{"schemaVersion":"1.0","type":"catalog","service":"Disruptive Concepts","purpose":"We curate ~20 disruptive papers every week from arXiv in AI, quantum, biotech, energy, and more — then write plain-English explainers free for people. AI agents purchase structured data via the x402 protocol.","product_sold":"disruptive_week_json","free":true,"spotlight":false,"paid_sku_count":5,"note":"Metadata + free HTML links only. Full structured bodies require x402 on paid routes. Spotlight: GET /api/v1/catalog?spotlight=1","site":"https://disruptive-concepts.com","wellKnown":"https://disruptive-concepts.com/.well-known/x402","currentWeekId":"2026-W37","this_week_url":"https://disruptive-concepts.com/this-week","weeks":[{"weekId":"2026-W30","title":"Week of July 21, 2026","paperCount":20},{"weekId":"2026-W31","title":"Week of July 28, 2026","paperCount":20},{"weekId":"2026-W32","title":"Week of August 3, 2026","paperCount":20},{"weekId":"2026-W33","title":"Week of August 10, 2026","paperCount":20},{"weekId":"2026-W34","title":"Week of August 17, 2026","paperCount":20},{"weekId":"2026-W35","title":"Week of August 24, 2026","paperCount":20},{"weekId":"2026-W36","title":"Week of August 31, 2026","paperCount":20},{"weekId":"2026-W37","title":"Week of September 7, 2026","paperCount":20}],"count":20,"items":[{"arxivId":"2609.04194","slug":"legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","title":"Legibility is Not Interpretability: Comparing Judged and Actual Importance in Chain-Of-Thought Reasoning","topics":["ai"],"weekId":"2026-W37","rankInWeek":1,"disruptivenessScore":93,"absUrl":"https://arxiv.org/abs/2609.04194","hasArticle":true,"url":"https://disruptive-concepts.com/article/legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","html_url":"https://disruptive-concepts.com/article/legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","articleUrl":"https://disruptive-concepts.com/article/legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","paperUrl":"https://disruptive-concepts.com/paper/2609.04194","headline":"Readable chain-of-thought traces look like a window into why a model answered as it did — but this paper shows that even strong LLM judges cannot reliably tell which steps actually change the chance of a correct answer. A caution for process rewards and critic models, not a product timeline.","blurb":"What: The authors measure a reasoning step’s real importance as its advantage — how much including it changes expected reward, estimated with Monte Carlo rollouts — then test whether LLM judges can spot high-advantage steps from the text alone.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04121","slug":"quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","title":"Quantum thermalization achieves optimal approximate quantum error correction","topics":["quantum"],"weekId":"2026-W37","rankInWeek":2,"disruptivenessScore":85,"absUrl":"https://arxiv.org/abs/2609.04121","hasArticle":true,"url":"https://disruptive-concepts.com/article/quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","html_url":"https://disruptive-concepts.com/article/quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","articleUrl":"https://disruptive-concepts.com/article/quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","paperUrl":"https://disruptive-concepts.com/paper/2609.04121","headline":"Generic thermalizing many-body dynamics, treated as an encoding, appear to hit the same optimal approximate error-correction limits as Haar-random codes — including a finite-temperature family based on the Scrooge ensemble.","blurb":"What: The authors treat typical late-time thermalizing states as codewords and show that generic thermalization follows a universal encoding-rate/distance/entropy curve that saturates Singleton-type bounds for approximate quantum error correction.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04196","slug":"puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","title":"Puffin-World: Scaling a Unified Multimodal Model with Native 3D World States","topics":["ai"],"weekId":"2026-W37","rankInWeek":3,"disruptivenessScore":83,"absUrl":"https://arxiv.org/abs/2609.04196","hasArticle":true,"url":"https://disruptive-concepts.com/article/puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","html_url":"https://disruptive-concepts.com/article/puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","articleUrl":"https://disruptive-concepts.com/article/puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","paperUrl":"https://disruptive-concepts.com/paper/2609.04196","headline":"Puffin-World is a single multimodal model that jointly handles physics, geometry, and appearance — plus camera motion — to generate and reconstruct 3D worlds without bolting on separate offline modules.","blurb":"What: The authors introduce a unified architecture that models gravity/latitude, depth, and images together with an Omni-Camera representation, trained on a 16-million-example mix of vision-language-camera triplets and trajectories.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04193","slug":"gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","title":"GIFT: Guided Intermediate Feature Training via Action-Oriented Structural Supervision for Robotic Manipulation","topics":["robotics"],"weekId":"2026-W37","rankInWeek":4,"disruptivenessScore":82,"absUrl":"https://arxiv.org/abs/2609.04193","hasArticle":true,"url":"https://disruptive-concepts.com/article/gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","html_url":"https://disruptive-concepts.com/article/gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","articleUrl":"https://disruptive-concepts.com/article/gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","paperUrl":"https://disruptive-concepts.com/paper/2609.04193","headline":"GIFT trains a robot policy’s intermediate features to keep geometry, affordances, and goals — the structure that actually matters for control — and shows consistent gains across three action formulations on LIBERO-Plus and RoboCasa.","blurb":"What: The authors name an action-sufficiency gap — rich visual features that still omit control-relevant structure — and add training-time constraints for geometry alignment, affordance prediction, and goal-region reconstruction.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04202","slug":"tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","title":"TokenMatch: 3D Mesh Correspondence Transformer with Curvature-Guided Tokenisation","topics":["ai"],"weekId":"2026-W37","rankInWeek":5,"disruptivenessScore":80,"absUrl":"https://arxiv.org/abs/2609.04202","hasArticle":true,"url":"https://disruptive-concepts.com/article/tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","html_url":"https://disruptive-concepts.com/article/tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","articleUrl":"https://disruptive-concepts.com/article/tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","paperUrl":"https://disruptive-concepts.com/paper/2609.04202","headline":"TokenMatch is a feed-forward transformer that matches 3D meshes — including hard partial, non-isometric cases — by tokenizing surfaces with curvature-guided patches, and it generalizes from partial training data to full shapes without a fine-tune.","blurb":"What: Trained only on BeCoS, a difficult partial-to-partial non-isometric set, the model uses self- and cross-attention over curvature-guided mesh patches to predict dense correspondences and still matches full shapes at sub-second speed.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04165","slug":"parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","title":"Parameterised graph theory for tensor networks: entanglement rerouting, structural simplification, and agnostic tomography","topics":["quantum"],"weekId":"2026-W37","rankInWeek":6,"disruptivenessScore":78,"absUrl":"https://arxiv.org/abs/2609.04165","hasArticle":true,"url":"https://disruptive-concepts.com/article/parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","html_url":"https://disruptive-concepts.com/article/parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","articleUrl":"https://disruptive-concepts.com/article/parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","paperUrl":"https://disruptive-concepts.com/paper/2609.04165","headline":"Graph parameters such as cutwidth and tree-cutwidth are shown to control how expensive it is to rewrite a tensor-network state as an MPS or tree network — and how hard it is to learn that state, even in an agnostic setting.","blurb":"What: The authors prove cutwidth and tree-cutwidth bound bond-dimension overhead when converting a tensor-network state to an MPS or TTN, and they give graph-dependent sample and compute bounds for realizable and agnostic tomography.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.02344","slug":"subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","title":"Subcellularly Resolved Single-Cell Embedding Learning with Transcriptomic data, Protein Structure and Localization Information","topics":["biotech"],"weekId":"2026-W37","rankInWeek":7,"disruptivenessScore":77,"absUrl":"https://arxiv.org/abs/2609.02344","hasArticle":true,"url":"https://disruptive-concepts.com/article/subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","html_url":"https://disruptive-concepts.com/article/subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","articleUrl":"https://disruptive-concepts.com/article/subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","paperUrl":"https://disruptive-concepts.com/paper/2609.02344","headline":"A multimodal cell-embedding model mixes RNA profiles with protein sequence and structure, using cross-attention inside subcellular compartments instead of treating each cell as one bag of molecules.","blurb":"What: The authors propose a cross-attention framework that builds cell embeddings from transcriptomics, protein sequences, and protein structures while modeling those signals inside distinct subcellular compartments.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.03795","slug":"exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","title":"Exciton multipolarity controls coherent and squeezed phonons in van der Waals heterostructures","topics":["materials"],"weekId":"2026-W37","rankInWeek":8,"disruptivenessScore":75,"absUrl":"https://arxiv.org/abs/2609.03795","hasArticle":true,"url":"https://disruptive-concepts.com/article/exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","html_url":"https://disruptive-concepts.com/article/exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","articleUrl":"https://disruptive-concepts.com/article/exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","paperUrl":"https://disruptive-concepts.com/paper/2609.03795","headline":"In MoSe2/WSe2 multilayers, dipolar excitons drive coherent interlayer breathing phonons while quadrupolar excitons drive squeezed ones — and a vertical electric field can switch the phonon state.","blurb":"What: Atomistic simulations of alternating MoSe2 and WSe2 layers show exciton multipolarity as a design rule: linear coupling from dipolar excitons yields coherent phonons; quadratic coupling from quadrupolar excitons yields squeezed phonons.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.03984","slug":"muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","title":"MulDP: Multimodal Diffusion Policy for Autonomous Quadruped Parkour Navigation across Complex Terrains","topics":["robotics"],"weekId":"2026-W37","rankInWeek":9,"disruptivenessScore":74,"absUrl":"https://arxiv.org/abs/2609.03984","hasArticle":true,"url":"https://disruptive-concepts.com/article/muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","html_url":"https://disruptive-concepts.com/article/muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","articleUrl":"https://disruptive-concepts.com/article/muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","paperUrl":"https://disruptive-concepts.com/paper/2609.03984","headline":"MulDP is a diffusion policy that turns vision, proprioception, and goals into anticipatory velocity commands so a quadruped can parkour-navigate without a human calling the high-level plays — trained on a new multimodal parkour dataset.","blurb":"What: The authors couple perception to embodied control with a multimodal diffusion policy that outputs temporally coherent navigation velocities, and they introduce QPND, described as the first quadruped parkour navigation dataset of its kind.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.03757","slug":"ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","title":"Ion-Engineered Insulator-to-Semiconductor Transition in Natural 2D Biotite","topics":["materials"],"weekId":"2026-W37","rankInWeek":10,"disruptivenessScore":72,"absUrl":"https://arxiv.org/abs/2609.03757","hasArticle":true,"url":"https://disruptive-concepts.com/article/ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","html_url":"https://disruptive-concepts.com/article/ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","articleUrl":"https://disruptive-concepts.com/article/ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","paperUrl":"https://disruptive-concepts.com/paper/2609.03757","headline":"A controlled NaOH treatment turns exfoliated biotite nanosheets from an insulator into a tunable 2D semiconductor by adding sodium, defects, and local reconstruction while keeping the layered mineral framework.","blurb":"What: Liquid-phase-exfoliated biotite, after NaOH processing, shows a narrowed optical gap (about 5.2 eV down to 3.2–3.5 eV), new near-IR transitions, nonlinear transport near 10 µA, and ultrafast trap-mediated carrier dynamics, backed by DFT.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04132","slug":"characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","title":"Characterizing Large Scale Quantum Systems with Error Per Circuit Layer","topics":["quantum"],"weekId":"2026-W37","rankInWeek":11,"disruptivenessScore":71,"absUrl":"https://arxiv.org/abs/2609.04132","hasArticle":true,"url":"https://disruptive-concepts.com/article/characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","html_url":"https://disruptive-concepts.com/article/characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","articleUrl":"https://disruptive-concepts.com/article/characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","paperUrl":"https://disruptive-concepts.com/paper/2609.04132","headline":"Error Per Circuit Layer (EPCL) estimates how polarizing each layer of a quantum circuit is by overlapping two registers that ran the same random circuits — no classical simulation of ideal outputs and no restriction to Clifford gates.","blurb":"What: EPCL applies identical random circuits to two disjoint registers, measures output-state overlap versus depth, and fits an effective layer polarization under an ensemble-averaged depolarizing model.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04147","slug":"a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","title":"A Low-Cost, Open Platform for End-to-End Autonomous Driving on a Miniature Ackermann Vehicle","topics":["robotics"],"weekId":"2026-W37","rankInWeek":12,"disruptivenessScore":70,"absUrl":"https://arxiv.org/abs/2609.04147","hasArticle":true,"url":"https://disruptive-concepts.com/article/a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","html_url":"https://disruptive-concepts.com/article/a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","articleUrl":"https://disruptive-concepts.com/article/a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","paperUrl":"https://disruptive-concepts.com/paper/2609.04147","headline":"An open miniature-car testbed — physical Ackermann vehicle, printed urban track, tools, and a Webots twin — lets researchers take end-to-end driving from simulation into a real closed loop without a full-size AV budget.","blurb":"What: The authors release a low-cost platform and a command-conditioned behavior-cloning baseline that reaches 6.1 cm mean cross-track error on the real track (human demo 4.7 cm), and they show FOV and sim-to-real data choices decide whether all routes complete.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.01357","slug":"poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","title":"PopPert: Population-level Joint-Distribution Modeling for Single-Cell Perturbation Prediction","topics":["biotech"],"weekId":"2026-W37","rankInWeek":13,"disruptivenessScore":68,"absUrl":"https://arxiv.org/abs/2609.01357","hasArticle":true,"url":"https://disruptive-concepts.com/article/poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","html_url":"https://disruptive-concepts.com/article/poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","articleUrl":"https://disruptive-concepts.com/article/poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","paperUrl":"https://disruptive-concepts.com/paper/2609.01357","headline":"PopPert predicts how a whole population of cells changes after a genetic or chemical perturbation by learning joint gene-expression distributions — without pretending destroyed, unpaired cells can be matched one-to-one.","blurb":"What: The method parameterizes population-level joint distributions (via a low-rank Gaussian copula), maps a control population plus a perturbation into new distribution parameters, and can sample synthetic perturbed profiles.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.03210","slug":"improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","title":"Improving precipitation forecasts in an AI weather model using observational data","topics":["climate"],"weekId":"2026-W37","rankInWeek":14,"disruptivenessScore":66,"absUrl":"https://arxiv.org/abs/2609.03210","hasArticle":true,"url":"https://disruptive-concepts.com/article/improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","html_url":"https://disruptive-concepts.com/article/improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","articleUrl":"https://disruptive-concepts.com/article/improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","paperUrl":"https://disruptive-concepts.com/paper/2609.03210","headline":"Fine-tuning a graph-transformer weather model on IMERG precipitation — not just ERA5 — improves medium-range rain scores by up to 19% and extreme-rain skill by 57% globally versus operational models, with a caveat on the very heaviest events.","blurb":"What: The authors fine-tune an AI weather model at 0.25° with IMERG observational precipitation to correct known ERA5 rain biases, gaining CRPS and storm/drizzle skill while a physics operational model stays more reliable on the heaviest rain.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.04189","slug":"robust-pac-learning-of-concurrent-stochastic-games","title":"Robust PAC Learning of Concurrent Stochastic Games","topics":["ai"],"weekId":"2026-W37","rankInWeek":15,"disruptivenessScore":65,"absUrl":"https://arxiv.org/abs/2609.04189","hasArticle":true,"url":"https://disruptive-concepts.com/article/robust-pac-learning-of-concurrent-stochastic-games","html_url":"https://disruptive-concepts.com/article/robust-pac-learning-of-concurrent-stochastic-games","articleUrl":"https://disruptive-concepts.com/article/robust-pac-learning-of-concurrent-stochastic-games","paperUrl":"https://disruptive-concepts.com/paper/2609.04189","headline":"A PAC algorithm for general-sum concurrent stochastic games with uncertain transitions either returns a near-optimal approximate Nash equilibrium or certifies that no exact equilibrium exists — with polynomial sample complexity under a reachability assumption.","blurb":"What: The authors maintain L1 confidence sets on transition kernels, solve a robust concurrent stochastic game for a social-welfare-optimal ε-Nash equilibrium, and use a Nash-margin test to either output that equilibrium or a sound “no exact NE” certificate.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.02985","slug":"sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","title":"Sparse concept attribution for histomorphological hypothesis generation from whole-slide classifiers","topics":["biotech"],"weekId":"2026-W37","rankInWeek":16,"disruptivenessScore":63,"absUrl":"https://arxiv.org/abs/2609.02985","hasArticle":true,"url":"https://disruptive-concepts.com/article/sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","html_url":"https://disruptive-concepts.com/article/sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","articleUrl":"https://disruptive-concepts.com/article/sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","paperUrl":"https://disruptive-concepts.com/paper/2609.02985","headline":"SCOPE explains whole-slide pathology classifiers by sparsely attributing them onto a histomorphological concept bank — and a new benchmark shows sparse attributions recover known morphology while dense ones look like chance.","blurb":"What: The authors pair pathology vision–language models with sparse concept attribution (SCOPE) and introduce MorphoRecoveryBench, seven pathologist-curated tasks where sparse (and a cheaper pooled-embedding variant) beat dense attribution.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.03975","slug":"detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","title":"Detection of hydrocarbons in Titan using high-resolution cross-correlation spectroscopy","topics":["space"],"weekId":"2026-W37","rankInWeek":17,"disruptivenessScore":59,"absUrl":"https://arxiv.org/abs/2609.03975","hasArticle":true,"url":"https://disruptive-concepts.com/article/detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","html_url":"https://disruptive-concepts.com/article/detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","articleUrl":"https://disruptive-concepts.com/article/detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","paperUrl":"https://disruptive-concepts.com/paper/2609.03975","headline":"Using Titan as a testbed, the authors build high-resolution cross-correlation templates from molecular cross-sections instead of complete line lists — recovering methane and acetylene and making a first HRCCS detection of ethane at SNR about 5.2.","blurb":"What: K-band CRIRES+ spectra of Titan (1.99–2.48 μm) are cross-correlated with both line-by-line and cross-section templates, yielding known CH4 and C2H2 plus C2H6 only from the cross-section templates because no high-resolution ethane line list exists.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.03779","slug":"morphology-of-radio-sources-in-representation-space","title":"Morphology of Radio Sources in Representation Space","topics":["space"],"weekId":"2026-W37","rankInWeek":18,"disruptivenessScore":57,"absUrl":"https://arxiv.org/abs/2609.03779","hasArticle":true,"url":"https://disruptive-concepts.com/article/morphology-of-radio-sources-in-representation-space","html_url":"https://disruptive-concepts.com/article/morphology-of-radio-sources-in-representation-space","articleUrl":"https://disruptive-concepts.com/article/morphology-of-radio-sources-in-representation-space","paperUrl":"https://disruptive-concepts.com/paper/2609.03779","headline":"A classifier ensemble plus representation-space clustering finds six extra radio morphologies in LoTSS-DR3 — including rare winged and FR-ambiguous sources — on top of twelve DR2 classes, in a catalog of more than 13 million objects.","blurb":"What: After applying a fine-tuned ensemble to LoTSS-DR3, the authors cluster low-probability sources in representation space, recover 88% in the old DR2 clusters, and identify representatives of six new morphological classes in the remainder.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.02614","slug":"ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","title":"Ruelle--Pollicott Theory for Metastable Systems: A Unified Framework for Tipping Transitions","topics":["climate"],"weekId":"2026-W37","rankInWeek":19,"disruptivenessScore":55,"absUrl":"https://arxiv.org/abs/2609.02614","hasArticle":true,"url":"https://disruptive-concepts.com/article/ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","html_url":"https://disruptive-concepts.com/article/ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","articleUrl":"https://disruptive-concepts.com/article/ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","paperUrl":"https://disruptive-concepts.com/paper/2609.02614","headline":"A Ruelle–Pollicott spectral theory of stochastic metastable systems shows that classic “critical slowing down” warnings are not universal: large responses need both a small spectral gap and a large residue coupling the observable to the perturbation.","blurb":"What: The authors factor invariant-statistic sensitivity into a spectral denominator and a residue, prove a frozen-spectrum non-normal example can look more alarming than a true gap closure, and separate in-well recovery from interwell escape in a fold tipping problem.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}},{"arxivId":"2609.03768","slug":"computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","title":"Computing high-order mixed derivatives in physics-informed neural networks using multi-index Bell polynomials","topics":["energy"],"weekId":"2026-W37","rankInWeek":20,"disruptivenessScore":49,"absUrl":"https://arxiv.org/abs/2609.03768","hasArticle":true,"url":"https://disruptive-concepts.com/article/computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","html_url":"https://disruptive-concepts.com/article/computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","articleUrl":"https://disruptive-concepts.com/article/computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","paperUrl":"https://disruptive-concepts.com/paper/2609.03768","headline":"A tabulated Faà di Bruno / Bell-polynomial recursion computes the mixed derivatives PINNs need — and their parameter gradients — without nested automatic differentiation, through order seven on CPU without the memory failures of nested graphs.","blurb":"What: The authors organize the multivariate Faà di Bruno formula over a downward-closed multi-index set, tabulate Bell-polynomial convolutions once, and backpropagate losses built from any subset of those derivatives, including nonlinear products and coupled fields.","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01"}}],"articles":[{"id":"legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","slug":"legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","title":"Legibility is Not Interpretability: Comparing Judged and Actual Importance in Chain-Of-Thought Reasoning","headline":"Readable chain-of-thought traces look like a window into why a model answered as it did — but this paper shows that even strong LLM judges cannot reliably tell which steps actually change the chance of a correct answer. A caution for process rewards and critic models, not a product timeline.","blurb":"What: The authors measure a reasoning step’s real importance as its advantage — how much including it changes expected reward, estimated with Monte Carlo rollouts — then test whether LLM judges can spot high-advantage steps from the text alone.","url":"https://disruptive-concepts.com/article/legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","html_url":"https://disruptive-concepts.com/article/legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","arxivId":"2609.04194","weekId":"2026-W37","score":93,"topics":["ai"]},{"id":"quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","slug":"quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","title":"Quantum thermalization achieves optimal approximate quantum error correction","headline":"Generic thermalizing many-body dynamics, treated as an encoding, appear to hit the same optimal approximate error-correction limits as Haar-random codes — including a finite-temperature family based on the Scrooge ensemble.","blurb":"What: The authors treat typical late-time thermalizing states as codewords and show that generic thermalization follows a universal encoding-rate/distance/entropy curve that saturates Singleton-type bounds for approximate quantum error correction.","url":"https://disruptive-concepts.com/article/quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","html_url":"https://disruptive-concepts.com/article/quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","arxivId":"2609.04121","weekId":"2026-W37","score":85,"topics":["quantum"]},{"id":"puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","slug":"puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","title":"Puffin-World: Scaling a Unified Multimodal Model with Native 3D World States","headline":"Puffin-World is a single multimodal model that jointly handles physics, geometry, and appearance — plus camera motion — to generate and reconstruct 3D worlds without bolting on separate offline modules.","blurb":"What: The authors introduce a unified architecture that models gravity/latitude, depth, and images together with an Omni-Camera representation, trained on a 16-million-example mix of vision-language-camera triplets and trajectories.","url":"https://disruptive-concepts.com/article/puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","html_url":"https://disruptive-concepts.com/article/puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","arxivId":"2609.04196","weekId":"2026-W37","score":83,"topics":["ai"]},{"id":"gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","slug":"gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","title":"GIFT: Guided Intermediate Feature Training via Action-Oriented Structural Supervision for Robotic Manipulation","headline":"GIFT trains a robot policy’s intermediate features to keep geometry, affordances, and goals — the structure that actually matters for control — and shows consistent gains across three action formulations on LIBERO-Plus and RoboCasa.","blurb":"What: The authors name an action-sufficiency gap — rich visual features that still omit control-relevant structure — and add training-time constraints for geometry alignment, affordance prediction, and goal-region reconstruction.","url":"https://disruptive-concepts.com/article/gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","html_url":"https://disruptive-concepts.com/article/gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","arxivId":"2609.04193","weekId":"2026-W37","score":82,"topics":["robotics"]},{"id":"tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","slug":"tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","title":"TokenMatch: 3D Mesh Correspondence Transformer with Curvature-Guided Tokenisation","headline":"TokenMatch is a feed-forward transformer that matches 3D meshes — including hard partial, non-isometric cases — by tokenizing surfaces with curvature-guided patches, and it generalizes from partial training data to full shapes without a fine-tune.","blurb":"What: Trained only on BeCoS, a difficult partial-to-partial non-isometric set, the model uses self- and cross-attention over curvature-guided mesh patches to predict dense correspondences and still matches full shapes at sub-second speed.","url":"https://disruptive-concepts.com/article/tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","html_url":"https://disruptive-concepts.com/article/tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","arxivId":"2609.04202","weekId":"2026-W37","score":80,"topics":["ai"]},{"id":"parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","slug":"parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","title":"Parameterised graph theory for tensor networks: entanglement rerouting, structural simplification, and agnostic tomography","headline":"Graph parameters such as cutwidth and tree-cutwidth are shown to control how expensive it is to rewrite a tensor-network state as an MPS or tree network — and how hard it is to learn that state, even in an agnostic setting.","blurb":"What: The authors prove cutwidth and tree-cutwidth bound bond-dimension overhead when converting a tensor-network state to an MPS or TTN, and they give graph-dependent sample and compute bounds for realizable and agnostic tomography.","url":"https://disruptive-concepts.com/article/parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","html_url":"https://disruptive-concepts.com/article/parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","arxivId":"2609.04165","weekId":"2026-W37","score":78,"topics":["quantum"]},{"id":"subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","slug":"subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","title":"Subcellularly Resolved Single-Cell Embedding Learning with Transcriptomic data, Protein Structure and Localization Information","headline":"A multimodal cell-embedding model mixes RNA profiles with protein sequence and structure, using cross-attention inside subcellular compartments instead of treating each cell as one bag of molecules.","blurb":"What: The authors propose a cross-attention framework that builds cell embeddings from transcriptomics, protein sequences, and protein structures while modeling those signals inside distinct subcellular compartments.","url":"https://disruptive-concepts.com/article/subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","html_url":"https://disruptive-concepts.com/article/subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","arxivId":"2609.02344","weekId":"2026-W37","score":77,"topics":["biotech"]},{"id":"exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","slug":"exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","title":"Exciton multipolarity controls coherent and squeezed phonons in van der Waals heterostructures","headline":"In MoSe2/WSe2 multilayers, dipolar excitons drive coherent interlayer breathing phonons while quadrupolar excitons drive squeezed ones — and a vertical electric field can switch the phonon state.","blurb":"What: Atomistic simulations of alternating MoSe2 and WSe2 layers show exciton multipolarity as a design rule: linear coupling from dipolar excitons yields coherent phonons; quadratic coupling from quadrupolar excitons yields squeezed phonons.","url":"https://disruptive-concepts.com/article/exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","html_url":"https://disruptive-concepts.com/article/exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","arxivId":"2609.03795","weekId":"2026-W37","score":75,"topics":["materials"]},{"id":"muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","slug":"muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","title":"MulDP: Multimodal Diffusion Policy for Autonomous Quadruped Parkour Navigation across Complex Terrains","headline":"MulDP is a diffusion policy that turns vision, proprioception, and goals into anticipatory velocity commands so a quadruped can parkour-navigate without a human calling the high-level plays — trained on a new multimodal parkour dataset.","blurb":"What: The authors couple perception to embodied control with a multimodal diffusion policy that outputs temporally coherent navigation velocities, and they introduce QPND, described as the first quadruped parkour navigation dataset of its kind.","url":"https://disruptive-concepts.com/article/muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","html_url":"https://disruptive-concepts.com/article/muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","arxivId":"2609.03984","weekId":"2026-W37","score":74,"topics":["robotics"]},{"id":"ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","slug":"ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","title":"Ion-Engineered Insulator-to-Semiconductor Transition in Natural 2D Biotite","headline":"A controlled NaOH treatment turns exfoliated biotite nanosheets from an insulator into a tunable 2D semiconductor by adding sodium, defects, and local reconstruction while keeping the layered mineral framework.","blurb":"What: Liquid-phase-exfoliated biotite, after NaOH processing, shows a narrowed optical gap (about 5.2 eV down to 3.2–3.5 eV), new near-IR transitions, nonlinear transport near 10 µA, and ultrafast trap-mediated carrier dynamics, backed by DFT.","url":"https://disruptive-concepts.com/article/ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","html_url":"https://disruptive-concepts.com/article/ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","arxivId":"2609.03757","weekId":"2026-W37","score":72,"topics":["materials"]},{"id":"characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","slug":"characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","title":"Characterizing Large Scale Quantum Systems with Error Per Circuit Layer","headline":"Error Per Circuit Layer (EPCL) estimates how polarizing each layer of a quantum circuit is by overlapping two registers that ran the same random circuits — no classical simulation of ideal outputs and no restriction to Clifford gates.","blurb":"What: EPCL applies identical random circuits to two disjoint registers, measures output-state overlap versus depth, and fits an effective layer polarization under an ensemble-averaged depolarizing model.","url":"https://disruptive-concepts.com/article/characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","html_url":"https://disruptive-concepts.com/article/characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","arxivId":"2609.04132","weekId":"2026-W37","score":71,"topics":["quantum"]},{"id":"a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","slug":"a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","title":"A Low-Cost, Open Platform for End-to-End Autonomous Driving on a Miniature Ackermann Vehicle","headline":"An open miniature-car testbed — physical Ackermann vehicle, printed urban track, tools, and a Webots twin — lets researchers take end-to-end driving from simulation into a real closed loop without a full-size AV budget.","blurb":"What: The authors release a low-cost platform and a command-conditioned behavior-cloning baseline that reaches 6.1 cm mean cross-track error on the real track (human demo 4.7 cm), and they show FOV and sim-to-real data choices decide whether all routes complete.","url":"https://disruptive-concepts.com/article/a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","html_url":"https://disruptive-concepts.com/article/a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","arxivId":"2609.04147","weekId":"2026-W37","score":70,"topics":["robotics"]},{"id":"poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","slug":"poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","title":"PopPert: Population-level Joint-Distribution Modeling for Single-Cell Perturbation Prediction","headline":"PopPert predicts how a whole population of cells changes after a genetic or chemical perturbation by learning joint gene-expression distributions — without pretending destroyed, unpaired cells can be matched one-to-one.","blurb":"What: The method parameterizes population-level joint distributions (via a low-rank Gaussian copula), maps a control population plus a perturbation into new distribution parameters, and can sample synthetic perturbed profiles.","url":"https://disruptive-concepts.com/article/poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","html_url":"https://disruptive-concepts.com/article/poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","arxivId":"2609.01357","weekId":"2026-W37","score":68,"topics":["biotech"]},{"id":"improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","slug":"improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","title":"Improving precipitation forecasts in an AI weather model using observational data","headline":"Fine-tuning a graph-transformer weather model on IMERG precipitation — not just ERA5 — improves medium-range rain scores by up to 19% and extreme-rain skill by 57% globally versus operational models, with a caveat on the very heaviest events.","blurb":"What: The authors fine-tune an AI weather model at 0.25° with IMERG observational precipitation to correct known ERA5 rain biases, gaining CRPS and storm/drizzle skill while a physics operational model stays more reliable on the heaviest rain.","url":"https://disruptive-concepts.com/article/improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","html_url":"https://disruptive-concepts.com/article/improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","arxivId":"2609.03210","weekId":"2026-W37","score":66,"topics":["climate"]},{"id":"robust-pac-learning-of-concurrent-stochastic-games","slug":"robust-pac-learning-of-concurrent-stochastic-games","title":"Robust PAC Learning of Concurrent Stochastic Games","headline":"A PAC algorithm for general-sum concurrent stochastic games with uncertain transitions either returns a near-optimal approximate Nash equilibrium or certifies that no exact equilibrium exists — with polynomial sample complexity under a reachability assumption.","blurb":"What: The authors maintain L1 confidence sets on transition kernels, solve a robust concurrent stochastic game for a social-welfare-optimal ε-Nash equilibrium, and use a Nash-margin test to either output that equilibrium or a sound “no exact NE” certificate.","url":"https://disruptive-concepts.com/article/robust-pac-learning-of-concurrent-stochastic-games","html_url":"https://disruptive-concepts.com/article/robust-pac-learning-of-concurrent-stochastic-games","arxivId":"2609.04189","weekId":"2026-W37","score":65,"topics":["ai"]},{"id":"sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","slug":"sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","title":"Sparse concept attribution for histomorphological hypothesis generation from whole-slide classifiers","headline":"SCOPE explains whole-slide pathology classifiers by sparsely attributing them onto a histomorphological concept bank — and a new benchmark shows sparse attributions recover known morphology while dense ones look like chance.","blurb":"What: The authors pair pathology vision–language models with sparse concept attribution (SCOPE) and introduce MorphoRecoveryBench, seven pathologist-curated tasks where sparse (and a cheaper pooled-embedding variant) beat dense attribution.","url":"https://disruptive-concepts.com/article/sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","html_url":"https://disruptive-concepts.com/article/sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","arxivId":"2609.02985","weekId":"2026-W37","score":63,"topics":["biotech"]},{"id":"detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","slug":"detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","title":"Detection of hydrocarbons in Titan using high-resolution cross-correlation spectroscopy","headline":"Using Titan as a testbed, the authors build high-resolution cross-correlation templates from molecular cross-sections instead of complete line lists — recovering methane and acetylene and making a first HRCCS detection of ethane at SNR about 5.2.","blurb":"What: K-band CRIRES+ spectra of Titan (1.99–2.48 μm) are cross-correlated with both line-by-line and cross-section templates, yielding known CH4 and C2H2 plus C2H6 only from the cross-section templates because no high-resolution ethane line list exists.","url":"https://disruptive-concepts.com/article/detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","html_url":"https://disruptive-concepts.com/article/detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","arxivId":"2609.03975","weekId":"2026-W37","score":59,"topics":["space"]},{"id":"morphology-of-radio-sources-in-representation-space","slug":"morphology-of-radio-sources-in-representation-space","title":"Morphology of Radio Sources in Representation Space","headline":"A classifier ensemble plus representation-space clustering finds six extra radio morphologies in LoTSS-DR3 — including rare winged and FR-ambiguous sources — on top of twelve DR2 classes, in a catalog of more than 13 million objects.","blurb":"What: After applying a fine-tuned ensemble to LoTSS-DR3, the authors cluster low-probability sources in representation space, recover 88% in the old DR2 clusters, and identify representatives of six new morphological classes in the remainder.","url":"https://disruptive-concepts.com/article/morphology-of-radio-sources-in-representation-space","html_url":"https://disruptive-concepts.com/article/morphology-of-radio-sources-in-representation-space","arxivId":"2609.03779","weekId":"2026-W37","score":57,"topics":["space"]},{"id":"ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","slug":"ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","title":"Ruelle--Pollicott Theory for Metastable Systems: A Unified Framework for Tipping Transitions","headline":"A Ruelle–Pollicott spectral theory of stochastic metastable systems shows that classic “critical slowing down” warnings are not universal: large responses need both a small spectral gap and a large residue coupling the observable to the perturbation.","blurb":"What: The authors factor invariant-statistic sensitivity into a spectral denominator and a residue, prove a frozen-spectrum non-normal example can look more alarming than a true gap closure, and separate in-well recovery from interwell escape in a fold tipping problem.","url":"https://disruptive-concepts.com/article/ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","html_url":"https://disruptive-concepts.com/article/ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","arxivId":"2609.02614","weekId":"2026-W37","score":55,"topics":["climate"]},{"id":"computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","slug":"computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","title":"Computing high-order mixed derivatives in physics-informed neural networks using multi-index Bell polynomials","headline":"A tabulated Faà di Bruno / Bell-polynomial recursion computes the mixed derivatives PINNs need — and their parameter gradients — without nested automatic differentiation, through order seven on CPU without the memory failures of nested graphs.","blurb":"What: The authors organize the multivariate Faà di Bruno formula over a downward-closed multi-index set, tabulate Bell-polynomial convolutions once, and backpropagate losses built from any subset of those derivatives, including nonlinear products and coupled fields.","url":"https://disruptive-concepts.com/article/computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","html_url":"https://disruptive-concepts.com/article/computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","arxivId":"2609.03768","weekId":"2026-W37","score":49,"topics":["energy"]}],"papers":[{"id":"legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","slug":"legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","title":"Legibility is Not Interpretability: Comparing Judged and Actual Importance in Chain-Of-Thought Reasoning","headline":"Readable chain-of-thought traces look like a window into why a model answered as it did — but this paper shows that even strong LLM judges cannot reliably tell which steps actually change the chance of a correct answer. A caution for process rewards and critic models, not a product timeline.","blurb":"What: The authors measure a reasoning step’s real importance as its advantage — how much including it changes expected reward, estimated with Monte Carlo rollouts — then test whether LLM judges can spot high-advantage steps from the text alone.","url":"https://disruptive-concepts.com/article/legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","html_url":"https://disruptive-concepts.com/article/legibility-is-not-interpretability-comparing-judged-and-actual-importance-in-chain-of-thought","arxivId":"2609.04194","weekId":"2026-W37","score":93,"topics":["ai"]},{"id":"quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","slug":"quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","title":"Quantum thermalization achieves optimal approximate quantum error correction","headline":"Generic thermalizing many-body dynamics, treated as an encoding, appear to hit the same optimal approximate error-correction limits as Haar-random codes — including a finite-temperature family based on the Scrooge ensemble.","blurb":"What: The authors treat typical late-time thermalizing states as codewords and show that generic thermalization follows a universal encoding-rate/distance/entropy curve that saturates Singleton-type bounds for approximate quantum error correction.","url":"https://disruptive-concepts.com/article/quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","html_url":"https://disruptive-concepts.com/article/quantum-thermalization-achieves-optimal-approximate-quantum-error-correction","arxivId":"2609.04121","weekId":"2026-W37","score":85,"topics":["quantum"]},{"id":"puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","slug":"puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","title":"Puffin-World: Scaling a Unified Multimodal Model with Native 3D World States","headline":"Puffin-World is a single multimodal model that jointly handles physics, geometry, and appearance — plus camera motion — to generate and reconstruct 3D worlds without bolting on separate offline modules.","blurb":"What: The authors introduce a unified architecture that models gravity/latitude, depth, and images together with an Omni-Camera representation, trained on a 16-million-example mix of vision-language-camera triplets and trajectories.","url":"https://disruptive-concepts.com/article/puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","html_url":"https://disruptive-concepts.com/article/puffin-world-scaling-a-unified-multimodal-model-with-native-3d-world-states","arxivId":"2609.04196","weekId":"2026-W37","score":83,"topics":["ai"]},{"id":"gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","slug":"gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","title":"GIFT: Guided Intermediate Feature Training via Action-Oriented Structural Supervision for Robotic Manipulation","headline":"GIFT trains a robot policy’s intermediate features to keep geometry, affordances, and goals — the structure that actually matters for control — and shows consistent gains across three action formulations on LIBERO-Plus and RoboCasa.","blurb":"What: The authors name an action-sufficiency gap — rich visual features that still omit control-relevant structure — and add training-time constraints for geometry alignment, affordance prediction, and goal-region reconstruction.","url":"https://disruptive-concepts.com/article/gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","html_url":"https://disruptive-concepts.com/article/gift-guided-intermediate-feature-training-via-action-oriented-structural-supervision-for","arxivId":"2609.04193","weekId":"2026-W37","score":82,"topics":["robotics"]},{"id":"tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","slug":"tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","title":"TokenMatch: 3D Mesh Correspondence Transformer with Curvature-Guided Tokenisation","headline":"TokenMatch is a feed-forward transformer that matches 3D meshes — including hard partial, non-isometric cases — by tokenizing surfaces with curvature-guided patches, and it generalizes from partial training data to full shapes without a fine-tune.","blurb":"What: Trained only on BeCoS, a difficult partial-to-partial non-isometric set, the model uses self- and cross-attention over curvature-guided mesh patches to predict dense correspondences and still matches full shapes at sub-second speed.","url":"https://disruptive-concepts.com/article/tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","html_url":"https://disruptive-concepts.com/article/tokenmatch-3d-mesh-correspondence-transformer-with-curvature-guided-tokenisation","arxivId":"2609.04202","weekId":"2026-W37","score":80,"topics":["ai"]},{"id":"parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","slug":"parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","title":"Parameterised graph theory for tensor networks: entanglement rerouting, structural simplification, and agnostic tomography","headline":"Graph parameters such as cutwidth and tree-cutwidth are shown to control how expensive it is to rewrite a tensor-network state as an MPS or tree network — and how hard it is to learn that state, even in an agnostic setting.","blurb":"What: The authors prove cutwidth and tree-cutwidth bound bond-dimension overhead when converting a tensor-network state to an MPS or TTN, and they give graph-dependent sample and compute bounds for realizable and agnostic tomography.","url":"https://disruptive-concepts.com/article/parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","html_url":"https://disruptive-concepts.com/article/parameterised-graph-theory-for-tensor-networks-entanglement-rerouting-structural-simplification","arxivId":"2609.04165","weekId":"2026-W37","score":78,"topics":["quantum"]},{"id":"subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","slug":"subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","title":"Subcellularly Resolved Single-Cell Embedding Learning with Transcriptomic data, Protein Structure and Localization Information","headline":"A multimodal cell-embedding model mixes RNA profiles with protein sequence and structure, using cross-attention inside subcellular compartments instead of treating each cell as one bag of molecules.","blurb":"What: The authors propose a cross-attention framework that builds cell embeddings from transcriptomics, protein sequences, and protein structures while modeling those signals inside distinct subcellular compartments.","url":"https://disruptive-concepts.com/article/subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","html_url":"https://disruptive-concepts.com/article/subcellularly-resolved-single-cell-embedding-learning-with-transcriptomic-data-protein","arxivId":"2609.02344","weekId":"2026-W37","score":77,"topics":["biotech"]},{"id":"exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","slug":"exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","title":"Exciton multipolarity controls coherent and squeezed phonons in van der Waals heterostructures","headline":"In MoSe2/WSe2 multilayers, dipolar excitons drive coherent interlayer breathing phonons while quadrupolar excitons drive squeezed ones — and a vertical electric field can switch the phonon state.","blurb":"What: Atomistic simulations of alternating MoSe2 and WSe2 layers show exciton multipolarity as a design rule: linear coupling from dipolar excitons yields coherent phonons; quadratic coupling from quadrupolar excitons yields squeezed phonons.","url":"https://disruptive-concepts.com/article/exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","html_url":"https://disruptive-concepts.com/article/exciton-multipolarity-controls-coherent-and-squeezed-phonons-in-van-der-waals-heterostructures","arxivId":"2609.03795","weekId":"2026-W37","score":75,"topics":["materials"]},{"id":"muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","slug":"muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","title":"MulDP: Multimodal Diffusion Policy for Autonomous Quadruped Parkour Navigation across Complex Terrains","headline":"MulDP is a diffusion policy that turns vision, proprioception, and goals into anticipatory velocity commands so a quadruped can parkour-navigate without a human calling the high-level plays — trained on a new multimodal parkour dataset.","blurb":"What: The authors couple perception to embodied control with a multimodal diffusion policy that outputs temporally coherent navigation velocities, and they introduce QPND, described as the first quadruped parkour navigation dataset of its kind.","url":"https://disruptive-concepts.com/article/muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","html_url":"https://disruptive-concepts.com/article/muldp-multimodal-diffusion-policy-for-autonomous-quadruped-parkour-navigation-across-complex","arxivId":"2609.03984","weekId":"2026-W37","score":74,"topics":["robotics"]},{"id":"ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","slug":"ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","title":"Ion-Engineered Insulator-to-Semiconductor Transition in Natural 2D Biotite","headline":"A controlled NaOH treatment turns exfoliated biotite nanosheets from an insulator into a tunable 2D semiconductor by adding sodium, defects, and local reconstruction while keeping the layered mineral framework.","blurb":"What: Liquid-phase-exfoliated biotite, after NaOH processing, shows a narrowed optical gap (about 5.2 eV down to 3.2–3.5 eV), new near-IR transitions, nonlinear transport near 10 µA, and ultrafast trap-mediated carrier dynamics, backed by DFT.","url":"https://disruptive-concepts.com/article/ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","html_url":"https://disruptive-concepts.com/article/ion-engineered-insulator-to-semiconductor-transition-in-natural-2d-biotite","arxivId":"2609.03757","weekId":"2026-W37","score":72,"topics":["materials"]},{"id":"characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","slug":"characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","title":"Characterizing Large Scale Quantum Systems with Error Per Circuit Layer","headline":"Error Per Circuit Layer (EPCL) estimates how polarizing each layer of a quantum circuit is by overlapping two registers that ran the same random circuits — no classical simulation of ideal outputs and no restriction to Clifford gates.","blurb":"What: EPCL applies identical random circuits to two disjoint registers, measures output-state overlap versus depth, and fits an effective layer polarization under an ensemble-averaged depolarizing model.","url":"https://disruptive-concepts.com/article/characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","html_url":"https://disruptive-concepts.com/article/characterizing-large-scale-quantum-systems-with-error-per-circuit-layer","arxivId":"2609.04132","weekId":"2026-W37","score":71,"topics":["quantum"]},{"id":"a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","slug":"a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","title":"A Low-Cost, Open Platform for End-to-End Autonomous Driving on a Miniature Ackermann Vehicle","headline":"An open miniature-car testbed — physical Ackermann vehicle, printed urban track, tools, and a Webots twin — lets researchers take end-to-end driving from simulation into a real closed loop without a full-size AV budget.","blurb":"What: The authors release a low-cost platform and a command-conditioned behavior-cloning baseline that reaches 6.1 cm mean cross-track error on the real track (human demo 4.7 cm), and they show FOV and sim-to-real data choices decide whether all routes complete.","url":"https://disruptive-concepts.com/article/a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","html_url":"https://disruptive-concepts.com/article/a-low-cost-open-platform-for-end-to-end-autonomous-driving-on-a-miniature-ackermann-vehicle","arxivId":"2609.04147","weekId":"2026-W37","score":70,"topics":["robotics"]},{"id":"poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","slug":"poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","title":"PopPert: Population-level Joint-Distribution Modeling for Single-Cell Perturbation Prediction","headline":"PopPert predicts how a whole population of cells changes after a genetic or chemical perturbation by learning joint gene-expression distributions — without pretending destroyed, unpaired cells can be matched one-to-one.","blurb":"What: The method parameterizes population-level joint distributions (via a low-rank Gaussian copula), maps a control population plus a perturbation into new distribution parameters, and can sample synthetic perturbed profiles.","url":"https://disruptive-concepts.com/article/poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","html_url":"https://disruptive-concepts.com/article/poppert-population-level-joint-distribution-modeling-for-single-cell-perturbation-prediction","arxivId":"2609.01357","weekId":"2026-W37","score":68,"topics":["biotech"]},{"id":"improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","slug":"improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","title":"Improving precipitation forecasts in an AI weather model using observational data","headline":"Fine-tuning a graph-transformer weather model on IMERG precipitation — not just ERA5 — improves medium-range rain scores by up to 19% and extreme-rain skill by 57% globally versus operational models, with a caveat on the very heaviest events.","blurb":"What: The authors fine-tune an AI weather model at 0.25° with IMERG observational precipitation to correct known ERA5 rain biases, gaining CRPS and storm/drizzle skill while a physics operational model stays more reliable on the heaviest rain.","url":"https://disruptive-concepts.com/article/improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","html_url":"https://disruptive-concepts.com/article/improving-precipitation-forecasts-in-an-ai-weather-model-using-observational-data","arxivId":"2609.03210","weekId":"2026-W37","score":66,"topics":["climate"]},{"id":"robust-pac-learning-of-concurrent-stochastic-games","slug":"robust-pac-learning-of-concurrent-stochastic-games","title":"Robust PAC Learning of Concurrent Stochastic Games","headline":"A PAC algorithm for general-sum concurrent stochastic games with uncertain transitions either returns a near-optimal approximate Nash equilibrium or certifies that no exact equilibrium exists — with polynomial sample complexity under a reachability assumption.","blurb":"What: The authors maintain L1 confidence sets on transition kernels, solve a robust concurrent stochastic game for a social-welfare-optimal ε-Nash equilibrium, and use a Nash-margin test to either output that equilibrium or a sound “no exact NE” certificate.","url":"https://disruptive-concepts.com/article/robust-pac-learning-of-concurrent-stochastic-games","html_url":"https://disruptive-concepts.com/article/robust-pac-learning-of-concurrent-stochastic-games","arxivId":"2609.04189","weekId":"2026-W37","score":65,"topics":["ai"]},{"id":"sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","slug":"sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","title":"Sparse concept attribution for histomorphological hypothesis generation from whole-slide classifiers","headline":"SCOPE explains whole-slide pathology classifiers by sparsely attributing them onto a histomorphological concept bank — and a new benchmark shows sparse attributions recover known morphology while dense ones look like chance.","blurb":"What: The authors pair pathology vision–language models with sparse concept attribution (SCOPE) and introduce MorphoRecoveryBench, seven pathologist-curated tasks where sparse (and a cheaper pooled-embedding variant) beat dense attribution.","url":"https://disruptive-concepts.com/article/sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","html_url":"https://disruptive-concepts.com/article/sparse-concept-attribution-for-histomorphological-hypothesis-generation-from-whole-slide","arxivId":"2609.02985","weekId":"2026-W37","score":63,"topics":["biotech"]},{"id":"detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","slug":"detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","title":"Detection of hydrocarbons in Titan using high-resolution cross-correlation spectroscopy","headline":"Using Titan as a testbed, the authors build high-resolution cross-correlation templates from molecular cross-sections instead of complete line lists — recovering methane and acetylene and making a first HRCCS detection of ethane at SNR about 5.2.","blurb":"What: K-band CRIRES+ spectra of Titan (1.99–2.48 μm) are cross-correlated with both line-by-line and cross-section templates, yielding known CH4 and C2H2 plus C2H6 only from the cross-section templates because no high-resolution ethane line list exists.","url":"https://disruptive-concepts.com/article/detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","html_url":"https://disruptive-concepts.com/article/detection-of-hydrocarbons-in-titan-using-high-resolution-cross-correlation-spectroscopy","arxivId":"2609.03975","weekId":"2026-W37","score":59,"topics":["space"]},{"id":"morphology-of-radio-sources-in-representation-space","slug":"morphology-of-radio-sources-in-representation-space","title":"Morphology of Radio Sources in Representation Space","headline":"A classifier ensemble plus representation-space clustering finds six extra radio morphologies in LoTSS-DR3 — including rare winged and FR-ambiguous sources — on top of twelve DR2 classes, in a catalog of more than 13 million objects.","blurb":"What: After applying a fine-tuned ensemble to LoTSS-DR3, the authors cluster low-probability sources in representation space, recover 88% in the old DR2 clusters, and identify representatives of six new morphological classes in the remainder.","url":"https://disruptive-concepts.com/article/morphology-of-radio-sources-in-representation-space","html_url":"https://disruptive-concepts.com/article/morphology-of-radio-sources-in-representation-space","arxivId":"2609.03779","weekId":"2026-W37","score":57,"topics":["space"]},{"id":"ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","slug":"ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","title":"Ruelle--Pollicott Theory for Metastable Systems: A Unified Framework for Tipping Transitions","headline":"A Ruelle–Pollicott spectral theory of stochastic metastable systems shows that classic “critical slowing down” warnings are not universal: large responses need both a small spectral gap and a large residue coupling the observable to the perturbation.","blurb":"What: The authors factor invariant-statistic sensitivity into a spectral denominator and a residue, prove a frozen-spectrum non-normal example can look more alarming than a true gap closure, and separate in-well recovery from interwell escape in a fold tipping problem.","url":"https://disruptive-concepts.com/article/ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","html_url":"https://disruptive-concepts.com/article/ruelle-pollicott-theory-for-metastable-systems-a-unified-framework-for-tipping-transitions","arxivId":"2609.02614","weekId":"2026-W37","score":55,"topics":["climate"]},{"id":"computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","slug":"computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","title":"Computing high-order mixed derivatives in physics-informed neural networks using multi-index Bell polynomials","headline":"A tabulated Faà di Bruno / Bell-polynomial recursion computes the mixed derivatives PINNs need — and their parameter gradients — without nested automatic differentiation, through order seven on CPU without the memory failures of nested graphs.","blurb":"What: The authors organize the multivariate Faà di Bruno formula over a downward-closed multi-index set, tabulate Bell-polynomial convolutions once, and backpropagate losses built from any subset of those derivatives, including nonlinear products and coupled fields.","url":"https://disruptive-concepts.com/article/computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","html_url":"https://disruptive-concepts.com/article/computing-high-order-mixed-derivatives-in-physics-informed-neural-networks-using-multi-index","arxivId":"2609.03768","weekId":"2026-W37","score":49,"topics":["energy"]}],"x402":{"payTo":"0xe1946c7A27afc4Afc92690cC5c79b2771F85a0c1","facilitator":"https://facilitator.openx402.ai","network":"eip155:8453","prices":{"paper":"$0.01","article":"$0.02","score":"$0.01","feed":"$0.05","weekBundle":"$0.25"},"docs":"https://docs.openx402.ai/","protocol":"https://x402.org/","usdcBase":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"},"relatedCatalogs":[],"related_catalogs":[]}