{"id":"34256e71-9e45-4163-a54a-7ff084c7c8f4","arxiv_id":"2412.04157","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Introduces regional excitation and proves high-probability, non-asymptotic RLS error bounds for sub-exponentially unstable nonlinear closed-loop systems, with O(sqrt(ln t/t)) convergence under global excitation.","lead":"This paper derives non-asymptotic error bounds for least-squares parameter estimation in unstable nonlinear stochastic systems under closed-loop control. It introduces a checkable regional excitation condition and proves bounds that hold while the trajectory stays in an informative region, or for all times when the whole state space is exciting.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1 is conditional on Assumption 3 and condition (11); Assumption 3 excludes exponentially unstable systems and (11) can fail for small exciting regions, so the 'unstable' scope is narrower than the title suggests.","rationale":"The paper's main argument is technically sound: it builds on standard martingale and covering arguments, and the regional excitation condition is a genuine, checkable relaxation of global PE. The proof of Theorem 1 is internally consistent as far as I can verify, and the two examples illustrate how the hypotheses can be checked and how condition (11) can hold. The reader's weakest assumption correctly identifies the central limitation: Assumption 3 restricts the instability class to sub-exponential growth, and condition (11) may fail, making the theorem vacuous for small exciting regions. This is a scope limitation rather than a proof error, but it is load-bearing because the title and abstract claim 'unstable' systems, and the basic linear exponentially unstable plant is outside the theory. The paper does state 'sub-exponentially unstable' in the abstract, so the mismatch is modest; however, the absence of any general guarantee that (11) holds means the result is a conditional, case-by-case bound rather than a broadly applicable guarantee. This does not invalidate the contribution, but it justifies the CONDITIONAL verdict and suggests the title should be adjusted or the condition highlighted more prominently. I found no internal inconsistency or mathematical gap that would require rejection.","tokens_in":33357,"tokens_out":31189,"duration_ms":276743,"concrete_test":"Recompute the deterministic trajectory bound for the scalar unstable linear system X(t+1)=ρ X(t)+W with ρ=1.1, x0=1, u=0, w=0: any χ1 in Assumption 3 must satisfy χ1(t) ≥ ρ^t. Since ln(χ1(t)) ≥ t ln ρ = Θ(t), χ1 cannot be K1-SE (which requires ln χ1(t)=o(t)). Thus Assumption 3 is violated, demonstrating that the theorem excludes exponentially unstable systems and the title overstates the class.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is Theorem 1: under Assumptions 1-6 and if T_burn-in(δ,x0) ≤ T_excited(δ,x0), the RLS error is bounded by e(t,δ,x0) on the PE interval with probability at least 1−δ. The most load-bearing condition is Assumption 3. It requires χ1, χ3, σ2, χ4 to be K1-SE/K2-SE, meaning ln(χ1(t)) = o(t) and ln(χ4(r²)) = o(r). This limits the instability of the closed-loop dynamics to sub-exponential growth. A canonical unstable linear system X(t+1)=ρ X(t)+w with ρ>1 has minimal χ1(t) = ρ^t, for which ln χ1(t)=Θ(t), not o(t); hence Assumption 3 fails and the theorem does not apply to the most basic unstable systems. The title's 'unstable' is therefore broader than the actual scope, which the abstract does narrow to 'sub-exponentially unstable'. Second, even inside this class, Theorem 1 is vacuous whenever condition (11) fails, i.e., when the burn-in time needed to accumulate excitation exceeds the time the one-step predicted state can be guaranteed to remain in the exciting region XPE. T_excited is finite for any bounded XPE, and T_burn-in grows with d and 1/pPE; there is no general result guaranteeing (11), only case-by-case verification in the two examples. For a small XPE, T_excited can be less than T_burn-in, making the PE interval empty and the bound trivial. This is not an internal inconsistency, but it substantially narrows the practical reach of the claimed non-asymptotic guarantee.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies regularised least-squares identification of the linearly parameterised discrete-time nonlinear stochastic system (1) in closed loop with a known, not necessarily stabilising policy. It introduces a regional excitation condition (Def. 2) that is checkable from the basis functions, policy, and noise distributions, and proves in Thm. 1 that, whenever the burn-in time in (10) does not exceed the excited time in (9), the RLS estimation error is bounded by e(t, δ, x0) in (13) uniformly on the PE interval with probability at least 1−δ. Under global excitation, Cor. 1 extends this to all times past burn-in and gives asymptotic decay; under polynomial reachability growth, Cor. 2 gives the O(sqrt(ln t / t)) rate. Two examples are analysed: a PWA system that fails the BMSB condition but satisfies regional excitation, and a double integrator controlled by an arbitrary bounded policy.","tokens_in":33627,"tokens_out":15128,"duration_ms":154965,"significance":"If the proofs are correct, this is a genuinely new finite-sample guarantee for closed-loop identification of a class of unstable nonlinear systems, avoiding mixing or boundedness assumptions. The regional excitation notion is a useful, verifiable alternative to BMSB, and the PWA example makes the advantage concrete. The proofs are detailed and combine standard self-normalized martingale inequalities, Chernoff bounds, and covering arguments; the examples verify the assumptions explicitly. The main caveats are that Assumption 3 restricts instability to sub-exponential growth and that condition (11) can be vacuous, so the actual scope is narrower than the word ‘unstable’ in the title suggests.","major_comments":[{"comment":"Assumption 3 rules out exponentially unstable systems, including the canonical linear system X(t+1)=ρX(t)+W(t) with ρ>1: the minimal comparison function is χ1(t)=ρ^t, for which ln χ1(t)=Θ(t), not o(t), so χ1 is not K1-SE. Consequently Theorem 1 and Corollary 2 do not apply to the most basic unstable linear plants, and the title’s ‘unstable nonlinear systems’ overstates the scope. The abstract is careful (‘sub-exponentially unstable’), but the introduction and conclusions should state this restriction prominently and explain why it is inherent; otherwise readers will likely misapply the theorem.","section":"Assumption 3, Theorem 1"},{"comment":"Theorem 1 is vacuous when Tburn-in(δ,x0) > Texcited(δ,x0), and no general sufficient condition for (11) is provided; the paper only verifies it case-by-case in two examples. Since Texcited is finite for any bounded XPE and Tburn-in grows with d and 1/pPE, there are natural systems for which the PE interval is empty and the claimed non-asymptotic bound does not exist. The discussion after (11) acknowledges this qualitatively, but a result intended as a non-asymptotic guarantee needs either a quantitative sufficient condition for (11) or an explicit statement that the theorem covers only systems for which the PE interval is nonempty.","section":"Condition (11), Section 3.1"}],"minor_comments":[{"comment":"The error bound e(t,δ,x0) contains γ^{1/2}|θ*|_F, so it is not fully data-independent in the sense claimed in the paragraph after Lemma 3; to compute a numerical confidence interval one must know a bound on |θ*|. The paper should replace |θ*| by a known bound B in the statement, or state |θ*| ≤ B as an assumption, and correct the word ‘data-independent’.","section":"Equation (13), after Lemma 3"},{"comment":"The notation in Example 1 is confusing because x denotes both the state variable and the threshold parameter; for instance XPE = (−∞, 0.9x] and the constants bw, bs in Prop. 3 mix the two roles. Renaming the threshold, say x̄ or c, would make the example much easier to check.","section":"Section 4.1"},{"comment":"The definition of little-o is misstated: after defining f(r)=O(g(r)), the text says ‘h(r)=o(r) if lim f(r)/g(r)=0’, which should read h(r)=o(g(r)) with lim h(r)/g(r)=0.","section":"Notation, Section 1"},{"comment":"There are several small typos: ‘simualtions’ in Sec. 4.1.2, ‘inequlity’ in the proof of Thm. 1, and an extra comma in Cor. 2’s statement ‘e(t, , x0)’. These should be fixed in revision.","section":"Throughout"},{"comment":"The displayed equation for e(t,δ,x0) has a malformed line break with the equation number inserted mid-formula; the formatting should be corrected so that the bound is legible as a single expression.","section":"Equation (13)"}],"recommendation":"major_revision","confidential_remarks":"The proofs appear sound and the examples are verified carefully, so this is not a reject. My main reservation is that the title and framing promise more than the sub-exponential stability assumption and condition (11) deliver; both issues are fixable by rewriting the claims and adding a discussion of when (11) can be guaranteed. The dependence of the final bound on |θ*| should also be stated honestly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick read: this is a solid contribution. It gives the first non-asymptotic high-probability error bounds for RLS identification of a class of unstable nonlinear stochastic systems in closed loop, using a new regional excitation condition that is checkable offline from the basis functions, policy, and noise distributions. The proof machinery is standard but carefully assembled, and the PWA example shows regional excitation can hold when the BMSB condition fails. That is a real advantage.\n\nThe main caveat is scope. Assumption 3 limits the closed-loop state growth to sub-exponential; the title says 'unstable' but the result does not cover exponentially unstable linear systems, the first thing many readers will check. The abstract is accurate, and I take the title as a minor overreach rather than a hidden flaw. The second caveat is condition (11): T_burn-in ≤ T_excited is not assured and can fail for small exciting regions. The paper says so plainly, which I appreciate, but it means the theorem is vacuous for exactly some systems where regional excitation is weak. That is a real limitation, not an internal contradiction.\n\nThe bound also depends on |θ*| through the regularization term, which is standard for RLS finite-sample bounds and not a serious defect. The numerical example is illustrative only; no code, no error bars, but that is normal for a theory paper.\n\nI did not find circularity: regional excitation is defined entirely in terms of known quantities, the PE lower bound follows from it, and the error bound follows from self-normalized martingale concentration. The claims in the paper match what is proven.\n\nThis paper deserves a serious referee. I would send it to review, and I would expect a revise-round that asks for a tightened title and a more explicit discussion of when (11) is actually verifiable. I'd bring it to reading group if anyone in the group works on identification or adaptive control.","headline":"Genuine extension of non-asymptotic identification to sub-exponentially unstable nonlinear closed-loop systems, with a checkable regional excitation condition; the title overreaches and condition (11) can be vacuous, but the core argument holds.","tokens_in":34248,"tokens_out":3455,"would_cite":true,"duration_ms":32480,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["93E12","93E35"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper derives non-asymptotic, high-probability error bounds for least-squares identification of a class of unstable nonlinear closed-loop stochastic systems, under a regional excitation condition.","keywords":["system identification","non-asymptotic bounds","least squares estimation","unstable nonlinear systems","closed-loop identification","persistency of excitation","regional excitation","stochastic systems"],"falsifier":"Run many Monte-Carlo trials of the double integrator example with δ=0.1 and check the empirical frequency of the event that |θ̂(t)−θ*| ≤ e(t,δ,x0) for every t ≥ T_burn-in; a frequency below 0.9 would directly refute Theorem 1. Alternatively, construct a system satisfying the assumptions with polynomial growth of high degree and check numerically whether the bound e(t) truly behaves as O(√(ln t/t)).","tokens_in":33091,"feed_emoji":"📉","tokens_out":11866,"duration_ms":100392,"temperature":0.7,"pith_summary":"This paper aims to establish that regularized least-squares estimation can recover the unknown parameters of a nonlinear, stochastic, possibly unstable system from a single state trajectory, with high-probability error bounds that hold at every time step. The central device is a new 'regional excitation' assumption: if a region of the state space yields informative regressors, then once the trajectory has spent enough time in that region, the parameter error is bounded uniformly over the interval during which the one-step predicted state stays inside it. If the entire state space is informative, the bounds hold for all times and converge to zero, at rate O(√(ln t/t)) when the instability is at most polynomial. The paper matters because earlier non-asymptotic identification guarantees either assumed stability or bounded trajectories, or relied on mixing conditions that automatically exclude unstable and even marginally stable plants.","feed_headline":"Unstable nonlinear systems now have finite-time identification bounds","feed_subtitle":"Regional excitation certifies least-squares estimates with high probability, matching the linear-system rate.","key_machinery":"Regional excitation (Definition 2): for every direction ζ of the regressor space, the projected regressor ζ⊤ψ(x+W, α(x+W,S,ϑ)) has probability at least pPE of having magnitude at least cPE, uniformly over the region X and over all parameter guesses ϑ. Because it is defined only through the known basis functions ψ, the control policy α, and the noise distributions µs, µw, it can be verified without knowing the true parameter θ*. It supports a single-direction persistency-of-excitation lemma (Lemma 7), and an ε-covering argument over the unit sphere converts that into a high-probability, linearly growing lower bound on the minimum eigenvalue of the regularized Gramian G(t). The other load-bearing piece is the sub-exponential input-to-state bound (Assumption 3), which limits the growth of the closed-loop trajectory and ensures the burn-in time is finite.","core_discovery":"Theorem 1 is the central claim. Under Assumptions 1–6 (measurability, independent sub-Gaussian process noise, a sub-exponential input-to-state bound on the closed-loop trajectory, bounded controls, polynomially growing basis functions, and regional excitation over a set XPE), the paper defines two offline-computable times: T_burn-in, when persistency of excitation begins, and T_excited, a conservative lower bound on how long the one-step predicted state remains inside XPE. If T_burn-in ≤ T_excited, then with probability at least 1−δ the estimation error |θ̂(t)−θ*| is bounded by the explicit, time-dependent quantity e(t,δ,x0) for every t in the interval. Corollary 2 adds global excitation and polynomial instability, giving e(t,δ,x0) = O(√(ln t/t)) and convergence to zero for all times; the authors state this matches the rate for linear systems with spectral radius at most one, and that to their knowledge no comparable non-asymptotic guarantee existed for this class of unstable nonlinear closed-loop systems.","pith_inferences":["A testable extension is to weaken Assumption 3 to exponentially growing trajectories; the burn-in time would likely diverge, suggesting that segment-based or restart-based identification would be needed for genuinely unstable plants.","Regional excitation can be read as a design constraint: shaping the policy α or the exploratory noise µs to maximize cPE and pPE would directly shrink the certified error e(t), though the paper does not address this optimization.","The confidence region from Theorem 1 could feed a robust adaptive controller with finite-time guarantees, a use the paper does not explore."],"forward_implications":["Under global excitation, the RLS estimate enters and remains inside any arbitrarily small ball around the true parameter with probability at least 1−δ, for every initial state.","The rate O(√(ln t/t)) matches the benchmark for linear systems with spectral radius at most one, so the identified nonlinear class is certified at the same speed as that known linear case.","The bounds and the condition T_burn-in ≤ T_excited are verifiable offline from known objects, so a designer can decide before running an experiment whether the planned controller and noise will yield informative data.","In the merely regional case the bound decreases over the PE interval and stops improving after the trajectory leaves the exciting region, matching the simulated behavior of the piecewise affine example.","The regional excitation condition can be verified for the PWA example even though the block martingale small-ball condition (which requires future regressors to be non-degenerate on average given the past) fails, enlarging the set of systems with certified finite-sample identification."],"supporting_citations":[{"why":"Defines the block martingale small-ball condition and gives the linear-system rate O(√(ln t/t)) that Corollary 2 matches; the regional excitation condition is positioned against it.","marker":"[27]"},{"why":"Provides the self-normalized martingale tail bound (Lemma 16) used in proving the data-dependent least-squares error bound (Lemma 3).","marker":"[1]"},{"why":"Cited as the precursor of the data-dependent least-squares error bound appearing as Lemma 3.","marker":"[2]"},{"why":"Supplies the asymptotic consistency benchmark for linear stochastic regression with Gramian growth, which motivates the non-asymptotic burn-in analysis.","marker":"[14]"},{"why":"Defines forward completeness, the continuous-time analogue of Assumption 3's sub-exponential input-to-state bound.","marker":"[3]"},{"why":"Establishes measurability and well-posedness of discrete-time stochastic closed-loop sequences, used in Lemma 2.","marker":"[8]"},{"why":"Gives non-asymptotic identification for linear systems with nonlinear policies under a uniform trajectory bound; the double integrator example contrasts with it.","marker":"[17]"}],"fun_headline_variants":["Finite-time identification bounds for unstable nonlinear closed loops","Regional excitation yields finite-time least-squares error bounds","High-probability error bounds for unstable closed-loop identification","First non-asymptotic guarantees for unstable nonlinear closed loops","Regional excitation gives finite-time guarantee for unstable systems"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the closed-loop trajectory grows at most sub-exponentially in time (Assumption 3); if the system is exponentially unstable or otherwise grows faster, the burn-in time can be infinite or the condition T_burn-in ≤ T_excited can fail, in which case Theorem 1 issues no guarantee.","fun_headline_variants_meta":{"raw":{"variants":["Finite-time identification bounds for unstable nonlinear closed loops","Regional excitation yields finite-time least-squares error bounds","High-probability error bounds for unstable closed-loop identification","First non-asymptotic guarantees for unstable nonlinear closed loops","Regional excitation gives finite-time guarantee for unstable systems"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000731,"raw_usage":{"total_tokens":3226,"prompt_tokens":855,"completion_tokens":2371,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":471,"completion_tokens_details":{"reasoning_tokens":2294}},"tokens_in":471,"tokens_out":2371,"duration_ms":16198,"temperature":1.0,"reasoning_tokens":2294,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:41:43.185359+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run many Monte-Carlo trials of the double integrator example with δ=0.1 and check the empirical frequency of the event that |θ̂(t)−θ*| ≤ e(t,δ,x0) for every t ≥ T_burn-in; a frequency below 0.9 would directly refute Theorem 1. Alternatively, construct a system satisfying the assumptions with polynomial growth of high degree and check numerically whether the bound e(t) truly behaves as O(√(ln t/t)).","supporting_citations":[{"cited_title":"Learning without mixing: Towards a sharp analysis of linear system identification","cited_arxiv_id":null,"evidence_quote":"Defines the block martingale small-ball condition and gives the linear-system rate O(√(ln t/t)) that Corollary 2 matches; the regional excitation condition is positioned against it."},{"cited_title":"Improved algorithms for linear stochastic bandits.Advances in neural information processing systems , 24, 2011","cited_arxiv_id":null,"evidence_quote":"Provides the self-normalized martingale tail bound (Lemma 16) used in proving the data-dependent least-squares error bound (Lemma 3)."},{"cited_title":"Regret bounds for the adaptive control of linear quadratic systems","cited_arxiv_id":null,"evidence_quote":"Cited as the precursor of the data-dependent least-squares error bound appearing as Lemma 3."},{"cited_title":"Least squares estimates in stochastic regression models with applications to identification and control of dynamic systems.The Annals of Statistics, 10(1):154–166, 1982","cited_arxiv_id":null,"evidence_quote":"Supplies the asymptotic consistency benchmark for linear stochastic regression with Gramian growth, which motivates the non-asymptotic burn-in analysis."},{"cited_title":"Forward completeness, unboundedness observability, and their lyapunov characterizations.Systems & Control Letters, 38(4- 5):209–217, 1999","cited_arxiv_id":null,"evidence_quote":"Defines forward completeness, the continuous-time analogue of Assumption 3's sub-exponential input-to-state bound."},{"cited_title":"Discrete-time stochastic control systems: A continuous lyapunov function implies robustness to strictly causal perturbations","cited_arxiv_id":null,"evidence_quote":"Establishes measurability and well-posedness of discrete-time stochastic closed-loop sequences, used in Lemma 2."},{"cited_title":"Non-asymptotic System Identification for Linear Systems with Nonlinear Policies","cited_arxiv_id":"2306.10369","evidence_quote":"Gives non-asymptotic identification for linear systems with nonlinear policies under a uniform trajectory bound; the double integrator example contrasts with it."}],"review_version":1}