Papers

The goal of this work is to calculate the mapping class groups of solenoidal manifolds. We introduce the dynamical commensurator group associated to an odometer action, and develop methods to calculate the mapping class group using the dynamical commensurator. Our results generalize the previous results of Odden for the mapping class group of the universal solenoid over a Riemann surface, and of Bering and Studenmund for the pointed mapping class group of the universal solenoid over a compact aspherical manifold. We include a selection of examples.

Embedding materials in optical cavities has emerged as a powerful strategy for controlling their physical properties. In this context, increasing attention has been devoted to the thermal Purcell effect, which describes how the radiative heat transfer between a material and its photonic environ- ment can be modified by the electromagnetic boundary conditions imposed by the cavity. Here, we investigate near-field effects in cavity-mediated radiative heat transfer in planar cavities composed of realistic, dissipative gold mirrors, extending previous work that considered ideal mirrors. We iden- tify the conditions under which near-field effects lead to a substantial enhancement of heat transfer between the material and a photonic thermal bath relative to free space, owing to the strong modi- fication of the photonic local density of states by evanescent modes within the cavity. In particular, we find that radiative heat flow can be enhanced by several orders of magnitude in cavities ranging from nanometer to micrometer scales, depending on the absorber frequency and dipole orientation. These findings suggest that near-field effects may be relevant to polariton chemistry experiments and other systems operating within this cavity length-scale regime. Our results add to the idea that cavities can modify material properties not only by engineering vacuum fluctuations in the strong light-matter coupling regime, but also by controlling radiative dissipation and heat transfer.

In 2019, Liu proved that the cop number of any $(P_k, claw)$-free graph is at most $k-3$. In 2026, Clow and Meger improved and generalized the result of Liu by showing that every connected $(P_k, E)$-free graph has cop number at most $\lceil \frac{k-1}{2} \rceil + 3$, where the graph $E$ is obtained by subdividing two edges of a claw. In this paper, we show that every $(P_k, claw)$-free graph $G$ has cop number at most $\lceil \frac{k-1}{2} \rceil$. This improves the result of Clow and Meger on claw-free graphs. We further show that this bound can be improved to $\lceil \frac{k}{3}\rceil$, when $G$ contains a hole of length at least $k-1$. We show that the cop number of any $(P_k, fork)$-free graph is at most $\lceil \frac{2(k-1)}{3} \rceil$. This also improves the result of Clow and Meger on fork-free graphs for some values of $k$. We further show that every $(P_7, fork)$-free graph has cop number at most $3$, and this bound is tight.

A comprehensive understanding of complex transition-metal oxides is key to designing next-generation spintronic materials with tailored magnetic and transport functionalities. We investigate the interplay among structural disorder, frustrated magnetism, and localized charge transport in the 3d-4d hybrid double perovskite Sr2FeRuO6. The compound adopts a monoclinic structure with significant B-site antisite disorder that drives its unconventional behavior. Magnetic measurements uncover a robust Griffiths phase extending well above the ordering temperature, evidencing stable short-range ferromagnetic clusters within a disordered matrix. Electrical transport remains strongly localized, transitioning from small-polaron hopping at high temperatures to variable-range hopping at intermediate temperatures. These results position Sr2FeRuO6 as a representative 3d-4d hybrid system in which quenched disorder stabilizes spin inhomogeneity, while polaronic and hopping transport are also observed.

Identification of optically active defects in hexagonal boron nitride (hBN) remains challenging because different defect candidates can produce similar calculated photoluminescence (PL) lineshapes. Here, we demonstrate that force-constant embedding methodology can be extended to calculate high-resolution, vibrationally resolved PL lineshapes in the polar layered material hBN. Benchmarking against the carbon dimer defect, responsible for the 4.1 eV emission, we show that the methodology accurately reproduces the low-energy acoustic phonons and fine spectral features that are inaccessible when using direct supercell calculations. Applying the methodology to neutral carbon trimers ($\mathrm{C_2 C_B}$ and $\mathrm{C_2 C_N}$), we find that the calculated PL lineshape and excitation energy of $\mathrm{C_2 C_N}$ closely match the low-temperature spectrum of the emitter with a zero-phonon line at 1.54 eV. These results establish high-resolution phonon-sideband modeling as a powerful tool for discriminating among candidate defects in polar materials.

Impulsive control, in the context of continuous-time Markov decision processes (CTMDPs), refers to actions that instantaneously update the state of the process, leading to theoretical complications. Previous work has overcome these complications, but at the expense of approachability and ease of implementation: those models are mathematically complex and incompatible with pre-existing solution methods. This paper presents an alternative formulation of impulsive control for CTMDPs that is easier to employ in the modelling of real-world problems and can be solved by pre-existing solution methods. This formulation is then extended to allow for multiple objectives and decomposability. A decomposition-based heuristic method is developed that leverages Dantzig-Wolfe decomposition to find approximate Pareto fronts for large multi-objective decomposable impulsive CTMDPs. The modelling framework and solution methodology are applied to a dynamic maintenance problem to demonstrate their computational efficiency, solution space coverage, and solution quality.

We study the route-set allocation problem between end nodes in capacity-constrained flow networks with transfer costs. We introduce an origin–destination (OD) bidding language, in which a participant bids for bounded conic combinations of inlet–outlet–quantity licences and the mechanism performs the routing through a network-flow optimising winner-determination problem. We then study payment rules for the OD mechanism. Under a restricted model with openly and uniformly priced sources and bidders who are indifferent to the source of the gas, the Vickrey–Clarke–Groves (VCG) payment is core-stable; the key structural fact is that in this regime some optimal aggregate flows never use a physical pipe in both directions, so the winner-determination problem reduces to a single-commodity network flow whose value is submodular in the set of active bidders. If source access is allowed to depend on the individual consumer – a realistic feature, e.g. when gas deals are embedded in bilateral political bargaining – the reduction fails and VCG payments may lie strictly outside the core. For this general model we propose a dual (nodal) pricing rule derived from the shadow prices of the flow-conservation constraints, and prove it is always core-stable, at the cost of dominant-strategy incentive compatibility. Illustrative examples accompany each result.

Human mesh recovery (HMR) provides a geometric foundation for understanding human motion, but accurately reconstructing both body motion and fine hand articulation from videos remains challenging, especially in real-world scenes where occlusion and partial visibility are common. Video generation pretraining offers rich spatial and temporal priors for this task, but its focus on visual synthesis leaves a gap between these representations and explicit human geometry. We present DreamHMR, a general framework that adapts video generation priors for complete body-and-hand reconstruction. It follows a global-to-local strategy, first grounding video representations in human geometry and then refining body and hand detail with reliable local evidence. Global Body Representation Transfer trains the video backbone to predict complete human renderings, turning its native prediction task into a means of learning surface correspondence and posed geometry. Local Joint Feature Enhancement combines contextual estimates with reliable local evidence to recover fine body and hand detail, while limiting interference from occluders and background. Experiments on five datasets with four video backbones demonstrate strong reconstruction performance across backbone families. Under a unified evaluation protocol, our best model outperforms the best existing methods on all reported metrics of 3DPW, EMDB-1, and RICH, reducing body MPJPE by 4.2%, 4.7%, and 6.1%, respectively. In more challenging occlusion settings on held-out BEDLAM and EgoBody sequences, DreamHMR reduces overall body and hand MPJPE by 7.0% and 10.0%, respectively. Our code will be made publicly available.

In this paper, we solve the direct monodromy problem for the Borel-summed $0$-parameter solution of Painlevé~I. Using the auxiliary time equation, we first construct formal WKB wave matrices whose amplitudes are regular at the double turning point. Rigorous Borel summation then yields exact WKB matrices in each Stokes region. Uniform estimates on moving action strips provide a common exact wave matrix in a full neighbourhood of the double point, proving that all four incident jumps are identities for consistent column normalisations. We formulate the resulting normalised Riemann–Hilbert problem and prove its existence and uniqueness on the Borel domains of the construction. Finally, after Painlevé scaling, the five Borel-summed solutions coincide with the five rotated tritronquée solutions of Painlevé~I.

Large language models can reduce vocabulary mismatch in sparse retrieval by generating alternative views of a query. Once these views are searched independently, however, their ranked lists may be complementary, redundant, or affected by query drift, making fixed fusion suboptimal. We present XViewRank, a lightweight retrieval-side method for conditionally fusing BM25 results from an original query, a hypothesis view, and a step-back view. For each pooled candidate, XViewRank retains three normalized BM25 scores and three query-level overlaps between the retrieved result sets. A small nonlinear scorer uses the overlaps as context that changes how the view-specific scores are interpreted, without jointly encoding query and passage text. Under a matched-input protocol, XViewRank improves nDCG@10 and MAP numerically over reciprocal rank fusion on both TREC DL 2019 and 2020, with a statistically significant MAP gain on DL19. Additional analyses show that the clearest benefit occurs when the view rankings have low agreement, while the evidence on DL20 is weaker. The results support overlap-conditioned score fusion as a compact retrieval-side alternative to fixed fusion, while also delimiting where its advantage is empirically established.

Rapid progress in AI agents and robotics makes it a matter of the near future that humans are served by a team of robots. Such a team will receive several goals at once, each specified in whatever form is convenient for the human, such as a category name, a photo, or a spoken description. Existing benchmarks study multi-agent, multi-modal, and multi-object goal navigation in isolation, and a benchmark that places all three in one episode is missing. To fill this gap, we introduce HM3D-M3. A team of K agents starts at distributed locations in a photorealistic indoor scene and searches for an unordered set of M object goals, each given as an open-vocabulary category, an instance image, or an instance description. HM3D-M3 provides 1,000 validation episodes over 3,926 goals from 379 open-vocabulary categories, 100K observation-assignment training samples, and two team-level metrics, goal-level SR and makespan-normalized SPL. We further contribute two baselines. HiGA is a trained hierarchical goal assigner on top of a frozen open-vocabulary navigation policy, and LaViRA-M3 is a training-free team of VLM agents with shared mapping and progress analysis. Adapted multi-robot and zero-shot methods reach 16–27% SR, while our baselines reach 38–41% with complementary strengths and distinct failure modes. Furthermore, we deploy LaViRA-M3 on a heterogeneous three-robot team. Project page: https://robo-lavira.github.io/hm3d-m3/

We present several advances on neural operators by viewing the action of operator layers as the minimizers of Bregman regularized optimization problems over Banach function spaces. The proposed framework allows interpreting the activation operators as Bregman proximity operators from dual to primal space. This novel viewpoint is general enough to recover classical neural operators as well as a new variant, coined Bregman neural operators, which includes the inverse activation operator and features the same expressivity of standard neural operators. Numerical experiments support the added benefits of the Bregman variant of Fourier neural operators for training deeper and more accurate models.

Stellar line-of-sight velocity distributions are a powerful tool for studying galaxy kinematics and formation history, with multiple-peaked LOSVDs often indicating the presence of complex kinematic components. We extract stellar LOSVDs across the disc of a simulated spiral galaxy viewed edge-on, and study how often and why we see multiple-peaked LOSVDs. We find that the multiple-peaked LOSVDs are mainly found close to the midplane because they result from the presence of spiral arms. This is not because the spiral arms have significantly different kinematics to the disc, but because they correspond to over-densities. Variations in the number of arms present along a line-of-sight create variations in the number of peaks in LOSVDs. This can be used to quantify the strength of the spiral arms above the midplane, and how the thickness of spiral arms vary across the disc. We find that the commonly used fourth order Gauss-Hermite moment, $h_4$, is negative in the regions where there are multiple-peaked LOSVDs, and hence traces spiral arms. Finally, we discuss the observability of the multiple-peaked regions and their associated $h_4$ measurements with current integral field spectrographs. We find that the spectral resolution of MUSE might not be high enough to uncover multiple peaks in LOSVDs when they result from spiral arms, but should allow for identification of the regions of negative $h_4$. On the other hand, higher resolution spectrographs like KOALA and WEAVE may be able to recover multiple peaks in LOSVDs, representing a promising avenue for characterising spiral arms in edge-on disc galaxies.

Multi-agent path finding (MAPF) solvers are evaluated on the schedules they produce, not on how well those schedules can be executed. Specifically, continuous-time optimal solvers can achieve even lower cost than their discrete counterparts by scheduling agents with near-zero temporal/spatial separation, and by exploiting graphs with higher connectedness; however, such schedules may prove hard to execute under a realistic controller. We integrate AOC-CBS, an anytime-optimal continuous-time MAPF solver, into a fleet coordination framework in which each agent tracks its timed path using model predictive control, and we evaluate the resulting system on congested MovingAI benchmark instances with 21 to 40 agents. Without intervention, only a fraction of the instances complete execution. We then apply T-robustness by adding a time-margin, i.e., a minimum temporal gap between agents at shared locations, as well as spatial inflation of the agent footprint. Increasing either margin raises execution success, for a relatively small increase in the schedule's sum-of-costs; it also improves how closely agents adhere to the schedule: when tracking each agent's position throughout execution and comparing it against the position it should occupy according to the schedule, the mean discrepancy falls significantly as the margin increases. In general, the two margins have different effects; temporal margin decreases deviation, giving the controller room to stay on schedule. Spatial margin affects deviation less and instead increases the clearance between agents, so that the same deviation no longer leads to a blocking stall. They are nonetheless substitutes rather than complements: spatial inflation is almost free in schedule cost, but its effect on execution success is small and does not compound with a temporal margin already in place.

KANLab is a browser-based visualization tool designed to interactively teach the core concepts of Kolmogorov-Arnold Networks (KAN), forked from the widely successful TensorFlow Playground. To maximize educational clarity, KANLab omits residual connections and learnable scaling parameters found in modern KAN implementations, ensuring that displayed B-spline activation curves directly correspond to their underlying trainable control points. Users can inspect and modify activation functions, initialization schemes and observe gradient descent correcting manual disruptions. Furthermore, user can explore phenomena such as activation variance, interactive pruning, and re\-/training within a compact 2D machine learning setting. KANLab is open-source and available at https://kanlab.app/

Long coding agent sessions outgrow the model's context window, so agents compact: older history is replaced by something shorter, usually a model-written summary. A summary selects and rewords, and the agent cannot see what was lost. Existing work has explored a range of compaction techniques, but comparatively little attention has been paid to their effectiveness with lightweight models, including those deployed locally or used for high-volume workloads. Here, we present transparent-compact, a simple deterministic method for the open-source Pi coding agent, designed specifically for use with such models and with the purpose of maximising the transparency of the agent's view of its own compacted context. It guarantees that user and assistant prose in its context is unchanged by a model or algorithm, and explicitly flags where compaction has stripped information. We compare it with Pi's default model-written summary compaction and with pi-vcc, a popular deterministic method with similar methods but without verbatim retention of prose. Our benchmarks run on four models of varying performance levels, with emphasis on Qwen3.8-27B: a lightweight model at the top end of what consumer hardware can serve locally. On sets probing commitments (rules the user set which exist only in the conversation) and recall of facts, transparent-compact matched or outperformed both alternatives. In particular, it showed an advantage in avoiding confident blunders, i.e. assertion of false information as opposed to recognizing that something was not recalled. Even with a frontier model, its advantage was present but smaller. We argue that transparent-compact spares lightweight models from reasoning or searching their way around what compaction lost, and results in improved recall and application of information as a result, without the need for task or context-specific structuring of the session history.

We introduce a novel extension of reflection calculus, $\mathsf{RC_{BI}}$, to analyze a family of extensions of $\mathsf{ACA}_0$ generated by uniform $\omega$-model and syntactic reflection principles, together with several other related operators. The union of these theories coincides with the theory of Bar Induction ($\mathsf{BI}$). We prove that $\mathsf{RC_{BI}}$ is arithmetically sound, and that it is sound and complete with respect to a natural class of Kripke models. Leveraging these modal semantics, we demonstrate that the class of theories denoted by $\nabla$-free closed formulas is a pre-wellordering with respect to $\Pi^1_1$-soundness and consistency strength. This allows us to establish an ordinal analysis for $\mathsf{BI}$ and its fragments, analogous to Beklemishev's analysis of $\mathsf{PA}$. The structural properties of $\mathsf{RC_{BI}}$ naturally give rise to a novel ordinal notation system for the Bachmann-Howard ordinal which differs substantially from existing representations.

Speech Emotion Recognition (SER) models and Audio LLMs fail when vocal tone contradicts textual semantics. Extensive training on congruent data engenders a rigid semantic prior, causing "Semantic Dominance", a severe text bias during cross-modal conflicts. We introduce the ASPIRE benchmark of adversarial audio-text pairs across four conflict types, alongside two metrics: Semantic Overconfidence Penalty (SOP) and Latent Decoupling Degree (LDD). Unlike fusion methods that merely re-weight collapsed features, our Acoustic Conflict Resolution Network (ACR-Net) uses Cross-Modal Attention and Contrastive Decoupling Loss to disentangle contradictory representations into orthogonal latent spaces. Experiments show existing models fail under polarity conflicts due to extreme semantic overconfidence. Conversely, ACR-Net preserves acoustic fidelity, maintaining high LDD to drive SOP near zero, achieving superior Acoustic Accuracy (ACC) in both adversarial and standard scenarios.

Let $\mathbb{F}$ be a finite field and let $G$ be a group such that every finite subgroup of $G$ is cyclic. In this paper we classify the division gradings by cyclic groups on matrix algebras over $\mathbb{F}$. This yields the classification of the $G$-gradings on these algebras. Moreover if $G$ is abelian we prove that two matrix algebras over $\mathbb{F}$ with $G$-gradings satisfy the same graded identities if and only if they are isomorphic as graded algebras.

Prior work on implicit bias shows that gradient descent selects the Frobenius max-margin direction on linearly separable problems, whereas Fan et al. (2025) established spectral max-margin behavior for spectral-norm steepest descent methods, including idealized Muon with exact orthogonalization. In practice, however, Muon approximates orthogonalization using finite Newton-Schulz iterations and a positive stabilization constant. We prove that these implementation choices cause Muon to select the Frobenius rather than the spectral max-margin direction in linear classification with separable data. We also establish this behavior for both exponential moving average and Nesterov-style momentum. The key technical ingredient is a summable perturbation argument showing that the resulting parameter trajectories remain at bounded distance from those of gradient descent. We then disentangle the roles of Newton-Schulz iteration count, the stabilization constant, and the Newton-Schulz polynomial, revealing distinct asymptotic behaviors, including directions that achieve neither Frobenius nor spectral max-margin. Finally, we show that Muon can recover its spectral max-margin bias with Newton-Schulz iterations growing logarithmically over training.

Distilling pretrained video diffusion models into few-step autoregressive (AR) generators typically combines student initialization with distribution matching distillation (DMD). An intermediate AR teacher can provide causally compatible targets for ODE initialization. However, causally compatible targets alone do not ensure high-quality generation under a short sampling schedule. We analyze the posterior-mean predictions favored by direct latent regression and empirically find that causal ODE initialization alleviates blur but still leaves fine detail unresolved. To address this limitation, we propose Representation Forcing, which learns few-step causal initialization through representation-space supervision rather than intermediate-teacher trajectory regression. The student is trained with teacher forcing by comparing clean predictions and training targets through normalized features of a frozen video diffusion transformer. This phase reuses the synthetic videos generated by a bidirectional model for AR teacher training and replaces both teacher training and causal ODE distillation, without generating additional causal ODE trajectories. The subsequent asymmetric DMD procedure remains unchanged. Empirical results show that, at four sampling steps, RepForcing better preserves generation quality than intermediate-teacher-based initialization before DMD. After DMD, our four-step chunk-wise model improves VBench Total from 84.04 to 84.66 (+0.62 points) and VisionReward from 6.326 to 10.074 (+59.25%) over the reported causal-ODE-based baseline.

We study the representation of bilinear forms associated with Riesz fractional gradients by symmetric integral kernels. Motivated by [SS18, Open Problem 1.10], we give a complete description of the constant coefficient case. We prove that any symmetric representing kernel is uniquely determined and we derive its explicit formula in terms of the constant matrix $A$. This yields a sharp characterization of the matrices for which the kernel is either nonnegative or uniformly positive. In particular, [SS18, Open Problem 1.10] has a negative answer (even for constant matrices). We also identify a class of nonconstant antisymmetric divergence free perturbations that are invisible to the fractional bilinear form and hence leave the representing kernel unchanged. The general case of nonconstant matrices exhibits a substantially different behavior and will be the main subject of our forthcoming paper [DSV26].

Data movement between memory and processing units remains a major bottleneck in data-centric systems. Memristive Stateful Logic (SL) addresses this problem by performing Boolean operations directly on stored resistive states. However, arithmetic efficiency depends jointly on the device-supported logic primitive set, the dependency structure of the resulting operation sequence, and its mapping to datapath architectures. While prior designs typically optimize device, logic, and architecture layers separately, we jointly consider them to co-design a stateful adder. We experimentally demonstrate a mixed MAGIC and FELIX logic primitive set on printed Electrochemical Metallization (ECM) memristors and derive a full adder sequence optimized for inter-bit carry propagation. Mapping this sequence to bit-serial, bit-parallel, and bit-pipelined datapaths reduces the latter two to 2n+6 cycles for an n-bit addition. For 32-bit addition, this requires 60-88% fewer cycles than prior SL adders. Across four representative workloads, we achieve 3.7x-16.4x speedup over prior. These results show how cross-layer co-design can translate into improved workload-level performance.

Neural symbolic regression (SR) models require diverse datasets of mathematical expressions for robust training. However, existing datasets rely predominantly on stochastic sampling methods, which suffer from high structural redundancy and incomplete functional coverage. To address this, we introduce an exhaustive dataset of univariate canonical skeletons generated via a deterministic constrained context-free grammar. A canonical skeleton acts as a generic structural representation that subsumes more specific functional forms, eliminating redundant expressions and avoiding repeated computational efforts. Although our dataset consists of univariate skeletons, it is designed to support the decomposable SR paradigm to solve multivariate SR problems without assuming linear separability. Furthermore, we pair this dataset with a sampling strategy that generates clean numerical data without triggering numerical instabilities or relying on artificial protected operators. Canonicalizing two existing neural SR datasets reveals that our dataset is substantially more compact, eliminates structural repetitions, and achieves a uniform operator distribution free of simplicity bias. Finally, we establish a benchmark of 150 SR problems across three tiers, on which state-of-the-art SR methods are shown to struggle significantly.

Web agents built on vision language models decide every action by generating text such as click(12), although most of that text encodes a choice among a few action types and a few dozen marked elements. We read these choices directly from the hidden states that the generator computes while it reads the goal and the screenshot. Plug-in decision heads on a frozen generator predict the action type, pool the visual tokens inside each numbered box to pick the target, and point at a listed text option or at a WRITE slot that leaves the writing of new text to the generator. A confidence cascade executes the heads' action when they are sure and otherwise lets the generator decode from the KV cache of the same forward pass. With Qwen3-VL 4B and 8B generators trained on 213k web steps, a strict threshold keeps the cascade between 0.3 points below and 1.4 points above the generator on seven offline test sets while the heads decide 12 to 64% of the steps. A looser threshold raises step success on unseen Mind2Web websites by 3.1 and 1.8 points, costs up to 2 points on GUIAct and MiniWoB++, and lifts WebArena task success from 12.5 to 13.8% (4B) and from 13.6 to 14.6% (8B) with less than half of the generated tokens. Training the heads jointly with the generator lowers the generator's own step success by up to 4.9 points; heads on a frozen generator avoid this cost.

We approximate the solution operator $(a,f)\mapsto u$ of $-\operatorname{div}(a\nabla u)=f$ by operator networks whose only primitives are heat-semigroup functions of a fixed reference operator and pointwise products. The main result is an emulation theorem on strongly local Dirichlet spaces with a spectral gap and a Bakry–Émery bound $\mathrm{BE}(-K,\infty)$: an explicit scheme reaches $\dot H^1$-error $\varepsilon$ uniformly over coefficients with bounded carré du champ and bounded generator image and over $L^2$ data, with parameters depending only on the coefficient bounds, the accuracy and $K$; on the torus this gives a $C^2$-type coefficient class with constants independent of the dimension. The torus version refines this to Wiener classes of fractional regularity $A^{1+\theta}$, with an explicit rate attained by the scheme's product-defect block. The scheme unrolls the Laplacian-preconditioned Richardson iteration, controls each inexact step in the operator norm of $\dot H^1$ without Galerkin truncation, and synthesises the coupling $\nabla a\cdot\nabla u$ from the product defect of one heat layer. All parameters are explicit; the results are approximation theorems for continuum operators.

Modern audio watermarks achieve strong robustness with learned encoder-decoder models, but their behavior under new processing is established empirically. CochlearMark is introduced as a training-free audio watermark that uses auditory-nerve phase-locking as its feature representation. A small waveform perturbation is optimized against a keyed random projection of this representation, and the same projection is used for detection, with no trained encoder or decoder. Because phase-locking represents temporal fine structure, the resulting robustness has a direct physiological interpretation: the mark is retained when this structure is preserved and lost when it is disrupted or rebuilt. Across speech, music, and environmental audio, CochlearMark achieves signal-edit robustness and imperceptibility comparable to trained neural baselines while requiring only a lightweight computational footprint. Among the evaluated training-free methods, CochlearMark is the only one to retain substantial detectability under neural-codec re-encoding.

Autoregressive transformers trained on limit order book (LOB) message streams are increasingly proposed as foundation models for market microstructure. Because the generative process of real markets is unknown, these models can be judged only by the realism of their output, leaving open what they actually learn. We make the question answerable by inverting the setup: we train GPT-style models on order flow from simulators whose latent parameters are known, so that the Bayes-optimal predictor, the generating state, and the exact data entropy serve as references. Measured against them, next-token pretraining on order flow implements amortized Bayesian inference over market regimes. The learned predictive matches the exact Bayes optimum to within a millinat per token and approaches the oracle floor even on a nonstationary family whose latent state drifts within an episode, a regime that requires filtering. The latent regime is linearly decodable from frozen activations, well above baselines, so a frozen-model read-out calibrates simulator parameters more accurately than windowed maximum likelihood at a fraction of the cost. In our experiments these capabilities are not produced by scale: capacity and token volume saturate early, and it is the diversity of training regimes, not the token count, that determines generalization to unseen markets. On real NASDAQ data, zero-shot transfer is set by how well the simulator prior covers real flow: an enriched mixture nearly closes the gap to an in-domain baseline, and pretraining beats from-scratch training at every budget across held-out tickers. Known ground truth turns average next-token loss, which is uninformative near the entropy floor, into an exact comparison against the Bayes-optimal and oracle predictors.

Maximizing the absorption of exciton-polaritons in transition-metal dichalcogenides (TMDCs) under critical coupling conditions is crucial for developing high-efficiency polariton-based photodetectors, solar cells, and perfect absorbers. Here, we propose a polaritonic critical self-hybridized coupling between Brillouin zone folding driven quasi-bound states in the continuum (BZF-QBIC) mode and A excitons in a bulk WS2 metasurface. This polaritonic critical coupling shows robust behavior with respect to the asymmetric parameters of the involved BZF-QBIC mode, in contrast to the asymmetry-sensitive conventional QBIC-based critical coupling. By changing the height of WS2 metasurfaces, we observe a typical anti-crossing behavior with a Rabi splitting of 246.5 meV. Moreover, the maximum absorption value of 0.5 for exciton-polaritons is achieved at the polaritonic critical coupling condition. The large population of excitons provided by bulk WS2 is essential for achieving polaritonic critical coupling and can further enhance polaritonic absorption. This work enables the realization of robust, high-absorption self-hybridized polaritonic critical coupling in TMDCs, and further expedite their applications in high-absorption low-dimensional polaritonic devices.

We investigate two aspects of multilingual pre-training with limited data, namely how language exposure is distributed across English, Dutch, and Chinese, and how that exposure is ordered during training. In particular, we compare balanced multilingual exposure to language-dominant exposure, where one language is exposed with more training data than the others. We also compare mixed training, where multilingual training data is used throughout training, with curriculum training, where monolingual exposure of the dominant language is followed by balanced multilingual exposure. Our results show that multilingual BabyLM performance varies with both the amount and order of language exposure. Mixed training outperforms curriculum learning across all 15 matched configurations, indicating that simultaneous multilingual exposure is more effective than staged exposure in our setting. Several language-dominant mixed training configurations achieve higher scores than the balanced configuration, suggesting that a balanced distribution is not necessarily the highest-performing configuration in our setting. Interestingly, the Dutch-primary configurations achieve the best overall performance.