I am considering the following equation $$m=\frac{n^2}{(n-m-1)\lambda+n}$$ where $n,m,\lambda$ are postive integers, $1\le\lambda \le n-1$ and $m\mid n$. If $m=n$, then $$\frac{n^2}{(n-n-1)\lambda+n}=\frac{n^2}{n-\lambda}\ne n.$$ So we can assume that $m<n$.

for m in [1..100] do
for k in [2..100] do
for r in [1..(n-1)] do
  if p mod q eq 0 then
  if p div q eq m then
    print r,m,n,p,q;
  end if;
  end if;
end for;
end for;
end for;

I run the above Magma code and it outputs

4 2 6 36 18
3 3 6 36 12

So I conjecture that

Except the above two example, i.e., $(n,m,\lambda)=(6,2,4)$ and $(n,m,\lambda)=(6,3,3)$, there does not exist three positive integers $n,m,\lambda$ such that $1\le\lambda \le n-1$, $m\mid n$ and $$m=\frac{n^2}{(n-m-1)\lambda+n}.$$

Can anyone proof the conjecture or give a counterexample?


My trying: For $\lambda=1$, I can proof that there exist two positive integers $n,m$ such that $m\mid n$ and $m=\frac{n^2}{2n-m-1}$.

Proof. Assume that there exist two positive integers $n,m$ such that $m\mid n$ and $m=\frac{n^2}{2n-m-1}$. Then we have $$ m(2n-m-1)=n^2\tag{1}$$ and $$n=km\tag{2}$$ where $k$ is an integer. Put $\mathrm (1)$ into $\mathrm (2)$ and obtain $$m(2km-m-1)=k^2m^2\tag{3}.$$ Since $m$ is positive, dividing both sides of $\mathrm (3)$ by $m$ and we have $$(2k-1)m-1=k^2m\tag{4}.$$ Obviously, the right hand side of $\mathrm (4)$ is a multiple of $m$ while the other side is not. It is a contradiction. So there could not exist such two integers $n$ and $m$. QED.

In a similar way, we can proof that if such integers exist, then $m\mid \lambda$. To be continued...


The conjecture is true.

Proof :

Let $n=mk$ where $k\ge 2\in\mathbb Z$. Then, we have $$(mk-m-1)\lambda=mk(k-1)$$ If $m=1$, then $\lambda=\frac{k(k-1)}{k-2}=k+1+\frac{2}{k-2}$. So, we have to have $k-2\mid 2$ implying $k=3,4$. So, we have $(n,m,\lambda)=(3,1,6),(4,1,6)$, but these don't satisfy $\lambda\le n-1$.

If $m\ge 2$, then since $m\not\mid mk-m-1$, we can write $\lambda=ma$ where $a$ is a positive integer such that $a\le k-\frac 1m$, i.e. $a\le k-1$. Then, we have

$$ma(k-1)-a=(k-1)k\implies a=(k-1)(ma-k)$$ from which we have $ma-k=1$ and $a=k-1$. So, eliminating $k$, we have $a(m-1)=2$.

  • $(a,m-1)=(1,2)$ implies $(n,m,\lambda)=(6,3,3)$

  • $(a,m-1)=(2,1)$ implies $(n,m,\lambda)=(6,2,4)$

Therefore, the only solutions are $$(n,m,\lambda)=(6,3,3),(6,2,4)$$

| cite | improve this answer | |

"OP" requested a counter-example numerical solution:


above satisfies the given equation

& also y=(n-1)=5 & (m divides n)

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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