Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Recurso digital |
| Lingua: | |
| Pubblicazione: |
Zenodo
2026
|
| Accesso online: | https://doi.org/10.5281/zenodo.20373198 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
Sommario:
- <p>Legendre's conjecture asserts that for any positive integer $N$, there is at least one prime in the interval $[N^2, (N+1)^2]$. This paper presents a rigorous proof of this conjecture within the framework of the multi-sieve compensation method. Using the square interval property, the problem is reduced to finding numbers in the interval $A=[0, L-1]$ (where $L = 2N+2$) that are not divisible by any prime $\le N+1$. We construct a large translation interval $U = B \cup C$, where $B = [0, mQ_t-1]$ ($m = P_{t+1}$) consists of complete residue systems modulo $Q_t$, and $C = [mQ_t, mQ_t+L-1]$ is translation equivalent to $A$. Applying the multi-sieve compensation method on $U$, leveraging the huge length of $U$ to ensure complete periods at every step, we prove $N(A) > \frac{2N+2}{6} \prod_{i=3}^t (1-2/P_i)$. Using explicit lower bounds from Mertens' theorem, we show $N(A) > 0$ for $N \ge 10^6$, and $N(A) \to \infty$. For $N < 10^6$, the conjecture can be verified directly. This completes the proof of Legendre's conjecture.</p>