← All problems

Bondesson's open closure question for Thorin's class

Let XX be a positive random variable whose law belongs to Thorin's class, equivalently the class of generalized gamma-convolution laws, and let q>1q>1. Is the law of XqX^q necessarily in Thorin's class?

References

Source: arXiv

Progress summary

Refreshed
Open · progress reported

A new 2026 preprint reports partial progress, but it does not settle whether taking powers preserves Thorin’s class.

Bondesson’s question asks whether the Thorin, or generalized gamma-convolution, class remains closed when an independent variable is raised to a power q>1q>1.

Known results

  • For Y∼Gamma⁡(k)Y\sim\operatorname{Gamma}(k) and independent XX with a hyperbolically monotone density of order ll, 0<k≤l0<k\le l, both YXYX and Y/XY/X are generalized gamma convolutions (2018).

October 2026 characterization

Thomas K. Kloster reports closure under multiplication by independent variables with hyperbolically completely monotone densities, together with additional inclusions in Thorin’s class. The abstract describes this as a partial settlement, not a solution of the general power question.

Current status (as of November 2026): A partial multiplication-closure result is claimed, while the general power-closure question remains open and no complete resolution is independently confirmed.

Sources

Solutions 0

No solutions have been posted yet.