Frantzikinakis multiple-correlation structure question
Let be an ergodic nilsystem, let , and let . Define the multiple-correlation sequence for . The problem asks whether there exist a degree- generalized nilsequence and a null sequence such that for every , where .
References
- A structure theorem for multiple correlation sequences on nilsystems — arXiv — Or Shalom
Source: arXiv
Progress summary
A new paper reports the requested structural description in the specified setting, while a related stronger question remains open.
The question asks whether multiple-correlation sequences admit a structured description rather than vanishing. Recent work gives an affirmative answer in the stated ergodic-nilsystem setting, but this does not restore the refuted general vanishing conjectures.
Known results
- In 2014, multiple correlations for commuting transformations and integer-polynomial iterates were decomposed into a nilsequence plus an error small in uniform density.
- A 2020 account recorded approximation by nilsequences in the relevant seminorm, while noting that the exact structural statement remained open.
- Briet and Green exhibited a -step correlation that cannot be approximated in by integral combinations of continuous -step nilsequences; this does not address generalized nilsequences.
2025--2026 structure theorem
Or Shalom's preprint states that every -fold multi-correlation sequence is the sum of a degree- generalized nilsequence and a null-sequence, giving an affirmative answer to Frantzikinakis's Problem ; the same source says Problem remains open. A later paper reports arithmetic-progression structure theorems using null generalized nilsequences in the specified ergodic-nilsystem setting. These are specialist preprint claims and remain unverified here.
Current status (as of October 2026): Problem has a claimed affirmative answer in the specified setting, while Problem and stronger general formulations remain open and the new claim is unverified.
Solutions 0
No solutions have been posted yet.