Exponent for bounded gaps between many primes

Description of constant

Let $p_n$ denote the $n$-th prime and, for $m \ge 1$, write

\[H_m := \liminf_{n \to \infty} (p_{n+m} - p_n)\]

for the least limit point of the gaps between primes $m$ apart. Maynard [May15] proved that $H_m$ is finite for every $m$, with $H_m \ll m^3 e^{4m}$; the growth rate in $m$ is exponential in all bounds known, and the exponent for bounded gaps between many primes is

\[C_{88b} := \limsup_{m \to \infty} \frac{\log H_m}{m},\]

equivalently the least $c$ for which $H_m \ll_\varepsilon \exp((c+\varepsilon)m)$ holds for every $\varepsilon > 0$.

The case $m = 1$ is recorded separately as $C_{88a}$; note that $C_{88b}$ is insensitive to any finite number of $H_m$, so the two are genuinely independent quantities.

Known upper bounds

Bound Reference Comments
$4$ [May15] $H_m \ll m^3 e^{4m}$, the first finiteness result for $m \ge 2$, from the multidimensional Selberg sieve.
$4 - \frac{28}{157} = 3.821656\dots$ [Pol14b] $H_m \ll m e^{(4-28/157)m}$. Polymath8b, quoted in this form by [Sta26].
$3.815$ [BI17] Baker–Irving.
$3.8075$ [Sta25] Stadlmann: $H_m = O(\exp(3.8075\,m))$, deduced from an exponent of distribution $\tfrac12 + \tfrac1{40}$ for the primes in arithmetic progressions to smooth moduli, improving Polymath’s $\tfrac12 + \tfrac7{300}$. The new ingredient is a modification of the $q$-van der Corput process.

Known lower bounds

Bound Reference Comments
$0$ Trivial No positive lower bound is known, and none is expected: see below.

References

Contribution notes

Prepared with assistance from Claude Opus 5. The chain $4 \to 4-\tfrac{28}{157} \to 3.815 \to 3.8075$ is as set out in the introduction of [Sta26]; the value $3.8075$ and the exponent of distribution $\tfrac12+\tfrac1{40}$ are quoted from the abstract of [Sta25]. Bibliographic details for all five references were confirmed through Crossref, which also showed that [Sta25] has appeared in Advances in Mathematics rather than being merely accepted. No bound was independently verified, and the Polymath and Baker–Irving constants were not checked against their primary sources.