{"id":"c7f389c5-96d5-47ec-afb7-f408b32b24b7","arxiv_id":"2607.07713","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":8,"one_line_summary":"Checkerboard rib–garter weft knits show geometry-tunable auxetic unfolding that a linear spring-network model predicts, and partial plating embeds capacitive strain sensors with programmable sensitivity–range tradeoffs.","lead":"A spring-network model predicts how checkerboard rib–garter knits fold, unfold, and expand sideways under stretch, and the same knits can be knitted with built-in capacitive strain sensors. The work gives a practical design path for programmable auxetic smart textiles without post-processing.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"Transfer of pure-strip spring parameters to checkerboard kinematics is the load-bearing soft spot, but multi-geometry agreement already constrains it tightly.","rationale":"The Reader correctly isolates the transferability of pure-strip parameters as the weakest link in the strongest claim. Multi-geometry experimental–numerical agreement on both relaxed wavelengths (Fig. 5D) and the non-monotonic \nu(ε) trends (Fig. 8) already supplies substantial independent evidence that the assumption holds in the unfolding window that matters for auxeticity. The post-flattening discrepancy is acknowledged by the authors and does not undercut the core kinematic story. Sensing results are secondary and do not alter the mechanical claim. Consequently the existing CONDITIONAL verdict (useful contribution with caveats that should be tightened before treating the model as a black-box design tool) remains appropriate; no stronger or weaker verdict is warranted by the present evidence. The concrete leave-one-n-out test above would settle residual quantitative doubt without requiring new fabrication.","tokens_in":13467,"tokens_out":631,"duration_ms":6168,"concrete_test":"Re-fit only the pure-strip tables on a held-out subset of n values, then predict the full \nu(ε) curves for the three experimental checkerboard geometries already shown in Fig. 8 ((4,16), (4,32), (8,48)). If the predicted location of min \nu shifts by more than ~0.05 strain or the depth of min \nu changes by more than ~0.15 relative to the published curves, the transferability assumption is weaker than claimed and the design-tool claim should be caveated more strongly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that ks, kt and \theta0 fitted only to uniaxial stress–strain of pure n\times n rib and garter strips (Tables 1–2), together with the fixed 1:2 stitch aspect that pairs n\times n rib springs with 2n\times2n garter springs, transfer without re-fitting to the 2-D checkerboard relaxation (sinusoidal boundaries, Fig. 5) and to the subsequent unfolding kinematics that produce the tunable negative Poisson’s ratio (Figs. 7–8). The paper itself notes that once the textile is fully flattened the sharp-fold idealization fails and simulated εxx drops more abruptly than experiment (§5). That post-flattening discrepancy is real, yet the pre-flattening regime that actually generates the auxetic peak (minimum \nu between εapplied ≈ 0.1–0.2) is the regime the model is asked to predict, and the multi-geometry wavelength collapse plus qualitative \nu trends already provide non-trivial support for transferability. The residual risk is therefore quantitative rather than qualitative: whether the magnitude and location of min \nu remain accurate enough for design once the pure-strip parameters are used off-distribution.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript develops a reduced-order spring-network model of linear extensional and torsional springs for weft-knit textiles built from checkerboard arrangements of rib and garter patches. Parameters (ks, kt, θ0) are fitted to uniaxial stress–strain curves of pure n×n rib and garter strips (Tables 1–2) and then reused, without re-fitting, to predict (i) the relaxed corrugated geometry of 2D checkerboards (sinusoidal patch boundaries) and (ii) local strain evolution under tension. Combining simulations with experiments, the authors show that programmed unfolding of the corrugations produces tunable auxetic response, with both the magnitude of the negative Poisson’s ratio and the applied strain at which the minimum occurs controlled by unit-cell geometry (nstripe, nsubunit). They further introduce partial plating of conductive yarns during knitting to embed capacitive strain sensors in a single manufacturing step, and report programmable tradeoffs between capacitance sensitivity and operating strain range for a small set of geometries.","tokens_in":13900,"tokens_out":1798,"duration_ms":22228,"significance":"If the central claims hold, the work supplies a practical design route for mechanically programmable, sensorized knits that is more accessible than yarn-level contact simulations and more predictive than purely empirical device-scale fits. Strengths include: multi-geometry experimental validation of relaxed wavelengths via DFT (Fig. 5), qualitative agreement of local strain maps (Fig. 7) and non-monotonic εxx and ν(ε) trends (Fig. 8), an explicit reduced-order energy formulation (Eqs. 1–9), and a fabrication method (partial plating) that eliminates post-processing for knitted capacitors. The open data statement (GitHub) further supports reproducibility. The combination of folding-driven auxeticity with in-fabric capacitive sensing is of clear interest for soft robotics and wearables.","major_comments":[{"comment":"§3–§5 and Tables 1–2: The load-bearing modeling claim is that ks, kt, and θ0 fitted solely to pure n×n rib/garter uniaxial curves transfer without re-fitting to checkerboard relaxation and unfolding kinematics (including the fixed ~1:2 stitch aspect that pairs n×n rib with 2n×2n garter springs). Multi-geometry wavelength collapse (Fig. 5D) and qualitative ν trends (Fig. 8) already constrain this transfer, but the manuscript does not report quantitative error metrics (e.g., RMSE or peak-location error on min ν and on εxx(εapplied) across the (nstripe, nsubunit) family). Please add such metrics, and a brief sensitivity check on how ± variations in the pure-strip parameters affect predicted min ν and its strain location, so that the design utility of the transferred parameters is bounded rather than asserted.","section":"§3–§5, Tables 1–2, Figs. 5 and 8"},{"comment":"§5 (text around Fig. 8): The authors correctly note that after full flattening the sharp-fold idealization fails and simulated εxx drops more abruptly than experiment. The auxetic peak (min ν between εapplied ≈ 0.1–0.2) lies in the pre-flattening regime the model is meant to capture, so this is not fatal, but the manuscript should state more explicitly the strain window over which the model is considered predictive for design (e.g., up to the εxx maximum / completion of unfolding) and avoid implying quantitative accuracy of ν after that point.","section":"§5, Fig. 8"},{"comment":"§6 and Fig. 10: The claim of “programmable tradeoffs between strain sensitivity and sensing range” and the prediction that varying nstripe and nsubunit will tune the capacitance–strain slope (final paragraph of §6) rest on four measured geometries plus mechanical kinematics, without a capacitance model or a systematic sensing map over the same (nstripe, nsubunit) family used in Fig. 8. The experimental curves are valuable; please either (i) measure capacitance for additional unit cells that span the auxetic family, or (ii) temper the language so that “programmable” is clearly prospective where only four samples are shown, and separate measured tradeoffs from model-based extrapolation.","section":"§6, Fig. 10"}],"minor_comments":[{"comment":"Eq. (1) and surrounding text: Δℓ_0i and Δθ_0i are written with a subscript 0 that is easy to misread as a rest-length index; clarify notation (e.g., Δℓ_i, Δθ_i relative to rest values ℓ_0, θ_0).","section":"§3.1, Eq. (1)"},{"comment":"Fig. 1D–E: Stress–strain axes and normalization (force per what width/thickness?) should be stated in the caption or Methods so that fitted ks units [N/mm] can be interpreted consistently.","section":"Fig. 1D–E, Methods"},{"comment":"§3.2: The choices k_wall = 10^2 N/mm, z_max = 6.0 mm, ρ_s = 0.11 kg/m^2, and the viscous damping c are introduced without a short justification or sensitivity note; a sentence that in-plane kinematics are insensitive to these within a stated range would help.","section":"§3.2, Eqs. (3)–(9)"},{"comment":"Fig. 7: Color-scale limits for ε^i_xx should be identical across experiment and simulation panels (and across A–C if possible) to make visual comparison fair.","section":"Fig. 7"},{"comment":"Typos / wording: Abstract and title use “Folding-Driven”; body has occasional spacing issues (e.g., “FOLDING-DRIVENAUXETIC…” in the header block), “stich” for “stitch” in the Introduction, and “for for viscous” in §3.2. Clean these in production.","section":"Introduction, §3.2"},{"comment":"Methods: State how many unit cells / samples underlie the averaged εxx, εyy, and ν curves in Fig. 8, and whether error bars or sample-to-sample variation were assessed.","section":"Methods, Fig. 8"}],"recommendation":"minor_revision","confidential_remarks":"Fit for a soft-matter / multifunctional-materials venue is good. The parameter-transfer issue is real but already partially stress-tested by multi-geometry wavelength and ν trends; I would not require a full re-fit to checkerboards if quantitative error bounds and a clear predictive window are added. Novelty of partial plating as a single-step knitted capacitor is a genuine plus relative to post-processed textile sensors. No concerns about scope or citation pattern that would affect the recommendation."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The useful core here is a linear spring network (extension + torsion) that, after fitting only pure n×n rib and garter strips, predicts both the relaxed corrugated geometry of checkerboard rib–garter patches and the unfolding kinematics that produce a tunable negative Poisson’s ratio. They also show partial plating of conductive yarn so capacitive pathways are knit in one step, with a clear sensitivity-versus-range tradeoff tied to the same unit-cell geometry.\n\nWhat is actually new is not auxetic knits or fabric capacitors—those exist—but the combination: a calibrated reduced-order model that maps (nstripe, nsubunit) onto min ν and the strain at which it appears, plus fabrication that embeds sensing without post-processing. The multi-geometry wavelength check (DFT of unit-cell boundaries) is a real independent test of the relaxation prediction, and the local strain maps and non-monotonic εxx/ν curves line up qualitatively across several cells. The authors flag that once the fabric is fully flat the sharp-fold idealization fails and simulated εxx drops too abruptly; that honesty helps.\n\nThe soft spot is transferability. ks, kt, and θ0 come from uniaxial pure-strip curves, then get reused with a fixed ~1:2 stitch aspect pairing. That is not circular—wavelengths and kinematics are out-of-sample—but it is load-bearing. Pre-flattening (where the auxetic peak lives, ~0.1–0.2 applied strain) is where the model is asked to work, and the multi-geometry agreement already constrains it. Residual risk is quantitative accuracy of min ν for design, not whether auxetic unfolding exists. Missing error bars and a vague “on github” data pointer are minor hygiene issues, not conceptual ones.\n\nThis is for people who design machine-knit actuators, wearables, or soft sensors and want a design handle rather than yarn-level simulation. Math is elementary energy minimization / ODEs; data are solid enough for the claims; citations cover the right prior art without pretending the building blocks are new. I would send it to peer review. Engage if you care about programmable knit mechanics; skim the sensing section even if you only want the auxetic map.","headline":"Usable reduced-order model plus single-step sensing for rib–garter checkerboard knits; transfer of pure-strip spring fits is the real caveat, not a deal-breaker.","tokens_in":14543,"tokens_out":558,"would_cite":true,"duration_ms":8935,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A linear spring network predicts how checkerboard rib–garter knits unfold into tunable auxetic fabrics and built-in capacitive sensors.","keywords":["weft knitting","auxetic textiles","rib and garter","spring-network model","capacitive strain sensing","partial plating","programmable corrugations","negative Poisson's ratio"],"falsifier":"Fabricate a new (nstripe, nsubunit) pair, measure its relaxed unit-cell wavelength by Fourier analysis of the boundaries and its Poisson’s-ratio-versus-strain curve under the same acrylic-plate tensile protocol, and check whether the spring model with the original tables predicts both quantities within the same tolerance shown for the existing geometries.","tokens_in":14363,"feed_emoji":"🧶","tokens_out":981,"duration_ms":9153,"temperature":0.7,"pith_summary":"Machine knitting can program geometry and function stitch by stitch, but designers still lack simple tools that predict large-deformation mechanics. This paper shows that a checkerboard of rib and garter patches relaxes into 3D corrugations whose programmed unfolding under tension produces negative Poisson’s ratio, and that both the strength of that auxetic effect and the strain at which it appears are set by two unit-cell integers. A reduced-order network of only linear extensional and torsional springs, with parameters fitted solely to pure rib and garter strips, recovers the relaxed corrugated shapes and the local strain fields measured in experiments. The same architecture is fabricated with conductive pathways plated in during knitting, turning the unfolding ridges into capacitors whose sensitivity-versus-range trade-off is likewise controlled by geometry. The result is a concrete route from stitch pattern to designed shape-morphing and sensing textiles without post-processing.","feed_headline":"Knitted checkerboards unfold into tunable auxetic sensors","feed_subtitle":"Linear springs predict geometry and strain; conductive yarns add capacitance without post-processing","key_machinery":"The reduced-order spring network: a rectangular grid of linear extensional springs (weft, warp, and weaker diagonals) linked by linear torsional springs whose rest angles encode the preferred curvature of rib versus garter ridges. Energy minimization first relaxes the flat grid into the corrugated state; the same network is then stretched to obtain local strain maps and effective Poisson’s ratio.","core_discovery":"Programmed unfolding of corrugations in checkerboard arrangements of rib and garter patches generates tunable auxetic behavior whose magnitude and onset strain are both governed by the unit-cell geometry (nstripe, nsubunit). A spring-network model that uses only linear extensional and torsional springs, with parameters taken from pure n×n rib and garter fits, predicts both the relaxed corrugated configuration and the subsequent evolution of local strains under tension, matching experiment well enough to design the response.","pith_inferences":["Because the model already captures interface frustration between rib and garter patches, it should transfer with little change to other non-periodic arrangements (stripes, gradients, or origami-like flat-face folds) once the corresponding pure-pattern parameters are measured.","The abrupt simulated drop in transverse strain after full unfolding suggests that adding a mild nonlinear stiffening term to the extensional springs would extend quantitative accuracy into the post-unfolding regime without increasing mesh density.","The same plating geometry that forms capacitors could be re-routed as resistive or inductive pathways, allowing multi-modal sensing on a single knitted architecture."],"forward_implications":["Unit-cell integers can be chosen so that a knitted textile exhibits a prescribed negative Poisson’s ratio at a chosen applied strain.","Partial plating of conductive yarns during knitting yields capacitive strain sensors whose sensitivity and operating range are set by the same geometric parameters that control auxeticity.","The same spring-network framework can be used as an inverse design tool to target prescribed shapes, auxetic windows, or sensing curves without re-deriving yarn-level mechanics.","Wider stripes and smaller unit cells systematically strengthen auxeticity, giving a simple design rule for more negative Poisson’s ratios."],"fun_headline_variants":["Checkerboard knits unfold into geometry-tuned auxetic sensors","Linear springs predict corrugated knit relaxation and auxetic strain","Rib-garter unit cells program auxetic magnitude and onset","Partial-plated yarns add capacitance to unfolding auxetic knits","Spring networks design knit shape-morphing and capacitive range"],"cache_read_input_tokens":128,"weakest_assumption_plain":"Stiffnesses and rest angles fitted only to the uniaxial curves of pure rib and garter strips transfer unchanged to the two-dimensional checkerboard, even after the sharp-fold idealization stops matching the flattened fabric.","fun_headline_variants_meta":{"raw":{"variants":["Checkerboard knits unfold into geometry-tuned auxetic sensors","Linear springs predict corrugated knit relaxation and auxetic strain","Rib-garter unit cells program auxetic magnitude and onset","Partial-plated yarns add capacitance to unfolding auxetic knits","Spring networks design knit shape-morphing and capacitive range"]},"model":"grok-4.5","effort":"low","cost_usd":0.005774,"raw_usage":{"total_tokens":1553,"prompt_tokens":790,"num_sources_used":0,"completion_tokens":66,"cost_in_usd_ticks":57740000,"prompt_tokens_details":{"text_tokens":790,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":697,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":790,"tokens_out":66,"duration_ms":8410,"temperature":1.0,"reasoning_tokens":697,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-12T08:03:47.866025+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Fabricate a new (nstripe, nsubunit) pair, measure its relaxed unit-cell wavelength by Fourier analysis of the boundaries and its Poisson’s-ratio-versus-strain curve under the same acrylic-plate tensile protocol, and check whether the spring model with the original tables predicts both quantities within the same tolerance shown for the existing geometries.","supporting_citations":[],"review_version":1}