arXiv:2607.13089v1 Announce Type: new
Abstract: This paper reviews the various reconstructions of Konrad Zuse's Z3 computer, built in 1941. The first is the reconstruction made by Zuse himself in the 1960s, which is now housed at the Deutsches Museum in Munich. Second is the 2001 reconstruction, made in Berlin, and which has a didactic purpose. Third is the Z3r, a Z3 reconstruction built by Horst Zuse in honor of his father's 100th birthday. Fourth is the reconstruction by Christoph Traber in Switzerland. All of these machines strive to preserve the cultural heritage that the Z3 represents.
Science Journals
arXiv:2607.13686v1 Announce Type: new
Abstract: We study the contextual combinatorial semi-bandit (CCSB) problem with general reward function approximation. At each round, the learner observes a context, selects a combinatorial action consisting of a subset of basic arms, and receives the reward of each selected arm; the goal is to maximize the cumulative reward over time. We propose SquareCB.Comb, a computationally efficient algorithm that, at each round, solves a convex optimization problem to sample a combinatorial action that balances exploration and exploitation. SquareCB.Comb scales to large arm sets and imposes no structural assumptions on the action set beyond a cardinality bound of $m$ on each combinatorial action. We prove that SquareCB.Comb achieves a minimax optimal regret bound of $O(\sqrt{m A T \log |\mathcal{F}|})$, where $A$ is the number of arms, $m$ is the maximum number of arms in a combinatorial action, $T$ is the time horizon, and $\mathcal{F}$ is the reward function class. In the realizable setting, this bound matches the state-of-the-art regret guarantees achieved by policy search-based algorithms in the more restricted slate recommendation settings, while simultaneously generalizing to arbitrary combinatorial action structures and general reward function approximation.
arXiv:2607.13111v1 Announce Type: new
Abstract: Distinguishing semantic-preserving commits from changing ones remains an open challenge in software repository mining. While existing approaches detect refactoring commits accurately, they cannot ensure that a commit is purely semantic-preserving, without any interleaving behaviour-changing modification. This limitation can impact several tasks, such as debugging, fault localisation, bug dataset construction, rollback analysis, and bug fixes backporting. To fill this gap, we propose SemaDiff, a novel approach for identifying semantic-preserving commits through behaviour-based analysis; comparison of similar test execution on pre- and post-commit versions. As code impacted by the refactoring is often hard to test and different accross both versions, we propose generating additional calling methods to that code, which serve as testing target. Given a commit, SemaDiff analyses the diff to identify modified code and extracts unchanged dependent code that calls it. It then generates an additional dependent class using a large language model to exercise the changed code in both versions, and automatically generates tests for the dependent code. This way, we obtain the same tests for the different code versions, enabling the behavioural-difference detection. The commit is classified as semantic-preserving only if all generated tests produce identical outcomes across the two versions. To evaluate SemaDiff, we construct and annotate manually a dataset of 183 commits, gathered from well-known open-source Java projects. The obtained results show that SemaDiff distinguishes accurately semantic-preserving from -- changing commits in about 76% of the cases, with a 100% precision in semantic-changing commit detection.
arXiv:2607.13672v1 Announce Type: new
Abstract: We propose a dual-pulse micronozzle acceleration scheme that enables phase-locked acceleration of laser-driven protons, mitigating the trade-off between maximum proton energy and laser-to-proton conversion efficiency. A delay-tuned synchronization window injects a compact proton front generated by a shaping prepulse into a quasistatic axial electric field driven by a delayed main pulse in a micronozzle cavity. Phase locking preserves the relative phase between the proton bunch and the accelerating field, suppresses thermal debunching, and prolongs the acceleration stage. At main-pulse intensities of order 10^21 W/cm^2, sub-GeV proton cutoff energies are obtained with a total laser-to-proton conversion efficiency of about 20%. The efficiency for protons above 100 MeV exceeds about 13%, indicating preferential energy loading into a compact proton population. Simulations with an unconfined dual-pulse hydrogen rod show that the improvement results from temporal synchronization and geometric confinement, which sustain a long-lived axial accelerating channel. An analytical synchronization model agrees with the simulations. Three-dimensional particle-in-cell simulations confirm that phase locking and spectral hardening are preserved in slit-nozzle geometries, with cutoff energies about 60% higher than those of an unconfined hydrogen rod. These results establish phase-locked acceleration as a practical design principle for compact, high-yield sub-GeV proton drivers for secondary-particle applications.
arXiv:2607.13129v1 Announce Type: new
Abstract: Operator equations (OEs) underpin quantitative modeling across science and engineering. Finite-element (FE) methods discretize continuous OEs into finite-dimensional algebraic systems, whereas tensor networks (TNs) provide flexible variational representations of correlated discrete systems. Here, we develop a framework that connects FE with TN for analytic OEs. The power of this method comes from its ability to convert highly non-linear partial differential equations into linear matrix equations. In particular, we show that FE discretization induces a hierarchy of multilinear interaction tensors, through which differential, integral, nonlinear, memory, and delay equations can be expressed within a common algebraic structure. The resulting systems are reformulated as weighted-residual optimization problems over TN degrees of freedom. Matrix-product-state calculations for one-dimensional linear and nonlinear diffusion reproduce conventional solutions with controlled error while preserving continuity and Neumann boundary conditions. The framework provides a common variational language for analytic OEs and establishes a direct connection between FE numerical formalism and TN variational algorithms, offering a general foundation for TN-based and quantum-inspired approaches to solving OEs.
arXiv:2605.12244v3 Announce Type: replace
Abstract: State and parameter estimation, along with fault detection, are three crucial estimation problems within the control systems community. Although different approaches have been proposed for each type of problem, the modulating function method proposes a more unified approach to all three problem classes, being used for state and parameter estimation of lumped systems, fault detection, and estimation of distributed and fractional systems. At the core of the method is the modulating function: a function that evaluates to 0 at the left or right boundaries up to a certain order of derivatives. By selecting the modulating functions, one directly determines the filter characteristics, and, for that reason, different function families have been proposed over the years. Nevertheless, many families of modulating functions are given in a rather similar mathematical structure. In light of these structures, this paper formally discusses the algebraic properties of modulating functions, and, after formalizing the closedness and group properties of modulating functions, a simple algorithm to construct new modulating functions is proposed, discussed, and illustrated with the construction of the newly introduced logarithmic modulating function families and 3 non-analytic modulating function families. Moreover, the fact that total modulating functions form a vector space and an algebra is exploited to construct orthonormal modulating functions, which are then used for the parameter estimation of a boat's roll dynamics, effectively avoiding matrix inversion issues.
arXiv:2605.13181v2 Announce Type: replace
Abstract: Precipitation nowcasting remains challenging due to the highly localized, rapidly evolving, and heterogeneous nature of atmospheric dynamics. Although recent methods increasingly adopt attention-based architectures in both unimodal and multimodal settings, they mainly emphasize stronger representation learning and prediction capacity, while paying less attention to the stability of attention responses across samples. In this work, we show that cross-sample instability of attention-response energy is an important and previously underexplored source of forecasting unreliability. Empirically, inaccurate forecasts are associated with larger attention-response energy variance across heads and layers. Theoretically, we show that cross-sample variability can propagate through self-attention, and enlarge a lower bound on prediction error. Based on this insight, we propose HARECast, a Head-wise Attention Response Energy-regulated framework for precipitation nowcasting. HARECast explicitly models head-wise attention-response energy and stabilizes it through a group-wise regularization objective that reduces cross-sample fluctuations. The proposed formulation is generic and applicable to both unimodal and multimodal nowcasting architectures. We instantiate HARECast in a standard forecasting pipeline with reconstruction branches and a diffusion-based predictor, and evaluate it on commonly used benchmarks--SEVIR and MeteoNet. Experimental results demonstrate that HARECast achieves state-of-the-art performance.
arXiv:2605.13866v2 Announce Type: replace
Abstract: Humans increasingly delegate consequential decisions to language models, yet whether these systems reproduce or reshape human patterns of discrimination remains unclear. Here, across 29 models and 177 occupations covering nearly half of U.S. employment, we show that language models incorporate demographics into hiring decisions, advantaging female and Black candidates while penalising disabled candidates, with effect sizes comparable to six months to one year of additional education. While pre-trained models show small demographic effects, post-training alignment, which adapts models to human norms and preferences, amplifies advantages for female and Black candidates by 396% and 413% and worsens the disability penalty by 152%. Compared with human employers in past correspondence experiments, language models reverse racial discrimination, substantially attenuate the disability penalty, and amplify the female advantage. Investigating the mechanisms, we find behavioural patterns that parallel statistical discrimination in human labour markets, but disability consistently fares worst across all channels. We trace this asymmetry to the composition of alignment data, where disability is structurally underrepresented, and to models' internal representations, where alignment shifts the encoding of disability most negatively among the three marginalised groups.
arXiv:2605.14477v2 Announce Type: replace
Abstract: We introduce EvoLib, a test-time learning framework that enables large language models to accumulate, reuse, and evolve knowledge across problem instances without parameter updates or external supervision. Instead of adapting model parameters, our approach maintains a shared library of knowledge abstractions, including modular skills and reflective insights, automatically extracted from the model's own inference trajectories. To support continual improvement, we introduce a principled weighting and consolidation mechanism that jointly optimizes for immediate utility and long-term value. This allows simple, instance-specific abstractions to evolve into more general and reusable ones over time. Across challenging benchmarks in mathematical reasoning, code generation, and multi-turn agentic environments, EvoLib improves substantially over the top test-time scaling and learning methods without ground-truth feedback.
arXiv:2605.15026v2 Announce Type: replace
Abstract: Online OS tuning can improve long-running services, but existing controllers are poorly matched to live hosts. They treat scheduler, power, memory, and I/O controls as black-box variables and optimize a scalar reward. This view ignores cross-knob policy structure, breaks down when application metrics are unavailable, and can send a running service into degraded regions that persist after the bad setting is removed. We present TuxBot, a host-side framework for steady-state OS tuning with bounded language-model guidance. TuxBot turns knob schemas, telemetry, current configuration, recent action--response history, and retrieved prior runs into a compact decision context. A fast loop proposes low-latency updates, a slower loop periodically revises the search strategy, and every proposed change passes through typed validation before reaching kernel or sysctl interfaces. This lets the controller reason about OS-control meaning and indirect performance signals while keeping model cost, latency, and authority constrained. We evaluate TuxBot on 13 live workloads from five benchmark suites while tuning up to 41 Linux parameters. Across the suite, TuxBot improves stable-phase performance by 72.5% over default settings and by 153.3% relative to the strongest non-LLM baseline. A 30-window session costs about $0.20 in model calls. With only host-level metrics, TuxBot still outperforms baselines given direct application objectives by 93.7 percentage points, while avoiding severe degraded regions reached by structure-blind exploration.
arXiv:2605.16366v2 Announce Type: replace
Abstract: Video MLLMs face a persistent tension between spatial fidelity and temporal coverage: preserving fine-grained visual details requires many spatial tokens, while capturing short-lived events requires dense temporal sampling. We propose \textbf{Fre-Res}, a budget-adaptive dual-track video-token compression framework that separates these two forms of evidence. Fre-Res preserves sparse high-fidelity spatial anchors and represents dense temporal evolution through compact residual-frequency tokens. Specifically, it applies temporal 1D-DCT to inter-frame residual trajectories in vision-latent space, where we observe strong low-frequency concentration. To align frequency-domain dynamics with native visual embeddings, Fre-Res introduces a Spatial-Guided Absorber that injects temporal residual information into spatially corresponding anchor tokens. Across fine-grained short-video and long-video reasoning benchmarks, Fre-Res achieves a favorable accuracy--efficiency trade-off, matching or approaching full-token performance while substantially reducing visual-token length. Extensive ablations further show that temporal-frequency residuals preserve causal transition cues, while spatial anchors remain essential for fine-grained object and layout reasoning.
arXiv:2605.19867v3 Announce Type: replace
Abstract: Neural operators are increasingly used as accelerators inside classical numerical methods, but it is rarely clear which architectural ingredients matter for which application. We answer this question for one important use case: the coarse-space correction inside a two-level preconditioner for discretised linear partial differential equations. We systematically vary four DeepONet-like architectures along two design axes: input discretisation (sampling versus integration against a basis) and source-term linearity. In doing this, we show that the favourable corner of this 2$\times$2 design is occupied by a single architecture, the Neural Green's Operator (NGO), and that moving away from it produces predictable failure modes: structurally non-symmetric preconditioned spectra, breakdown of preconditioned conjugate gradients on self-adjoint problems, and stagnation on non-self-adjoint ones. Used as a coarse-space correction, the NGO matches the iteration count of an exact coarse solve on diffusion and advection-diffusion problems. The principle generalises: integrating inputs against the basis used for the output is what allows a neural operator to serve as a Galerkin-type coarse-space correction.
arXiv:2605.20689v2 Announce Type: replace
Abstract: High-dimensional language-model embeddings increase storage and search costs, while supervised compressors can overfit when relevance labels are scarce. We present DIVE (Dimensionality reduction with Implicit View Ensembles), a residual compression adapter codesigned with a self-limiting hinge loss, geometry distillation, and head-wise NT-Xent over implicit coordinate views. The hinge stops updating satisfied ranking constraints, while the dense objectives stabilize the compressed representation; only the first head is retained at inference. Under query-disjoint evaluation with two LLM2Vec backbones, five BEIR benchmarks, 128d and 256d outputs, and six baselines, DIVE is the strongest adapter on all five primary benchmarks. It also outperforms PCA and an autoencoder in comparisons against unsupervised compressors.
arXiv:2607.13394v1 Announce Type: new
Abstract: Generative Flow Networks (GFlowNets) offer a promising alternative to reward-maximizing reinforcement learning (RL) for large reasoning models, encouraging diverse reasoning paths by matching reward distributions rather than collapsing to dominant modes. Recent work shows promise on math and code, but scaling GFlowNet-style RL to modern post-training pipelines remains difficult: as model size, rollout horizon, reward noise, and distributed-systems complexity grow together, a learned prompt-conditional partition function becomes a source of gradient instability and engineering overhead rather than a useful normalizer. Through systematic analysis, we find that the learned partition function, previously treated as essential, can be replaced by an in-batch Monte Carlo estimate computed from the rollout group already required for training. We propose GFlowRL, a streamlined GFlowNet-style RL algorithm that removes the auxiliary partition network entirely while preserving the reward-distribution-matching objective, completed by two stabilizers: importance-sampling correction for rollout/trainer drift and asymmetric flow-gap clipping for outlier residuals. GFlowRL exceeds all counterparts on math, code, and adversarial red-teaming benchmarks, reaching a Codeforces rating of 2048 at the 14B scale (within 25 Elo of o3-mini) and attaining the highest average ASR@1 on AdvBench and HarmBench, outperforming the previous SOTA multi-turn attacker in a regime where FlowRL, a prior GFlowNet-style method, diverges. The same recipe transfers to all evaluated MoE configurations up to 235B parameters, where FlowRL again fails to converge. To our knowledge, GFlowRL is the first GFlowNet-style RL algorithm to scale stably across both dense and sparse architectures. Code will be at: https://github.com/microsoft/gflowrl
arXiv:2605.21268v3 Announce Type: replace
Abstract: Land Use Scene Classification (LUSC) from remote sensing imagery plays a critical role in environmental monitoring, urban planning, and sustainable resource management. In recent years, deep learning methods have significantly advanced the state of the art, with Convolutional Neural Networks (CNNs) dominating the field because of their strong ability to capture local spatial features. However, the emergence of Vision Transformers (ViTs) has introduced a new paradigm that models long-range dependencies through self-attention mechanisms, potentially enabling improved global context understanding. This paper presents a comparative assessment of Vision Transformers and CNN-based architecture for remote sensing land use scene classification. Representative CNN models, such as AlexNet, is evaluated alongside the Vision Transformer (ViT) using benchmark remote sensing datasets, including the UC Merced Land Use and EuroSAT Land Use datasets. The study examines classification accuracy, precision, recall, F1-score, and computational complexity to provide a comprehensive performance comparison. Experimental results demonstrate that CNNs perform robustly on datasets with limited training samples and strong local texture characteristics, whereas Vision Transformers exhibit superior performance in capturing global spatial relationships in complex scenes when sufficient training data are available. However, ViTs typically require greater computational resources and larger training datasets to achieve optimal performance. The findings of this study provide insights into the strengths and limitations of both architectures and offer guidance for selecting appropriate models for remote sensing land use scene classification applications.
arXiv:2605.21753v2 Announce Type: replace
Abstract: The Erd\H{o}s-Ginzburg-Ziv theorem states that every sequence of 2n - 1 integers contains a subsequence of length n whose sum is divisible by n. Choi, Kang, and Lim gave a simple deterministic O(n log n) algorithm for finding such a subsequence, and Leung recently improved this to O(n log log log n). We give a deterministic linear-time algorithm. The core is a linear-time algorithm for the following prime target subset-sum problem: given p - 1 nonzero residues in Z_p and a target residue, find a subset with the prescribed sum. Our algorithm maintains a compact arithmetic-progression representation of reachable sums. When two progressions intersect, a bounded Frobenius interval in their sum allows them to be merged into one longer progression, with enough growth to pay for the update. When the representation either contains a full progression or covers all nonzero residues, the target residue is recovered constructively. The standard multiplicative reduction then extends the prime algorithm to arbitrary moduli.
arXiv:2605.23045v3 Announce Type: replace
Abstract: Video representation learning has seen tremendous progress in recent years. This has been driven by many factors, including the scale of training and the success of visual models trained contrastively with language. While these factors have pushed the boundaries of what video models can do, they also introduce their own set of limitations: first, scaling video models can reach prohibitive costs and second, learning from language restricts the range of concepts that can be learned to those in captions. As a result, video models still struggle with temporal understanding. In this paper we propose a novel approach that uses motion as the central modality for video representation. In particular, given the motion in a video in the form of point-tracks, we use a masked-autoencoder to mask some of the tracks and train the autoencoder to reconstruct the missing tracks. This allows us to learn a representation in a self-supervised manner. We show that using motion to represent videos actually addresses both of the core limitations of video technology. First, it allows us to massively reduce the scale of training data, as motion is inherently appearance-independent and hence needs fewer examples to generalize well. Second, motion allows us to bypass the language-dependent training paradigm, learning better fine-grained concepts. The result is an embedding that we call TIME (Temporally Informed Motion Embedding), a representation trained exclusively on synthetic motion data. We test this embedding on a wide set of tasks in a zero-shot manner. We observe that without bells and whistles, performance is on par with state-of-the-art models using up to 4 orders of magnitude less training data. This is a stepping stone towards a new paradigm of video models that are both more temporally aware as well as more scalable.
arXiv:2605.23763v2 Announce Type: replace
Abstract: Two-dimensional (2D) spectroscopy combines high temporal and spectral resolution, allowing the observation of ultrafast energy transfer and the separation of homogeneous and inhomogeneous broadening. Typically, 2D spectroscopy is dominated by the lowest-order nonlinear signal for a given phase-matching configuration while signals of higher order are present but difficult to access separately. Recently, we introduced a technique to separate nonlinear orders in 2D spectroscopy by systematically varying the intensity of the pump pulses and appropriate post-processing. Here, we unravel the full potential of higher-order 2D spectroscopy by separating multiple nonlinear orders at different multi-quantum positions. As an example, we investigate a squaraine dimer. Using a theoretical model, we find excellent qualitative and quantitative agreement throughout all nonlinear orders and multi-quantum positions. Our simulations demonstrate the sensitivity and information content hidden in the higher-order spectra such as transition dipole moments and energy levels even of highly excited states. Our results pave the way for establishing higher-order spectroscopy as a unique extension of multidimensional spectroscopy, providing access to highly excited states and their properties encoded in successive orders of nonlinearity.
arXiv:2605.24395v2 Announce Type: replace
Abstract: Alignment plays a fundamental role in many machine learning problems, such as multi-network analysis, multimodal learning, and point cloud registration. Recent works increasingly leverage optimal transport (OT) for distributional alignment, whose effectiveness largely depends on sparse supervision that is hard or costly to obtain in practice. Existing works, however, largely overlook how to actively acquire high-quality supervision to improve their alignment performance under OT frameworks. In this paper, we propose a principled active alignment framework for optimal transport alignment called AvAtar. We quantify the informativeness of a candidate by measuring its gradient-based impact on the global alignment result, computed as the gradient propagation from the global alignment result to all possible supervisions of the candidate through the entropy-regularized OT formulation. While differentiating through OT is challenging given its constrained nature, we leverage the adjoint-state method to reformulate the computation to a linear system solvable by the conjugate gradient method with linear complexity and guaranteed convergence. By encoding the global alignment result via effective utility functions, AvAtar is applicable to general alignment problems under the OT framework. Extensive experiments on three representative alignment tasks demonstrate the effectiveness, scalability, and generalizability of the proposed AvAtar.
arXiv:2605.24602v5 Announce Type: replace
Abstract: Multimodal large language models (MLLMs) frequently suffer from object hallucinations, yet the visual perceptual mechanism underlying this failure remains poorly understood. In this work, we reveal that hallucinations are strongly associated with a human-like attention distraction phenomenon, where humans under divided focus experience degraded visual clarity and produce inaccurate descriptions, while in models the same mechanism manifests as spatial inconsistency in multi-head attention and temporal fading of attention to image tokens during decoding. We further provide theoretical insights that attention dispersion increases model complexity and degrades classification generalization. Motivated by these findings, we propose an Attention-Focused Approach for Improved Image Perception (AFIP), which corrects attention distraction via cross-head attention enrichment and reinforces visual grounding through dynamic historical attention enhancement. Extensive experiments on multiple benchmarks and models validate the effectiveness of AFIP without additional training. Code is available at: https://github.com/MIKUZ12/AFIP.
arXiv:2605.24902v2 Announce Type: replace
Abstract: Reasoning-enabled LLMs perform strongly on medical reasoning benchmarks, but it remains unclear whether these gains transfer to structured clinical documentation; we investigate this question using SOAP note generation from clinical dialogue in a source-aware benchmark spanning OMI Health, ACI-Bench, and PriMock57. We evaluate GPT-5.4, DeepSeek-V4-Flash, and Gemma-4-E4B in a controlled 2x2 design that independently toggles provider-native reasoning and same-source retrieval-augmented generation (RAG). Outputs are assessed using seven automatic metrics alongside two reference-aware LLM judges. Both evaluation approaches agree that a non-reasoning GPT-5.4 configuration achieves the highest overall quality, while DeepSeek-V4-Flash performs best among reasoning-enabled configurations. Enabling reasoning significantly degrades GPT-5.4 performance across all three datasets, whereas same-source RAG yields smaller, model-dependent improvements. Overall, the findings indicate that stronger reasoning capability should not be assumed to improve fidelity-sensitive SOAP note generation without dedicated, task-specific evaluation.
arXiv:2605.31272v2 Announce Type: replace
Abstract: As predictive models are increasingly deployed in high-stakes settings such as credit approval, there is a growing need for post-hoc methods that provide recourse to affected individuals. Many such models operate on tabular data, where features correspond to real-world attributes. Recently, in-context learning (ICL) has enabled large language models to perform tabular prediction by conditioning on labeled examples at inference time, without explicit training. However, algorithmic recourse for tabular decision-making under ICL remains largely unexplored. In this work, we present the first study of algorithmic recourse for tabular data under ICL. We carry out a theoretical analysis, showing that recourse remains well-defined and bounded, and we characterize how recourse converges toward classical solutions as the context size increases. In practice, we propose a novel zeroth-order recourse framework, Adaptive Subspace Recourse for In-Context Learning (ASR-ICL), that efficiently generates actionable and sparse recourse for black-box ICL models. The proposed framework naturally extends to multi-class tabular tasks. Experiments across multiple real-world datasets and models demonstrate that ASR-ICL achieves recourse quality comparable to existing methods with fewer queries and empirically confirm the predicted convergence behavior, supporting our theoretical analysis.
arXiv:2606.00244v2 Announce Type: replace
Abstract: In terrestrial networks, especially in urban areas, cell-edge users often face significant capacity limitations due to high path loss, shadowing, and inter-cell interference (ICI). This paper proposes integrating a high-altitude platform station (HAPS) into terrestrial networks, where terrestrial base stations (BS) can alleviate these issues by relaying data intended for cell-edge users via HAPS, thereby leveraging line-of-sight (LoS) links. We formulate an energy-efficiency (EE) maximization problem to jointly design beamforming vectors at the BS and HAPS with the goal of improving cell-edge user performance. Since the resulting problem is non-convex, we develop an online optimization framework based on a graph neural networks (GNN), which effectively captures the network topology. Numerical results show that the proposed HAPS-assisted architecture improves network performance, particularly by increasing the 5th-percentile EE, thereby enhancing service for cell-edge users.
arXiv:2606.01004v2 Announce Type: replace
Abstract: A recent pedagogical paper by Strauch concretely demonstrated how interaction-mediated entanglement can suppress fringe visibility in a one-dimensional model of the electron double-slit experiment. Here we extend that framework to model actual experimental data from electron biprism interferometry. Kerker et al. showed that the macroscopic QED model of Scheel and Buhmann successfully describes their measured results. We show that this Scheel-Buhmann model can be recovered from Strauch's simplified framework by employing a Markov approximation and including thermal effects. The resulting decoherence rate is expressed in terms of mode-resolved scattering probabilities familiar from electron energy-loss spectroscopy (EELS), directly relating EELS to decoherence. The thermal dependence is significant in its own right, as recent theoretical work suggests that visibility reduction could serve as a non-invasive thermal probe for nanoscale systems. This progression from a toy model, to a quantitative account of real data, to a measurement application offers a case study in how simplified models can be made experimentally relevant.
arXiv:2606.02911v2 Announce Type: replace
Abstract: Current research primarily focuses on model performance, while comparatively less attention has been devoted to uncertainty estimation, particularly in settings where LLMs are increasingly used to generate annotated data. We introduce a framework combining conformal prediction with Collaborative Filtering-style annotators' representation to model LLM behavior in relation to human annotators and to analyze patterns of agreement and disagreement. Using Non-Conformity Scores, we introduce the Ghost Prediction metric and the Ghost Annotator representation to quantify cases in which model predictions diverge from all available human annotations. We compute cosine similarity measures to explore differences in model behavior across sociodemographic axes. We evaluated four LLMs of different size and families across four content moderation datasets. Our finding shows that while we find that all models uncertainty increases with annotator disagreement, larger models tend to be more confident in the classification of texts that are not aligned with any human annotation. Finally, the Ghost Annotator framework reveals a consistent and robust pattern of demographic misalignment, suggesting a structural bias likely rooted in pretraining corpora.