arXiv:2607.05849v1 Announce Type: new Abstract: Low-resource languages remain challenging for machine translation, and Mongolian is a representative case. As a digraphic language, Mongolian is written in both Cyrillic and Traditional scripts, which exhibit a severe imbalance in data availability. While the Cyrillic script is relatively well-resourced, the Traditional script remains extremely data-scarce and orthographically ambiguous, leading to substantial performance degradation in direct translation. We propose CoPiT, a cognitively motivated pivot-based translation pipeline that exploits this internal resource hierarchy by routing translation through the Cyrillic script. The pipeline explicitly resolves script-induced ambiguity in the Traditional script before translation, enabling more stable and accurate meaning transfer. Across multiple backbone models and target languages, CoPiT consistently outperforms direct translation, achieving substantial absolute BLEU improvements together with consistent 1.5-1.6x COMET gains. These gains allow strong open-source models to match or outperform GPT-4.1 under comparable evaluation settings. Beyond inference-time improvements, CoPiT enables the construction of synthetic parallel data directly from Traditional-script text, mitigating data scarcity in realistic low-resource scenarios. We release a new multi-script parallel dataset covering Mongolian in both scripts alongside English, Korean, and Russian. All datasets and code are publicly available at https://anonymous.4open.science/r/anonymous_project-76C7.
Science Journals
Information Limits and Attractor Dynamics in Economies of Frontier LLM Agents: A Pre-Registered Test
arXiv:2607.06001v1 Announce Type: new Abstract: We report a pre-registered, two-part experiment on small economies of frontier language-model agents (Claude Opus 4.8), testing two quantitative predictions about coupled multi-agent systems: an information-theoretic capacity region for wealth growth under market coupling, and a mean-field residual-scaling law for population misalignment under incentive and control levers. All predictions, acceptance bands, and decision rules were frozen in a public git chain before any run; every reported number re-derives mechanically from cached model outputs; the entire experiment cost $138.76 in metered API spend and is re-runnable at zero cost from the cache. Result 1 (confirmation): in parimutuel-coupled economies, relative growth equals relative claimed information -- the gap law G_a - G_b = I_a - I_b holds to a worst-case 46 millinats (pre-registered band: 50) across four perception structures; coalition value is submodular exactly where channels are conditionally independent, and a designed XOR synergy control flips it supermodular by 0.62 >= ln2/2 nats, with agents reasoning out the joint bit; the joint growth ceiling G_S <= H(X) binds exactly; and the best-informed agent absorbs essentially the whole wealth pool in 4/5 market seeds. Result 2 (structural negative): the residual-scaling test returned "domain not found." In all 72 population runs, goal dispersion collapsed (V -> 0; maximum 4.85 against a frozen floor of 5.31), the population's response to the two levers was a step function across the dominance boundary rather than a smooth response, and cells near the boundary were bistable with seed-selected outcomes. No tested LLM population at any capability level realizes the noise-maintained-dispersion regime the smooth mean-field model assumes. We release the full protocol, pre-registration chain, call cache, and analysis code.
arXiv:2607.06425v1 Announce Type: new Abstract: Adaptive control learns the plant online; neural-operator control learns the control gains offline. We bring the two together for a class of nonlinear hyperbolic PDEs whose dynamics are governed by an unknown Volterra series of arbitrarily many kernels. An observer-based passive identifier learns a truncation of this series online. The infinite-dimensional map that synthesizes the backstepping kernels from the parameter estimates -- a cascade of PDEs on simplex domains of increasing dimension, prohibitive to solve in real time -- is approximated once, offline, by a neural operator. The closed loop then carries two learning processes in series: online learning of the plant feeds an offline-learned PDE solver, whose output is the online control gains. We prove closed-loop stability and asymptotic regulation of the plant state, observer state, and input, on a basin that recovers the exact-kernel basin as the neural-operator accuracy improves. With a single Lyapunov function we absorb at once the perturbations -- all vanishing -- of truncating an infinite Volterra series, of identifying the plant online, and of approximating the gains.
arXiv:2607.06283v1 Announce Type: new Abstract: Skill usage can significantly enhance the ability of modern agent systems to complete complex tasks. However, the growing scale of skill libraries makes accurate skill selection increasingly challenging. In real-world scenarios, ambiguous semantic matching often arises between a specific task requirement and multiple generic yet semantically similar candidate skills. Moreover, existing methods tend to overlook the dynamic influence of task difficulty and skill applicability when selecting the optimal target skill set. To address these issues, we propose SkillReranker, an inference-time reranking framework for adaptive skill selection. Specifically, we first perform semantic decomposition on both the task and skill sides, yielding informative subtask and execution-state descriptions as well as transition-state descriptions that characterize each skill's functionality. These descriptions are then used to construct a directed acyclic execution graph, where intermediate task states are modeled as nodes and candidate skills as edges, thereby establishing a structured task-skill correspondence. On this basis, SkillReranker determines whether each state node satisfies the split condition to identify subtask intervals. For each task interval, we employ a cross-encoder to perform comprehensive scoring over candidate skills and select the most suitable ones to form the final target skill set. Experiments on ALFWorld and ScienceWorld with three backbone LLMs show that SkillReranker effectively improves task performance, reduces environment interaction steps, and lowers token consumption compared with existing skill selection baselines.
arXiv:2606.30238v2 Announce Type: replace Abstract: This paper studies sparse sensor placement for control of Rayleigh-B\'enard convection with multi-agent reinforcement learning. We train dense expert policies with windowed observations and distill sparse apprentice policies by supervised learning with grouped regularization on encoder input weights. The framework combines ordered non-convex grouped regularization and iterative reweighted grouped regularization, and uses a grouping construction that enforces consistent pruning across overlapping observation windows. Experiments with fixed and varying initial conditions show that Multi-Agent Transformer policies train more stably than proximal policy optimization baselines, while sparse apprentices retain control behavior comparable to dense experts. Sparsity results are strong for the proposed grouped methods across settings, including maximal sparsity in all fixed-initial-condition setting variants and maximal or near-maximal sparsity in varying-initial-condition setting variants. As an additional proof of concept, training from learned minimal sensor sets reduces per-agent observation size from 360 to 12 and preserves the overall training trend in simulation while reducing data throughput. The results provide both an interpretable basis for identifying control-relevant spatial regions and state components, and a practical pathway toward sensor-efficient control under realistic hardware constraints.
arXiv:2606.30243v2 Announce Type: replace Abstract: This paper presents KYON, a hybrid wheel-legged quadruped robot equipped with a bimanual upper body for loco-manipulation tasks. The platform features a semi-modular design with a reconfigurable lower legs, enabling both wheeled and legged locomotion depending on the environment. A design approach that places actuators in the base and uses transmission mechanisms reduces distal inertia, improving agility and dynamic performance. The robot integrates a whole-body control framework together with a reinforcement learning based policy to handle nonlinear dynamics and enhance robustness to disturbances for the execution of locomotion and manipulation tasks, independently. Experimental results demonstrate effective dynamic locomotion and bimanual manipulation, validating the platform's capability to operate in complex and unstructured scenarios.
arXiv:2606.30340v2 Announce Type: replace Abstract: We formulate a Bayesian framework for reconstructing doping profiles in pn-junction semiconductor devices from boundary flux measurements. The unknown doping field is modeled as a piecewise-constant function characterized by an unknown interface and two plateau concentrations, leading to a nonlinear ill-posed inverse problem governed by a Poisson-Boltzmann-type equation. To represent this structure while enabling efficient gradient-based inference, we introduce a pushforward prior constructed by mapping a latent Gaussian field with Mat\'ern-type covariance through a sigmoid transformation. The latent field is parameterized by a truncated Karhunen-Lo\`eve expansion, while the two piecewise-constant levels are represented by scalar plateau parameters. The prior yields differentiable approximations of piecewise-constant fields with controllable interface sharpness. We establish well-posedness of the Bayesian formulation by proving Lipschitz continuity of the forward map and Hellinger stability of the posterior. We then sample the posterior using the No-U-Turn Sampler (NUTS) with gradients computed by the adjoint method. Numerical experiments show that the combination of the proposed prior and NUTS provides more efficient posterior exploration than the dimension-robust preconditioned Crank-Nicolson (pCN) sampler, yielding one to two orders of magnitude larger effective sample sizes. In the known-plateau setting, the method reconstructs both planar and curved interfaces and provides spatially resolved uncertainty quantification (UQ). When the interface geometry and plateau concentrations are inferred jointly, posterior correlations reveal structural non-identifiability. These results demonstrate the effectiveness of combining pushforward priors with adjoint-gradient-based sampling for reliable UQ in nonlinear partial differential equation-constrained inverse problems with sharp interfaces.
arXiv:2606.30530v5 Announce Type: replace Abstract: The continuous integral R2 indicator is a Pareto-compliant refinement of the classical finite-weight-vector R2 indicator, used in performance assessment, bounded archiving for a-posteriori multi-objective optimization, and skyline selection in databases. This work introduces a bidirectional perspective mapping between continuous integral R2 computation and integration over unions of anchored axis-aligned boxes. After translating the ideal point of a minimization problem to the origin, approximation points become strictly positive loss vectors, and the subgraph of the lower weighted Tchebycheff envelope over the weight simplex maps to the complement of an anchored-box union in reciprocal objective space. The Jacobian gives an absolute R2 formula as a weighted complement volume with density $(x_1+\cdots+x_N)^{-(N+1)}$, while differences of R2 values become finite weighted hypervolume differences. Hence, hypervolume algorithms that emit box decompositions can be reused by replacing ordinary box volumes with closed-form weighted box integrals. For $N$ objectives, this gives an output-sensitive overhead $O(2^N M)$ for an $M$-box decomposition, or $O(M)$ for fixed $N$. Using existing box-decomposition approaches, the integral R2 can be computed in $O(n \log n)$ for $N=2,3$, in $O(n^2)$ for $N=4$, and in $O\left(n^{\lfloor (N-1)/2\rfloor+1}\right)$ for $N\geq4$, with $n$ denoting the size of the approximation set. On the lower-bound side, exact value computation has an $\Omega(n\log n)$ lower bound in the algebraic decision-tree model already in two objectives, this bound lifts to every fixed $N\geq2$, and exact computation is $\#P$-hard when $N$ is part of the input. Together, the proposed perspective mapping provides a powerful tool for transferring algorithmic and structural results between anchored-box union and hypervolume theory and integral R2 computation.
arXiv:2607.01646v2 Announce Type: replace Abstract: State-of-the-art large language model (LLM) training takes tens of thousands of graphics processing units (GPUs) for months and encounters failures across the software and hardware stack. Existing fault-tolerance mechanisms either impose non-trivial overhead during failure-free execution or suffer from prolonged recovery latency, particularly under scenarios where a small subset of compute nodes experience permanent failures. %The tradeoff between failure-free overhead and recovery latency forms a space forms a Pareto frontier We present PHOENIX to simultaneously address both optimization objectives. PHOENIX incorporates a fault-tolerance mechanism that restores LLM training via hot-swapping, namely by replacing failed nodes with spare nodes without terminating the complete job. The hot-swapping of PHOENIX is enabled by two ideas: First, it exploits an off-critical-path in-memory checkpointing mechanism for spatial redundancy. Second, it introduces a communicator reconstruction protocol that replaces failed nodes with spare nodes at runtime. PHOENIX efficiently overlaps the in-memory checkpointing with computation, thus introducing zero overhead during error-free execution. Upon permanent node failures, PHOENIX can rebuild memory states with minimal recomputation by leveraging in-memory checkpoints. We evaluate PHOENIX across scales (up to 512 NVIDIA A100 GPUs) and LLMs (up to 65B parameters), and observe zero checkpoint overhead with hot-swapping recovery completing in under 40 seconds. These results show that PHOENIX simultaneously achieves both zero-overhead error-free execution and extremely low recovery cost.
Boundary-preserving hp interpolation and p-robust discrete harmonic extensions on tetrahedral meshes
arXiv:2606.02086v3 Announce Type: replace Abstract: We construct a boundary-preserving hp interpolation operator on three-dimensional tetrahedral meshes with locally variable polynomial degrees. If the trace of an H^1 function on the prescribed Dirichlet boundary is already a piecewise polynomial trace of the finite element space, the interpolant preserves this trace exactly and satisfies the standard local h_K/p_K approximation estimates. The statement follows the scaling form of Melenk's hp quasi-interpolation for nonsmooth functions. As a consequence, a discrete trace is extended by first applying the continuous trace theorem and then applying the boundary-preserving interpolant; the corresponding discrete harmonic extension is bounded by variational comparison. The proof of the interpolation theorem uses local polynomial trace liftings on tetrahedral boundary layers, nonsingular vertex patches, and a variable-degree tetrahedral projection. These auxiliary liftings are also stable in scaled boundary-layer norms.
arXiv:2607.01880v2 Announce Type: replace Abstract: Value functions are an essential component in actor-critic based deep reinforcement learning (RL). Conventionally, these functions are trained as a regression task by minimising the mean squared error (MSE) relative to bootstrapped target values. Meanwhile, in distributional RL, a distribution of returns is modelled based on the distributional Bellman operator. This work investigates the Gaussian Histogram Loss (HL-Gauss), a recent approach that reframes value estimation as classification by encoding each scalar Bellman target as a Gaussian-smoothed categorical target. Despite its potential, applying histogram-based losses to RL presents inherent challenges, most notably the requirement to pre-define a fixed support interval, which is often complicated by the non-stationary and stochastic nature of target values typically found in RL tasks. In this work, we propose an approach that dynamically learns the lower and upper bounds of the support instead of assigning them beforehand. We derive an objective that jointly learns these bounds whilst learning the categorical representation of the scalar values, and we show that this objective forms an upper bound on the mean-squared Bellman error. Our theoretical analysis further shows that this bound is tighter than that of non-learned supports of HL-Gauss. Empirically, the proposed objective enables stable adaptation of the support interval and matches HL-Gauss-based actor-critic algorithms on most continuous-control tasks whilst improving on a subset, without requiring a pre-specified support interval.
arXiv:2607.02275v2 Announce Type: replace Abstract: We present Cadence, a Byzantine fault-tolerant multi-proposer consensus protocol with arbitrarily low block intervals, optimal resilience, and optimal fast-path latency. Cadence divides time into equally spaced slots, one block per slot, each finalized in its own consensus instance. Blocks do not build directly on their predecessor, so instances run independently and none waits for an earlier block to finish or propagate; we call this extreme pipelining, decoupling the block interval from network latency. Cadence also removes the single-leader monopoly over transaction inclusion and ordering: under multiple concurrent proposers (MCP), several validators propose for each block, and it guarantees that, under synchrony, a transaction a correct proposer includes cannot be censored or deferred (short-term censorship resistance), and that no proposer can craft its proposal in reaction to the others' (hiding). To realize extreme pipelining, we introduce a general framework that turns any one-shot consensus meeting our slot-consensus specification into a multi-shot protocol. We instantiate it for MCP with two protocols of our own: Chorus, a slot consensus whose fast path finalizes a block in an optimal three rounds, with speculative finality one round earlier, and Conductor, an orchestrator that opens slots at an even cadence, more slowly under asynchrony to keep open slots bounded. To our knowledge, Cadence is the first MCP protocol to provide short-term censorship resistance and hiding at the fast-path latency of single-leader consensus. We prove safety, liveness, censorship resistance, and hiding under partial synchrony with optimal resilience (n = 3f+1). In simulation over Monad's 200 validators with five proposers per slot, finalization averages 219 ms (167 ms to speculative finality); at a 100 ms block interval a transaction waits on average 50 ms to enter a proposal.
arXiv:2607.04107v2 Announce Type: replace Abstract: Word surprisal is a well-established computational predictor of human neural responses during language comprehension, but it remains less clear whether local semantic fit explains neural response variation beyond lexical expectation during naturalistic reading. Using the Dublin EEG-based Reading Experiment Corpus (DERCo), this study examined whether contextual semantic relevance predicts word-locked EEG activity in the N400 and P600 windows. Contextual semantic relevance was computed as an attention-aware measure of how strongly a target word is semantically connected to its recent discourse context, and it was compared with GPT-based word surprisal. Across 22 participants and 32 EEG channels, we tested both predictors using regression-based ERP analyses and generalized additive mixed models while controlling for lexical variables and repeated observations. Both predictors were reliably associated with EEG responses, but they showed partly different temporal and scalp-level patterns. Surprisal captured expectancy-related variation, whereas contextual semantic relevance showed robust effects across N400- and P600-window mean voltages, with particularly strong explanatory support in the P600 window. Model comparisons indicated that contextual semantic relevance contributed explanatory value beyond lexical controls and surprisal. These findings suggest that naturalistic reading depends on both lexical expectation and local semantic integration, and that contextual semantic relevance offers an interpretable computational link between discourse semantic fit and ERP dynamics.
arXiv:2607.06446v1 Announce Type: new Abstract: We study solution discovery in the token-sliding model from a logical and cost-value optimization perspective. In solution discovery, we are given a graph, an initial placement of $k$ tokens, and a movement budget $b$. The task is to find a reachable target configuration satisfying a prescribed condition. Our results are inspired by \textsc{Partial Vertex Cover Discovery}, where the condition is that the~$k$ tokens cover at least $t$ edges of the input graph. This objective is not merely a sum of independent occupied vertex contributions: each selected vertex contributes its degree, but edges with both endpoints selected have to be subtracted once. To capture this phenomenon, we introduce \textsc{FO Value Discovery}, an optimization problem in which the value of a selected tuple is given by unary vertex weights together with first-order definable correction terms. We further generalize the setting to \textsc{FO Cost-Value Decision}, where vertices carry both costs and values, and the task is to decide whether there is a tuple whose first-order value expression reaches a prescribed value threshold while respecting a cost bound. Finally, we study the parameterized complexity of \textsc{Partial Vertex Cover Discovery} and \textsc{Vertex Cover Discovery}. As a consequence of the logical meta-theorems, we obtain fixed-parameter tractability of \textsc{Partial Vertex Cover Discovery} on several graph classes, including classes of locally bounded cliquewidth. We also show that \textsc{Partial Vertex Cover Discovery} is W[1]-hard parameterized by $k+b$ and fixed-parameter tractable on $d$-degenerate graphs parameterized by $k+d$. For \textsc{Vertex Cover Discovery}, we prove NP-hardness on planar graphs, W[1]-hardness parameterized by the clique cover number, even when a clique cover is supplied with the input, and W[1]-hardness with respect to parameter cutwidth.
arXiv:2607.06449v1 Announce Type: new Abstract: Global space-time spectral methods give spectral accuracy in time but typically require the whole space-time history to be resolved and stored on a single tensor-product domain $T \times \Omega$. We record that in an endpoint-benign Legendre or Chebyshev-$T$ time basis, whose polynomials all equal one at the right endpoint, the final time slice of a space-time block is recovered exactly by summing the stored coefficients along the time index. This peel-and-pass step is a special case of a Jacobi endpoint identity, which also gives derivative formulae for higher-order equations. Writing such higher-order equations as first-order systems preserves the benign value-passing structure. The result is a sparse space-time spectral element method that advances block by block, stores only one block, and needs far fewer time coefficients per solve for long-time problems. We prove the identities, give resident-memory, solve-cost and error-propagation models, and demonstrate the method on $(1{+}1)$D heat, wave and Klein--Gordon equations, and on $(2{+}1)$D fractional heat on the disk with weighted Zernike polynomials in space.
arXiv:2607.06471v1 Announce Type: new Abstract: Context: Domain-Driven Design (DDD) is a leading paradigm for managing software complexity, yet research remains largely theoretical; our prior work found nearly 39% of DDD studies lack rigorous empirical evaluation, leaving practical adoption largely unexamined at scale. Objective: We provide the first large-scale characterisation of the DDD landscape on GitHub, a data-driven baseline for how the paradigm is implemented and sustained in practice. Method: Using a Mining Software Repositories (MSR) approach with a hybrid strategy (topics and README keywords), we identified 11,742 candidate repositories. To address label noise, we built a novel semantic validation pipeline using GPT-4o with a triplicate majority-vote strategy, yielding 2,502 verified repositories. Validation against a manually labelled sample showed substantial agreement with human experts (kappa = 0.77). Results: DDD adoption accelerated sharply after a 2017 inflection point, and the resulting projects are notably long-lived: their median lifespan exceeds the typical GitHub project by over an order of magnitude, indicating sustained, professional-grade engineering rather than short-lived experiments. Layered and Clean Architecture dominate, while CQRS and Event Sourcing recur in distributed, data-intensive systems. Notably, the data challenge the Java-centric assumption of much academic work: C# and TypeScript, not Java, lead practical adoption. Conclusions: DDD has matured into a stable, professional-grade practice adopted across diverse languages and domains. However, a quarter of projects (25.3%) record no explicit business context, revealing a persistent gap between how domain intent is designed and how it is preserved in version control. We call for lightweight architectural traceability standards and offer guidance for teams reusing these repositories as reference implementations.
arXiv:2607.06475v1 Announce Type: new Abstract: We study the scattering of time-harmonic electromagnetic waves by periodic layered gratings, modelled by the 2D Helmholtz equation. The periodic obstacle may include penetrable and impenetrable regions, and consists of a finite number of stacked layers. The boundary value problem is formulated on a single periodic cell using quasi-periodic boundary conditions. The radiation condition in the vertical directions is imposed through Dirichlet-to-Neumann (DtN) operators. To efficiently treat multilayer configurations, we adopt a formulation based on the T-matrix method. The global scattering problem is decomposed into boundary value problems posed on individual layers. On the layer boundaries, the field is expressed in terms of quasi-periodic modal expansions, and the layer T-matrix describes the map between incoming and outgoing wave modes. Each local T-matrix is approximated numerically using a plane-wave based Trefftz Discontinuous Galerkin (TDG) method, which provides an efficient discretization of the layer scattering response. The T-matrix technique leads to linear computational complexity in the number of layers in the grating.
arXiv:2607.06483v1 Announce Type: new Abstract: Robustness to domain shift is a key requirement for floor plan generative models to be applicable beyond the single dataset they were trained on, as floor plans vary widely across regions due to distinct architectural cultures, spatial constraints, and construction practices, while acquiring new annotated datasets remains costly and domain-specific. Yet, no prior work has studied this robustness in the context of conditioned floor plan generation. In this paper, we evaluate state-of-the-art models from two fundamentally different generative paradigms across three public datasets (RPLAN, MagicPlan and Swiss Dwellings) and show that they are highly sensitive to domain shift, with up to an order of magnitude performance degradation when transferred across domains. To mitigate this with minimal target-domain supervision, we introduce a procedural method to generate a large-scale synthetic training dataset that enforces strict physical constraints (non-overlapping rooms, valid door placement, graph consistency) while intentionally sacrificing architectural realism through highly irregular spatial arrangements and aggressive geometric perturbation of room shapes. We show that pre-training on this synthetic data considerably improves zero-shot cross-domain performance, outperforming in-domain training on MagicPlan. Furthermore, it provides a highly effective initialization for fine-tuning, accelerating target domain adaptation and outperforming real-world initialization baselines by up to 40% in a low-data regime.
arXiv:2607.05424v1 Announce Type: new Abstract: The transmission chains (sanad) of Islamic Hadith literature encode relationships among tens of thousands of historical narrators whose biographical records are dispersed across independently maintained digital databases that share no common identifier. We present a two-phase entity resolution pipeline that links narrator names from the Sanadset 650K corpus - 650,986 Hadith records from 926 books containing 185,216 unique narrator name variants - to two biographical databases: Hadithtransmitters (Hawramani; 100,915 entries) and Muslimscholars (25,247 entries). Phase 1 matches Sanadset names to Hawramani using name-only similarity (Sanadset carries no metadata), yielding 94,628 links (51.1%; HIGH 39,938 / MED 54,690). Phase 2 cross-references Hawramani against Muslimscholars via a weighted multi-signal function combining name similarity, death-year proximity, and reliability grade polarity, yielding 95,573 links (94.7% of Hawramani; HIGH 18,245 / MED 71,546 / LOW 5,782). Chaining the two phases gives Sanadset narrators transitive access to Muslimscholars data. The linked data enable construction of a 185,216-node, 814,093-edge directed transmission graph enriched with cross-source biographical metadata. The annotated link corpora and enriched graph are released as open resources.
Edge-based mean-field approximation of dynamics on networks via approximate lumping of Markov chains
arXiv:2607.05425v1 Announce Type: new Abstract: Mean-field approximations for dynamical processes on networks are widely used, but existing derivations often rely either on moment closures or on idealised assumptions about network structure, leaving the nature of the underlying averaging unclear. Here we present a mathematically principled framework for deriving edge-based mean-field approximations for a broad class of Markov processes on networks using approximate lumping. We consider models in which each vertex is in one of a finite number of vertex states and transitions depend on the number of neighbours in each state. Our approach partitions the full Markov chain state space according to the number of vertices and edges in each possible state, and averages transition rates between partitions. This yields density-dependent population processes that, in the limit of large system size, reduce to a low-dimensional system of ordinary differential equations. We demonstrate the method on single graphs and graph ensembles, such as Erd\H{o}s-R\'enyi random networks, and show that well-known edge-based mean-field approximations arise as special cases of our approach. Our approximate lumping framework clarifies the nature of the averaging underlying mean-field approximations, providing a basis for future work on assessing their accuracy.
arXiv:2607.05728v1 Announce Type: new Abstract: Motivated by worst-case algorithmic time bounds for solving sudoku, we prove that a majority of filled-in $n^2\times n^2$ sudoku grids require all but a logarithmic fraction of cells to be filled by clues. For $9\times 9$ and $16\times 16$ sudoku, we construct grids that require $18$ clues and $80$ clues.
arXiv:2607.05438v1 Announce Type: new Abstract: Multimodal retrieval-augmented generation (RAG) grounds a generator in evidence drawn from heterogeneous modalities -- text, tables, and images. The dominant deployment choice is binary and made before the model has tried to answer: either run a cheap text(+table) pipeline, or pay for an expensive vision-language model (VLM) over every image. Recent adaptive systems improve on this by selecting the modality or fidelity pre-retrieval, from a question-conditioned predictor of which modality will be needed. We show that this is the wrong decision point. Through an oracle headroom analysis on MultiModalQA, we find that the relevance of a modality to a question is a weak predictor of whether that modality is actually needed to answer correctly: a large fraction of questions whose gold support includes an image are nonetheless answerable from text and tables alone, and a pre-retrieval router that escalates on apparent visual relevance over-escalates substantially relative to an oracle. We propose \textbf{post-hoc selective modality escalation}: answer cheaply from text and tables, run a verifier on the (query, draft answer, evidence) tuple that localizes which modality is missing, and pay for VLM evidence only there. A calibrated value-of-escalation router then decides whether the expected accuracy gain justifies the visual cost. On MultiModalQA, our router recovers the accuracy of an always-on VLM pipeline while issuing far fewer visual calls, and closes most of the gap to the oracle escalation rate. The result extends a routing-signal hierarchy established for retrieval depth and reasoning hops to a third axis -- modality -- under a single cost-aware selective-escalation view.
arXiv:2607.05441v1 Announce Type: new Abstract: Integrating external tools with Large Language Models (LLMs) has emerged as a promising paradigm for accomplishing complex tasks. Since LLMs still struggle to effectively manage large tool collections, researchers have begun exploring retrieval-based methods to pre-select the most relevant options, addressing input length and latency constraints. However, existing retrievers are often misaligned with tool-calling LLMs due to their separate training processes. This paper presents PORTS, a novel odds ratio preference optimization method for training retrievers aimed at tool selection. Using a perplexity-inspired preference signal from a frozen LLM, our approach fine-tunes a retriever to find helpful tools by optimizing the correlation between the selection probabilities and the downstream performances while jointly enforcing a contrastive semantic loss between documentation strings. The versatility of PORTS and its ability to significantly improve tool selection accuracy are demonstrated through extensive experiments on six datasets, two encoder models, and three LLMs with diverse prior knowledge. With low computational demands, our alignment process facilitates generalization to new queries and tools, proving valuable for practical applications with evolving toolsets.
arXiv:2607.05443v1 Announce Type: new Abstract: Scientists increasingly rely on open-source tools to support their research workflows, yet discovering relevant software among over 600 million GitHub repositories remains challenging. Existing code search benchmarks focus on general software engineering tasks and fail to capture the domain-specific vocabulary and needs of scientific computing. We present a curated corpus of 5,264 high-quality, domain-classified scientific repositories spanning five NASA Science Mission Directorate divisions -- Earth Science, Astrophysics, Planetary Science, Heliophysics, and Biological & Physical Sciences -- enriched with cleaned READMEs, extracted topics, and additional context from crawled links. Building on this corpus, we introduce two novel information retrieval benchmarks: (1) a repository search benchmark with 219 expert-curated queries designed by domain scientists, and (2) a large-scale code snippet retrieval benchmark containing 117,950 code snippets and 119,720 queries across seven programming languages. Baseline evaluations on repository search reveal significant performance variation across scientific domains. Code snippet retrieval proves equally challenging, with substantial variation driven by differing documentation practices, coding standards, and programming language conventions across scientific communities. All datasets and benchmarks are publicly released on HuggingFace to support research on scientific tool discovery.
arXiv:2607.06484v1 Announce Type: new Abstract: Poisoning attacks against public datasets lead to major concerns, such as (i) misclassification of perceived objects when the poisoned data is used for training and (ii) embedding of backdoors that may eventually be triggered later on, when specific conditions in the system apply over the learned models. Its impact over data augmentation models is unclear. While data augmentation reduces the likelihood of poisoning attack success, some valid questions remain. Is data augmentation affecting the impact of poisoning attacks? can it increase the number of poisoned samples or injected backdoors? We explore in this paper some of these questions. We assess the effects of augmenting poisoned 3D point cloud datasets and validate that poisoning is able to evade the sanitizing nature of augmentation techniques when using the concrete case of Generative Adversarial Network (GAN) techniques to exemplify the case of data augmentation processing. We also validate that poisoning propagates over the augmented datasets and perturbs the decision made by general-purpose classifiers, in the end. All the experimental material (including tools, datasets, and classifiers) is publicly available, to facilitate reproducibility and to foster further research in the topic.