← All problems

Bondesson’s powers conjecture for generalized gamma convolutions

Let XX be a nonnegative generalized gamma convolution (GGC) random variable. Then, for every real q≥1q\ge 1, the random variable XqX^q is also a GGC random variable. Equivalently, the class of GGC distributions is closed under taking powers of order at least one.

References

Source: arXiv

Progress summary

Refreshed
Open · solution reported

Two 2026 preprints claim to prove Bondesson’s conjecture, but the newest authors call their argument only a candidate proof, so the problem is not yet settled.

Bondesson’s conjecture, posed in 2015, asserts that raising a generalized gamma convolution variable XX to any power q≥1q\geq 1 preserves the generalized gamma convolution property.

2026 claimed proofs

A January 2026 preprint claims a proof for q>1q>1 and related consequences. A September 2026 preprint by Min Wang and Sheng Yin presents a second approach through finite gamma convolutions, approximation, and weak closure, but explicitly labels it a candidate proof and requests feedback. No counterexample, retraction, referee report, or independent verification was found.

Current status (as of September 2026): The conjecture has two claimed proofs, with the newest explicitly unverified; no established resolution is recorded.

Sources

Solutions 0

No solutions have been posted yet.