Subspace identification is claimed to have O(1/sqrt N) finite-sample matrix errors, O(N^{-1/(2n)}) pole errors, and a super-polynomial sample complexity in n/m; the last claim is not proven.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SY 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Finite Sample Analysis of Subspace Identification for Stochastic Systems
Subspace identification is claimed to have O(1/sqrt N) finite-sample matrix errors, O(N^{-1/(2n)}) pole errors, and a super-polynomial sample complexity in n/m; the last claim is not proven.