is not of the form $2^2pq$ where $p,q$ are distinct odd primes such that $(p,q)\not\equiv (1,1)\pmod 4$.

  • $D(N)\ge 90$


Claim 1 : If $N=2^k$, then $\frac{N}{D(N)}$ is an almost perfect number.

Proof : We have $D(N)=2^{k+1}-(2^{k+1}-1)=1$ and $$\sigma\bigg(\frac{N}{D(N)}\bigg)-\bigg(\frac{2N}{D(N)}-1\bigg)=\sigma(2^k)-\sigma(2^k)=0\qquad\blacksquare$$


Claim 2 : If $N = 2^k(2^{k+1} + 2^t - 1)$ where $2^{k+1} + 2^t - 1$ is an odd prime and $0\lt t\le k$, then $\frac{N}{D(N)}$ is not an almost perfect number.

Proof : If $N = 2^k(2^{k+1} + 2^t - 1)$ where $2^{k+1} + 2^t - 1$ is an odd prime, then we get $$D(N)=2^{k+1}(2^{k+1} + 2^t - 1)-(2^{k+1}-1)(2^{k+1} + 2^t)=2^t$$ from which we have $$\begin{align}&\sigma\bigg(\frac{N}{D(N)}\bigg)-\bigg(\frac{2N}{D(N)}-1\bigg) \\\\&=\sigma\bigg(2^{k-t}(2^{k+1} + 2^t - 1)\bigg)-(2^{k+1}-1)(2^{k-t+1}+1) \\\\&=(2^{k-t+1}-1)(2^{k+1}+2^t)-(2^{k+1}-1)(2^{k-t+1}+1) \\\\&=(2^{k-t+1}+1)(1-2^t) \\\\&\not=0\qquad\blacksquare\end{align}$$


Claim 3 : If $N$ is not of the form $2^k$ and $N\le 49024$, then $\frac{N}{D(N)}$ is not an almost perfect number.

Proof : In the following, red numbers are of the form $2^k$, and blue numbers are of the form $2^k(2^{k+1} + 2^t - 1)$ where $2^{k+1} + 2^t - 1$ is an odd prime and $t\le k$ where $a_n$ is the $n$-th deficient-perfect number and $$f(N)=\sigma\bigg(\frac{N}{D(N)}\bigg)-\bigg(\frac{2N}{D(N)}-1\bigg)$$Note here that $$\text{$\dfrac{N}{D(N)}$ is an almost perfect number}\iff f(N)=0$$

$$\begin{array}{|c|c|c|c|c|c|c|c|c|c|c|c|c|c|} \hline n&1&2&3&4&5&6&7&8&9&10&11&12&13&14&15\\ \hline a_n&\color{red}1&\color{red}2&\color{red}4&\color{red}8&\color{blue}{10}&\color{red}{16}&\color{red}{32}&\color{blue}{44}&\color{red}{64}&\color{red}{128}&\color{blue}{136}&\color{blue}{152}&\color{blue}{184}&\color{red}{256}&\color{red}{512}\\ \hline f(a_n)&0&0&0&0&\not=0&0&0&\not=0&0&0&\not=0&\not=0&\not=0&0&0 \\\hline \end{array}$$ $$\begin{array}{|c|c|c|c|c|c|c|c|c|c|c|c|c|c|} \hline n&16&17&18&19&20&21&22&23&24&25&26\\ \hline a_n&\color{blue}{752}&884&\color{red}{1024}&\color{red}{2048}&\color{blue}{2144}&\color{blue}{2272}&\color{blue}{2528}&\color{red}{4096}&\color{red}{8192}&\color{blue}{8384}&\color{blue}{12224} \\ \hline f(a_n)&\not=0&-189&0&0&\not=0&\not=0&\not=0&0&0&\not=0&\not=0 \\\hline \end{array}$$

$$\begin{array}{|c|c|c|c|c|c|c|c|c|c|c|c|c|c|} \hline n&27&28&29&30&31&32&33&34&35\\ \hline a_n&\color{red}{16384}&17176&18632&18904&\color{red}{32768}&\color{blue}{32896}&\color{blue}{33664}&\color{blue}{34688}&\color{blue}{49024} \\ \hline f(a_n)&0&-111&-1863&-2205&0&\not=0&\not=0&\not=0&\not=0 \\\hline \end{array}$$


Claim 4 : If $\frac{N}{D(N)}=2^k$ where $k\ge 1$ with $D(N)\not=1$, then

  • $D(N)$ is even.

  • $D(N)$ has at least three distinct prime factors.

  • $D(N)$ is not of the form $2pq$ where $p,q$ are distinct odd primes.

  • $D(N)$ is not of the form $2^2pq$ where $p,q$ are distinct odd primes such that $(p,q)\not\equiv (1,1)\pmod 4$.

  • $D(N)\ge 90$

Proof : Let $D(N)=m$. Then, we have $$\begin{cases}2N-\sigma(N)=m \\\\\frac{N}{2N-\sigma(N)}=2^k\end{cases}\implies\sigma(2^km)=m(2^{k+1}-1)\tag1$$ for some $k\ge 1$.

Suppose that $m\ge 3$ is odd. Then, we have $$(1)\implies\sigma(2^k)\sigma(m)=m(2^{k+1}-1)\implies \sigma(m)=m$$ which is impossible.

Suppose that $m=2^s$ where $s\ge 1$. Then, we have $$(1)\implies 2^{k+s+1}-1=2^s(2^{k+1}-1)$$which is impossible since LHS is odd while RHS isn't.

Suppose that $m=2^st^u$ where $s\ge 1$ and $t$ is an odd prime with $u\ge 1$. Then, we have $$(1)\implies (2^{k+s+1}-1)(1+t+\cdots +t^u)=2^st^u(2^{k+1}-1)$$ There exists a positive integer $v$ such that $$1+t+\cdots +t^u=2^sv\tag2$$ to have $$(2^{k+s+1}-1)v=t^u(2^{k+1}-1)$$ There eixsts a positive integer $w$ such that $$2^{k+1}-1=vw\tag3$$ to have $$2^{k+s+1}=t^uw+1\tag4$$ Multiplying the both sides of $(3)$ by $2^s$ and using $(2)(4)$ give $$1-2^s=(1+t+\cdots +t^{u-1})w$$ which is impossible since LHS is negative while RHS isn't.

So, we see that $D(N)$ has at least three distinct prime factors.

Suppose that $m=2pq$ where $p,q$ are distinct odd primes. Then, we get $$(1)\implies (2^{k+1}-1)(p+1)(q+1)=2pq(2^{k+1}-1)$$ which is impossible since LHS is divisible by $4$ while RHS isn't.

Suppose that $m=2^2pq$ where $p,q$ are distinct odd primes and $(p,q)\not\equiv (1,1)\pmod 4$. Then, we get $$(1)\implies (2^{k+3}-1)\cdot\frac{p+1}{2}\cdot\frac{q+1}{2}=pq(2^{k+1}-1)$$ which is impossible since LHS is even while RHS isn't.

Hence, it follows that $D(N)\ge 2\cdot 3^2\cdot 5=90$. $\quad\blacksquare$

Related Question