Determinant Cumulants from Three Readings: Sharp Risk and Communication Limits
This paper studies a determinant contraction of the fourth joint cumulants of a random real three-by-three matrix. Three rank-one readings per independent matrix identify the functional through a finite family of measurement settings, while two readings are insufficient even for smooth signal laws with fixed mean and covariance. A design with 60 settings yields an unbiased estimator under unknown independent channel errors or jointly Gaussian readout errors. The paper gives sharp minimax risk rates under weak tail bounds and characterizes the tradeoff between communication and estimation risk at the weak eighth-tail boundary. Three sensors attain this order by sending simultaneous messages based on their own readings. The paper also gives bounds for adaptation to an unknown tail exponent and efficiency results for unquantized measurements.
Authors
- Akihiro Koide (ORCID: https://orcid.org/0009-0002-8391-3480)
Publication Details
- Journal
- Figshare
- Published
- 2026-11-21
- DOI
- https://doi.org/10.6084/m9.figshare.35081320
- Primary Topic
- Distributed Sensor Networks and Detection Algorithms
- Type
- preprint