Papers

In the wake of the increasingly public discussion about AI alignment, recent work has tried to propose specific AI architectures that behave safely. However, the proposed arguments that seemingly demonstrate proved alignment mostly neglect the environment the agent needs to act in. We discuss theoretical bounds for agent-agnostic safety guarantees in environments that can only be partially observed and within which an action is required within limited time. We introduce two realistic scenarios, one with an infinite state space and one with signal mixture. In these scenarios, we prove that even a perfect agent cannot guarantee safe behaviour. It will be argued that for any proof of AI safety or alignment, the environment and associated safe actions need to be specifically considered together with the agent.

The early stage of high-energy heavy-ion collisions undergoes transient dynamics, evolving non-perturbatively from a gluon-rich to a chemically equilibrated quark-gluon plasma. We develop an effective hydrodynamic approach that casts the microscopic quark equilibration dynamics as the relaxation evolution of an effective macroscopic bulk viscous pressure. The evolution of this effective bulk viscous pressure captures quark chemical equilibration and the breaking of conformal symmetry, thereby providing a smooth onset of the lattice QCD equation of state after a few fm/$c$ in the hydrodynamic evolution. We investigate phenomenological effects of this macroscopic description on final-state observables in $\mathrm{Pb}+\mathrm{Pb}$, $\mathrm{O}+\mathrm{O}$, and $\mathrm{p}+\mathrm{Pb}$ collisions at the Large Hadron Collider.

Humans can grasp everyday objects with almost perfect success rates using fingertip tactile feedback, yet much of the robotic grasping literature emphasizes vision-based grasp selection with parallel grippers. In this work, we systematically investigate how high-resolution, dynamic tactile sensing contributes to grasp stability prediction and model-guided grasping in dexterous robotic hands. To this end, we collected a dataset of 10,000 grasp trials across 200 objects using a multi-fingered robotic hand equipped with four Digit 360 tactile sensors, recording external vision, proprioception, and tactile streams throughout each grasp. With this dataset, we trained end-to-end temporal multimodal models to predict post-lift stability from pre-lift grasp observations and compared sensing modalities and encoding backbones. Experimental results and controlled input ablations show that incorporating touch, and particularly high-resolution, dynamic touch, improves grasp stability prediction. Finally, we deployed the learned predictor as an online stability gate on the real robot, where visuo-tactile model-guided regrasping improved the success rate among executed lifts by 10.5 percentage points over a non-tactile gate. These results show how rich fingertip sensing and expressive temporal models that capture the dynamics of touch can support learned grasping with multi-fingered hands without explicit contact or force modeling, providing a scalable data-driven path from tactile experience toward stable dexterous manipulation. The dataset is publicly available at https://lasr-lab.github.io/dexterous-grasp-stability/.

We present a deterministic reduction from min-sum subset convolution to min-plus matrix product. We show that if the min-plus product of two $D\times D$ matrices with $\beta$-bit integer entries can be computed in $D^{3-\delta}\operatorname{poly}(\beta,\log D)$ time for a fixed rational $0<\delta<1$, then min-sum subset convolution on an $n$-element universe can be solved in $(2+2^{-\delta})^n 2^{O(\sqrt n\log(n+1))}\operatorname{poly}(n,\beta)$ time. Instantiating this reduction with the recent breakthrough on subcubic min-plus matrix product by Alman and Vassilevska Williams gives a Las Vegas algorithm with expected running time $O^*(2.9987^n)$ and a deterministic algorithm with running time $O^*(2.9997^n)$, strictly breaking the longstanding $3^n$ computational barrier. Notably, these speedups translate directly to database query optimization, yielding the same expected and deterministic running-time bounds for join ordering under the $C_{\mathrm{out}}$ cost function.

A task abstraction can specify the intended events while its spatial layout prevents a fixed agent and controller from completing them. Starting from a supplied structured task record, we compile whole-task tracking, clearance, and actuation requirements into auditable affine layout constraints. We repair only declared continuous coordinates, preserving event order, timing, topology, and the controller. A most-violated-row update admits conditional finite-certification and net-displacement bounds; a same-compiler quadratic projection separates the representation from the optimizer. On three researcher-authored task abstractions, both backends certify all three layouts and complete all 300 fresh paired rollouts per backend. A risk-target sweep also exposes fixed event tests that the chosen certificate cannot satisfy through layout edits alone.

We study the low-frequency scattering of gravitational and pseudoscalar waves off a non-rotating black hole in dynamical Chern-Simons (dCS) gravity, where the axial (odd-parity) gravitational sector is irreducibly coupled to the dynamical pseudoscalar while the polar sector remains general relativistic. Working to first order in the dimensionless coupling zeta = (ell_CS/M)^4 and using matched asymptotic expansions, we obtain an exact closed form for the graviton-to-pseudoscalar conversion probability, d_2 = 3(pi^2-9)^2/8, and a high-precision reconstruction, in a proven transcendental basis, of the graviton greybody (absorption) asymmetry, kappa_2 = (21/80) pi^4 + (27/8) pi^2 - 89833/1536, for the quadrupole (ell=2) channel. Both reduce to combinations of pi^4, pi^2, and rational numbers, with no ln2 or odd-zeta terms in the reconstructed basis; the transmission-phase sector, by contrast, contains an analytically derived zeta(3)-9/8 static integral, while its overall matching prefactor is identified numerically. The leading reflection-phase asymmetry vanishes, so that no polarization rotation is imprinted on the reflected wave at leading low-frequency order, while a subleading O(omegazeta) asymmetry remains in the transmitted phase. The normalized flux asymmetries are numerically consistent with a 1/(Momega)^2 falloff in the geometric-optics limit, where the leading remaining flux effect is conversion loss. The matched-asymptotic results are checked against direct two-channel integration, while the one-dimensional integral determining kappa_2 is evaluated to high precision within an analytically controlled transcendental basis. The results provide analytic targets linking dCS parity violation in absorption and mode conversion to the axial/polar quasinormal-mode splitting.

Deep thermalization indicates the emergence of universal ensembles of pure states in a subsystem after measuring its environment. We investigate this phenomenon in a free-fermion chain using local measurements that preserve Gaussianity and break particle-number conservation. Although the measurement outcomes provide partial information about local conserved charges within a finite subsystem, we argue that this information becomes negligible in the thermodynamic limit. This effective non-revelation underlies the emergence of a fermionic Gaussian Scrooge ensemble, determined by the subsystem's generalized Gibbs ensemble. We construct this universal ensemble directly at the subsystem level and provide numerical evidence that it describes the late-time projected ensemble at finite temperature. Our results extend the description of thermalization in non-interacting integrable systems beyond the reduced density matrix to universal statistics of conditional pure states.

We propose a neural algorithm for sampling from distributions specified by unnormalized Boltzmann densities. Our approach is based on the Jordan–Kinderlehrer–Otto scheme for the Kullback–Leibler divergence in the Wasserstein–Fisher–Rao geometry (WFR JKO scheme). Our contributions are twofold. First, we prove that, for any fixed step size, the exact WFR JKO iterates converge exponentially fast to the target as the number of iterations tends to infinity. Notably, this result requires no structural assumptions on the target, such as log-concavity or a logarithmic Sobolev inequality. Second, we develop a neural implementation of the WFR JKO scheme that parametrizes its transport and reaction components using reweighted normalizing flows. Numerical experiments on challenging multimodal targets demonstrate the promising performance of the proposed method.

Statistical-mechanical analyses predict that MacKay-Neal codes can achieve channel capacity at fixed degrees. We prove this prediction for the uncoupled $(\ell,3,3)$ MN socket ensemble for every fixed integer $\ell\ge4$. For each binary-input memoryless symmetric channel of capacity strictly greater than $3/\ell$, the ensemble-average block error probability under maximum a posteriori (MAP) decoding is $O(\log N/N)$, where $N$ is the transmitted blocklength. The actual transmitted rate converges to $3/\ell$. The result includes both punctured and transmitted variables and does not condition the sparse square matrix on invertibility. The proof establishes an entropy inequality for six bits subject to even parity by an analytic argument that reduces the domain by symmetry, restricts its interior stationary points, where both partial derivatives vanish, to the diagonal, and controls the resulting scalar functions by explicit polynomial bounds. Exact configuration counts then bound the conditional entropy on the binary symmetric channel. A standard comparison by mutual information extends the entropy bound to general symmetric channels; independent output erasures and a minimum-distance estimate for the transmitted code yield the bit and block error bounds.

An LLM safety patch can pass a benchmark while still being a poor repair. This risk is especially acute for jailbreak repairs, where the goal is to correct a specific unsafe behaviour without changing unrelated behaviours. A patch may block exact evaluation prompts yet fail on close harmful variants, or suppress harmful behaviour by over-refusing benign prompts that share its wording or structure. Existing protocols primarily test whether models can be broken, while aggregate metrics (attack success, refusal rates, global capability) cannot distinguish selective repairs from broader local suppression. To address this gap, we introduce PatchBench, a benchmark of empirically observed model-specific jailbreak failures inducing actionable harmful answers. Starting from 27,870 prompts from 37 public datasets, we curate 15,314 English prompts and query 8 open-source instruction-tuned models. Combining WildGuard filtering, pairwise Elo ranking, and manual verification, we retain a curated bank of 400 high-confidence jailbreak failures. We further introduce PatchBench-Local, an evaluation protocol testing whether a patch is behaviourally precise. For each harmful source prompt, PatchBench-Local generates three families of local neighbours: harmful variants preserving malicious intent, benign prompts with matched structure, and benign prompts reusing key harmful terms. It evaluates harmful-neighbour correction and benign-neighbour preservation, distinguishing selective repair from broader local suppression. Evaluating four activation steering methods with PatchBench-Local and MMLU shows that global capability can remain nearly unchanged while local benign regressions are severe, confirming aggregate metrics miss important collateral damage. PatchBench-Local provides a more precise basis for developing and comparing jailbreak repair methods.

We propose interaction-mediated control via Active Inference as a general architectural approach to high-dimensional control problems in Human–Computer Interaction (HCI). This architecture recasts user interaction as the provision of evidence about a latent task objective, rather than the direct specification of plant-control inputs. The mediating function is distributed between an interaction broker, which selects informative user queries and performs Bayesian inference over user preferences, and an Active Inference controller, which autonomously plans and acts under the resulting preference information to control the plant. This division of labour decouples the semantics of user interaction from those of low-level plant control. We instantiate the architecture in a simulated, multi-link robotic arm to perform a simultaneous whole-arm target-coverage task. A simulated user communicates exclusively through a clutch-style binary evaluative channel, without specifying joint-torque commands. Across increasing arm dimensionalities, the architecture achieves successful interaction-mediated control, although task success is lower than when the controller receives the true target preferences directly. Exact target-subset identification also remains imperfect, highlighting the distinction between preference inference and successful task completion. The experimental findings provide an initial computational demonstration of the proposed architecture under controlled, matched-model assumptions and motivate its further investigation in broader HCI applications.

Token-based world models enable fine-grained latent planning, but repeatedly processing large spatial token grids makes action search expensive. We introduce COSTGRAD, a training-free, goal-conditioned selector that ranks spatial tokens by the gradient norm of the planning cost with respect to each input token. By deriving importance from the downstream control objective, COSTGRAD targets tokens that matter for planning rather than merely for prediction. On AdaLN-conditioned predictors at $50\%$ sparsity, COSTGRAD matches or exceeds full-token planning on three of four continuous-control benchmarks, while giving a measured $2.6\times$ wall-clock speedup per environment planning step. Combining token sparsity with reduced CEM search increases this to a $\sim 5\times$ total speedup while still exceeding the full-token baseline. We also identify an architecture-dependent failure mode: in a matched AdaLN-vs-concat comparison, concat maintains comparable full-token performance but pure COSTGRAD loses its advantage over random selection. This difference tracks action-pathway drift: gradient-selected removal produces less drift than random removal on AdaLN, but more on concat. These results highlight selector-architecture compatibility as a design axis for sparse world-model planning. Project page and demos: https://ycxuyingchen.github.io/costgrad/

Despite its widespread use, Proximal Policy Optimization with clipping (PPO-Clip) remains difficult to tune, and the interactions among critic learning, clipping, and rollout reuse remain incompletely understood. We develop a non-asymptotic analysis of PPO-Clip as a closed-loop actor–critic system. It captures actor–critic coupling, nonsmooth probability-ratio clipping, finite-batch reuse, and predictable early stopping under explicit coverage and critic regularity assumptions, using raw GAE and Monte Carlo critic targets. Our synchronous and asynchronous guarantees jointly characterize policy stationarity and the tracking accuracy of the learned critic, with explicit dependence on algorithmic parameters. A sufficient coupling condition gives optimization, critic tracking, clipping, and finite-batch errors a common amplification bound. The asynchronous result also requires a delay-dependent critic stepsize restriction; violating these conditions does not establish divergence. For finite layered MDPs with tabular critics, a uniform bound on the actual clipped-gradient class replaces complete-trajectory counting. A verified growing-horizon family has polynomial sample complexity, and a two-time-scale schedule gives $O(T^{-2/5})$ stationarity and critic-tracking bounds with explicit fresh-rollout accounting. These results together advance our understanding about PPO and provide theoretical guidance in tuning.

We study the minimum monetary subsidy required for exact envy-freeness in allocating $m_n$ indivisible goods to $n$ agents. Valuations are additive, item values are i.i.d. from a distribution on $[0,1]$ with density bounded above and away from zero, and all agents value money equally. We consider $n\to\infty$ with $m_n=qn+r_n$, where $q\ge0$ is a fixed integer and $0\le r_n<n$. For the nonzero-remainder case where $1\le r_n<n$, we give a cubic-time algorithm based on a maximum-weight matching of balanced bundles and complementary dual prices. It returns an envy-free outcome for every instance. Under the random model, both its payment and the unrestricted minimum are $n-r_n+o_p(n)$ for $q\ge1$. When $q=0$ and $r_n/n$ converges to a limit below one, the error term improves to $O_p(1)$. For exact divisibility, where $r_n=0$, prior work gives zero subsidy with high probability for $q\ge2$. The square case $q=1$ is exceptional. Its minimum subsidy is $\Theta_p(\log n)$ and, with high probability, equals the optimum over allocations in which every agent receives exactly one good. This restricted optimum can be computed in cubic time using a maximum-weight perfect matching and all-pairs shortest paths. Together, these results provide a unified asymptotic characterization of the minimum subsidy.

Interval localization in multiple change-point detection aims to identify intervals containing true change-points while controlling false discoveries. We formulate this task as a multiple testing problem over detector-reported intervals and develop an algorithm-agnostic and distribution-free framework that controls the interval-level FDR, the expected proportion of retained intervals containing no true change-points. The proposed permutation-based order-preserving interval selection (POIS) uses order-preserving sample splitting to separate detection from permutation testing. To improve power, POIS+ reuses the detection-side evidence and introduces detection-conditional $e$-values as unnormalized weights for permutation $p$-values. Both procedures achieve finite-sample FDR control. Under the stated conditions, POIS attains asymptotically full power while POIS+ has asymptotically no lower power than POIS. Simulation results show that POIS maintains high power and POIS+ further improves power with empirical FDR below the nominal level for both procedures. Two real data applications demonstrate the practical effectiveness of the proposed methods.

World action models (WAMs) have emerged as an important class of generalist robot policies, aiming to transfer video prediction priors to action learning. However, we find that existing WAMs frequently produce incorrect motion predictions in open-ended environment, leading to erroneous actions. We attribute this limitation to two factors: (1) base video models are not optimized for manipulation, and (2) naively incorporating action components into video models can substantially degrade their generalization capabilities. We introduce Video Prediction Policy 2 (VPP2), a WAM that enables strong zero-shot generalization in both video prediction and action generation. First, we curate a large-scale, diverse dataset of manipulation videos to continue pretraining the base video foundation model. We annotate video clips with detailed captions and perform event-level video pretraining to promote generalization across open-ended manipulation tasks. Second, we post-train and distill the video model into a single-step visual planner with fixed prediction horizon. Finally, we introduce action module via a mixture-of-transformers (MoT) architecture to learn implicit inverse dynamics model. Experiments demonstrate three key results: (1) VPP2-14B outperforms Cosmos3-64B by 11.0% points in video prediction instruction-following success rate on open-ended tasks; (2) VPP2 surpasses the strongest baseline by 18.5% points in success rate on real-world zero-shot ALOHA manipulation tasks; and (3) following benchmark-specific post-training, VPP2 achieves the highest success rates among evaluated methods on the challenging LIBERO-Pro, LIBERO-OOD, and RoboDojo benchmarks.

The Melvin–Morton–Rozansky theorem relates the large-colour semiclassical limit of the coloured Jones polynomial to the inverse Alexander polynomial. Bar-Natan and Garoufalidis [BNG96] proved the theorem using weight systems and, more generally, established the corresponding formula for arbitrary complex semisimple Lie algebras. We give a direct proof of this semisimple Lie algebra generalization by analysing a state sum. For a complex semisimple Lie algebra $\mathfrak{g}$ and a dominant integral weight $\lambda$, we use the embedding $V_{d\lambda}\hookrightarrow V_{\lambda}^{\otimes d}$ to analyse the state sum. Under the specialization $q=e^{h/d}$, we show that, as $d\to\infty$, only the identity terms and those proportional to $E_{\alpha}\otimes F_{\alpha}$ contribute at leading order. It follows that the leading-order contribution decomposes over the positive roots, with each root contribution evaluated as in the $\mathfrak{sl}_{2}$ case. Writing $t_{\alpha}=e^{-2(\lambda,\alpha)h}$, the limit is therefore given by the product of $\Delta_{K}(t_{\alpha})^{-1}$ over the positive roots $\alpha$.

Electronic structure descriptors have been widely used to rationalize catalytic activity, but their application to CO$_2$-to-CH$_4$ conversion in perovskite oxides remains relatively limited. Here, we investigate six electronic structure descriptors related to band centers and bandwidths. We find that the unoccupied transition-metal (TM) $3d$-band center, charge-transfer energy, and the ratio of the unoccupied TM $3d$ to occupied O $2p$ bandwidths, $W_{3d}/W_{2p}$, exhibit negative correlations with CH$_4$ activity. These qualitative trends are further quantified using a parametric brute-force searching (BFS) approach to obtain explicit fitting equations. Notably, the selected equations depend only on $W_{3d}/W_{2p}$, providing an electronic structure perspective on the enhanced activity in double perovskites. We apply the selected equation to screen 24 La-based double perovskites. This screening suggests that La$_2$CoGaO$_6$, La$_2$CoAlO$_6$, and Ti-containing compositions are promising candidates. The present approach can help identify descriptor-activity relationships and guide materials screening in perovskite oxides.

Cosmic noon ($z\sim2$) marks the peak epoch of cosmic star formation and black hole accretion activity, where strong accretion and outflows are expected. We report two candidate absorption features in the XMM-Newton spectrum of the quasar WISEA~J094835.95+432302.6 at $z=1.893$. Two redshifted features occur at rest-frame energies of $\sim4.65$ and $\sim5.75$keV with confidence levels of 99.3% and 94.1%, respectively, estimated from Monte Carlo (MC) simulations considering the look-elsewhere effect. These features can be explained by the Fe XXV/XXVI K-shell transitions from one ultra-fast inflow (UFI) with an apparent velocity of $0.37^{+0.02}_{-0.01}c$. If confirmed by follow-up XMM-Newton observations, these detections would extend reported UFI candidates from the local Universe to cosmic noon, offering a rare laboratory for black hole accretion at the peak of their growth.

Dense self-expression matrices and full-affinity spectral clustering limit the scalability of subspace clustering. We introduce the Latent Orthogonal Optimization Model for Subspace Clustering (LoomSC), a framework that addresses both bottlenecks through projector factorization and exact spectral reduction. Motivated by the spectral structure of least-squares regression, LoomSC jointly learns latent features and a projector self-representation through two thin factors. Alternating Procrustes and least-squares updates preserve the sample factor's orthogonality while keeping the coefficient matrix implicit. We construct a nonnegative quadratic affinity that preserves the projector's support. An exact feature map then reduces its normalized spectral problem to an eigenproblem whose dimension depends only on the factor width. Neither the full affinity nor the sample Laplacian needs to be formed. Our analysis quantifies the projector approximation and identifies conditions for subspace preservation and within-subspace connectivity. For fixed dimensions and iteration budgets, the complete pipeline has linear time and memory complexity in the number of samples. Across five image-clustering benchmarks, LoomSC ranks first or second in all 15 dataset-metric comparisons against 9 state-of-the-art baselines. Its mean accuracy exceeds the highest baseline mean by 6.66 percentage points. Synthetic experiments scale to 500,000 samples while maintaining at least 99.8% accuracy.

Personal memory for language agents is usually judged by whether the final an- swer is correct. That score hides errors that arise before generation: the memory block may contain an obsolete value, a fact about the wrong person, or no use- ful fact before the serving deadline. We measure these failures directly. Using Personal Fact Memory (PFM) as a reference layer, we find that temporal validity is primarily a property of memory construction in our setting. On a controlled revision benchmark, serving only the active value of each correctly keyed slot eliminates observed stale exposure; without update resolution, 70.3% of prompts expose a superseded value. Once retrievers share the same active store and par- ticipant information, participant-aware BM25 is equivalent to the reference ranker within a prespecified 0.02 margin. The harder problem is assigning revisions to the right slot. Missed merges leave stale values active, whereas false merges silently remove current values; four LLM key assigners achieve higher key re- call than a rule extractor yet produce lower clean-retrieval rates, and open-domain merge recall on LongMemEval never exceeds 0.062. Misattribution survives va- lidity filtering: an entity posterior reduces same-name exposure on controlled data but cannot distinguish identically named speakers in LoCoMo. Two frozen lan- guage models reproduce prompt errors in generated text. Retrieval latency varies across rankers, but prompt prefill dominates turn-level latency on our hardware. These results argue for evaluating agent memory before generation, separating stored-state validity, identity resolution, abstention, and serving latency.

Multimedia editing requires generative models to interpret complex, compositional instructions and modify only the desired elements across modalities, a capability largely beyond existing systems. Current approaches are typically limited to simple, single-attribute edits within one modality, since diverse, high-quality editing pairs are hard to obtain, especially for cross-modal tasks where aligned audiovisual (AV) data is scarce. We present CrossEdit, a unified omni-modal editing model for images, audio, and video that performs AV movie scene edits zero-shot through cross-modal transfer. Our key observation is that complex instruction-following, once learned in any modality, generalizes to others. We exploit the additive nature of acoustic signals to procedurally generate large-scale audio editing pairs with compositional instructions, and show that fine-tuning on this synthetic data and self-supervised AV masked reconstruction, alongside a targeted set of curated cross-modal tasks, induces instruction-following that transfers zero-shot to unseen modality and instruction combinations. To evaluate this capability, we release CrossEditBench, a human-annotated benchmark of AV edits on movie scenes, and propose AV-FES, a metric that jointly scores instruction following and consistency. We show that our proposed techniques improve performance on zero-shot AV movie scene editing, while maintaining or improving performance on lip-synced speech editing and standard image, video, and audio editing benchmarks. Demos are available at https://wanchichen.github.io/crossedit/.

As coding agents advance from bounded software engineering tasks toward long horizon development, dynamic concurrency offers a promising way to scale complex development tasks. Under this policy, agents decide during execution whether and how to spawn concurrent sub-agents. Model capability largely determines outcomes on shorter tasks, whereas long horizon development makes orchestration central to task completion. Existing work, focused on coding agent failures on shorter tasks or collaboration in predefined multiagent workflows, offers little insight into dynamic concurrency in frontier agents across task complexity. We study dynamic concurrency as an execution policy through controlled comparisons of matched Codex, Claude Code, and Kimi Code executions with the policy enabled or disabled. Across 354 tasks and 2,124 executions spanning a range of task complexities and execution horizons, we evaluate its end to end effects and scheduling behavior, and analyze matched trajectories to characterize 13 concurrency specific failure modes, 28 observable patterns, and the conditions under which it provides an advantage.

Kotelskiy, Watson and Zibrowius show that the Bar-Natan invariant $\unicode{x0414}(T)$ associated to a $4$-ended oriented tangle $T$ extends to a twisted complex $\unicode{x0414}^\infty(T)$ over an $A_\infty$-enhancement of the Bar-Natan algebra. We prove that $\unicode{x0414}^\infty(T)$ itself is a tangle invariant, settling a conjecture from their work. We deduce this from an invariance statement for a matrix multifactorization $\mathcal M(T)$, generalizing a construction of Ballinger to the coefficient ring $\mathbb Z[G]$. This multifactorization is well defined up to a notion of 1-homotopy equivalence. The proof rests on explicit formulas for special deformation retracts of Koszul matrix factorizations, which realize delooping in this setting and may be of independent interest.

Context: Once defined a taxonomy of stages structuring Machine Learning (ML) pipelines (e.g. Data Preprocessing, Modeling...), extracting these stages from source code is key for better understanding ML practices. However, the diversity caused by the constant evolution of ML (e.g., algorithms, datasets) makes this task challenging. Existing approaches either rely on non-scalable manual labeling or on classifiers that do not properly support domain's diversity. These limitations call for more reliable solutions. Objective: We evaluate whether Small Language Models (SLMs) can leverage their code understanding and classification abilities to address these limitations, and enhance our understanding of practices in ML. Method: We conduct a confirmatory study based on two relevant reference works representing current limitations in the state-of-the-art. We first compare several SLMs using Cochran's Q test, then evaluate the best-performing model against reference studies via two McNemar's tests. An additional Cochran's Q test examines how taxonomy definition variations affect the SLM performance. Finally, goodness-of-fit tests compare ML practice insights from SLM classification with those from prior studies. Results: First, we found that the taxonomy wording significantly impacts classification performance. Second, the best performing SLM yielded good results, yet, without outperforming other classifiers. Third, the three classification methods led to significantly different insights, with varying effect sizes, when exploring practices of data scientists. Conclusions: Limitations of existing classification methods bias our understanding of ML practices. While current SLMs show promising results without prior fine-tuning, they still exhibit common limitations, in addition to inference high costs challenging their applicability in large-scale studies.

Intrusion detectors can confidently misclassify attacks that were not seen during training. Human review can correct these errors, but only a limited number of cases can be checked. Uncertainty-based review may overlook confident errors, while anomaly scores alone do not show whether changing the review plan will correct more errors. We introduce PairAudit to find overlooked errors and improve review under a fixed budget. Its graph tokens capture prediction patterns across connected nodes. Rather than building another predictor through feature aggregation, PairAudit uses unusual relational patterns to uncover potential errors in existing predictions. Human feedback then helps decide whether these findings justify changing review priorities. Experiments across security tasks show that PairAudit corrects more errors on average than uncertainty-based review, including more errors on unseen attacks. These gains account for all review costs and do not require retraining the detector.

Non-Markovian open quantum networks are among the most challenging problems in quantum theory, as they suffer from both, the exponentially scaling many-body state space as well as long memory times when the sites are strongly coupled to structured environments. Recent efforts to address them by combining tensor networks with real-time path integrals, such as TEMPO and process tensors, were however limited to systems with rather weak system-environment couplings because strong couplings lead to large bond dimensions. Here, we devise a tensor network scheme (CTEMPO) resulting in particularly small bond dimensions for single-site problems. We then extend this to a practical scheme to simulate non-Markovian quantum networks by time-evolving a corresponding comb tensor network. Benchmarking the single-site algorithm on the spin-boson model we find orders-of-magnitude improvement over established methods. The multi-site algorithm is shown to solve the notoriously difficult problem of the 7-site Fenna-Matthews-Olson photosynthetic complex, whose spectral density contains 62 peaks. We then apply our approach to explore the scaling behavior of the superradiance of up to $N=20$ quantum dots individually coupled to super-Ohmic phonon baths. Our approach provides a turnkey solution to non-Markovian quantum networks and is compatible with more general tensor network topologies.

Quantum libraries are now critical infrastructure for quantum algorithm development, yet their correctness remains difficult to test. Existing testing techniques mainly rely on failure-based or comparison-based oracles, exposing bugs only when executions fail, violate runtime checks, or disagree with another implementation. Their applicability is limited when suitable execution-based oracles are unavailable, leaving some silent bugs undetected. Such missed bugs can produce incorrect results that propagate into experimental conclusions, simulation studies, and algorithmic designs. Here we present QuSema, an autonomous testing agent for finding silent bugs in quantum libraries. QuSema uses constraints from quantum semantics and documentation as a source-level semantic oracle to assess whether implementation logic can produce invalid outputs from valid inputs. It operates through an agentic loop that repeatedly inspects library API documentation and source code, reasons about the intended behavior of quantum operations, identifies potential semantic deviations, and validates them by generating executable tests through library APIs. Guided by quantum-domain reasoning, QuSema turns high-level behavioral mismatches into concrete, user-triggerable bug reports, enabling it to uncover non-crash defects. We implement QuSema for Qiskit and PennyLane. On a benchmark of 20 historical silent bugs, QuSema achieves higher mean bug relocation counts than Claude Code and Codex, with the DeepSeek configuration costing less than Claude Code. QuSema also discovers 40 previously unknown bugs confirmed by the developers, including 30 silent bugs.

Entropy-regularized optimal transport (EOT) has become a key tool in optimal transport, providing both theoretical insight and efficient computation via the Sinkhorn algorithm. We study the first-order \(\Gamma\)-expansion of EOT as the regularization parameter \(\varepsilon\) vanishes, by rewriting EOT as the minimization of a relative entropy with respect to a Gibbs measure \(\gibbs\) associated with the duality gap energy. Using a mild extension of the Laplace method, we prove the convergence of the Gibbs measures and derive this \(\Gamma\)-expansion for non-degenerate ground costs, under suitable assumptions on the duality gap, the contact set and the marginals. This yields at the same time an entropic selection principle and an expansion of the EOT cost up to \(o(\varepsilon)\) for this class of ground costs, generalizing existing results.

Reality may establish that an outcome occurred without identifying which evolving procedure produced it or why. This distinction matters in production ML systems whose code, configuration, and artifacts change while external feedback accumulates. We examine it in a human-directed, agent-engineered quantitative trading system, using oracle to mean an external source of realized outcomes rather than a complete correctness specification. Across one year, the account gained and outperformed a broad market index, while annual alpha was not statistically distinguishable from zero under the main retrospective specification. Retrospectively selected subperiods include adverse relative performance and conditional candidate-level weakness under declared approximate references. Engineering records document changes during the episode, and complete recommendation-to-runtime binding is unavailable. The archive does not establish a common frozen instance or a unique cause. The case motivates an outcome–diagnosis gap: outcome evidence, evaluated-object identity, and causal explanation support distinct claims. We distinguish frozen instances, pre-specified adaptive procedures, and ad-hoc development; organize archive-relative claim identifiability and an evidence hierarchy; and propose a prospective production-binding protocol. An illustrative compatible-history example shows how factual binding can resolve a recommendation's referent without supplying its counterfactual effect. The protocol is proposed rather than prospectively validated. External feedback constrains outcome claims, while provenance and additional identification structure determine the resolution of diagnosis.