{"id":"d612b6de-51c1-49a6-917b-3c8aaac0b895","arxiv_id":"2607.22090","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"A new algorithm computes an 'irreducible' oblique supercell size for cubic CSL grain boundaries and uses low values of that size to predict structurally special boundaries, tested on aluminum mixed boundaries.","lead":"This paper proposes an algorithm that finds a smaller, oblique (monoclinic) supercell for any cubic grain-boundary character, and uses that cell size as a pre-simulation score to flag 'special' boundaries. A materials-simulation reader would care because the method promises to cut computational cost in high-throughput grain-boundary studies and to prioritize which boundaries deserve expensive atomistic calculations.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'irreducible' supercell size is not proven: Eq. 4's finite N trial search has no completeness bound, so S_IR may not be minimal and particularity rankings built on it are ungrounded.","rationale":"The reader's weakest assumption is exactly the unproven completeness of the finite N search. I agree that this is the most load-bearing issue: the paper repeatedly uses 'irreducible' for the supercell and builds the structural-particularity ranking on S_IR, so if the search is not exhaustive, the central quantitative claim collapses, independent of any later fitting or simulation. The separate error in Eq. (6) (cos instead of sin for a parallelogram area) is also concrete and damaging, but it is a formula-level defect that could be corrected; the missing completeness proof is fundamental to the algorithm's stated capability. I considered whether to recommend REJECT independently, but since the reader's verdict is already REJECT and my concern supports it, no adjustment is needed. The paper has a plausible heuristic and some suggestive qualitative energy/mobility correlations, but the central 'irreducible' premise remains unverified, and the released data/code are not provided, so the cautious verdict is unchanged.","tokens_in":11754,"tokens_out":3673,"duration_ms":42277,"concrete_test":"For a sample of CSL GB characters (including the Table A1 Sigma5 and Sigma9 mixed sets), compute the true minimal 2D supercell area by exhaustive enumeration of all in-plane integer vector pairs up to a rigorous bound (e.g., all (alpha,beta) with lattice length below a cutoff derived from the CSL basis, or via Smith normal form / 2D lattice reduction). Compare the reported S_IR values and the first/second pairs from Eq. (4) with these exact minima. If any reported S_IR exceeds the true minimal area, or the selected pairs differ from the primitive basis, the finite-N search is not exhaustive and the 'irreducible' claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative output is S_IR, defined in Eq. (6) and used to rank GBs by 'structural particularity.' The algorithm selects {w,v} as the first and second shortest candidate pairs among N integer trials in Eq. (4), then labels them 'irreducible' via constraint (III), Eq. (3.3). But the text gives no bound on N, no exhaustive enumeration of the in-plane lattice, and no proof that the first two nonparallel vectors obtained this way are the true primitive basis of the coincident-site 2D lattice. For a high-index GB, the true second Minkowski vector or primitive basis may correspond to a larger (alpha,beta) combination that lies outside the finite trial set. Without such a guarantee, S_IR is only a heuristic upper bound, not the irreducible size; the claimed quantitative prediction of 'special' GBs from low S_IR is not founded. This is the load-bearing gap in the paper's central argument.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an automated algorithm to construct a so-called irreducible macroscopic representation of CSL grain-boundary characters in cubic lattices. The algorithm searches for short in-plane lattice vector pairs by integer linear combinations of the two grain orientations, defines a 2D 'irreducible supercell size' S_IR, and claims this parameter can pre-screen which GB characters are structurally particular, i.e., likely to sit at energy cusps or mobility transitions. The method is applied to 14 GBs from the Homer dataset for supercell-size and energy comparisons, and to two families of aluminum mixed GBs (Σ5 <001> and Σ9 <011>) to test correlations between S_IR and GB energy/mobility. The paper concludes that low S_IR identifies special GBs and presents a weighted sampling strategy for high-throughput GB exploration.","tokens_in":12036,"tokens_out":4698,"duration_ms":55064,"significance":"If correct, the algorithm would be a cheap geometric pre-screening tool that reduces supercell sizes and prioritizes potentially special GBs before expensive atomistic structure searches, which is a useful contribution to GB engineering and high-throughput simulation. The comparison with the Homer dataset and the independent LAMMPS energy calculations are credible in spirit, and the idea of using a purely geometric size parameter as a particularity indicator is attractive. However, the central quantitative definition of S_IR contains an apparent area-formula error, and the claim that the finite search yields a true irreducible representation is not supported by any completeness proof. The validation is also weakened by the fact that the same S_IR criterion selects the GBs that are then used to demonstrate the criterion's predictive power. These issues are load-bearing because the paper's main claims—that the method finds the irreducible representation and that S_IR predicts structural particularity—rest on them.","major_comments":[{"comment":"The area of a 2D parallelogram with side lengths L_w and L_v and included angle θ is L_w L_v sin(θ), not L_w L_v cos(θ). As written, Eq. (6) gives zero for perpendicular w and v, which are the most common orthogonal supercells and are used elsewhere in the paper. The nonzero entries in Table A1 and the S_IR curves in Figure 3 cannot therefore be the values defined by Eq. (6). This is not a typo of merely cosmetic importance: S_IR is the central parameter used for all rankings and for the structural-particularity predictions. The manuscript must correct this definition and verify that the numerical results, Table A1, and Figure 3 are recomputed consistently.","section":"Algorithm procedure, Eq. (6)"},{"comment":"The term 'irreducible' is used repeatedly, but no proof is given that the first and second shortest vector pairs obtained from the finite N-trial search in Eq. (4) form a primitive basis of the coincident-site 2D lattice. The text does not state how N is chosen, how the integer pairs (α_m, β_m) are enumerated, or why the two shortest nonparallel candidate pairs from this finite set must coincide with the true reduced basis for every CSL character. Without a completeness bound or a proper lattice-reduction argument, S_IR is at best a heuristic upper bound on the minimal 2D cell area, and the particularity rankings built on it are not grounded. This is the key theoretical gap in the paper's central claim.","section":"Equations (3.1)–(3.4) and Eq. (4)"},{"comment":"The validation of the structural-particularity prediction is weaker than claimed because the same S_IR criterion is used to select which GBs are called 'special', and then those same GBs are inspected to confirm that they are energy minima or mobility transitions. This selection procedure can inflate the apparent success rate even if S_IR has no predictive power beyond the training set. A stronger test would be to compute energies and mobilities for all 25 interpolated GBs in each family (or for a held-out set) and report correlations between S_IR and energy/mobility without pre-filtering. In addition, the demonstration is limited to two 1D angular scans (Σ5 <001> and Σ9 <011>), whereas the abstract claims a prediction across the 'vast 5D space'; additional sampling dimensions or a clear statement of the intended scope are needed.","section":"Structural particularity section, Figure 3"}],"minor_comments":[{"comment":"The phrase 'across the vast 5D space' overstates what is demonstrated: the structural-particularity tests in Figure 3 vary only the inclination angle for two fixed disorientation axes and Σ values. Please qualify the claim to reflect the actual validation.","section":"Abstract and Introduction"},{"comment":"The operation in Eq. (4) is called 'rotating' x^p and x^q, but the formula is an integer linear combination α x + β y. The terminology is misleading; suggest 'generating rational in-plane lattice directions' or similar.","section":"Eq. (4)"},{"comment":"The algorithm is implemented in 'private Fortran code GB-SG.f90' available only 'upon reasonable request'. For a methods paper whose central contribution is an algorithm, this is not sufficiently reproducible. The code, or at least a pseudocode listing with the exact enumeration of N and (α_m, β_m), should be included or deposited.","section":"Data availability"},{"comment":"The table has two separate series both numbered 1–25 with identical column headers. It needs a clear split or a single ID column to avoid ambiguity about which rows belong to Σ5 and which to Σ9.","section":"Table A1"},{"comment":"The user-defined weight W is introduced without any guidance or sensitivity analysis. Since the weighted sampling scheme is a secondary contribution, a brief discussion of how W affects the selected GB set would be useful.","section":"Eq. (7)"},{"comment":"The mobility data show relatively large error bars at several points, and the claim that the marked GBs represent 'transitions between different mobility trends' is visually suggestive but not statistically quantified. Please add a quantitative measure, e.g., local slope changes or a fit comparison.","section":"Figure 3(c)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript addresses an important and timely problem, and the basic idea of using a geometric supercell-size parameter as a cheap particularity indicator is appealing. However, the Eq. (6) area error and the missing completeness proof for the 'irreducible' claim are serious enough that the paper should not be accepted in its present form. Both issues appear fixable within the scope of the work, but the revisions will need to recompute or re-derive the central parameter and strengthen the validation. The author should also be encouraged to provide the code or a detailed algorithm specification to enable independent checking of the finite-search claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core idea here is worth a look: use the minimal oblique supercell size as a cheap geometric pre-simulation score for structural particularity. That specific proposal is new, as far as I can tell, and the energy/mobility data in Figures 3b and 3c do show low-S_IR boundaries sitting at energy cusps and mobility transitions. The qualitative alignment is suggestive, and the comparison with the Homer dataset plus the parity plot in Figure 2b gives some confidence that the supercell representation does not change the computed GB energy. The weighted sampling strategy in Section 3 is also sensible. So the paper earns credit for the idea and for doing actual LAMMPS simulations to test it.\n\nBut there are two load-bearing problems. First, Eq. (6) defines S_IR as L_w L_v cos(angle), which is wrong. The area of a parallelogram is L_w L_v sin(angle). For the common case where w and v are perpendicular, cos is zero and S_IR becomes zero — absurd for a cell size. This error propagates into Table A1 and Figure 3, so all quantitative rankings built on S_IR are suspect. It looks like the author may have computed the correct thing elsewhere, but the paper as written defines the central parameter incorrectly.\n\nSecond, the 'irreducible' claim is not backed up. The algorithm selects the two shortest candidate pairs from a finite set of N integer trials, but there is no proof that the true primitive basis of the CSL 2D lattice lies within that finite set, nor a bound on N. Without that, S_IR is only a heuristic upper bound, not the irreducible size. The term 'irreducible' is overclaimed. This matters because the structural-particularity rankings built on S_IR are the whole point of the paper.\n\nThere is also a milder circularity issue: the same algorithm is used to select the boundaries that are then taken to confirm the particularity hypothesis, so the confirmation is not blind. And no code or data are released; the Fortran program is 'available upon reasonable request,' which is not real openness.\n\nDespite these problems, the paper is not incoherent. The idea is plausible, the errors are fixable, and the qualitative trends are worth investigating with a corrected formula and a rigorous search guarantee. I would not cite it in its current form, but I would bring it to a reading group to discuss whether the pre-screening concept can be salvaged.\n\nFor peer review: yes, send it out. A good referee could help the author fix the formula, prove or properly bound the search, and add a random-sampling baseline. It deserves that attention, even though major revision is likely.","headline":"Clever pre-screening idea for GB simulations, but a mis-defined formula and an unproven 'irreducible' claim undercut the results as written.","tokens_in":12457,"tokens_out":1889,"would_cite":false,"duration_ms":23881,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["61.72.Mm"],"model":"deepseek-v4-flash","headline":"For any CSL grain-boundary character in a cubic crystal, the paper defines an irreducible monoclinic supercell whose size, computed without building atoms, identifies boundaries that sit at energy cusps and mobility transitions.","keywords":["grain boundary","coincident site lattice","irreducible supercell","CSL character","structure-property relationship","atomistic simulation","structural particularity","aluminium"],"falsifier":"For a fixed CSL character, increase the number of trial rotations N until the reported S_IR stops changing, then compute an independent rigorous reduction of the 2D lattice generated by the boundary-plane vectors; if any valid parallelogram cell with smaller area exists, the algorithm's 'irreducible' label fails. A direct test is to compare S_IR against the true primitive-cell area for a set of high-Sigma characters; any mismatch refutes the claim of minimality.","tokens_in":11609,"feed_emoji":"🔬","tokens_out":5840,"duration_ms":64735,"temperature":0.7,"pith_summary":"The paper tries to establish that every coincidence-site-lattice (CSL) grain-boundary character in a cubic crystal has an irreducible macroscopic representation: the smallest parallelogram supercell that satisfies both grains' periodicity, computable directly from the geometric character without building atoms. It argues that this minimal supercell size can be used as a pre-screening number: a small irreducible cell signals a structurally simple boundary, and such boundaries are the special ones in the five-dimensional grain-boundary landscape. The claim is tested on a series of aluminium mixed grain boundaries, where the flagged characters indeed sit at local energy minima and mark transitions in mobility trends. If correct, the method lets researchers rank the immense space of boundary characters for computational cost and scientific interest before running any expensive atomistic simulation.","feed_headline":"Smallest supercell predicts which grain boundaries are special","feed_subtitle":"Computing a CSL boundary's irreducible cell costs nothing and flags boundaries that sit at energy cusps and mobility transitions.","key_machinery":"The central object is the irreducible macroscopic representation: a pair of in-plane lattice-vector systems {w,v} for the two grains that generate the smallest parallelogram supercell simultaneously compatible with both crystals. The workhorse operation is an integer trial loop that forms candidate vectors as combinations of the two orthogonal in-plane vectors, rotates both grains in lockstep, and sorts the candidates by length so that the two shortest non-parallel pairs can be assigned to w and v. Three constraints filter the candidates: the CSL constraint (equal lengths in both grains), the rotation-invariance constraint (equal in-plane angles), and the irreducible-representation constrain","core_discovery":"The paper's central claim is that every CSL grain-boundary character in a cubic lattice admits an irreducible macroscopic representation: two pairs of shortest valid lattice vectors, one pair for each grain, that together define a monoclinic supercell with the smallest possible in-plane area. The algorithm finds these vectors by rotating the two grains' in-plane lattice orientations in lockstep over a list of integer trial combinations, then selecting the two shortest non-parallel candidate pairs that satisfy the CSL length equalities and rotation-invariance constraints. The resulting irreducible supercell size S_IR is shown to be smaller than the conventional orthogonal supercell and often","pith_inferences":["Editorial inference: if the minimality guarantee is made rigorous, S_IR could serve as a universal geometric descriptor for structure-property machine-learning models, reducing reliance on expensive atomistic labels.","Editorial inference: the finite-N integer scan could be replaced by a proven lattice-basis reduction; this would convert the algorithm from a heuristic search to an exact computation and would remove the main robustness concern.","Editorial inference: the same pre-evaluation logic could extend beyond CSL boundaries and cubic lattices, where a similarly defined minimal-cell size might still correlate with structural complexity and property anomalies.","Editorial inference: because S_IR is material-independent after normalization, rankings built on aluminium data may transfer to other fcc metals, but the energy and mobility links would need re-validation for each interatomic potential."],"forward_implications":["Supercells for atomistic simulations of CSL grain boundaries can be made smaller than orthogonal representations, directly reducing the computational cost for complex mixed boundaries.","Because S_IR is computed before any atomistic run, it enables pre-evaluation of the vast five-dimensional grain-boundary character space, flagging characters that deserve detailed structural study.","The aluminium mixed-boundary results indicate that low-S_IR characters tend to sit at energy cusps and at transitions in mobility trends, so prioritizing them in high-throughput sampling should capture the main features of structure-property relationships.","A weighted sampling strategy built on S_IR can balance comprehensive coverage of the character space against the need to focus resources on structurally particular boundaries.","The normalized, material-independent form of S_IR allows the geometric ranking to be transferred across cubic materials without re-running the geometry search."],"fun_headline_variants":["Irreducible supercell flags special grain boundaries cheaply","Tiny supercell predicts boundary anomalies in aluminum","Zero-cost cell size spots special grain boundaries","Smallest cell identifies special grain boundaries in 5D space"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the finite list of integer trial rotations always contains the two truly shortest lattice-vector pairs that generate a primitive cell for the given CSL character; the paper gives no bound on the number of trials and no proof that these two vectors form the irreducible basis.","fun_headline_variants_meta":{"raw":{"variants":["Irreducible supercell flags special grain boundaries cheaply","Tiny supercell predicts boundary anomalies in aluminum","Zero-cost cell size spots special grain boundaries","Smallest cell identifies special grain boundaries in 5D space"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000293,"raw_usage":{"total_tokens":1501,"prompt_tokens":656,"completion_tokens":845,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":400,"completion_tokens_details":{"reasoning_tokens":783}},"tokens_in":400,"tokens_out":845,"duration_ms":6620,"temperature":1.0,"reasoning_tokens":783,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T05:48:39.540590+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a fixed CSL character, increase the number of trial rotations N until the reported S_IR stops changing, then compute an independent rigorous reduction of the 2D lattice generated by the boundary-plane vectors; if any valid parallelogram cell with smaller area exists, the algorithm's 'irreducible' label fails. A direct test is to compare S_IR against the true primitive-cell area for a set of high-Sigma characters; any mismatch refutes the claim of minimality.","supporting_citations":[],"review_version":1}