Equivalences of Goldbach conjecture - Added value?

76 Views Asked by At

Let $I(n)$ be the set of the first $n$ odd integers, $S_p(n)$ be a subset of the last $p$ elements of $I(n)$ such that $p$ is some prime number, and $a_{n-\left(\frac{p-1}{2}\right)}$ the $n-\left(\frac{p-1}{2}\right)_{th}$ element of $I(n)$.

Thinking on the Goldbach conjecture, I came to the following equivalence:

Goldbach conjecture is equivalent to affirm that every set $I(n)$ has some subset $S_p(n)$ such that $a_{n-\left(\frac{p-1}{2}\right)}$ is prime.

The proof is pretty straightforward. The sum of the elements of $I(n)$ is $n^2$, and the sum of the first $n-p$ elements of $I(n)$ is $\left(n-p\right)^2$. Therefore, the sum of the $p$ remaining elements is $n^2-\left(n-p\right)^2=-p^2+2np=p\left(2n-p\right)$.

Note that $a_{n-\left(\frac{p-1}{2}\right)}=2n-p$. Therefore, if both $p$ and $2n-p$ are prime numbers, Goldbach conjecture holds.

I am not sure that this approach adds any value for solving the conjecture. I share some of my thoughts:

Each prime $q_i$ generates $i$ subsets $S_i(n_i)$ such that $i$ is prime and $q_i=a_{n_i-\left(\frac{i-1}{2}\right)}$; so each prime $q_i$ guarantees that the conjecture holds for $i$ distinct values of $n$. Therefore, the first $i$ primes jointly guarantee that the conjecture holds for $1+2+3+...+i=\frac{î^2+i}{2}$ values of $n$. However, those values of $n$ are not necessarily distinct. The though part is proving that no value of $n$ is excluded.

Note also that some prime $q_i$ is the element $a_{\left(\frac{q-1}{2}\right)+1}$ of any set $I(n)$, and if $I(n)$ has some subset $S_p(n)$ such that $a_{n-\left(\frac{p-1}{2}\right)}$ is prime, then, noting that $a_{\left(\frac{q-1}{2}\right)+1}=a_{n-\left(\frac{p-1}{2}\right)}$, we have that $\left(\left(\frac{q-1}{2}\right)+1\right)+\left(\frac{p-1}{2}\right)=n$

Therefore, note that Goldbach conjecture would be equivalent to show that $\mathbb N-\{1,2\}\in P+Q$, where $P=\{x:\frac{p-1}{2}\}$, $Q=\{y:x+1\}$, and $P+Q=\{x+y:x\in P, y\in Q\}$. But this is kind of a tautology, as (I understand) does not add any information to solve the conjecture (if $p+q =2n$, then dividing by two $\frac{p}{2}+\frac{q}{2}=n$, and substracting $1$ from each numerator, and adding $1$, we get $\left(\left(\frac{q-1}{2}\right)+1\right)+\left(\frac{p-1}{2}\right)=n$.

I would like to know if this approach has something interesting, or does not add anything new to the problem.

Thanks!

1

There are 1 best solutions below

2
On BEST ANSWER

Goldbach's conjecture postulates that every sufficiently large even number can be represented as the sum of two primes: $2n=p_1+p_2$. This is no different than saying that every sufficiently large integer is the average of two primes: $n=\frac{p_1+p_2}{2}$, which means $n$ lies equidistant between two primes, viz: $p_1+a=n=p_2-a$, or $p_1=n-a$ and $p_2=n+a$.

From this, we see $(n-a)(n+a)=p_1p_2=n^2-a^2$. But $a=n-p_1$ so $a^2=(n-p_1)^2$ and $n^2-a^2=n^2-(n-p_1)^2 =p_1(2n-p_1)$, which is just the formula OP arrived at in his statement of proof.

As I suggested in a comment, OP has rediscovered this known relationship in a clever new way.