{"id":"c4a19b90-4125-4d45-b873-43073329f196","arxiv_id":"2608.05388","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"SC-UCG uses message passing to self-consistently infer internal state probabilities during coarse-grained MD, and with the MISC training scheme it captures a chiral symmetry-breaking phase transition from single-temperature training.","lead":"Researchers built a new coarse-grained simulation method that lets molecules' hidden internal states change automatically during simulation, without hand-picked coordinates. They show it can reproduce a temperature-driven chiral phase transition in a model fluid even though it was trained at only one temperature.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"One-ISC-pass RLE is unvalidated exactly in the critical regime; without an SCE convergence check, the reported phase transition may be an artifact of the update schedule rather than equilibrium SC-UCG sampling.","rationale":"The reader's weakest_assumption is RLE, and that is the right load-bearing spot. The theory and code appear coherent, and the ISS-RDF agreement is independent supportive evidence, but the numerical evidence for the headline claim cannot distinguish equilibrium RLE sampling from the dynamics of the update schedule. A secondary issue is that the fine-tuning iterations 20 and 40 are arbitrarily chosen; this needs seed-averaged statistics, but the RLE/SCE convergence check is more fundamental. Because this concern is testable and the verdict is already CONDITIONAL, I recommend no change to the reader's verdict.","tokens_in":21549,"tokens_out":9367,"duration_ms":94328,"concrete_test":"At T=4.3, take 100 stored SC-UCG CG configurations, initialize p_i uniformly, and iterate Eq. (12) (or Eqs. (22)-(23) for Bethe) until max_i ||p^(k+1)-p^(k)||_∞ < 1e-6, recording the number of iterations and the converged p. Compare converged p and the implied forces to the one-pass values, and check whether the fixed point is unique. If convergence typically requires more than a few iterations, rerun the T-sweep with SCE-converged updates and see whether the bimodal ee histograms and T_c persist; if they shift or disappear, the phase-transition claim is an artifact of the one-pass update schedule.","verdict_should_be":"UNCHANGED","load_bearing_attack":"SC-UCG's central temperature-transferability demonstration is made at T=4.2-4.6 around the critical point T_c≈4.4, where internal-state fluctuations are the collective slow modes (critical slowing down). The method's RLE premise, stated in Section V, requires internal states to reach local equilibrium within an MD timestep, implemented as exactly one ISC pass per step. The paper provides no convergence test for the self-consistent equation (12) or the Bethe updates (22)-(23), and it explicitly presents multi-pass SCE only as an ideal. Near a second-order transition, the self-consistent iteration can converge slowly; one pass may be far from the fixed point, so the sampled distribution is not the RLE mixed-ensemble Hamiltonian of Eq. (3)/(17). The observation that collective switching is faster than in the AA reference (Section IV.C) is consistent with non-equilibrium driving, not with RLE equilibrium. If the one-pass update schedule generates the bimodal ee distribution while SCE-converged sampling does not, the headline claim fails.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes Self-Consistent Ultra-Coarse-Graining (SC-UCG), an extension of RLE-UCG in which internal state probabilities of CG beads are inferred during MD by message-passing iterations driven directly by the UCG interaction matrix, eliminating hand-selected collective variables for the internal state update. The mixed-ensemble Hamiltonian is extended with a Bethe-Peierls entropy term to capture pairwise internal-state correlations, and a new training loss, Multilayer Internal State Consistency (MISC), is derived from the relative entropy principle for the joint distribution of positions and internal states. The method is tested on a one-site-per-molecule CG model of a chiral tetramer fluid that exhibits a second-order symmetry-breaking transition. The authors report that a model trained on a single subcritical temperature (T=4.3) reproduces collective chiral switching and recapitulates a phase transition across T=4.2-4.6 with a critical temperature near the all-atom value T_c≈4.365.","tokens_in":21758,"tokens_out":9004,"duration_ms":71796,"significance":"The methodological core is attractive: the variational derivation of the self-consistent inference equation (Eq. 12) is clean, the Bethe extension (Eqs. 16-23) is a useful step beyond mean-field RLE, and the MISC loss (Eq. 37) gives a principled, simulation-free training objective that reduces to pseudolikelihood in the one-layer, one-hot limit. The software and data are openly available. However, the demonstration of the flagship claim—temperature transferability and phase-transition reproduction—is conditional on an arbitrary choice of fine-tuning checkpoint and lacks a validation of the rapid-local-equilibrium assumption near criticality. These issues, rather than the derivation, are the main barriers to accepting the paper's stronger conclusions.","major_comments":[{"comment":"The central temperature-transferability demonstration is based on forcefields taken from UCG-REM fine-tuning iterations 20 and 40, which the caption to Fig. 7 describes as 'arbitrarily chosen to represent the first and second cross-over regions.' Since the cross-over regions are identified by inspecting the very phenomenon that the temperature-transferability claim is meant to establish (frequent collective switching), the selection is dangerously close to circular: the phase-transition result may be a property of the selected checkpoint rather than of the SC-UCG/MISC training procedure. Indeed, at Iteration 0 the MISC forcefield with r_c=6 shows no collective chiral transitions (Figs. 5a-b), and for r_c=8 the transitions only appear after one UCG-REM iteration. The authors should either report the temperature-transferability results for the converged MISC forcefield, for the full set of fine-tuning iterations, or for a pre-defined stopping criterion (e.g., fixed number of iterations or convergence of the loss).","section":"IV.C / Fig. 7"},{"comment":"The rapid-local-equilibrium (RLE) premise is not validated in the regime where the method is claimed to work. The implementation performs exactly one ISC pass per MD timestep (Fig. 1b), whereas Eqs. (12) and (22)-(23) define fixed-point equations for the internal state probabilities and edge probabilities. Near the critical point, internal-state fluctuations are the slow collective modes, and a single message-passing pass may be far from the fixed point. The paper provides no comparison of one-pass sampling with multi-pass, converged SCE sampling, and no diagnostic of the residual (e.g., the change in p_i between successive ISC passes). The observation in Section IV.C that collective switching is faster than in the all-atom reference is consistent with a non-equilibrium update schedule rather than with equilibrium RLE. A concrete test would be to run multi-pass SCE-converged SC-UCG at T=4.3 and check that the bimodal distribution of the enantiomer excess persists; if it does not, the reported phase transition is an artifact of the one-pass update.","section":"V / II.A, Eq. (12)"},{"comment":"The statistical evidence for the phase-transition claim is under-powered. The training dataset is 500 frames from a single all-atom trajectory of 25,000 steps for N=1000 molecules. Each SC-UCG temperature appears to be a single simulation (no replicates are described), and the histograms in Figs. 6-8 are presented without error bars or bootstrap estimates. Given the acknowledged sensitivity of near-critical behavior to tiny forcefield changes (Section IV.C), the inferred critical temperature T≈4.4 has no reported uncertainty, and the discrepancies shown in Fig. 8 (too mild low-temperature extrapolation and spurious high-temperature mixed-state peak) are not assessed quantitatively. The authors should report uncertainty estimates or at least multiple independent runs to support the transferability claim.","section":"III.A / Figs. 6-8"}],"minor_comments":[{"comment":"The definition of J_ij as U_ij(0,0)+U_ij(1,1)-U_ij(0,1)-U_ij(0,0) repeats the first term; the last term should be U_ij(1,0) (or U_ij(0,1) with appropriate bookkeeping). As written, the coupling reduces to U(1,1)-U(0,1), which is not the standard Ising coupling.","section":"Eq. (20)"},{"comment":"The text refers to simulations of '2×10^w steps' and '2×10^w timesteps'; the exponent appears to be garbled (presumably 10^5). Please correct these instances.","section":"III.B, IV.C"},{"comment":"The interaction potential contains a garbled fragment '¬T𝑟!=%+X−𝑟!=%uV'; the intended Lennard-Jones-like form should be typeset correctly.","section":"Eq. (42)"},{"comment":"The assertion that one ISC layer per MD step 'falls within the fast-mixing RLE category' (paragraph after Eq. 13) would benefit from a quantitative justification or a reference to a timescale separation analysis.","section":"II.A"},{"comment":"The slow-mixing limit parameter N_G is mentioned but never defined; please clarify its meaning and how it is chosen.","section":"Fig. 1c"},{"comment":"The notion of a 'cross-over region' is used to select checkpoints; a quantitative definition (e.g., based on the frequency of ee sign changes or the width of the bimodal peak) would make the selection less ad hoc.","section":"IV.B"}],"recommendation":"major_revision","confidential_remarks":"To the editor: The paper is likely to be of interest to the coarse-graining community, and the methodological derivation appears sound. My major concern is that the headline numerical claim is built on an arbitrarily selected fine-tuning checkpoint and an unvalidated RLE assumption near criticality; both are fixable with additional analysis but currently make the stronger conclusions conditional. The paper would also benefit from tempering its abstract: 'recapitulates the phase transition across temperatures' is not yet demonstrated for a systematically chosen model."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Take a look at this one if you care about coarse-grained models with internal states. The method is a real step forward: it removes the hand-designed collective variables from RLE-UCG, adds pair correlations via a Bethe approximation, and introduces a training loss (MISC) that avoids iterative sampling of intermediate forcefields. The variational derivation is clean, the message-passing interpretation is natural, and the code and data are publicly available. The chiral tetramer application is a smart stress test.\n\nWhat is actually new: the self-consistent ISP update via graph message passing, the Bethe-enhanced RLE Hamiltonian with its edge update, and the MISC loss derived from relative entropy on the joint position/state distribution. These are not just rewordings of existing UCG theory; they solve real problems in the earlier framework. The paper is also unusually honest about its limitations—it flags the single-temperature training, the mean-field family's shortcomings near criticality, and the need for future multi-temperature training.\n\nThe soft spots are mostly in the numerical demonstration. The central claim—that SC-UCG recapitulates the phase transition across temperatures from single-temperature training—rests on one ISC pass per MD timestep. That is exactly the RLE premise, and near T_c the internal states are the slow collective modes. The paper never shows SCE-converged sampling to confirm that the one-pass schedule gives the same physics as the self-consistent fixed point. If the bimodal ee distribution is an artifact of the update schedule rather than equilibrium sampling, the headline claim fails. That is a real gap, not a quibble. Also, the selection of fine-tuning iterations 20 and 40 as representative \"cross-over\" points is arbitrary, and there are no error bars or repeated runs, so the phase-transition temperature is essentially read off a single trajectory per condition. The agreement with the all-atom data is qualitative, and the paper admits the low-T and high-T deviations.\n\nNone of this kills the paper. The method itself is a solid contribution to the coarse-graining toolkit, and the open questions are addressable with a few additional simulations: multi-pass SCE convergence checks, repeated runs with error bars, and a sensitivity analysis for the fine-tuning iteration choice. The derivation and framework are the paywall; the demonstration is suggestive rather than conclusive.\n\nWorth taking seriously. Send it to peer review. I would want the convergence question addressed before leaning on the phase-transition result, but the method deserves referee time. If I were working in this area, I would cite it.","headline":"SC-UCG is a genuinely useful extension of UCG—removing hand-designed CVs, adding Bethe correlations, and a training loss that avoids iterative sampling—but the phase-transition demonstration leans on an unvalidated one-pass RLE schedule near criticality.","tokens_in":22295,"tokens_out":2022,"would_cite":true,"duration_ms":21907,"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":"SC-UCG infers correlated internal states by message passing and reproduces a chiral-racemic phase transition from a single-temperature training set.","keywords":["ultra-coarse-graining","internal states","message passing","Bethe approximation","relative entropy minimization","chiral symmetry breaking","phase transition","coarse-grained molecular dynamics"],"falsifier":"Run the same chiral tetramer system with an SC-UCG variant that iterates the message-passing self-consistent equation to full convergence inside each MD step at $T=4.3$. If the bimodal enantiomer-excess distribution, the collective switching events, and the temperature-driven transition near $T_c \\approx 4.4$ disappear or shift substantially, then the one-pass rapid-local-equilibrium choice is carrying the reported result.","tokens_in":21319,"feed_emoji":"⚛️","tokens_out":7394,"duration_ms":64426,"temperature":0.7,"pith_summary":"Ultra-Coarse-Graining (UCG) lets a coarse-grained simulation carry discrete 'internal states' on each bead, but prior versions either switched those states slowly by Monte Carlo or assigned them through hand-chosen collective variables, which breaks symmetry and ignores correlations between neighboring beads. This paper claims that the internal-state probabilities can instead be inferred directly from the UCG forcefield itself: a variational minimization of the mixed free energy yields a self-consistent message-passing equation whose weights are the UCG interaction matrices, so no collective variables are needed in the coarse-grained ensemble. The authors add a Bethe approximation to the rapid-local-equilibrium Hamiltonian, giving explicit pairwise state correlations, and derive a training loss, Multilayer Internal State Consistency (MISC), from relative entropy minimization that avoids iterative resampling of intermediate forcefields. In a chiral tetramer fluid with a second-order symmetry-breaking phase transition, the method reproduces collective switching between D-rich and L-rich states and recapitulates the phase transition over a range of temperatures even though it was trained only on a single subcritical temperature.","feed_headline":"One-temperature training reproduces chiral phase transition","feed_subtitle":"Self-consistent internal states replace hand-made collective variables and capture cooperative switching.","key_machinery":"The key object is the self-consistent internal-state update: a temperature-controlled softmax equation of the form $\\mathbf{p}_i = \\sigma_\\beta(\\mathbf{h}_i + \\sum_{j \\in \\mathcal{N}(i)} \\mathbf{U}_{ij}\\mathbf{p}_j)$, which is at once a variational stationarity condition for the UCG free energy and an Internal State Convolution (ISC) message-passing layer in a graph neural network. The Bethe variant substitutes the Bethe entropy for the mean-field entropy, introducing an edge-level update through a closed-form pair-correlation function $\\xi_{ij}^*$ that couples the two-site probability matrix to the single-site probabilities. The training side is Multilayer Internal State Consistency (MISC), a cross-entropy loss over the outputs of successive ISC layers, derived from relative entropy minimization over the joint distribution of positions and internal states, which allows forcefield optimization without iterative sampling of intermediate forcefields.","core_discovery":"The central claim is that a bottom-up coarse-grained model can learn and simulate correlated discrete internal states without any user-designed collective variable. Minimizing the UCG free energy with respect to the single-site internal state probabilities gives a self-consistent equation in softmax form, $\\mathbf{p}_i = \\sigma_\\beta(\\mathbf{h}_i + \\sum_{j\\in\\mathcal{N}(i)}\\mathbf{U}_{ij}\\mathbf{p}_j)$, which the authors recognize as one layer of message-passing on the molecular graph, with the forcefield as the trainable weights. Replacing the mean-field entropy with the Bethe entropy adds explicit pair correlations through a closed-form edge update. Trained by the MISC loss on a single subcritical dataset at $T=4.3$, the resulting model shows bimodal enantiomer-excess distributions, collective chiral switching events, and a symmetry-breaking transition around $T_c \\approx 4.4$ when the temperature is changed, matching the all-atom critical behavior qualitatively.","pith_inferences":["A direct stress test of the method's premise: run SC-UCG at $T=4.3$ with multiple ISC passes per timestep until the self-consistent equation converges, and compare the apparent critical temperature to the one-pass result; the size and direction of the shift would show whether rapid local equilibrium or over-smoothing controls the transition.","Because the MISC loss with one-hot labels coincides with pseudo-likelihood maximization for an inverse Ising model, the learned UCG couplings are effectively inferred interaction strengths; this suggests the method could be used to extract effective cooperativity from experimental or simulation state-label data, not just to generate dynamics.","The reported temperature transferability is limited in the low- and high-temperature tails; coupling the MISC loss with multi-temperature training appears as the natural next step to get full transferability, as the authors note."],"forward_implications":["UCG models no longer need hand-designed collective variables for internal states, so symmetric forcefields stay symmetric unless the training data itself breaks the symmetry.","Pairwise correlations between internal states of neighboring beads enter through the Bethe-Peierls update, which improves internal-state-specific radial distribution functions and lets collective fluctuations appear.","A forcefield trained at one subcritical temperature can generate temperature-dependent behavior, crossing from the racemic fluid into chiral phases near $T_c \\approx 4.4$.","The preferred training strategy, pre-training with ISC/MISC loss followed by UCG-REM fine-tuning, is stable and largely insensitive to the choice of cutoff radius, layer count, and other hyperparameters.","The same machinery is anticipated to extend to correlated internal-state phenomena such as lipid bilayer ripple phases and proton transport."],"supporting_citations":[{"why":"Establishes the UCG theory of discrete internal states as auxiliary variables, which SC-UCG builds on.","marker":"[23]"},{"why":"Implements MC-UCG with Monte-Carlo switching of internal states, the slow-transition limit SC-UCG unifies.","marker":"[24]"},{"why":"Defines RLE-UCG with internal state probabilities conditioned on collective variables; the mean-field limit SC-UCG replaces.","marker":"[25]"},{"why":"Introduces Internal State Regression and entropy-based UCG training, the baseline forcefield training and ISP dependence SC-UCG overcomes.","marker":"[26]"},{"why":"Provides the chiral molecular model, the all-atom reference data, and the critical-temperature benchmark used to test SC-UCG.","marker":"[41]"},{"why":"Introduces the tetramer fluid model with chiral-racemic phase transition, the source of the mapped UCG system.","marker":"[40]"},{"why":"Supplies the relative entropy minimization principle from which MISC training is derived.","marker":"[13]"},{"why":"Supplies the Bethe free energy formalism used to add pairwise correlations to the RLE Hamiltonian.","marker":"[47]"}],"fun_headline_variants":["Self-consistent internal states replace hand-made collective variables","One-temperature training captures chiral phase transition","Message-passing forcefield learns internal states without CVs","Bethe-corrected coarse-graining predicts phase transition from single T","Graph message passing replaces user CVs for coarse-graining"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the rapid-local-equilibrium assumption: one message-passing pass per MD timestep is enough for the internal states to be effectively equilibrated with their neighbors. If internal-state dynamics are slow, or the self-consistent fixed point needs many iterations, the sampled state distributions and the predicted phase transition would be biased.","fun_headline_variants_meta":{"raw":{"variants":["Self-consistent internal states replace hand-made collective variables","One-temperature training captures chiral phase transition","Message-passing forcefield learns internal states without CVs","Bethe-corrected coarse-graining predicts phase transition from single T","Graph message passing replaces user CVs for coarse-graining"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00054,"raw_usage":{"total_tokens":2622,"prompt_tokens":1012,"completion_tokens":1610,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":628,"completion_tokens_details":{"reasoning_tokens":1531}},"tokens_in":628,"tokens_out":1610,"duration_ms":10679,"temperature":1.0,"reasoning_tokens":1531,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T14:05:57.824513+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same chiral tetramer system with an SC-UCG variant that iterates the message-passing self-consistent equation to full convergence inside each MD step at $T=4.3$. If the bimodal enantiomer-excess distribution, the collective switching events, and the temperature-driven transition near $T_c \\approx 4.4$ disappear or shift substantially, then the one-pass rapid-local-equilibrium choice is carrying the reported result.","supporting_citations":[],"review_version":1}