{"id":"e5529a47-55d9-4d0e-b97f-7debcdf88c59","arxiv_id":"2606.29083","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"Residual-guided neural dictionary learning reduces spectral pollution and improves forecast accuracy in Koopman approximations on benchmarks and sea-surface temperature data.","lead":"This paper trains neural-network dictionaries for Koopman approximation by minimizing Residual DMD residuals plus a condition-number penalty, rather than prediction error alone. A smart generalist might read it to see a concrete way to make spectral claims from data-driven dynamics more certifiable instead of relying on prediction accuracy.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Residual DMD residuals certify consistency within the learned dictionary span but lack a demonstrated bound linking them to the true infinite-dimensional Koopman spectrum","rationale":"The reader's weakest assumption matches the load-bearing point exactly. Because the full text was not supplied to the initial reader, the UNVERDICTED verdict remains appropriate; the concern above would be settled by the concrete test rather than by re-reading the abstract.","tokens_in":1807,"tokens_out":351,"duration_ms":30053,"concrete_test":"On a linear system whose Koopman spectrum is known exactly (e.g., the 2-D harmonic oscillator), compute the Residual DMD residual for both the true eigenvalues and for deliberately constructed spurious modes lying outside the true eigenspace; if any spurious mode can be made to have residual below the threshold used in the paper's selection criterion, the residual does not certify genuineness.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that minimizing Residual DMD residuals produces eigenvalues/modes that are genuine for the infinite-dimensional operator rather than artifacts of the finite dictionary. The residual is an a-posteriori quantity computed from the same finite lifted data matrix used to form the EDMD operator; small residual therefore indicates that the candidate mode is approximately invariant under the empirical operator, but does not automatically imply proximity to a true eigenfunction of the Koopman operator on the underlying function space. Without an a-posteriori error estimate that controls the distance to the true spectrum (or a proof that the residual vanishes only for true spectral objects), the certification claim rests on the empirical observation that the learned dictionaries perform better on benchmarks. This is the precise location where the argument is least secure.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript proposes Residual-Guided Dictionary Learning, in which neural-network dictionaries for EDMD are trained by minimizing Residual DMD residuals (operator-level a-posteriori errors) together with a penalty on the condition number of the lifted data matrix. The central claim is that this produces dictionaries that are expressive, stable, and spectrally disciplined, sharply reducing spectral pollution, improving residual pseudospectral inclusion, and lowering forecast error on conservative/dissipative benchmarks as well as on sea-surface temperature data.","tokens_in":1972,"tokens_out":323,"duration_ms":18240,"significance":"If the residual-based objective can be shown to furnish a reliable a-posteriori certificate that computed eigenvalues and modes are close to those of the infinite-dimensional Koopman operator (rather than merely consistent inside the learned finite span), the approach would supply a principled alternative to prediction-error-only dictionary learning and strengthen the trustworthiness of numerical Koopman spectra.","major_comments":[{"comment":"Abstract: the assertion that Residual DMD residuals 'test whether computed eigenvalues and modes are genuine Koopman spectral objects' is load-bearing for the central claim, yet the provided description supplies no a-posteriori error bound relating the finite-dictionary residual to the distance from the true infinite-dimensional spectrum; the residual is formed from the same lifted data matrix used to build the EDMD operator, so small residuals certify invariance under the empirical operator but do not automatically imply proximity to true eigenfunctions.","section":"Abstract"}],"minor_comments":[],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the careful reading and for highlighting this important distinction regarding the interpretation of Residual DMD residuals. We address the major comment below.","responses":[{"response":"We agree with the referee that the residual, being formed from the same lifted data matrix, certifies approximate invariance under the empirical EDMD operator and does not supply a rigorous a-posteriori bound on the distance to the spectrum of the infinite-dimensional Koopman operator. The manuscript does not derive or claim such a bound. Nevertheless, minimizing the residual during dictionary learning selects observables for which the finite-dimensional operator is more consistent with the observed data in a spectral sense; this is what produces the observed reduction in spectral pollution and improved pseudospectral inclusion in the experiments. We will revise the abstract (and the corresponding claim in the introduction) to state that the residuals provide a certificate of consistency with the empirical operator, which in practice yields dictionaries whose spectra are more reliable, without asserting that they directly test genuineness with respect to the infinite-dimensional operator.","revision_made":"yes","referee_comment":"[Abstract] Abstract: the assertion that Residual DMD residuals 'test whether computed eigenvalues and modes are genuine Koopman spectral objects' is load-bearing for the central claim, yet the provided description supplies no a-posteriori error bound relating the finite-dictionary residual to the distance from the true infinite-dimensional spectrum; the residual is formed from the same lifted data matrix used to build the EDMD operator, so small residuals certify invariance under the empirical operator but do not automatically imply proximity to true eigenfunctions."}],"tokens_in":1341,"tokens_out":342,"duration_ms":23933,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The paper's main move is to fold Residual DMD residuals directly into the training objective for neural dictionaries, along with a penalty on the condition number of the lifted matrix. This is a concrete way to push the learned observables toward ones where the computed modes are approximately invariant under the empirical operator.\n\nWhat works is the empirical side. On the benchmarks the method cuts spectral pollution and improves forecasts compared to fixed dictionaries, and the sea-surface temperature example shows cleaner diagnostics from noisy data. The conditioning term is a sensible addition to avoid degenerate representations.\n\nThe soft spot is the interpretation of the residuals as a certificate for genuine Koopman spectral objects. The residual measures how well a mode is invariant under the finite EDMD operator built from the learned dictionary. That is useful for consistency inside the span, but the paper does not supply a bound or argument showing that small residuals imply closeness to an eigenfunction of the true infinite-dimensional Koopman operator. The abstract states that the residuals 'test whether computed eigenvalues and modes are genuine Koopman spectral objects,' yet the stress-test note correctly flags that this link is not automatic. Without that, the reliability claim rests on the benchmark improvements rather than on a certified error estimate.\n\nThe work is aimed at people doing data-driven spectral analysis of nonlinear systems who want more disciplined dictionary learning. A reader interested in practical improvements to EDMD-style methods will find the experiments useful. It is coherent enough to deserve referee time, though the theoretical gap around the residual certificate should be addressed.\n\nI would send it to review.","headline":"The paper folds Residual DMD residuals plus a conditioning penalty into neural dictionary training for Koopman spectra, with decent benchmark gains, but the claim that residuals certify genuine infinite-dimensional spectral objects is not backed by a bound or argument.","tokens_in":2483,"tokens_out":400,"would_cite":false,"duration_ms":24561,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Neural dictionaries trained to minimize Residual DMD errors produce Koopman approximations with less spectral pollution and more reliable eigenvalues.","keywords":["Koopman operator","dictionary learning","residual DMD","spectral approximation","neural networks","dynamic mode decomposition","numerical stability","forecasting"],"falsifier":"On a benchmark system whose Koopman spectrum is known analytically, observe whether eigenvalues produced by the residual-trained dictionary still deviate from the true spectrum even when the reported residuals are small.","tokens_in":2687,"feed_emoji":"","tokens_out":606,"duration_ms":17784,"temperature":0.7,"pith_summary":"The paper trains neural-network dictionaries for Koopman operator approximation by minimizing Residual Dynamic Mode Decomposition residuals instead of prediction error alone. These residuals serve as operator-level a-posteriori checks that test whether computed eigenvalues and modes are genuine spectral objects of the infinite-dimensional operator. A penalty on the condition number of the lifted data matrix is added to prevent collapse into unstable coordinates. The resulting dictionaries are shown to reduce spectral pollution, tighten residual pseudospectral inclusion, and improve forecast accuracy on both benchmark systems and real sea-surface temperature data.","feed_headline":"Residuals train neural dictionaries for reliable Koopman spectra","feed_subtitle":"Minimizing a-posteriori DMD residuals plus a condition penalty yields cleaner eigenvalues and lower forecast error than fixed dictionaries.","key_machinery":"Residual DMD residuals, used as a-posteriori operator errors that certify whether eigenvalues and modes are genuine Koopman spectral objects, together with a condition-number penalty on the lifted data matrix.","core_discovery":"By making minimization of Residual DMD residuals the training objective and coupling it with a condition-number penalty, the learned dictionary becomes expressive, numerically stable, and spectrally disciplined, so that its finite-dimensional eigenvalues and modes more closely reflect the spectrum of the underlying infinite-dimensional Koopman operator.","pith_inferences":["The same residual objective could be applied to other linearization methods beyond EDMD to certify spectral objects.","If residuals remain small on new data streams, the dictionary may transfer across related dynamical regimes without retraining.","The conditioning penalty suggests a general template for dictionary learning that balances expressivity against numerical stability in other operator-learning settings."],"forward_implications":["Spectral pollution is sharply reduced on both conservative and dissipative benchmark systems.","Residual pseudospectral inclusion improves, tightening the set of candidate eigenvalues.","One-step forecast error decreases relative to standard fixed dictionaries.","Koopman diagnostics become cleaner and one-step forecasts improve on noisy sea-surface temperature observations without known governing equations."],"fun_headline_variants":["Residual-guided learning for accurate Koopman spectra","Neural dictionaries trained on DMD residuals for spectral accuracy","Residual and condition penalties for trustworthy Koopman spectra","Learning dictionaries to minimize residual DMD errors"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"That small Residual DMD residuals reliably indicate that the computed eigenvalues and modes are true spectral features of the infinite-dimensional Koopman operator rather than artifacts of the finite dictionary.","fun_headline_variants_meta":{"raw":{"variants":["Residual-guided learning for accurate Koopman spectra","Neural dictionaries trained on DMD residuals for spectral accuracy","Residual and condition penalties for trustworthy Koopman spectra","Learning dictionaries to minimize residual DMD errors"]},"model":"grok-4.3","cost_usd":0.004358,"raw_usage":{"total_tokens":2106,"prompt_tokens":672,"num_sources_used":0,"completion_tokens":54,"cost_in_usd_ticks":43578000,"prompt_tokens_details":{"text_tokens":672,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1380,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":672,"tokens_out":54,"duration_ms":15218,"temperature":1.0,"reasoning_tokens":1380,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-30T08:20:49.544549+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"On a benchmark system whose Koopman spectrum is known analytically, observe whether eigenvalues produced by the residual-trained dictionary still deviate from the true spectrum even when the reported residuals are small.","supporting_citations":[],"review_version":1}