Breiman's conjecture
Let and be independent sequences of i.i.d. random variables, where is integrable and nonconstant and with . Define . Breiman's conjecture asserts that if converges in distribution to a nondegenerate random variable, then the tail is regularly varying at infinity with index for some ; equivalently, for every , .
References
- Subsequential Version of Breiman's Conjecture — arXiv — Volodymyr Pavlenkov
Source: arXiv
Progress summary
A new result settles the slowest-tail case, but the broader conjecture remains unproved outside stronger moment assumptions.
Breiman's 1965 conjecture asks when convergence of a randomly weighted mean to a nondegenerate limit forces a specific heavy-tail structure for the weights. The known formulation concerns integrable, nonconstant marks and tail index .
Known results
- Mason and Zinn: the characterization holds when the mark has a finite moment of order .
- A later extension proves it for centered marks satisfying a characteristic-function condition with index .
- Continuous-time analogues establish sufficiency under integrability, but retain stronger moment assumptions for necessity.
October 2026 subsequential theorem
Volodymyr Pavlenkov's preprint proves the slowly varying endpoint and characterizes broader nondegenerate subsequential limits, including limits for . Its results isolate the remaining nonzero-index necessity issue; the claimed extension to mere integrability is not independently verified here.
Current status (as of October 2026): The slowly varying case is claimed resolved, while the full nonzero-index conjecture remains open or conditional beyond the finite-second-moment regime.
Solutions 0
No solutions have been posted yet.