DySMHO: Data-Driven Discovery of Governing Equations for Dynamical Systems via Moving Horizon Optimization

The paper defines DySMHO’s CV-thresholding rule (computed over the last ω coefficient estimates) and its stopping test (|Θ| unchanged for Ω consecutive thresholdings) and motivates CV as a variance-based discriminator between basic and non‑basic terms, but it does not provide a formal theorem or proof of finite-time identification under any separation hypothesis. The candidate solution, by contrast, gives a correct finite-time identification and termination argument once an explicit CV-separation assumption after some iteration i0 is imposed and minor regularity conditions are stated. Hence, the paper is incomplete on the theoretical claim, while the model’s proof is correct for the posed SOLVER_QUESTION. See the DySMHO algorithm description and stopping test in Algorithm S.2 and thresholding in Algorithm S.3, as well as the variance heuristic in Claim 1 .