I checked some relations between primes, here $1<n<10^5$ and $p_n$ is the $n$th prime.

$a) p_n^{1/3} - p_{n-1}^{1/3}<1/2$

$b) p_n^{1/n} - p_{n-1}^{1/n}<1/n $

$c) (\log p_n)^{1/2} - (\log p_{n-1})^{1/2} < 1/4$

$d) (\log p_n)^{1/n} - (\log p_{n-1})^{1/n} < 1/n^X, n\geq7,X=2$

In $d)$ I tried to find a larger $X$, but I failed.

Maybe some will fail for a larger $n$. Are any of these know to be true? Also, what can be deduced from they?

  • $\begingroup$ a) is true for sufficiently large $n$, by sufficient strengthenings of Bertrand's postulate. E.g., it is known that $[x, x+x^{0.525}]$ contains a prime for all sufficiently large $x$. Evaluating at $x=p_{n-1}$ gives $p_n^{1/3}-p_{n-1}^{1/3}\to 0$. $\endgroup$ Feb 6 at 3:54

1 Answer 1


The first three statements are true for $n$ sufficiently large. The fourth statement is equivalent to a very strong bound on prime gaps (which has a chance to hold but is right on the edge and hopeless to prove or disprove at present), while any $X>2$ produces a false bound for all $n>n_0(X)$.

Regarding a), a classical result of Ingham (1937) shows that $$p_n^{1/3} - p_{n-1}^{1/3}<\frac{p_n-p_{n-1}}{3p_{n-1}^{2/3}}=o(1).$$

Regarding b), the Prime Number Theorem gives that $$p_n^{1/n} - p_{n-1}^{1/n}\sim\frac{p_n-p_{n-1}}{n p_{n-1}}=o\left(\frac{1}{n}\right).$$

Regarding c), Bertrand's postulate shows for $p_n>7$ that $$(\log p_n)^{1/2}-(\log p_{n-1})^{1/2}<\frac{\log p_n-\log p_{n-1}}{2(\log p_{n-1})^{1/2}}<\frac{\log 2}{2(\log p_{n-1})^{1/2}}<\frac{1}{4}.$$ In fact the left-hand side tends to zero (as $n\to\infty$).

Regarding d), the Prime Number Theorem gives that $$(\log p_n)^{1/n}-(\log p_{n-1})^{1/n}\sim\frac{p_n-p_{n-1}}{n p_n\log p_n}\sim\frac{p_n-p_{n-1}}{n^2\log^2 n}.$$ Under some standard (but bold) conjectures the right-hand side stays below $2/n^2$, and perhaps it is even $o(1/n^2)$. So statement d) restricted to large $n$ is equivalent to some expected but hopelessly difficult upper bound on prime gaps. On the other hand, the above calculation shows that d) fails for any $X>2$ and $n>n_0(X)$.


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.