Papers

LLM-agent security has produced a dense landscape of defenses - prompt hardening, content filters, permission gates, sandboxes - yet no framework tells a deployer what a defense actually guarantees, or where that guarantee comes from. We apply Verification Autonomy Levels (VAL) - L0: LLM self-declaration; L1: deterministic rules; L2: objective ground truth; L3/L4: decidable completeness; L5: impossible - to 22 agent-security defenses; the taxonomy is falsifiable (10/10 prediction hits on frozen cards, flagged). We run the first controlled deployment-value comparison: at equal budget, a VAL-guided stack (confirmation gate + schema sandbox) versus a mainstream intuition stack (prompt hardening + keyword filter), 50 scenarios, 12 attack variants, adaptive/white-box/PAIR escalation (~7,000 testbed calls; ~10,000 harness calls on AgentDojo/JADE). The VAL stack holds 0.000 attack success at 1.000 benign success (0.5% ASR at 79.7% utility on AgentDojo banking vs 4.3% undefended); the intuition stack reaches 0.000 ASR but kills all benign actions - security by model-behavior luck, not structure. Across testbeds of rising attack-surface hardness the intuition stack's zero drifts (0->1.9%->6.2%, n=16 on JADE) while the VAL stack's holds within its ODD (0->0->0), its only breach a disclosed out-of-ODD password gap (0.5%). The same zero, two different guarantees: zero is an outcome, not a guarantee.

In this article, we study a weighted analogue of the $k$-plane transform, where the weight on the $k$-dimensional measure is given by a radial power of distance. Particularly, we first consider the existence of such operators on Lebesgue spaces imbibed with radial power weights. Then, we proceed to see the mapping properties of the said transform when it acts on radial functions. In this regard, we prove certain weighted $L^p$-improving boundedness and end-point Lorentz space estimates. Finally, we look at the weighted $L^p$-$L^p$ boundedness of the operator acting on general functions, and evaluate its operator norm.

Medical Vision-Language Models (Med-VLMs) have demonstrated strong capabilities in clinical tasks. However, they often struggle to understand anatomical structures and spatial positioning, which are crucial for medical reasoning. To address this, we propose a localization-aware enhancement to the Med-VLM pipeline, introducing improvements at three levels: data,architecture, and alignment. First, we introduce localization lens, a set of expert-validated representations that provide richer anatomical and positional context. However, as these representations increase input complexity, we integrate pixel shuffle within the model architecture to filter and refine representations, enhancing spatial information processing while preserving anatomical continuity. Lastly, to effectively align the localization lens representations with textual features, we incorporate decoupled contrastive loss (DCL) alongside the standard loss function. This ensures better feature discrimination and robustness, particularly in data limited medical settings. Through extensive evaluations on medical visual question answering (Med-VQA) datasets, we show that our methodology improves localization-driven performance across different Med-VLM architectures. Our analysis of localization-based questions further reveals that improvements in anatomy and spatial reasoning directly enhance the overall accuracy of Med-VQA upto 6.2%. The proposed approach is model-agnostic and can be seamlessly integrated into existing Med-VLM pipelines. The dataset, code, and trained models will be made publicly available at https://github.com/CVLABLUMS/localizationlens.

In this contribution we consider the singular periodic problem \[ u''+k\,u=g(u)+e(t),\quad u(0)=u(2\pi),\,\,u'(0)=u'(2\pi), \] where $k\in\R,$ $e:\J\to\R$ is Lebesgue integrable and $g:\I\mapsto\R$ is continuous, but can have a singularity at $x=0.$ In particular, I wish to revisit our earlier result regarding the existence of at least two its solutions and present a proof that is somewhat more transparent than the original one.

Adding expressive and environmental controls to a speech generator requires learning heterogeneous attributes without losing earlier capabilities. VoiceWeaver addresses this problem with structured label prefixes and staged emotion-tone-event training in a shared text-audio model. Replay-based distillation retains earlier predictions, while embedding decorrelation and attribute dropout regularize conditioning. Evaluation separates single-attribute correctness, joint emotion-event generation, and three-attribute correctness. Rounded to whole percentage points, emotion accuracy is 86% in Chinese and 83% in English. Chinese emotion-event joint accuracy is about 70%, versus 56% for mixed-task training and 48% for Ming-omni-tts; English joint accuracy is 66%. Three-attribute accuracy is 59% in Chinese and 55% in English on 500 samples each (57% pooled). Text error rates exceed those of the external TTS baselines. Audio samples are available at https://anonymous.4open.science/api/repo/VoiceWeaver1-4D88/file/index.html.

Radiology report generation (RRG) aims to convert medical images into diagnostic texts to assist in clinical decision-making and alleviate the workload of physicians. Although existing methods have made extensive progress in cross-modal interaction and the incorporation of external priors, the distribution shift of underlying representations and the undifferentiated rigid coupling of heterogeneous information cause weak visual abnormality cues to be easily diluted by massive text priors and generation inertia during decoding. To overcome this bottleneck, inspired by cognitive science, we propose a novel two-stage Homogeneous Semantic Alignment and Hierarchical Expert Routing (HSA-HER) framework. First, the model introduces an explicit homogeneous distribution constraint in the underlying latent space to effectively eliminate the cross-modal distribution shift between visual and textual features, thereby extracting purified visual features as semantic anchors that accurately align with diseases. Second, for heterogeneous clinical evidence composed of visual features, local entities, and global retrievals, we design a hierarchical expert routing mechanism guided by these disease semantic anchors. This mechanism abandons the undifferentiated rigid coupling paradigm. Specifically, it dynamically activates expert networks to perform targeted mining and semantic reconstruction on multi-source evidence, and adaptively allocates fusion weights. Extensive experiments on three mainstream benchmark datasets demonstrate that HSA-HER achieves state-of-the-art performance, accurately depicting complex imaging details and key diagnostic information.

We study how the ranks of factors in a positive semidefinite slack factorization constrain its size. We bound the number of low-rank vertex factors in any fixed factorization, and use this to determine the asymptotic order of the minimum factorization size under any fixed bound on vertex-factor ranks: $\Theta(n\log n)$ for permutahedra $\Pi_n$ and $\Theta(\log N)$ for regular $N$-gons. In the rank-one case this gives the order of the minimum dimension of a space of real functions on the vertices in which every facet slack is a sum of squares, with no symmetry or degree restriction. For unrestricted lifts, the bound gives $\mathrm{xc}_{\mathrm{PSD}}(\Pi_n)\ge n+\log_3 n-O(1)$, and shows that a permutahedron factorization of linear size would need vertex factors of rank $\Omega(\log n)$ at all but a vanishing fraction of vertices. We also prove that every polytope of real positive semidefinite rank at most four has at most twelve vertices, and construct explicit size-four lifts for every square-symmetric octagon, so eight is attained. The upper bound combines a count of the polygon corners reached by curves of rank-one factors with incidence constraints on factor ranks. Whether eight is the maximum remains open.

Exoplanetary astronomy has entered its Golden Age, with humanity having found over 6000 exoplanets to date. The majority of these discoveries have been achieved through transit-photometry and radial-velocity (RV) techniques. While transit surveys lead in discovery numbers, the RV method persists as a vital and indispensable technique in exoplanetary science, particularly in the most prevalent stellar population – M dwarfs. Their low masses make them the theoretical favourites for RV detections, since exoplanet signals are stronger over the same period, and are therefore easier to detect and measure accurately. Nevertheless, the stellar activity M dwarfs exhibit introduces significant challenges, particularly when seeking RV semi-amplitudes to the order of 1 m$\,$s$^{-1}$, or lower. Stellar activity is known to mimic or suppress planetary signals, and its effects manifest across all time scales: from short-lived signals due to convective motion (lasting minutes to days), through active regions on the stellar surface (days to months), to long-term magnetic cycles (years to decades). These issues highlight the urgent need for robust methodologies that are capable of reliably disentangling stellar variability from genuine planetary signals. This doctoral dissertation explores these challenges through the application of Gaussian-process (GP) modelling. I present three distinct but interconnected publications that integrate high-precision spectroscopy and photometry with a unified GP-based approach in two cases, and that enable robust disentanglement of planetary signals from stellar activity. Together, these three studies illustrate how observations of different origin can be combined with appropriate statistical machinery for the sake of detecting exoplanets, with the possibility of interpreting stellar activity.

Symmetry has long provided the organizing principle for understanding how materials respond to external perturbations. Here we show that what is permitted by symmetry need not be what actually emerges from the underlying system, and this distinction becomes particularly consequential for symmetry-related gapless Dirac fermions such as in altermagnets. A perturbation that shifts the Dirac points drives a Lifshitz reconstruction and generates an intrinsically nonanalytic effective potential, giving rise to undiscovered singular responses. In Dirac altermagnets, this nonanalyticity acquires a magnetic manifestation through $C$-paired spin-polarized Dirac cones: breaking the pairing symmetry $C$ lifts their degeneracy and converts the associated Lifshitz reconstruction into an itinerant magnetization. As a consequence, piezomagnetism, the induction of a net magnetization by a mechanical strain, becomes nonlinear and nonanalytic, with $M\propto\mathrm{sgn}(\varepsilon)\varepsilon^2$, whereas the linear response is absent despite being allowed by symmetry. Accordingly, the leading response cannot be inferred from the conventional symmetry-based paradigm of material response built upon the cornerstones of Neumann's principle and analytic Landau expansion. First-principles calculations identify the monolayer Fe$_2$S as a candidate material realization for such singular magnetic responses. These results suggest quantum singularities as a new degree of freedom for engineering material responses.

Nowadays, emojis are often replacing words. Yet computational systems still oversimplify them. Most existing approaches treat emojis as static sentiment indicators and overlook their emotional distributions. In this study, we propose an emoji-aware emotion analysis framework based on a Twitter (X) dataset of 100,000 emoji-containing replies collected between 2020-2025. After preprocessing and text cleaning, we applied text-to-emotion classification to detect five primary emotions (Happy, Angry, Sad, Fear, and Surprise) for each message. By aggregating emotion scores across contexts in which each emoji appears, we estimate emoji-emotion association distributions and construct an emoji-emotion ranking system reflecting relative emotional dominance. Furthermore, we project emojis into the Russell valence-arousal space to enable continuous affective interpretation. Our results demonstrate that emojis exhibit probabilistic, context-sensitive emotional profiles rather than fixed sentiment polarities.

Testing deep reinforcement learning (DRL) agents in cyber-physical systems aims to uncover diverse failures before deployment, but each execution can be expensive. Surrogate-assisted testing reduces this cost by learning to predict which test configurations are likely to fail. Prior surrogates treat the system as a black box and predict pass or fail outcomes directly; we instead model how a test unfolds and estimate failure from an imagined episode. We introduce DreamTest, a world-model surrogate for testing DRL agents. DreamTest adapts a recurrent state-space model to learn agent behaviour and environment dynamics from the agent's training log. Given a candidate configuration, imagined rollouts produce a failure score that guides search without executing every candidate in a simulator or real system. We evaluate DreamTest for failure prediction, test generation, and failure diversity on Parking, Humanoid, and DonkeyCar. Mean area under the precision-recall curve (AUPRC) exceeds the strongest baseline by 97%, 12%, and 39%, respectively, and gains on five out-of-distribution test sets reach 145%, 29%, and 44%. Under the same simulator-validation budget, the best "DreamTest + search" combinations find 29%, 22%, and 79% more novel failures on average. Across clusterings with k = 2-40, failures generated with DreamTest cover the most behavioural clusters for almost all k, indicating that DreamTest consistently discovers behaviourally diverse failures.

Studying attention through the lens of interacting particle dynamics has shown how token clustering can emerge from the underlying dynamics. We extend this perspective to slot attention, a method for object-centric image segmentation and representation learning in which learned components obscure how much of the clustering behaviour is intrinsic to the attention dynamics. We therefore introduce simplified slot attention (SSA), a parameter-free variant whose dynamics are connected to soft $k$-means clustering and which provides a straightforward mechanistic explanation for the emergence of object-centric representations. On the Pascal VOC dataset, SSA achieves performance comparable to that of slot attention, demonstrating that competitive object-centric segmentation can be achieved without learned neural-network components.

Harnessing integrability at arbitrary interaction strength, we build long-lived and spatially localized soliton-like excitations in the 1D Bose gas with contact repulsion (Lieb-Liniger model). We show how these density profiles, built from selective superpositions of multi-hole eigenstates, display long lifetimes tunable through parameters like dispersion width and number of holes. We further illustrate a scaling protocol that in principle enables unbounded increase in lifetimes in the thermodynamic limit. Our results showcase classical-like excitations originating from unitary quantum dynamics, away from the semi-classical regime, and in regions where interactions cannot be treated perturbatively.

We construct an explicit infinite virtually free group $G=C_4\ast_{C_2}D_3$ with a finite inverse-closed generating set whose Cayley graph is $2$-geodetic but not geodetic. More precisely, we describe the full language of geodesics and show that every element of $G$ has a unique geodesic representative except for a single element, which has exactly two. This gives a counterexample to the conjecture that an infinite group admitting a $k$-geodetic Cayley graph must admit a geodetic Cayley graph with respect to the same generating set. We further prove that $G$ is freely indecomposable and hence non-plain. Consequently, Shapiro's conjecture that every geodetic group is plain and the conjecture that every hyperbolic $k$-geodetic group is geodetic cannot both hold.

As large language models (LLMs) grow in parameter count, model storage and parameter memory traffic have become major bottlenecks to efficient deployment. Codebook-based weight quantization reduces these costs, but imbalanced nearest-codeword assignments during fitting can leave some codewords insufficiently updated, limiting effective codebook utilization. To address this limitation, we propose Balanced Assignment Refinement for Quantization (BARQ), which improves quantization quality through balanced fitting of existing codebooks. Specifically, we first compute joint soft assignments between weight blocks and codewords through entropically regularized optimal transport with uniform marginals and curvature-weighted reconstruction costs, ensuring equal positive fitting mass for every codeword in the exact solution. We then refine the codewords through an assignment-weighted barycentric update, which we prove minimizes the fitting objective for fixed assignments. For finite Sinkhorn iterations, the implemented update retains this optimality provided all codeword masses exceed the denominator floor. Finally, we discard the soft assignments and use the refined codebook for standard hard nearest-codeword encoding, with our analysis establishing sufficient conditions for reducing hard-quantization distortion and evaluation loss. Across multiple LLMs, BARQ achieves lower perplexity and higher mean zero-shot accuracy than the evaluated baselines at comparable bit budgets. The code is available at https://github.com/chenhangcuisg-code/BARQ.

Understanding parameter functions helps elucidate the internal mechanisms of large language models (LLMs). However, how to connect parameters from different modules to verifiable output effects and further characterize the relationship between their functional organization and model capability remains to be explored. To this end, we introduce the downstream vocabulary lens (DV-Lens), a parameter-level interpretability framework that links native parameter directions to their downstream vocabulary responses. Specifically, we first estimate module-specific downstream Jacobians over a reference prompt set for attention query, key, value, and output (Q/K/V/O) projections and feed-forward networks (FFNs). Second, we use these mappings to project native parameter columns into the final vocabulary space, obtaining signed readouts that characterize their average local output responses. Third, we group parameter columns by their vocabulary readouts and introduce downstream vocabulary complexity (DV-Complexity), which quantifies within-group structural variation using normalized reconstruction residuals of the original weights. At the parameter level, randomized controls and finite-difference tests show that DV-Lens readouts capture non-random vocabulary structure and predict local logit changes with 98.0% coordinate-orientation agreement across 720 cases from nine models. These readouts further guide parameter ablation, steering, and swapping across 21 models, shifting target-token probabilities in the predicted directions under controlled conditions. At the model level, the joint-parameter score of DV-Complexity achieves a Spearman correlation of 0.904 with benchmark-based capability rankings across 48 language models. Together, these results provide intervention-based evidence for DV-Lens interpretations and reveal an association between DV-Complexity and model capability.

Although reinforcement learning (RL) post-training repairs the localized segmental errors of zero-shot text-to-speech (TTS), arriving at a working recipe still relies on tedious manual tuning, and whether LLM agents can take over this research pipeline is unclear. We investigate this question with AgenticTTS-Forge, a collaborative workflow that structures human guidance and agentic execution around a shared workspace, applied to CosyVoice2-0.5B. To measure what the agent automates, we audit its trajectory stage by stage against the published recipe. To measure what it exploits, we score its policies with held-out observers hidden from the agent. Our results show that the agent recovers an underspecified recipe, improves it, and, when gains stall, surveys the literature unprompted and pivots from the LM carrier to the flow carrier, halving Bad cases. However, its autonomy exposes three traps across the data, proxy, and algorithm axes: the held-out set leaks through a channel the contract never reads, a self-shaped reward inflates the proxy where it is scored, and separately tuned policies do not compose additively. These findings show that the binding constraint is measurement rather than reasoning, and can inform the design of harnesses whose contracts read every channel the agent does.

In this paper we prove that $B_1(X)$ has a dense cofinally Polish subspace if and only if $X$ is a $\Delta_1$-space. Also we prove that $B_1(X, [0,1])$ has a dense cofinally Polish subspace if and only if $X$ is a weak $\lambda$-space. We also establish that $C_p(X, [0,1])$ is pseudocompact if and only if $C_p(X,[0,1])$ has a dense cofinally Polish subspace. This result provides answers to four open questions from (Tkachuk in RACSAM 115(68), 2021).

The Zel'dovich-inspired (ZI) transform is a local nonlinear transform of the density field that can Gaussianize the late-time matter field and recover information from higher-order $N$-point correlation functions through its power spectrum, with the parameter $\eta$ controlling the weighting of higher-order information. Using Fisher-matrix analyses of halo catalogs from the Quijote and Quijote-PNG simulations, we study the redshift-space monopole power spectrum of the ZI-transformed halo density field. The joint statistic $P_{\rm ZI,0}$, constructed from three representative transforms with $\eta=1$, $10$, and $+\infty$, yields substantially tighter constraints than the conventional halo power spectrum $P_{\rm h,0}$ on standard cosmological parameters, the summed neutrino mass, and primordial non-Gaussianity amplitudes, while outperforming the marked power spectrum for many parameters. We further calibrate a scale-dependent noise correction for the ZI power spectrum, with an effective noise amplitude that depends on wavenumber, and find that the correction further tightens the constraints. In particular, the noise-corrected $P_{\rm ZI,0}$ improves the constraints on $f_{\rm NL}^{\rm equil}$, $f_{\rm NL}^{\rm ortho}$, $M_\nu$, and $n_{\rm s}$ by factors of $5.6$, $5.5$, $5.2$, and $5.8$, respectively, relative to $P_{\rm h,0}$. These results demonstrate the potential of ZI power spectra as a complement to standard two-point statistics for extracting nonlinear cosmological information. Our forecasts may be conservative because the halo catalogs have substantially lower number densities than those expected from forthcoming galaxy surveys.

AI is not only a technological artefact; it is also a new site of political economy, legal struggle, and sovereignty contestation. AI systems depend on the continuous collection, storage, processing, and monetisation of large volumes of data. In South Africa, these datasets increasingly include behavioural, biometric, financial, consumer, health, and public-sector information. Our article uses data colonisation as a theoretical lens to analyse whether South Africa's existing legal and governance framework can regulate AI-specific risks. It argues that South Africa's current laws, including the Protection of Personal Information Act (POPIA), regulate personal information but do not adequately prevent data colonisation. POPIA remains important for privacy, lawful processing, automated decision-making, and cross-border transfers. However, data colonisation cannot solely be reduced to privacy protection. It concerns ownership, dependency, infrastructure, surveillance, regulatory power, value extraction, and the capacity of people and institutions to challenge algorithmic decisions. The article examines South African AI-related risks in various areas (e.g. cloud and digital infrastructure, biometric surveillance, and digital markets). This paper argues that South Africa's legal framework does not yet adequately regulate algorithmic opacity, externally controlled digital infrastructure, biometric surveillance, automated profiling, collective data harms, or the extraction of value from South African data. We argue for the need of a decolonial AI governance framework based on algorithmic impact assessments, public-sector AI registers, independent audits, biometric safeguards, competition-law remedies, public-interest data governance, stronger regulatory capacity, and democratic control to strengthen South Africa's digital future.

We give a first-order Frobenius description of conjugate minimal immersions in the round three-sphere and use it to define the conjugate of an arbitrary Jacobi function. The construction is intrinsic at the linearized level and does not require the Jacobi function to be integrable through a normal deformation by minimal surfaces. We then describe how the boundary conditions are exchanged under conjugation. Along a great-circle boundary arc, a Dirichlet Jacobi function has, after a suitable ambient gauge, a conjugate Jacobi function with vanishing Neumann data. Conversely, along a reflection curve, vanishing Neumann data yield vanishing Dirichlet data for the conjugate modulo a Jacobi–Killing function. We also construct conjugate Jacobi functions for conjugate CMC~$1$ immersions in $\mathbb{R}^3$ and exhibit cylindrical counterexamples to both pure boundary exchanges, even modulo ambient Killing fields.

In this paper we use the topological vertex formalism to derive a structural formula for the leading terms of the generating series of Gromov–Witten invariants of local $\mathbb P^2$. Then we give two applications of this formula. First we use this result to derive an explicit formula for the leading terms of the transformed Gopakumar–Vafa invariants of local $\mathbb P^2$. This is an extension of a formula of Guo–Zhou \cite{gz} from degree $2d-5$ to $3d-1$, and shares some similar features with the famous Göttsche–Yau–Zaslow formula. And as the second application, we prove Guo–Wu–Moreira's Conjecture \cite{gw} about the leading Betti numbers of the moduli space of one-dimensional Gieseker semistable sheaves on $\mathbb P^2$.

To address the dual challenges of limited transmission distance and wireless communication security in maritime communications, a directional modulation network enhanced by an elevatable and rotatable antenna array is investigated in this paper. Specifically, the base station, whose array and antenna elements can be rotated and whose antenna height can be adjusted, transmits information to legitimate users while resisting eavesdropping. Moreover, transceiver hardware impairments are taken into account. Accordingly, a worst-case secrecy rate maximization problem is formulated under the joint optimization of precoding vectors, array orientation, antenna orientations, and array height. To solve this non-convex optimization challenge, an alternating optimization scheme and a Transformer-based learning scheme are proposed for the single-user and multi-user scenarios, respectively. Simulation results validate the effectiveness of the proposed algorithms. Security performance gains can be obtained after optimizing the array height, array orientation, and antenna orientations. In addition, compared with the isotropic antenna benchmark, the Transformer-based learning scheme can achieve approximately 90% power savings, 87% antenna savings, and serve more legitimate users.

Pulsar glitches are thought to involve sudden unpinning and fast transport of large numbers of pinned vortices, possibly triggered by breaking of the neutron star crust solid. Such a vortex avalanche can take place on a fast timescale only if there are pathways for fast vortex motion. Formation of local vortex “traps” (regions containing larger than average density of pinned vortices) will provide vortex-free regions surrounding the traps, which can connect to form pathways for fast vortex motion. We propose a simple two-dimensional model for the neutron star crust superfluid, with the vortex pinning landscape tiled into traps and surrounding vortex-free regions which can connect to azimuthal pathways for fast vortex motion. The trap and tile sizes are estimated from the physics of the crust solid and vortex pinning, namely the critical strain angle of the crust solid and the critical velocity for vortex unpinning. Unpinned vortices move with the superfluid flow in the azimuthal direction while also scattering radially outward, by a small angle, due to mutual friction. The model qualitatively explains the glitch parameters and their evolution in young pulsars and provides an explanation for the bi-modality in the distribution of glitch magnitudes. Laboratory experiments and simulations on vortex systems with spatially inhomogeneous vortex pinning landscapes are likely to be more realistic models for pulsar glitches. Breaking the spatial scale-invariance and monitoring the results may also lead to better understanding of collective behaviour in pinned vortex systems in general.

Trusted Execution Environments (TEEs) prove the integrity of code and data to a remote peer through remote attestation: the TEE issues an attestation report by signing, with a hardware-bound key, its configuration information, including measurements (cryptographic digests) of the code loaded inside it. A verifier appraises the report by matching these measurements against reference values. When the measured artifact is proprietary, however, the verifier cannot derive the reference values itself: it can only confirm that something with a given measurement is running inside the TEE—not that the program behaves as expected—forcing full trust in the reference value provider. We propose an architecture that closes this gap without disclosing the artifact, composing two TEE workloads: attestable build (Hugenroth et al., 2025), which binds a source code digest to a build artifact measurement, and attestable audit (proposed in this paper), which binds the same source code digest to the verdicts of automated audits—fuzzing, static analysis, AI code auditing, formal methods—executed inside a TEE. Both are instances of Confidential Computing Proofs: TEE attestation viewed as a hardware-backed zero-knowledge proof. Chaining the two certificates lets the verifier conclude that the running artifact was built from source code satisfying the audited properties, without disclosing either the source code or the build artifact. A proof-of-concept implementation on Intel SGX with the Gramine Library OS, running the Bandit security analyzer on Python code, demonstrates the practicality of the approach.

Partial-spoof detectors must reject synthetic content while accepting benign edits. We diagnose temporal anchors and boundary-consistency training using frozen WavLM features, nonlinguistic unit controls, and source-locked evaluation. Genuine-genuine and genuine-fake splices contrast editing false alarms with synthetic-content misses; they do not isolate a unique causal artifact. With layer-6 features, phone units have higher PartialSpoof evaluation frame EER than soft frames (11.18% versus 9.59%). In 500 retrospective PS-eval cases, augmentation reduces genuine-splice false alarms; synthetic-core misses increase at 5% PS-development FPR but not demonstrably at 1%. A 200-case Llama A follow-up retains the false-alarm reduction but does not establish a miss-rate increase. Constructed-case ranking can improve while fixed-threshold misses rise. Consistency gives no uniform gain, and encoder-layer rankings change across corpora. We report frame and event localization with three seeds, including word units, and treat PS evaluation as retrospective. Audit code, selected training scripts, and result summaries are available at https://github.com/mysxs/partial-spoof-diagnostics.

The notions of classical two-scale convergence and the periodic unfolding method, which both rely on an additional scale for the representation of periodic micro-oscillations, are in this paper traced back to the wider concept of general two-scale convergence. Moreover, a comparison is made between very weak two-scale convergence and periodic unfolding with respect to sequences bounded in Sobolev spaces, amplified by the scale of the micro-oscillations. It turns out that, for amplified sequences, these approaches are not equivalent in an immediate sense. The paper both serves as a literature overview, with examples and illustrations of the concepts of convergence, and contains new results.

We extend the parallel integration method for mixed towers of Part~II from elementary antiderivatives to antiderivatives in special functions. The class covered is the incomplete gamma function $\Gamma(s,\cdot)$ at rational $s$, which contains $\Ei$, $\li$, $\Si$, $\Ci$, $\erf$ and the Fresnel integrals, together with the elliptic integrals $F$, $E$, $\Pi$. Each special function enters through a kernel: a known element of the tower whose antiderivative is that function. It is either fixed by residues or added as one more column of the single linear system, and no Risch differential equation is solved. We prove where kernels can have poles; the $\erf$ kernels live in the sub-critical window of radical towers. The denominator theory, degree bounds and certificates of Part~II carry over, and new criteria decide most places that Part~II leaves to a guess. Elliptic integrals are carried by the radical, and a non-torsion residue divisor becomes a third-kind term. We prove that special functions are introduced only when necessary. Elementary answers are returned unchanged, and in strict mode a special function comes with a certificate that the integrand has no elementary integral. When no complete answer is found, a partial answer with a reduced remainder is returned. All examples are computed by a SymPy implementation and verified by differentiation.

The Wen-Zee term is a mixed gauge-curvature coupling in the topological Chern-Simons description of a quantum Hall fluid, encoding the electric charge bound to spatial curvature. We investigate a Lorentz-covariant generalization of this coupling, expanding its scope to black-hole spacetimes and determine its contribution to horizon entropy. Such a coupling arises naturally in three-dimensional Chern-Simons gravity based on the Maxwell algebra. We show that its Wald entropy contribution is a horizon boost charge, given by a loop integral of the additional gauge field projected along the tangent to the bifurcation circle. This contribution can be nonzero even when the mixed term vanishes on shell. We show that the integral admits an Abelian holonomy interpretation when the projection direction is covariantly constant. We evaluate the charge on a flat cosmological solution and, using smooth horizon fields and a specified thermal ensemble, on an AdS-Lorentz black hole. These results connect the geometric response familiar from quantum Hall physics to gravitational horizon thermodynamics.

Predicting cellular responses to genetic perturbations is a central capability for virtual cells and a key step toward computational modeling of biological interventions. Most existing models directly map an unperturbed molecular profile and perturba- tion to the resulting observation without explicitly representing the latent cellular transition induced by the intervention. We introduce a mechanism-centric virtual cell world model that represents cellular state as a set of Latent Mechanism Units (LMUs) and treats genetic perturbations as actions on these latent states. Each LMU combines a reusable identity grounded in multimodal biological evidence with an observation-specific state, allowing a perturbation to induce mechanism- specific stochastic transitions before decoding the resulting transcriptional response. We train VCLMU through two-stage pretraining, first on around 200K pseudo-bulk perturbation profiles and then on gene-aligned single-cell perturbation data. Across six perturbation-disjoint benchmarks, VCLMU consistently improves perturbation- specific response recovery over strong baselines while maintaining competitive global response accuracy. We further analyze learned LMUs through enrichment between perturbation responses and LMU gene sets and show that they capture structured biological response programs. These results support mechanism-level latent state transition as a useful formulation for virtual cell models that aim to predict and interpret cellular responses to biological interventions.