{"id":"896e4cba-ce89-4883-a5f5-d3258d583547","arxiv_id":"2607.05266","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":10,"one_line_summary":"MXtalGFlow combines a canonical crystal parameterization with energy-based GFlowNet training to sample thermodynamic distributions of molecular crystals, recovering known polymorphs and predicting new competitive packing modes.","lead":"The paper builds a GFlowNet-based framework (MXtalGFlow) that samples a Boltzmann distribution over molecular crystal structures, rather than just producing ranked lists of candidates. It demonstrates the approach on two molecules, recovering known polymorphs and identifying unreported packing modes with competitive probabilities.","discovery_kind":"unclear","skeptic_critique":{"model":"glm-5.2","headline":"Basin thermodynamic rankings rely on un-reweighted sample densities from an imperfectly converged model, with no error bars or sensitivity analysis on the heuristic KDE parameters that determine them.","rationale":"The reader correctly identified the load-bearing concern: imperfect convergence combined with heuristic analysis choices and no error bars undermines the specific thermodynamic predictions. My analysis confirms this and sharpens it in two ways. First, the paper explicitly chooses not to reweight samples despite acknowledging systematic bias — this is not just an implicit assumption but a deliberate methodological choice that directly affects the basin comparisons. Second, the use of peak density P(x*) rather than integrated basin probability adds a further layer of unreliability: even with perfect convergence, peak density is not the same as basin thermodynamic weight unless basin shapes are comparable, which is not verified. The reader's verdict of CONDITIONAL is appropriate. The framework itself — canonical parameterization, energy-based GFlowNet training, distributional analysis — is genuinely novel and theoretically sound. The GFlowNet TB loss does converge to the Boltzmann distribution in principle (Eq. 2), and the training protocol (three-phase with mode coverage guarantees) is a reasonable engineering approach. The parameterization with Jacobian corrections is properly derived. These are real contributions. But the specific quantitative claims about thermodynamic preferences of new packing modes are not yet supported to the level the paper claims. The paper is transparent about its limitations (acknowledging convergence issues, omitting reweighting, noting these are provisional), which is commendable, but the headline claims in the abstract and conclusions still present the basin preferences as findings rather than hypotheses. No code or data release is referenced, which further limits independent verification. The verdict should remain CONDITIONAL: the framework is promising and the methodology is sound in principle, but the specific thermodynamic predictions require convergence improvement, importance reweighting, sensitivity analysis, and ideally experimental confirmation before they can be accepted.","tokens_in":22034,"tokens_out":2558,"duration_ms":73928,"concrete_test":"Recompute the basin probability rankings in Figures 5g and 6f using importance-reweighted samples, where each sample x_i is weighted by the TB residual w_i = R(x_i)·p_B(τ_i)/p_F(τ_i) (the per-sample estimate of the convergence error). Then recompute P(x) with these weights and compare the basin rankings. Additionally, vary d_cut (e.g., 10% and 20% quantiles) and σ (e.g., d_cut/4 and d_cut/2) to test sensitivity. If the basin rankings flip or become statistically indistinguishable under reweighting or parameter variation, the thermodynamic preference claims do not hold.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim — that certain basins are 'thermodynamically preferred' over known experimental polymorphs (Section 2.6, panel g) — rests on comparing peak probability densities P(x*) across basins. P(x) is a KDE over raw model samples (Eq. 4), computed without importance reweighting, despite the paper's own acknowledgment that 'GFN models of this type tend to over-weight higher energy states systematically' (Section 2.5) and that TB parity plots show visible deviation from the diagonal (Figure S7, with Pearson correlations reported only on the top 99% of samples). The paper explicitly states: 'We omit manual sample reweighting [9] and, instead, present the learned distributions as they are.' This means the basin probability comparisons directly inherit the model's convergence errors. Two compounding issues arise: (1) If the model over-weights certain regions of configuration space, the relative P(x*) values between basins are biased in ways that could flip marginal rankings — and the language 'slightly thermodynamically preferred' suggests the differences are indeed marginal. (2) P(x*) is the peak density, not the integrated basin probability ∫_basin p(x)dx; basins with different shapes or widths could have high peak density but low total thermodynamic weight. The paper does not report integrated basin probabilities, error bars, or sensitivity to the heuristic choices (σ = d_cut/3, d_cut at 15% quantile, d_step scanning, 15kBT energy cutoff). Without these, the specific prediction that new packing modes are thermodynamically preferred over known polymorphs is not yet reliable.","agreement_with_reader":"agree"},"referee_report":{"model":"glm-5.2","summary":"The manuscript presents MXtalGFlow, a framework for generative modeling of molecular crystal structures that combines a canonical crystal parameterization with energy-based GFlowNet training. The central claim is that the sampled distribution approximates the Boltzmann distribution over crystal structures, enabling thermodynamically meaningful comparison of structural basins. The authors demonstrate the approach on two molecular systems (MIPCAS and NEHZOR) under two energy functions (Lennard-Jones and UMA), recovering known experimental polymorphs and identifying new competitive packing modes. The framework rests on three pillars: a physically grounded parameterization, a three-phase GFlowNet training protocol, and distributional analysis via RDF-EMD-based probability density estimation.","tokens_in":22371,"tokens_out":1110,"duration_ms":167743,"significance":"The conceptual contribution is timely and well-motivated: framing molecular crystal structure prediction as distributional sampling rather than enumeration of disconnected candidates is a meaningful shift. The canonical parameterization (Section 4.1) is a solid building block, and the three-phase training protocol (Section 4.5) addresses real practical difficulties in GFlowNet convergence on rugged landscapes. The comparison of crystal distributions under different energy functions (Section 2.5) is illustrative. However, the quantitative basin thermodynamic predictions that constitute the most novel claims rest on un-reweighted sample densities from imperfectly converged models, which limits the reliability of the central results as currently presented.","major_comments":[{"comment":"Section 2.6, panel (g) discussion and Figure 6 panel (f): The claim that 'basin 1 is slightly thermodynamically preferred over the known experimental structure in basin 2' (for MIPCAS) and that 'basins 1 and 2 are thermodynamically preferred compared to the known polymorphs' (for NEHZOR) rests on comparing peak probability densities P(x*) across basins. However, P(x) as defined in Eq. (4) is a KDE over raw model samples without importance reweighting, and the paper explicitly acknowledges that 'GFN models of this type tend to over-weight higher energy states systematically' (Section 2.5). The paper states: 'We omit manual sample reweighting and, instead, present the learned distributions as they are.' This means the basin probability comparisons directly inherit the model's convergence errors. Since the differences are described as 'slightly' preferred, even modest systematic bias could翻","section":null},{"comment":"Section 2.3, Eq. (4): P(x*) is the peak density, not the integrated basin probability. Basins with different shapes or widths could have high peak density but low total thermodynamic weight. The paper does not report integrated basin probabilities, which would be the more appropriate thermodynamic quantity. This is a load-bearing distinction for the claim that certain basins are 'thermodynamically preferred.'","section":null},{"comment":"Section 2.3, Eq. (4) and surrounding text: The KDE parameters (sigma = d_cut/3, d_cut at 15% quantile) and the basin assignment threshold (0.8) are heuristic choices. No sensitivity analysis is provided to show that the basin probability rankings are robust to these choices. Given that the thermodynamic preference claims are described as marginal ('slightly preferred'), the absence of error bars or sensitivity analysis on these parameters undermines the reliability of the quantitative predictions.","section":null},{"comment":"Figure S7 and Section S7: The TB parity plots show visible deviation from the diagonal, and Pearson correlations are reported only on the top 99% of samples by reward. The paper should more transparently report the full convergence metrics (including the full-sample correlations) and discuss quantitatively how the residual convergence error propagates into the basin probability estimates.","section":null}],"minor_comments":[{"comment":"Section 2.1: The phrase 'specifically specifically' appears to be a typo in the first paragraph of Section 2.2.","section":null},{"comment":"Section 2.3: The definition of d_step and the scanning procedure over increasing d_step could be described more precisely — it is unclear what 'stable plateau' means quantitatively.","section":null},{"comment":"Figure 5, panel (g): The table would benefit from explicit numerical values for the integrated basin probabilities (or at least basin sample counts) in addition to peak densities.","section":null},{"comment":"Section 4.4: The LJ stiffness factor k=2.5 is stated without justification; a brief comment on sensitivity to this choice would help.","section":null},{"comment":"The paper would benefit from a clearer statement upfront (e.g., in the abstract or introduction) that the thermodynamic preference claims are conditional on the GFlowNet having converged, and that this convergence is imperfect.","section":null}],"recommendation":"major_revision","confidential_remarks":"The reader's report and stress-test note correctly identify the core weakness: the basin thermodynamic rankings depend on un-reweighted densities from an imperfectly converged model. I agree this is a load-bearing issue. The paper's framework is conceptually sound and the engineering is substantial, but the authors need to either (a) implement importance reweighting, (b) provide error bars and sensitivity analysis on the KDE parameters, or (c) substantially soften the thermodynamic preference claims. Option (c) alone may suffice for minor revision if the authors reframe the basin comparisons as qualitative rather than quantitative thermodynamic predictions. The integrated vs peak density issue (major comment 2) is also important and should be addressed."},"author_rebuttal":{"model":"glm-5.2","summary":"We thank the referee for a careful and constructive reading of our manuscript. The referee correctly identifies that the most novel quantitative claims—basin thermodynamic preferences—rest on analysis choices and convergence levels that warrant greater scrutiny than we provided. We address each comment below and outline concrete revisions.","responses":[{"response":"The referee is correct that our basin probability comparisons inherit the model's convergence errors, and that the differences we report for MIPCAS are described as marginal. We acknowledge that presenting these as thermodynamic predictions without error quantification or reweighting overstates the reliability of the result. We will revise the manuscript in two ways. First, we will implement importance reweighting of samples using the trajectory balance residual as an importance weight estimator (following the approach in the Boltzmann Generator literature, Ref. [9]), and re-evaluate the basin probability rankings under reweighting. If the marginal preference for basin 1 over basin 2 in MIPCAS does not survive reweighting, we will report that explicitly and soften the claim accordingly. Second, we will reframe the language throughout: rather than stating that basins are 'thermodynamically preferred,' we will state that the sampled distribution suggests competitive probability density, subject to the convergence limitations we document. The NEHZOR case, where basins 1 and 2 have substantially higher probability densities than the polymorph basins, is more robust to modest bias, but we will still apply reweighting and report both raw and reweighted results.","revision_made":"yes","referee_comment":"Basin probability comparisons use un-reweighted sample densities from imperfectly converged models; marginal differences ('slightly preferred') could be flipped by systematic bias."},{"response":"This is a valid and important distinction. We used peak density P(x*) as a computationally convenient proxy, but the referee is right that integrated basin probability is the more appropriate thermodynamic quantity. We will add integrated basin probabilities to the summary tables in Figures 5(g) and 6(f), computed by summing the normalized P(x) over all samples assigned to each basin (using the Gaussian basin membership weights already defined in Section 2.3). We will report both peak density and integrated probability, and we will base our thermodynamic preference claims on the integrated quantities. If the rankings change when using integrated probabilities rather than peak densities, we will report that transparently.","revision_made":"yes","referee_comment":"P(x*) is peak density, not integrated basin probability; basins with different shapes could have high peak but low total thermodynamic weight."},{"response":"We agree that a sensitivity analysis is needed, particularly given that the MIPCAS preference is described as marginal. We will add a sensitivity analysis varying d_cut (at the 10%, 15%, and 20% quantiles), sigma (at d_cut/2, d_cut/3, and d_cut/4), and the basin assignment threshold (at 0.7, 0.8, and 0.9). We will report how the basin probability rankings and integrated basin probabilities change across these parameter choices. This will be included as a new supplementary figure and discussed in the main text. If the marginal MIPCAS result is not robust to these variations, we will state so explicitly.","revision_made":"yes","referee_comment":"KDE parameters (sigma, d_cut) and basin assignment threshold (0.8) are heuristic; no sensitivity analysis provided."},{"response":"The referee is right that reporting convergence metrics only on the top 99% of samples by reward gives an incomplete picture. We will add the full-sample Pearson correlations to Figure S7 and the accompanying text, alongside the top-99% values, so that readers can assess the degree of residual convergence error. We will also add a quantitative discussion of how the TB residual (the deviation from the diagonal) propagates into basin probability estimates. Specifically, we will estimate the effective error on basin probability ratios by propagating the per-sample TB residuals through the KDE-based probability density, giving approximate error bars on the integrated basin probabilities. These error bars will be included in the revised summary tables.","revision_made":"yes","referee_comment":"TB parity plots show deviation from diagonal; correlations reported only on top 99% of samples; should report full-sample metrics and discuss error propagation."}],"tokens_in":21743,"tokens_out":1193,"duration_ms":77820,"standing_objections":[]},"desk_editor":{"model":"glm-5.2","letter":"The main thing to know: this paper introduces a real framework — canonical asymmetric-unit parameterization + energy-based GFlowNet training + RDF-EMD distributional analysis — that is genuinely new for molecular crystal CSP. The concept of moving from ranked lists to thermodynamic distributions is the right framing, and the execution is mostly sound. The specific quantitative predictions (basin 1 is preferred over the known polymorph) are not yet trustworthy, for reasons the paper partly acknowledges but does not fully grapple with.","headline":"MXtalGFlow: a genuinely new framework for distributional molecular crystal modeling, but the specific thermodynamic basin rankings are not yet reliable","tokens_in":22857,"tokens_out":808,"would_cite":false,"duration_ms":17381,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"glm-5.2","headline":"Generative model samples Boltzmann distribution of molecular crystals","keywords":[],"falsifier":"Train the GFlowNet to convergence on a system where the free energy ranking of polymorphs is known from independent methods (e.g., quasi-harmonic vibrational free energy calculations). If the basin probabilities from MXtalGFlow disagree with the known thermodynamic ranking, or if the predicted new preferred packing modes are destabilized upon DFT-level refinement, the distributional claims about thermodynamic preference are not reliable.","tokens_in":22264,"feed_emoji":"🔬","tokens_out":1269,"duration_ms":130825,"temperature":0.7,"pith_summary":"The paper introduces MXtalGFlow, a framework that pairs a canonical 12-dimensional parameterization of molecular crystal structures with energy-based GFlowNet training to sample an approximation of the Boltzmann distribution over crystal packings for a given molecule and space group. The central claim is that, unlike standard crystal structure prediction which produces ranked lists of isolated low-energy candidates, this approach yields a physically interpretable probability distribution over the entire crystal landscape, where the probability mass assigned to each structural basin reflects its thermodynamic weight. The paper demonstrates the framework on two molecules (MIPCAS and NEHZOR) under two energy functions (Lennard-Jones and the Universal Model for Atoms), recovering known experimental polymorphs within high-probability basins and identifying previously unreported packing modes that, under the UMA potential, carry competitive or higher basin probabilities than the experimentally observed structures. The load-bearing mechanism is the combination of a physically grounded parameterization (eliminating unit-cell degeneracy and mapping to a bounded latent space with Jacobian corrections) with a three-stage GFlowNet training protocol that separates mode discovery from distributional equilibration, enabling the model to approximate Boltzmann weights across a rugged, high-dimensional energy surface.","feed_headline":"Generative model samples Boltzmann distribution of molecular crystals","feed_subtitle":"A GFlowNet trained on a canonical crystal parameterization recovers known polymorphs and predicts new competitive packing modes from thermod","key_machinery":"The framework rests on three components: (1) a canonical 12-parameter crystal representation (box vectors, centroid position, rotation vector) reduced to the asymmetric unit and mapped to a bounded latent space with Jacobian corrections for volume and orientation; (2) a diffusion-sampler GFlowNet trained via a three-stage protocol (MLE support expansion, backward TB thermalization on a noised prior, forward-backward TB global equilibration) using a Huber-type trajectory balance loss; (3) a distributional analysis pipeline using radial distribution function earth mover's distance (RDF EMD) to define a kernel density estimate P(x), hill-climbing to find probability maxima, and Gaussian basin-7","core_discovery":"By training a diffusion-sampler GFlowNet on a canonical crystal parameterization with Boltzmann rewards, the authors produce a sampled distribution over molecular crystal structures where known polymorphs are recovered as basin maxima and previously unreported packing modes are identified with basin probabilities competitive with or exceeding the experimental structures, as observed for both MIPCAS (P-1) and NEHZOR (P21/c) under the UMA potential. The distributional view reveals that basins with higher minimum energies can still carry higher thermodynamic probability than lower-energy basins due to a larger configurational density of states, a distinction invisible to standard ranked-list晶体C","pith_inferences":["The claim that basin 1 of MIPCAS is thermodynamically preferred over the experimental structure in basin 2 depends critically on the GFlowNet having converged sufficiently to approximate Boltzmann weights; the paper itself acknowledges systematic over-weighting of high-energy states and reports Pearson correlations only on the top 99% of samples, suggesting the quantitative basin rankings should b","The kernel density bandwidth (sigma = d_cut/3, d_cut at 15% quantile) and basin assignment threshold (normalized weight > 0.8) are heuristic choices that could affect which packing modes are identified and how probability mass is partitioned; sensitivity analysis on these parameters would strengthen the robustness of the basin rankings.","If the framework is extended to flexible molecules (Z' > 1 or multiple conformers), the dimensionality increase may substantially worsen GFlowNet convergence, and the separation of mode discovery from equilibration may need to be revisited.","The comparison between LJ and UMA distributions being nearly disjoint for MIPCAS raises the question of whether the framework could be used to benchmark energy functions themselves: if an experimental polymorph falls in a low-probability region under a given potential, that potential may be missing key physics for that system."],"forward_implications":["The distributional framework could replace ranked-list CSP outputs with thermodynamic probability maps, enabling direct comparison of basin weights under different energy functions and identification of packing modes that are thermodynamically competitive but not energy-minimal.","Swapping energy functions (LJ vs UMA vs future machine-learned potentials) becomes a tool for understanding how specific physical interactions (hydrogen bonding, dispersion, sterics) reshape the crystal landscape, rather than merely re-ranking structures.","The three-stage training protocol (support expansion, prior thermalization, global equilibration) may transfer to other rugged, high-dimensional Boltzmann sampling problems beyond molecular crystals.","Conditional generalization to multiple molecules and space groups in a single model, combined with flexible intramolecular degrees of freedom, could yield an efficient lead generator and probability density estimator across broad chemical spaces.","The separation of mode discovery (classical optimization) from distributional equilibration (GFlowNet) acknowledges the complementary strengths of each approach and suggests hybrid pipelines as a practical path forward."],"fun_headline_variants":["GFlowNet samples thermodynamic distribution of molecular crystal polymorphs","Crystal GFlowNet recovers known polymorphs and predicts competitive packing modes","Distributional crystal model finds unreported packings competitive with known polymorphs","Generative flow network learns Boltzmann distributions over molecular crystal packings","Thermodynamic view of molecular crystals identifies unreported packing basins"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The quantitative basin probability rankings that predict new thermodynamically preferred packing modes depend on the GFlowNet having converged closely enough to the true Boltzmann distribution, but the paper acknowledges systematic over-weighting of high-energy states and reports convergence metrics only on the top 99% of samples, without error bars or sensitivity analysis on the basin probability comparisons.","fun_headline_variants_meta":{"raw":{"variants":["GFlowNet samples thermodynamic distribution of molecular crystal polymorphs","Crystal GFlowNet recovers known polymorphs and predicts competitive packing modes","Distributional crystal model finds unreported packings competitive with known polymorphs","Generative flow network learns Boltzmann distributions over molecular crystal packings","Thermodynamic view of molecular crystals identifies unreported packing basins","Higher-energy basins can carry higher thermodynamic probability than low-energy basins","MXtalGFlow samples Boltzmann distribution of molecular crystal structures","Treating molecular crystal polymorphs as a distribution finds unreported packing modes","Generative crystal model treats polymorphs as a distribution, not a ranked list","Boltzmann-trained GFlowNet maps packing basins around known molecular polymorphs"]},"model":"glm-5.2","effort":"high","cost_usd":0.0,"raw_usage":{"total_tokens":4666,"prompt_tokens":595,"completion_tokens":4071,"prompt_tokens_details":null},"tokens_in":595,"tokens_out":4071,"duration_ms":58879,"temperature":1.0,"reasoning_tokens":3382,"cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-07T20:33:46.963549+00:00","model_set":{"reader":"glm-5.2"},"falsifier":"Train the GFlowNet to convergence on a system where the free energy ranking of polymorphs is known from independent methods (e.g., quasi-harmonic vibrational free energy calculations). If the basin probabilities from MXtalGFlow disagree with the known thermodynamic ranking, or if the predicted new preferred packing modes are destabilized upon DFT-level refinement, the distributional claims about thermodynamic preference are not reliable.","supporting_citations":[],"review_version":1}