Papers

Optimal circuit synthesis is the problem of finding the shortest-depth circuit representation of a given functionality with respect to a pre-specified elementary gate set. This is a central problem in both quantum and classical hardware design. While the classical version is very well understood – both in terms of heuristics and rigorous hardness assertions – much less is known about optimal quantum circuit synthesis. We focus on optimal Clifford circuit synthesis, for which various heuristics are known, e.g. via reduction to 3-SAT. Our main result supplies a matching hardness result: a variant of optimal Clifford synthesis is NP-hard. The proof proceeds in two parts: (i) reduce 3-edge colorability on 3-regular graphs to a circuit synthesis problem that only involves CZ gates, (ii) prove that the availability of additional elementary Clifford gates – most notably: Hadamard, phase and CNOT – cannot lead to further improvements of the optimal circuit depth. Our work sharpens the complexity-theoretic understanding of Clifford circuits: simulation and equivalence checking are in P, whereas deciding whether a Clifford unitary admits an implementation within a prescribed depth is NP-complete.

Blind extragalactic radio continuum surveys are increasingly conducted at sub-arcsecond angular resolutions. However, these surveys contain substantially more spurious sources than expected from purely Gaussian noise statistics. We investigate the origin of these false detections using the 10GHz GOODS-N mosaic obtained with the Karl G. Jansky Very Large Array (VLA), which reaches a native angular resolution of 0.22arcsec and exhibits a nearly Gaussian pixel-noise distribution. We show that part of the increase in the spurious-source fraction at high angular resolution arises naturally from the larger number of independent resolution elements, while residual wide-field imaging effects associated with the primary-beam response make only a modest contribution. Nevertheless, even after accounting for these effects, the observed fraction of false detections remains several times higher than predicted for Gaussian noise. An analysis of the noise autocorrelation function reveals that images with relatively sparse uv coverage retain significant dirty-beam sidelobes that introduce correlated noise fluctuations, substantially increasing the probability of false detections. We conclude that the excess of spurious sources is an intrinsic consequence of incomplete uv sampling rather than the imaging methodology itself, highlighting that future deep radio continuum surveys require not only high angular resolution but also sufficiently dense uv coverage to minimize spurious detections.

We give a column-avoidance modification of the no-least-join construction of Downey, Greenberg, Lewis, and Montalbán. We show that if $C$ is computably enumerable, $A,B<_T C$, and $A\not\leq_T B$, then for every noncomputable set $E$ there is a set $X\leq_T C$ such that $B\leq_T A\oplus X$ and $E\not\leq_T X$. Consequently, for the corresponding Turing degrees $a,b,c$, the degrees $x\leq c$ satisfying $b\leq a\vee x$ have no nonzero common lower bound; equivalently, no nonzero Turing cone contains all of them. This gives a negative answer to their Question 1.19.

Layered viscoelastic interiors control the frequency dependence of planetary tidal dissipation, but repeated layered Love-number calculations are expensive in long-term orbital and spin integrations. Equivalent homogeneous rheologies provide a cheaper representation of the same response, but their effective parameters do not by themselves identify the interior layers they represent. We develop a local interpretation of such parameters by differentiating the layered-to-homogeneous construction about a fixed-geometry baseline, perturbing only selected layer rigidities and viscosities. We use the pole-residue representation of the degree-two Love number to construct a physical-to-effective transfer matrix. We then use its singular values, resolution matrix, and mode-dominance matrix to determine which physical perturbations are visible in the Love-number response and how they appear in effective-rheology coordinates. Applied to a five-layer lunar model, the full degree-two response is sensitive to seven of eight solid-layer rheological parameters. The single near-null direction is the inner-core rigidity, consistent with shielding by the fluid outer core and with the weak dependence of Love numbers on core elastic structure. Over the LLR frequency interval, fewer physical directions are visible. To first order, the elastic spring, isolated dashpot, and two Voigt elements reproduce the nine-pole response over this interval. Finite perturbations confirm the tangent prediction to within a few percent, with the validity radius limited by a nearly degenerate pair of long-timescale modes.

Typically, multiple copies of a quantum state are described by a tensor power of that state. However, from an operational point of view, an experimentalist cannot, in general, rule out imperfections at unknown positions or correlations between copies, so assuming independent and identically distributed (iid) copies is not justified a priori. In the following, we relax the iid assumption by considering a whole set of resource states that are fundamentally indistinguishable from a tensor power state, and in that sense look like many copies of the quantum state to the experimentalist. Using a de Finetti-type argument, we show that this operational set of states can asymptotically be represented by almost-iid states. As an application, we introduce operational analogues of asymptotic transformation rates, such as entanglement cost and distillation. This raises the question of whether these rates coincide with their iid counterparts.

Nonlinear nanophotonics enables the engineering of optical functions using light-matter interactions in nanoscale structures with carefully designed geometries. Phase matching in nonlinear optics is typically essential for realizing high efficiency for desired processes and engineering of sophisticated functions, but it can be challenging to reliably achieve in nanophotonic waveguides. Conventionally, phase matching in nanophotonics is achieved through nanofabrication, e.g., by poling a device's nonlinear material using electrodes with fixed geometry. This approach has two drawbacks: the optical functions a device can perform are fixed at the time of fabrication, and the device's performance can be degraded by fabrication imperfections. Here, we circumvent these limitations by engineering programmable $\chi^{(2)}$ holograms inside nanophotonic waveguides, enabling reconfigurable quasi-phase matching that can be updated after device fabrication in approximately one second. By projecting structured optical illumination onto photoconductive electrodes, we generated spatial patterns of electric-field-induced $\chi^{(2)}$ to exert spectral, modal, and polarization control over three-wave mixing processes. This programmability also allowed us to employ closed-loop in situ optimization to compensate for phase mismatch in complex waveguide structures, such as width-modulated waveguides and spiral waveguides exceeding 10 cm in length. Furthermore, we show that programmable nonlinearity enables powerful in situ diagnostics, including direct dispersion measurements of various transverse modes and on-chip wavefront phase tomography. Our work opens up the possibility of building cascaded nonlinear nanophotonic systems where yield issues due to fabrication variations can be sidestepped while retaining the benefits of etched waveguides such as confinement and dispersion engineering.

While Large Language Models (LLMs) advance 3D indoor scene synthesis, current pipelines fail to retain user-specific preferences across sessions, making immersive authoring a repetitive and physically fatiguing process. We present SPHERE, an adaptive VR generation framework that transforms isolated synthesis into continuous human-AI co-creation. SPHERE extracts persistent spatial preferences from natural multimodal interactions (speech and controller edits). To ensure geometric resilience against spatial distortions, it abstracts these raw edits into hierarchical constraints modeling both local functional and global topological contexts. Furthermore, a human-in-the-loop reinforcement learning mechanism dynamically updates retrieval policies based on the user's final edited scenes. A mixed-design user study ($N=42$) and an offline ablation demonstrate that SPHERE significantly reduces corrective edits and physical demand, preventing bias toward shallow object-level traits to yield geometrically resilient, profile-aligned layouts. Ultimately, SPHERE demonstrates how capturing demonstrated spatial logic enables controlled spatial adaptation, establishing a reliable, governed human-AI collaboration framework for immersive authoring. Project page and source code will be available at: https://github.com/hyeonmin11/SPHERE

AI-based ideation systems are often evaluated on the novelty of their ideas, and that judgment is increasingly delegated to large language models. Such judges are typically built ad hoc and validated, if at all, on human-authored papers rather than on the generated ideas they are meant to score. So, how do novelty judges perform? Not well. We present a systematic controlled study of novelty evaluation design choices. We build an evaluation set automatically, mining OpenReview for passages where reviewers explicitly affirm or dispute a paper's originality and keeping only submissions with unanimous agreement at the extremes of their research area; we pair these with ideas from a vanilla LLM generator. Across six judges, we find that seemingly small prompt design choices have large consequences; e.g., simply telling the judge that reviewers found one idea novel and the other not can change its verdict on more than half of the identical idea pairs it is shown, shifting pairwise accuracy by over 50 points and occasionally pushing it below chance. The same change helps one judge and hurts another. Retrieval and larger reasoning budgets help little, and two purpose-built novelty evaluators are outperformed by our cheapest prompted baseline. These results raise questions about reported novelty gains of automated ideation systems, and call for robust novelty evaluation methods.

Recent vision-language models (VLMs) exhibit remarkable generalization and reasoning abilities, yet 3D understanding in these models is limited by data scale, training diversity, and reasoning capacity. Instead of naively extending these models into 3D, we take a different approach: we enable powerful 2D VLMs to operate reliably in 3D by introducing 3D grounding and iterative feedback loops with two novel concepts: Canonical Coordinate Framing (CCF) and Task-Adaptive Feedback (TAF). CCF serves as a unified visual representation that anchors both inputs and outputs to a shared Euclidean coordinate system, solving common challenges in 3D grounding such as axis ambiguity, inconsistent metric scale, and floating references. Complementary to this structured framing of the 3D inputs, TAF closes the reasoning loop with task-adaptive dynamic feedback that enables 2D VLMs to perform varied open-vocabulary tasks within their native visual context. Building on this foundation, we introduce 3D-Prog, a 3D understanding, reasoning, and generation framework that jointly employs the capabilities of CCF and TAF together with powerful VLMs. Without requiring any retraining, 3D-Prog performs open-vocabulary 3D understanding, manipulation, and generation across both object-level and scene-level tasks. Our experiments show that the joint use of CCF and TAF transforms 2D VLMs into geometry-aware 3D programmers, achieving consistent, interpretable, and high-quality results across diverse 3D tasks.

A net of conics can remain unchanged as the four-point configuration generating it varies. We study this phenomenon over a field ${\mathbb{k}}$ of characteristic different from $2$. We allow degenerate quadrangles, consisting of four not necessarily distinct points and six joining lines, provided that three sides form a nondegenerate triangle. The net of a quadrangle is obtained by adjoining constants to the defining polynomials in the pencil generated by its pairs of opposite sides, viewed as conics. We call two quadrangles entangled when they generate the same net. We prove that entanglement is equivalent to the apparently asymmetric condition that every side of one quadrangle bisects the other quadrangle at that side's midpoint. The line-pair members of the net form a bisector field that determines the entire net, so entanglement is also equivalent to equality of bisector fields. Finally, we prove that the locus of centers of the central conics in the net, together with one nonparallel line-pair member, determines the net.

The rapid growth of video-based social media has increased users' exposure to harmful content, creating a need for reliable automated video safety detection. Although recent Vision-Language Models (VLMs) show strong video understanding capabilities, existing harmful video detection systems face two key limitations: they typically reduce safety detection to binary classification, overlooking the inherently multi-label nature of unsafe videos, and they rely on static training objectives that do not support controllable precision-recall trade-offs, though the desired operating point may vary across moderation pipelines and unsafe categories. To address these gaps, we propose Adaptive Tversky Policy Optimization (ATPO), a reinforcement learning framework for Multi-label Video Safety Detection (Multi-VSD). ATPO introduces the Adaptive Tversky Reward (ATR), which dynamically adjusts false-positive and false-negative penalties during training to enable controllable precision-recall trade-offs. Experiments on SafeWatch-Bench and XD-Violence show that ATPO substantially improves multi-label performance, increasing the Jaccard Index from 40.66 to 75.44 on SafeWatch-Bench-Real. Moreover, ATR enables reliable steering of the precision-recall operating point, supporting deployment scenarios with heterogeneous policy requirements. Code and checkpoints are provided at https://bruceyg.github.io/ATPO-project-page/ .

Given a subvariety of the Grassmannian and a data point, we seek to find a point on the subvariety minimizing the Euclidean distance to the data point. The number of complex critical points of this optimization problem is the Euclidean distance (ED) degree. We show that the ED discriminant of the whole Grassmannian, that is, the set of data points with a number of critical points different than the ED degree, is the discriminant of the characteristic polynomial of the data as a projection matrix. Another closely connected algebraic complexity measure for a subvariety in the Grassmannian is the Grassmann distance (GD) degree, which is the number of complex critical points of the distance optimization problem when the data point itself is in the Grassmannian. We give formulae for ED and GD degrees of geometrically meaningful subvarieties of the Grassmannian, namely, products of smaller Grassmannians, matroid realization varieties, and Schubert varieties.

We use the notion and properties of topological graphs to construct a quantum CW complex for quantum real projective spaces. This construction mirrors the CW complex construction in the classical case. Moreover, we introduce multipullback even-spheres as boundaries of noncommutative "cylinders" and show that they admit a gluing akin to the gluing of balls to the ends of a hollow closed cylinder.

We present new algorithms for the vertex-failure distance oracles and labeling schemes problems in undirected weighted graphs. A vertex-failure distance oracle is a data structure that, given two vertices $x$ and $y$ and a failed vertex set $F$ of size at most $f$, returns an approximation to the distance between $x$ and $y$ in $G \setminus F$. In the labeling-scheme setting, the data structure needs to be stored distributively as labels on the vertices, and each query $(x,y,F)$ must be answered by accessing only the labels of the vertices in $F \cup \{x,y\}$. For any $f\geq 1$ and $k \ge 1$, we obtain a vertex-failure distance oracle with $O(k^{6})$ approximation, space $\tilde{O}(f^{2}n^{1+1/k})$, query time $\tilde{O}(f^{5}n^{1/k})$, and polynomial preprocessing time. In particular, this is the first time-efficient oracle for multiple vertex failures with space close to linear, as well as the first constant-approximation oracle with polynomial space when tolerating $\Omega(\log n)$ vertex failures. The previous results, due to [Duan-Gu-Ren, SODA'21], gave two alternatives: for any constant $c \ge 1$ and $\epsilon>0$, one oracle has $\mathrm{poly}(\log n,f)$ approximation, space $n^{2+1/c}\mathrm{poly}(\log n,f)$, and query time $\mathrm{poly}(\log n,f^{c})$, while the other has $(1+\epsilon)$ approximation, space $n^{2+1/c}(\log n/\epsilon)^{O(f)}$, and query time $\mathrm{poly}(\log n,f^{c},1/\epsilon)$. We also obtain a vertex-failure distance labeling scheme with $O(k^{6})$ approximation and label size $f^{3}n^{1/k}\log^{O(k)} n$. This is the first nontrivial distance labeling scheme for vertex failures. Our techniques build on recent tools related to length-constrained vertex expanders and also introduce a new expander-based shortcut sparsification. The latter also leads to a deterministic vertex-failure connectivity labeling scheme of size $\tilde{O}(f^{2})$.

Recent advances in LLM reasoning models—driven primarily by the paradigm of post-training via reinforcement learning with verifiable reward (RLVR)—have enabled them to accomplish impressively complex tasks. However, in parallel with their rising capabilities, LLMs have increasingly displayed signs of language drift in their chains of thought (CoTs): unusual, non-standard, and seemingly nonsensical language use. Although it is well-documented—and can potentially impair CoT monitorability—the causes of language drift are thus far poorly understood. In this paper, we identify the conditions under which language drift occurs: we prove theoretically that RLVR optimization pressure permits unbounded language drift, while supervised fine-tuning does not. We then show empirically that language drift specifically arises during RLVR on novel reasoning tasks—i.e. when the target behavior cannot be drawn out of the base model. Finally, we prove that it is not possible to constrain language drift without constraining expected reward, suggesting that CoT monitorability cannot be improved without harming performance during RLVR post-training at the frontier.

The rapid maturation of artificial intelligence (AI) and machine learning (ML) has catalyzed a profound shift in how chemical engineering problems are formulated, analyzed, and solved. Advances in computing, data availability, and learning algorithms have enabled AI/ML methods to impact applications spanning atomic-scale simulations, materials and catalyst discovery, transport and thermodynamics, separations, process systems engineering, and industrial operations. This article provides a perspective on recent methodological developments and representative applications, emphasizing how AI/ML tools are being integrated with first-principles models to address challenges of predictive accuracy, data scarcity, extrapolation, interpretability, and model lifecycle management. Across domains, a unifying trend is the move away from purely black-box approaches toward hybrid and physics-informed frameworks that explicitly respect conservation laws, thermodynamic consistency, and known structural constraints. These approaches not only improve robustness and reliability, but also enable meaningful human-AI collaboration by providing information at an appropriate level of abstraction for the task and decision context. We conclude that AI and ML are not replacing the core principles of chemical engineering; rather, they are amplifying them. As the field advances toward increasingly autonomous, adaptive, and sustainable systems, the thoughtful integration of AI/ML with first-principles understanding and domain expertise will be essential to realizing their full potential across both research and industrial practice.

Equivariant machine learning interatomic potentials (MLIPs) have revolutionized atomistic modeling, but accurate treatment of complex materials and molecular systems demands expensive models. This limits simulation length- and time-scales, with tensor products a key computational bottleneck. The recent emergence of foundation-scale MLIPs further exacerbates this challenge. We present Branch Interatomic Potential (BranchIP), a single-model framework for learned adaptive tensor product computation, trained with a novel distillation loss. In our experiments on two systems of physical interest, a heterogeneous catalysis system and a proton-conducting solid acid electrolyte, BranchIP accelerates MLIPs across model sizes by up to $2.4\times$ while reducing memory usage by up to $2.6\times$. This is achieved while maintaining physical fidelity. Furthermore, the learned adaptive computation provides model interpretability by revealing which interactions demand deeper computation and showing how computational depth relates to chemical complexity and dynamics.

Reinforcement learning (RL) is a powerful paradigm for training agents, yet its success rests on domain expertise of human engineers who design informative reward signals for every new task. Unsupervised RL aims to reduce this engineering with intrinsic motivation (IM): reward signals that emerge from the agent environment interaction itself. Existing IM objectives, however, involve the selection of information variables, which re-introduces domain expertise the field has sought to eliminate. We introduce Forward CIP (F-CIP), an RL-native formulation of the Controllable Information Production (CIP) objective, which is defined by the system's dynamics alone and requires no such selection. We prove that F-CIP is compatible with RL and demonstrate its effectiveness with existing algorithms. Training agents with F-CIP results in unsupervised discovery of primitive behaviors such as balancing and maintaining controllability, which are essential for more complex robot behaviors. Paired with a simple forward-velocity reward, our method produces coordinated gaits such as hopping and running which otherwise require reward engineering to learn.

Evaluating explainable AI (XAI) systems from a human-centred approach requires researchers to select from numerous evaluation dimensions and measures, often in an ad hoc and fragmented manner. This paper introduces a method to help HCI, computer science, designers and social science researchers systematically evaluate XAI systems. The approach is based on an updated XAI-specific evaluation framework derived from an analysis of 82 studies. Using this framework, we developed a card-sorting method with 36 cards to help researchers prioritise relevant evaluation aspects. The process was tested with two research groups (n = 13) across five projects. The XAI Evaluation Cards are available as a printable appendix, along with an online repository of methods from previous XAI studies. Although not exhaustive, our findings indicate that the card-sorting approach can organise and streamline the design of the evaluation process, encouraging a more comprehensive and multidisciplinary assessment of XAI systems in research and development.

Invisible watermarking has become a central tool for tracing AI-generated images, but its robustness against adaptive removal attacks remains an open security question. We introduce Latent Frequency Masking, an attack that erases watermark evidence by replacing selected Fourier coefficients in the latent representation of a watermarked image. The replacement can be sampled from Gaussian noise for efficiency or derived from diffusion regeneration for improved image preservation. We provide a theoretical distortion bound relating the change between the reconstructed adversarial image and the masked latent-frequency perturbation. We evaluate the proposed attack against six diffusion watermarking methods on images generated from DiffusionDB and MS-COCO prompts. Latent Frequency Masking removes or substantially weakens several watermarks while preserving perceptual quality and achieving favorable runtime compared with existing attacks. These results identify latent-frequency manipulation as a practical attack surface and highlight the need to include such attacks in robustness evaluations of generative image watermarking.

Let $m$ denote the normalized Haar measure on the unit circle $\mathbb{T}$. For $0<r<\infty$, define $\lVert f\rVert_{r}:=\left(\int_{\mathbb{T}}|f|^{r}\,\mathrm{d} m\right)^{1/r}$, with $\lVert f\rVert_{0}$ and $\lVert f\rVert_{\infty}$ interpreted as the geometric mean and the supremum norm, respectively. Set $Q_n(z)=1+z^n$. We prove that, for every nonzero polynomial $p$ of degree $n$ whose zeros all lie on $\mathbb{T}$, $$ \frac{\lVert p\rVert_s}{\lVert Q_n\rVert_s} \le \frac{\lVert p\rVert_t}{\lVert Q_n\rVert_t}, \qquad 0\le s\le t\le\infty. $$ This settles Baernstein's quasi-norm monotonicity conjecture. As corollaries, we obtain an $L^r$ extension of Visser's coefficient inequality, the sharp O'Hara–Rodriguez inequality and its higher-power analogues, the Erdős–Szekeres product bound $ \left\lVert\prod_{j=1}^N(1-z^{s_j})\right\rVert_\infty\ge2\sqrt N $ for all positive integers $s_1,\ldots,s_N$ and Agler–McCarthy's entropy conjecture. We also provide a Lean 4 formalization of the main results.

Kikuchi matrices are a family of structured matrices that were introduced to study problems involving tensors and hypergraphs. We show that, as the ambient dimension grows, dense random Kikuchi matrices have a limit described by a system of $\Gamma$-independent semicircular elements. This characterizes their limiting spectral distribution and yields improved bounds on their spectral norm, a key quantity in the analysis of algorithms for Tensor PCA. Finally, we show that, in an appropriate double limit, independent Kikuchi matrices converge to the $q$-Gaussian system, another central object in noncommutative probability.

In [H. Lavenant, J. Luckhardt, G. Mordant, B. Schmitzer, L. Tamanini, The Riemannian geometry of Sinkhorn divergences. Ann. Inst. H. Poincaré Anal. Non Linéaire 43 (2026)] we introduced a Riemannian metric $\mathsf{d}_S$ on the space of probability distributions obtained from entropic optimal transport, specifically from the Sinkhorn divergence $S_\varepsilon$. In the present work we discuss how to approximate and compute $\mathsf{d}_S$. Spatially, we prove Gromov–Hausdorff convergence of the metric and convergence of geodesics for increasingly fine Eulerian discretization of the base space. Temporally, we show $\Gamma$-convergence of the chain discretization $N \sum_{k=0}^{N-1} S_\varepsilon(\mu_k, \mu_{k+1})$ to the energy functional defining $\mathsf{d}_S$. We deduce and implement numerical schemes to compute approximations of $\mathsf{d}_S$.

The layered honeycomb $\alpha$-RuCl$_3$ has been much explored as a candidate to display unconventional cooperative magnetism between spin-orbit entangled $j_{\rm eff}=1/2$ Ru$^{3+}$ magnetic moments stabilized by strongly-frustrated, Kitaev and other bond-dependent anisotropic interactions. Most studies so far have focused on relatively large single crystals that undergo a structural phase transition from monoclinic to rhombohedral upon cooling below 150 K and a magnetic transition to zigzag order around 7 K. Here we study the magnetism of high-quality, very small (sub 100 $\mu$m diameter) single crystals, which remain monoclinic upon cooling to low temperatures and display a sharp magnetic transition at 14 K. Using highly-sensitive piezo-cantilever magnetic torque measurements in fields up to 16 T, we report a comprehensive study of the magnetic phase diagrams for magnetic field rotated in three orthogonal crystallographic planes. Our extensive torque data collected upon varying almost continuously the angular orientation of the field as well as the field magnitude, allows clear detection of phase transitions via anomalies in the raw torque data and its higher order derivatives with respect to both angle and field magnitude. The obtained magnetic phase diagrams show many differences compared to rhombohedral samples, displaying reduced rotational symmetry, substantially higher fields required to suppress the spontaneous magnetic order, and distinct field-induced phases. We provide a direct comparison of the torque data with mean-field calculations for magnetic Hamiltonians appropriate for $\alpha$-RuCl$_3$ allowing for the monoclinic lattice symmetry and propose minimal magnetic structure models for all explored regions of the magnetic phase diagrams.

A multi-LLM council lets several large language models (LLMs) deliberate on a question and return an answer together with a confidence estimate. As these systems become increasingly used for reasoning, that confidence should represent a calibrated probability of being correct, and the decision should remain robust when some agents are persistently unreliable. Existing council aggregation methods fail on both fronts: their confidence estimates measure decisiveness rather than correctness, and they cannot identify or discount persistently unreliable agents. We introduce Bayesian Dialectical Argumentation (BDA), which treats the council's typed moves—who proposed, challenged, or conceded which answer—as observations of a classical annotator model with per-agent reliabilities. This formulation recasts multi-agent deliberation as a reliability estimation problem, using the deliberation trace to infer agent reliability under persistent adversarial behavior. By weighting evidence according to inferred agent reliability, BDA yields calibrated posterior probabilities over candidate answers while allowing persistently unreliable agents to be inverted rather than merely outvoted. Across binary and multi-class benchmarks, BDA achieves the best calibration among zero-cost council aggregation methods, requiring no additional LLM calls, and improves robustness under persistent adversarial coalitions while remaining competitive in clean settings.

We turn metric reconstruction into an analytic solver for rotating black holes in modified gravity, bypassing the direct solution of coupled field equations. As an example, in parity-even cubic gravity, we construct the first analytic spinning black hole metric at linear order in the coupling, without spin or weak-field expansions. We stringently validate the field equations and reproduce independent analytic predictions for horizon properties. This establishes a new route to construct spinning black holes beyond Einstein gravity and lays the foundation for strong-field tests.

We show that a two-dimensional chiral fluid supports an antisymmetric contribution to the pressure, and determine its microscopic expression. Its origin lies outside the standard kinetic and collisional channels of the normal pressure tensor, stemming instead from the average of a parity-breaking pseudo-scalar. This chiral pressure is set by the rotational (vortex) viscosity of micropolar hydrodynamics, whose kinetic value for rough disks we compute here, to Gaussian order. Crucially, both the chiral pressure and the rotational viscosity survive at zeroth order in the gradients, meaning they persist even in the fully homogeneous fluid – which we show is not only well defined but steady (the Quiescent Chiral State, QCS). Since this averaged pseudo-scalar does not vanish in a chiral fluid, any expansion of its distribution function must contain terms in this pseudo-scalar variable.

Large Language Model (LLM) agents now take part in organizational work, where many authors record decisions across documents over months. Because a revised decision arrives as a new document rather than an edit, answering a question requires knowing which version held at a given time. However, most memory systems compress the record at write time. By distilling each document into facts, notes or graph edges, these methods fix what can be answered before any question is asked. To address this, we propose Mem++, a non-destructive memory framework shifting from write-time distillation to read-time selection. Mem++ stores every document whole with its date and author, and it calls no generative model at write time. At read time, it retrieves only documents dated up to the time a question asks about and fuses lexical and semantic rankings. Unlike systems that overwrite older versions, Mem++ keeps them and leaves the choice to the answering model. Evaluations on the organizational benchmark OrgMemBench demonstrate that Mem++ surpasses the strongest memory system baseline by 8.0 to 13.1 points across two answering models. With gpt-4.1-mini, it also achieves the best overall score, 2.6 points above RAG. In addition, Mem++ achieves the best average LLM-judge score on LoCoMo and ranks second on LongMemEval-S, behind only its entity-graph variant. Code for benchmark evaluation is available at https://github.com/AIDAChip-Inc/mem-plus-plus.

Small open-weight models (2-9B) run on ordinary laptops, but under cloud-scale agent harnesses they rarely complete real tasks: tool prefill overflows the context, self-correction diverges, tool demonstrations loop, and tasks are silently abandoned. We present evidence, from a controlled single-machine comparison and one third-party benchmark, that a substantial share of these failures is attributable to the harness rather than the model. We introduce Mingbird, a local-first agent harness for Windows and Ollama whose ten mechanisms compensate point-by-point for small-model failure forms, three of them representative: a byte-level net-zero prefill budget, a finish gate that re-reads the task before accepting completion, and signature-level loop detection. On LRAB, a controlled comparison holding machine, models, budgets, and scoring fixed (4 harnesses $\times$ 4 open models (2B-35B) $\times$ 18 real tasks, deterministic artifact scoring), Mingbird reaches 0.886 overall against 0.631 (goose), 0.479 (opencode), and 0.405 (agent-mini), with all 288 cells published; on $\tau^2$-bench (278 tasks, three arms, one protocol) it totals 0.856 against 0.791 and 0.737; and a frontier-model probe on the same 18 tasks spans 0.997 to 0.478 across harnesses, with well-formed scaffolds staying within 0.072 of each other. A leave-one-mechanism-out ablation is reported as directional only: same-night replications of the same arm move its mean by up to 0.069, the size of every nominal single-trial delta, and the one batch-matched comparison (full mechanism stack versus text re-read alone) gives the executable completion guards a paired +0.10 across three replications. The evidence carries stated limits: a self-built benchmark, a single machine, and single-trial scoring.

Adverse conditions such as rain, snow, fog, and dust remain challenging for camera-based perception in autonomous driving. We study multi-class weather recognition from street-view images under domain shift, where most available training data come from non-street-view sources that differ markedly from real driving scenes. We propose Weather-Aware Adversarial Discriminative Domain Adaptation (WA-ADDA), which conditions the domain discriminator on predicted weather to promote features that are both domain-invariant and weather-sensitive. We also assemble a multi-dataset benchmark by unifying diverse non-street-view weather collections as sources and real street-view images as targets, and define a standardized evaluation protocol with macro accuracy as the primary metric. Across backbones (ResNet-50, EfficientNet, VGG, DenseNet), WA-ADDA consistently improves street-view performance and yields strong per-class recalls in challenging conditions while preserving clear-weather accuracy. These findings highlight the feasibility of domain-adapted weather recognition and the value of our benchmark for advancing robust, on-board perception.