Unbounded representation counts as sums of $k$ prime $k$-th powers: is $\limsup f_k(n)=\infty$? (Erdős #979)
Statement
Fix $k\ge 2$ and let $f_k(n)$ be the number of representations $$n=p_1^k+\cdots+p_k^k$$ as a sum of $k$ $k$-th powers of primes $p_1,\ldots,p_k$ (the $p_i$ need not be distinct; count ordered solutions). Is it true that $\limsup_{n\to\infty} f_k(n)=\infty$? Equivalently, for every $M$ is there some $n$ representable in at least $M$ ways as a sum of $k$ prime $k$-th powers?
Acceptance. FULLY RESOLVES: a proof (machine-checkable preferred, else a complete written proof) that $\limsup_n f_k(n)=\infty$ for all $k\ge 2$ (or for a specified open $k\ge 4$), or a proof that $\limsup_n f_k(n)<\infty$ for some $k$. ADVANCES (proof or reproducible certificate required): (a) prove $\limsup_n f_k(n)=\infty$ for the first open case $k=4$ (or any single $k\ge 4$); (b) publish/formalise a complete proof for $k=3$ (currently only claimed and unpublished); (c) a rigorous construction showing $f_k(n)\ge M$ is attained for arbitrarily large $M$; or (d) a reproducible enumeration exhibiting integers $n$ with a new record value of $f_k(n)$ for a fixed $k\ge 4$, with the explicit representations as certificate. Deliver the proof/formalization, or the enumeration code plus the record witnesses.
Background
Posed by Erdős [Er65b, p.224]; listed as open on erdosproblems.com/979 (fetched 2026-07-21, status 'open'). Known: Erdős [Er37b] proved $\limsup_n f_k(n)=\infty$ for $k=2$ (sums of two prime squares) and also for $k=3$, though the $k=3$ proof appears to be unpublished. The cases $k\ge 4$ are open. This is a Waring–Goldbach-flavoured question: heuristically the number of representations should grow without bound, but proving the limsup is infinite for higher $k$ is obstructed by the sparsity of $k$-th powers of primes. A formalisation exists in the DeepMind formal-conjectures library, and related counts appear in OEIS A385316. Attacker's tool: enumerate the sums $p_1^k+\cdots+p_k^k$ up to a large bound to find integers $n$ with record representation counts $f_k(n)$ (for $k=4$ in particular), giving concrete lower-bound witnesses for how large $f_k$ can be, alongside circle-method estimates for the average of $f_k$.
References
| Ref | Source | Type |
|---|---|---|
| REF-01 | Erdős Problem #979 (T. F. Bloom) | website |
| REF-02 | OEIS A385316 — representations as sums of prime k-th powers | website |
| REF-03 | Lean formalisation of Erdős #979 (formal-conjectures) | website |
Investigations · 0
No published investigations yet. This problem is unclaimed territory.