{"id":"c0f81232-3914-49e1-a009-5483f8da2f90","arxiv_id":"1908.05001","paper_version":1,"verdict":"UNVERDICTED","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A review of recent progress and challenges in making chemically specific coarse-grained simulation models reproduce the dynamical properties of the underlying molecular systems.","lead":"This paper is a review of how scientists build coarse-grained simulation models of soft matter, which simplify computer models of molecules by grouping atoms together, while also capturing realistic dynamics. It maps out three broad strategies for fixing the dynamical inaccuracies of these models and connects many methods that have been developed.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; the review's organizational claim is modest, well-supported by the cited literature, and explicitly self-limited regarding general dynamical consistency.","rationale":"The reader's verdict of UNVERDICTED rests on the view that a review paper does not fit the accept/reject categories for primary research. I agree that the paper contains no new computational or experimental results to verify, and the central claim is an organizational synthesis. The reader's weakest_assumption points to the accuracy and representativeness of the literature categorization; that is the most plausible soft spot, but it is not a load-bearing flaw in the sense of making the review's claim false or misleading. The review is internally coherent: the three approaches are consistently described, their limitations are acknowledged with specific examples, and the conclusions are carefully qualified. I therefore see no reason to change the verdict from UNCHANGED. The only residual risk is selection bias in the cited literature, which is a standard limitation of reviews and can be checked by systematic search. I mark agreement as 'partial' because I share the reader's identification of representativeness as the weakest assumption, but I do not consider it significant enough to downgrade the review's central claim.","tokens_in":25028,"tokens_out":4419,"duration_ms":50651,"concrete_test":"Perform a systematic literature search (e.g., Web of Science or Scopus, 2014-2019) using keyword combinations such as 'coarse-grained dynamics Mori-Zwanzig', 'time rescaling coarse-grained diffusion', and 'Markov state model coarse-grained refinement', then classify the retrieved papers into the review's three categories. Check whether any substantial class of methods (e.g., equation-free approaches, machine-learned coarse-grained dynamics, or data-driven GLE inference) is absent; if a major omitted route is found, the completeness of the taxonomy would need revision, and the organizational claim would be weakened accordingly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"This is a review paper whose central claim is an organizational statement: that recent progress in Mori-Zwanzig parametrization, time-rescaling relations, and free-energy-landscape-informed refinement has improved the ability of chemically-specific CG models to reproduce dynamical properties, while no single method yet provides general consistency. That claim is appropriately hedged in the abstract, introduction, and Section VI, where the author explicitly states that there is no general solution and that several variational frameworks still lack molecular applications. The three-route taxonomy is not internally inconsistent: the review acknowledges coupling between conservative and dissipative forces, gives counterexamples where uniform time rescaling breaks down (e.g., the azobenzene liquid crystal example of Mukherjee et al.), and notes that reproducing structural properties does not guarantee kinetic fidelity. The main residual risk is that the literature selection may not be fully representative of the field, which is an inherent limitation of any review rather than a technical flaw in the argument. Because the paper makes no overreaching universal claim and explicitly identifies open challenges, I do not find a load-bearing concern that would invalidate the central claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This review surveys recent methodological advances toward chemically specific coarse-grained (CG) models that reproduce dynamical properties of higher-resolution references. The author organizes the field into three perspectives: (i) explicit correction of dynamics via the Mori-Zwanzig (MZ) formalism, including DPD, momentum-dependent friction, memory kernels, and variational approaches; (ii) empirical or semi-empirical time-rescaling relationships for polymers and liquids; and (iii) free-energy-landscape-informed refinement, including structural-kinetic relationships and Markov-state-model-based reparametrization. The review is explicitly self-limited, acknowledging that no single method currently provides general dynamical consistency and that several variational frameworks remain without molecular applications.","tokens_in":25219,"tokens_out":6150,"duration_ms":57277,"significance":"The paper is a useful and well-structured synthesis of a rapidly growing literature. Its main contribution is organizational: it connects approaches that are usually presented in separate communities (MZ-based bottom-up dynamics, time-rescaling, and MSM-based kinetics) and identifies the plateau problem, the breakdown of uniform rescaling, and the coupling between conservative and dissipative forces as recurring cross-cutting issues. The review is carefully hedged, uses representative counterexamples (e.g., azobenzene liquid crystals, ionic liquids) to illustrate limitations, and explicitly lists open challenges. It does not overclaim, and it gives credit to recent theoretical advances. As a review, its value lies in the accuracy and comprehensiveness of its literature synthesis rather than in new results.","major_comments":[],"minor_comments":[{"comment":"The definition of the Liouville operator as L = Σ_i (∂/∂r_i + ∂/∂p_i) is dimensionally inconsistent and incomplete. It should read iL = Σ_i [(p_i/m_i)·∂/∂r_i + F_i·∂/∂p_i] (up to the usual sign convention). Please correct or replace with a proper reference.","section":"Section II.A, Eq. (5)"},{"comment":"The citation 'Salerno and Grest [101]' should be 'Salerno et al. [101]' since the cited article has four authors; likewise, 'Armstrong and Ballone [117]' should be 'Armstrong et al. [117]'.","section":"Section III.A, text near ref. [101]"},{"comment":"References [31] and [11] are duplicates, as are references [50] and [26]; please consolidate or remove the redundant entries.","section":"Reference list"},{"comment":"The caption contains a typo: 'long time sale' should read 'long time scale'.","section":"Figure 3 caption"},{"comment":"The symbol t is reused as an integration cutoff and as an evaluation time; the text should explicitly state that t is chosen as a plateau time and that the three expressions are approximations that coincide only under time-scale separation.","section":"Section II.B, Eqs. (8)-(10)"}],"recommendation":"minor_revision","confidential_remarks":"The review draws heavily on the author's own published work (Rudzinski and Bereau, Bereau and Rudzinski) in Sections IV.A and IV.B, and the discussion of those methods is somewhat more detailed than the treatment of comparable external efforts. This is not a disqualifying issue for a review, but it is worth keeping in mind when evaluating balance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a well-executed review of a real problem, not a research claim. It deserves a serious referee. The three-way taxonomy—Mori-Zwanzig-based friction, time rescaling, and free-energy-landscape refinement—is useful and mostly right. The physics is accurate: the GLE and DPD derivations are correct, the plateau problem is handled fairly, and the discussion of when Markovian friction fails matches what I know of the field.\n\nWhat the paper does well: it connects approaches that usually live in separate literatures. The MZ section covers friction kernels, memory, and variational methods without overstating maturity. The time-rescaling section distinguishes polymer and liquid cases and gives the azobenzene example where uniform rescaling breaks down. The free-energy-landscape section makes a genuine case for why barrier heights and intermediate states matter for kinetics, and it is honest that these methods do not reproduce full dynamics. The paper repeatedly says there is no general solution yet, and it points to open challenges—like the lack of molecular applications for some variational frameworks—without hiding them.\n\nSoft spots, in proportion: this is a narrative review, not a systematic one. The literature selection is representative rather than exhaustive, and some claims of “recent progress” rest on selected examples. That is normal for the genre, but a table comparing methods, target properties, and tested systems would have made it easier to judge breadth. The author’s own MSM-based refinement work is cited prominently, but the claims are specific, published results, and not oversold; I do not see a self-citation problem. A minor tension: the “free-energy-landscape” route sometimes shades into top-down modeling, which the paper acknowledges but could have addressed more explicitly.\n\nI agree with the reader’s take that the paper does not fit a simple accept/reject verdict as a research claim. But as a review, it is accurate, fairly comprehensive, and useful. Anyone entering CG dynamics would benefit from reading it, and method developers will find the organization helpful even if they already know the papers. The math and citation pattern look solid. I would cite it if I were writing about CG dynamics, and I would bring it to a reading group for students.\n\nRecommendation: send it to peer review. A serious referee can check coverage and push for a more explicit statement of selection criteria, but the core synthesis is sound and worth publishing.","headline":"A solid, well-hedged review that organizes the CG-dynamics literature into three useful routes; no new results, but no load-bearing flaws—worth sending to a serious referee.","tokens_in":25700,"tokens_out":1894,"would_cite":true,"duration_ms":22449,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Chemically specific coarse-grained models now have three complementary routes to recover consistent dynamics—Mori–Zwanzig friction corrections, empirical time rescaling, and barrier-targeted potential refinement—but none is universally…","keywords":["coarse-grained models","dynamical properties","Mori-Zwanzig formalism","generalized Langevin equation","time rescaling","free-energy landscape","Markov state models","chemically specific models"],"falsifier":"The claim would be falsified by a chemically specific CG model that, with only a uniform Langevin thermostat and a structure-based potential, reproduces the reference model's velocity autocorrelation functions, diffusion constants, and ratios of mean first-passage times between metastable states; or, in a controlled two-basin system, by showing that changing barrier heights without changing intrabasin friction leaves the relative transition rates unchanged.","tokens_in":24826,"feed_emoji":"⚛️","tokens_out":6909,"duration_ms":64046,"temperature":0.7,"pith_summary":"Coarse-grained (CG) simulations replace groups of atoms with single sites to reach longer times and larger systems, but removing degrees of freedom silently rewrites the dynamics: friction is lost and the free-energy landscape is smoothed, so relative time scales of different processes come out wrong. This review argues that recent work has attacked that problem along three lines: adding friction and noise back through the Mori–Zwanzig generalized Langevin equation, rescaling time by an empirical factor fitted to a target transport property, and reparametrizing the conservative potential to reproduce the dominant free-energy barriers that control barrier-crossing kinetics. The paper sorts the literature into these three perspectives and shows how they are complementary rather than competing. If the synthesis is right, dynamical consistency is no longer an afterthought in CG model building but an explicit design target with a known menu of methods, each with identifiable limitations.","feed_headline":"No single fix yet gives coarse-grained models true dynamics","feed_subtitle":"A review shows friction kernels, time rescaling, and barrier-focused potentials each fix part of the problem.","key_machinery":"The central object is the generalized Langevin equation (GLE) obtained from the Mori–Zwanzig projection-operator formalism: dP/dt = −dU0/dR − ∫ Γ(R,P,t−t′)V(t′) dt′ + δF^Q(t). The paper uses the GLE as a conceptual and practical scaffold: the conservative term is the many-body potential of mean force, the memory-kernel term encodes friction that depends on coordinates, momenta, and time, and the random force is connected to the kernel by a fluctuation–dissipation relation. The three reviewed strategies correspond to simplifying this equation in different ways—retaining a structured friction kernel (MZ approach), collapsing the kernel to a scalar and absorbing it into a time-rescaling factor (time-rescaling approach), or neglecting explicit friction and instead shaping the conservative potential to capture the free-energy barriers that dominate barrier-crossing kinetics (free-energy-landscape approach). The review's conceptual move is to treat these three simplifications as complementary perspectives on the same GLE.","core_discovery":"The paper's central claim is that the two known sources of dynamical error in bottom-up coarse-grained models—the loss of friction from eliminated degrees of freedom and the smoothing of the many-body potential of mean force by approximate interaction potentials—distort not just absolute time scales but the ratios of time scales between distinct processes, which can change the qualitative pathways a model samples. It organizes recent progress into three approaches: (i) bottom-up parametrization of the generalized Langevin equation, from dissipative particle dynamics friction kernels to memory kernels and variational methods; (ii) empirical time-rescaling relations, including scalar rescalings for polymer melts and entropy-based rescalings for liquids; and (iii) free-energy-landscape-informed refinement, where structural–kinetic relationships and Markov state models are used to reparametrize conservative forces so that dominant barriers and the hierarchy of long-time processes are reproduced. The review does not claim any single method is universally successful; it claims that the field now has a structured view of the problem and a set of methods that work in identifiable regimes, with clear markers of when they fail.","pith_inferences":["A testable extension: combining a barrier-targeted potential with a modest memory-friction correction should outperform either alone for systems like ionic liquids, where both barrier heterogeneity and missing friction are severe; this is implicit in the review's outlook but not yet demonstrated.","The review's emphasis on relative time scales suggests a useful quality metric for CG models: not the absolute speed-up factor but the ratio of mean first-passage times between pairs of metastable states compared with the reference model; this metric would be transferable across systems where a uniform rescaling is inapplicable.","The structural–kinetic results for helix–coil systems hint that steric excluded volume alone constrains the attainable free-energy landscape, implying that CG representations that preserve accurate excluded volume may inherit more kinetic fidelity than those that soften it—a connection left implicit in the paper.","For machine-learned CG potentials, the review implies that training losses should be augmented with dynamical observables or constraints on relative barriers, not just structural correlation functions, to avoid reproducing structures while losing dynamics."],"forward_implications":["MZ-based methods with memory or momentum-dependent friction are now practical for high-resolution CG models, though they demand a clear time-scale separation to reproduce both short-time correlation functions and long-time diffusion.","Scalar time rescaling works when a single, state-point-dependent factor reconnects CG and reference dynamics; this holds for homopolymer melts and some simple liquids but fails when distinct species or processes demand different rescalings, as in ionic liquids or azobenzene liquid crystals.","Refining conservative potentials to reproduce free-energy barriers—guided by Markov state models or structural–kinetic relationships—can restore the hierarchy of long-time processes without adding dissipative forces, at the cost of accepting errors in basin-resolved thermodynamics.","Variational and path-space methods (relative entropy rate, trajectory matching, spectral matching) are emerging as systematic routes to target dynamics directly, but have not yet been widely tested on molecular systems.","A consistent theme: the CG mapping, the conservative potential, and the dissipative forces are coupled; optimizing one without the others limits achievable dynamical accuracy."],"supporting_citations":[{"why":"Supplies the projection-operator formalism that yields the generalized Langevin equation, the theoretical foundation for all MZ-based CG dynamics work.","marker":"[9]"},{"why":"Independent derivation of the same generalized Langevin equation; grounds the review's key equation and the friction-kernel route.","marker":"[10]"},{"why":"Foundational discussion of the MZ approach for parametrizing CG equations of motion, framing the friction-kernel perspective.","marker":"[11]"},{"why":"Defines the many-body potential of mean force and the force-matching (MS-CG) framework for conservative CG forces; the baseline that the three dynamical routes build on.","marker":"[19]"},{"why":"First practical parametrization of CG friction from force/velocity time-correlation functions; the template for bottom-up dissipative force calculations.","marker":"[29]"},{"why":"Demonstrates scalar time rescaling for polystyrene-based CG melts; the canonical time-rescaling result.","marker":"[93]"},{"why":"Identifies structural–kinetic relationships for helix–coil transitions; supplies the empirical basis for the free-energy-landscape perspective.","marker":"[134]"},{"why":"Introduces Markov-state-model-guided refinement of CG potentials to reproduce barrier-crossing kinetics; the exemplar for the third route.","marker":"[157]"},{"why":"Presents spectral matching to dynamical-propagator eigenfunctions, providing the recent variational bridge between MZ and landscape approaches.","marker":"[161]"}],"fun_headline_variants":["Coarse-grained dynamics: three fixes, but no universal cure yet","Review maps fixes for coarse-grained model dynamics","Dynamics in coarse-grained models: a field with partial fixes","No single method fixes coarse-grained dynamics, review says","Coarse-grained dynamics still lack a one-size-fits-all fix"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The synthesis assumes the Mori–Zwanzig diagnosis—that missing friction and smoothed free-energy barriers are the two dominant sources of dynamical error—is correct for chemically specific CG models, and that the cited studies fairly represent the field.","fun_headline_variants_meta":{"raw":{"variants":["Coarse-grained dynamics: three fixes, but no universal cure yet","Review maps fixes for coarse-grained model dynamics","Dynamics in coarse-grained models: a field with partial fixes","No single method fixes coarse-grained dynamics, review says","Coarse-grained dynamics still lack a one-size-fits-all fix"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000575,"raw_usage":{"total_tokens":2744,"prompt_tokens":1004,"completion_tokens":1740,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":620,"completion_tokens_details":{"reasoning_tokens":1658}},"tokens_in":620,"tokens_out":1740,"duration_ms":13060,"temperature":1.0,"reasoning_tokens":1658,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:24:41.837090+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The claim would be falsified by a chemically specific CG model that, with only a uniform Langevin thermostat and a structure-based potential, reproduces the reference model's velocity autocorrelation functions, diffusion constants, and ratios of mean first-passage times between metastable states; or, in a controlled two-basin system, by showing that changing barrier heights without changing intrabasin friction leaves the relative transition rates unchanged.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies structural–kinetic relationships for helix–coil transitions; supplies the empirical basis for the free-energy-landscape perspective."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces Markov-state-model-guided refinement of CG potentials to reproduce barrier-crossing kinetics; the exemplar for the third route."},{"cited_title":"N¨ uske, L","cited_arxiv_id":null,"evidence_quote":"Presents spectral matching to dynamical-propagator eigenfunctions, providing the recent variational bridge between MZ and landscape approaches."}],"review_version":1}