Papers

Due to limited support for intra-tensor heterogeneous precision in conventional accelerators, neural network quantization remains largely restricted to per-tensor precision assignment. We present ShatterQuant, a hardware-software co-designed framework enabling mixed-precision quantization within each tensor by assigning independent bit-widths to blocks of a weight projection. ShatterQuant couples precision granularity with PE configuration, such that each precision determines an effective block height. We introduce (1) a hardware-aware post-training method that assigns intra-tensor precision based on block-level standard deviation and weight sensitivity; (2) the ShatterQuant Transformer Accelerator supporting 1/2/4/8-bit weight precision, precision-dependent PE configuration, block rescaling, and integrated softmax and piecewise-linear nonlinearities; and (3) an evaluation of model-hardware tradeoffs using an implementation in the TSMC 16nm PDK operating at 1 GHz, achieving 1.5 TOPS, 760 GOPS/$mm^2$ area efficiency, and 2.8 TOPS/W energy efficiency. On DeiT and ImageNet-1K, ShatterQuant achieves accuracy within $3.3\%$ of state-of-the-art mixed-precision techniques while using a 2 bit lower effective bitwidth, while for PixelDiT demonstrates comparable generation quality. ShatterQuant demonstrates how fine-grained intra-tensor mixed-precision can be realized through hardware-software co-design.

We challenge the traditional deterministic view of the classical mechanical universe by combining dynamical chaos with the distinction between physical phenomena and their mathematical models, and quantify the resulting indeterminism in several quasi-realistic examples.

Over twenty editions, the ICWSM community has examined social life online as platforms, interactions, and research methods have changed. What can this body of research tell us at this critical juncture, as AI increasingly reshapes how people communicate online? We analyzed 2,139 indexed contributions from 2007 to 2026, distinguishing topics identified through nonnegative matrix factorization from problem framings captured through explicit textual cues. We find that platform mentions shift from blogs toward Twitter and, more recently, Reddit. Online community research maintains a similar topic share (10.5% to 10.0%), but governance cues within it increase from 4.0% to 34.3%. Harm-related cues also increase after restricting abstracts to a fixed length. Our review also traces advances in sampling, measurement, and causal and experimental methods. We discuss how AI-mediated interactions complicate these questions and provide a reporting checklist to support research across changing platforms.

Visual-token compression for vision–language models is posed almost entirely as a selection problem: decide which tokens to keep and discard the rest. The criteria that work best rank tokens by the attention the language model pays them, which makes the ranking a function of the question being asked. That is invisible in a single-turn benchmark and decisive whenever a compressed representation is written once and read many times, as when it is cached across the turns of a conversation or transmitted between a device and a server. We take the other half of the classical transform-coding toolkit instead: keep every token and vary its rate. A transform code exposes each token's measured distortion–rate curve, and a fixed bit budget is distributed across tokens by exact integer rate–distortion optimisation on those curves. No text enters the pipeline, so one compressed representation serves any query. At equal bit budgets, on two datasets and two capacities, it preserves the model's output distribution and its answers better than uniform-rate coding, the closed-form water-fill and distortion-ranked pruning. It matches attention-ranked pruning on the question pruning was tuned for, and overtakes it once the compressed image must answer a different question about the same image.

A smooth 3D function can be efficiently and accurately approximated by a 3D lookup table of significantly lower resolution combined with tetrahedral interpolation. The function is sampled once on a coarse grid in a preprocessing step. For each input, the enclosing grid cube is located, one of the 6 tetrahedra in this cube is selected, and the values at its 4 corners are combined by a weighted sum. For RGB to CMYK color conversion, a particularly frequent application of this, a table with $n = 16$ cells per axis ($17^3$ nodes, 16-bit values) requires 38\,KiB. A full 8-bit table requires 64\,MiB. As shown in an experimental evaluation, the mean error of the $n = 16$ table is 0.08\,% ink per channel. For 8-bit input, the lookup uses integer arithmetic only (Section~\ref{sec:integer}). It requires no floating point and no division, and all intermediate values fit in 32 bits. Floating point is needed only to build the table, once and offline. The integer result differs from tetrahedral interpolation with exact fractions by at most 1 unit of the 16-bit output, and white and black are reproduced exactly. This makes the approach suitable for low-power devices and real-time applications where computational resources are limited.

Several recent pipelines build RLVR training data by masking a span of real corpus text and asking a language model to invent plausible wrong answers around it. The correct option is therefore genuine human prose; every distractor is synthetic. Correctness and provenance become entangled, and a policy could in principle learn the second instead of the first. We audit that possibility in GooseReason-0.7M. First we ask whether the asymmetry is visible at all: a classifier reading only five surface statistics (never the meaning) reaches AUROC 0.562 over 315,499 options, barely above chance. The aggregate hides something, though. Code sits at 0.416, below chance, and manual inspection explains why: code distractors turn out to be single-operator mutations of the gold answer rather than freely written alternatives, so the two classes are nearly identical by construction. Detecting a signal is not the same as showing a model uses it, so we then run an intervention. We build a paraphrase-matched control corpus, hold training-set size identical across arms, and train two policies under one fixed budget. The exploitation gap does not favour the unmodified-data arm: 0.021 against 0.027 for the control. Under our budget, in other words, a detectable artifact went unexploited. We think that dissociation, along with the domain-specific construction finding, is worth knowing for anyone curating corpora of this kind, and we release the audit as a mostly CPU-only protocol.

Parameterized quantum circuits require repeated measurements to estimate objective gradients, making the number of circuit evaluations a central resource. Simultaneous perturbation stochastic approximation (SPSA) is widely used because it obtains a full gradient estimate from only two cost evaluations, independently of the number of parameters. We analyze the random perturbation for the structured Fourier landscapes generated by quantum circuits. For a finite update width, we derive the exact mode-by-mode response of the mean Rademacher estimator and show that it selectively suppresses modes involving many parameters. For general generators, the filtered mean field need not be conservative. We derive its sampling variance, compare it with Gaussian random-direction smoothing, and formulate the resulting stochastic dynamics through a Fokker–Planck equation. The non-conservative field generates acceleration to the steady state and the global minimum of the cost function by gradually decreasing the parameters in SPSA.

An unresolved mixture of beta-decaying ground and isomeric states populates a common daughter nucleus through different excitation-energy and spin-parity distributions. A framework is developed to determine when such data admit conventional $\beta$-Oslo and Shape analyses and which quantities remain identifiable. The physical primary-$\gamma$ distribution is an average of separately normalized branching kernels; the global parent fraction does not, in general, equal the fraction in an excitation-energy bin or an accepted primary-$\gamma$ sample. Population invariance, Oslo factorization, and recovery of the total nuclear level density are separated. An exact log-linear compatibility test and a lower-cascade replacement identity isolate factorization and first-generation-subtraction effects. For Shape analysis, matched final-state spins and parities remove common angular-momentum factors but can retain excitation-dependent E1/M1 weights. A graph representation gives consistency conditions and exposes the information lost when only one fixed final-state separation is measured. Deterministic spin-parity-resolved cascade benchmarks reproduce the population-invariant limit and exhibit both exact factorization with an effective, rather than total, level density and small matrix residuals accompanied by appreciable shape errors. Evaluated $\rb\rightarrow\sr$ properties specify a relevant application, including two $2^+$ candidate final states and time-dependent parent mixing. The numerical examples are controlled synthetic tests, not an extraction from $^{90}$Rb data. The results provide criteria for interpreting unresolved-parent measurements and for identifying the additional observations required for population-aware inference.

Large language model search agents are often trained with synthetic questions whose difficulty is increased through larger evidence graphs, additional hops, and longer trajectories. These global properties, however, are only indirect proxies for the local retrieval capabilities required during search. To address this mismatch, we introduce latent anchor reasoning, which consists of resolving an unnamed retrieval anchor from descriptive specifications and transferring the recovered anchor into a subsequent information demand. This primitive retrieval unit decomposes deep search into chains of coupled operations and organizes question construction around anchor resolution and relation transfer, without prescribing a canonical search path. Based on this formulation, we propose PrimeSeeker, a capability-oriented framework that constructs web-grounded anchor structures and jointly derives a question and a reference evidence skeleton. The skeleton preserves supporting evidence from construction and guides expert generation through extractive highlights of current tool observations. These highlights are removed before supervised fine-tuning, while the skeleton is subsequently reused to audit reference-step coverage for reinforcement-learning rewards. We construct 9,221 expert trajectories, training a 30B search agent. Across five deep-search benchmarks, PrimeSeeker achieves strong performance, while reference-step optimization further improves the supervised policy. The resulting trajectories exhibit low retrieval redundancy, and fixed-budget evaluation shows strong solution coverage with substantially fewer tool calls than long-horizon systems.

Researchers across academia increasingly base significance claims on LLM judge scores and small-sample AI evaluations. Yet without well-calibrated confidence intervals (CIs), hypothesis tests, and judge-bias corrections, such claims are unreliable. We address these issues in several contributions. First, we find that running statistics over raw LLM judge scores leads to inflated false positives: counterintuitively, for many inter-rater agreement metrics, false positive risk peaks at "almost perfect" human-LLM agreement. To help researchers understand how to run statistics over LLM judges responsibly, we present guidance and tooling for the statistical analysis of mixed human-AI judge designs, and implement nine hypothesis tests via prediction-powered inference (PPI), including the first known PPI corrections for four rank-based tests (Wilcoxon signed-rank, Mann-Whitney U, and omnibus variants). To keep PPI++ stable with small human-labeled calibration sets, we introduce bootstrap-adaptive power tuning, which shrinks the estimated weight toward a target estimated from the labeled data, and accounts for that weight's own sampling variance. Second, through Monte Carlo simulations, we derive recommendations for what CI, p-value, and FWER correction methods to use for small-sample AI evaluations (N<100), and warn researchers against bootstrap CIs. We package these recommendations into evalstats, an open-source Python package that selects calibrated methods automatically, and demonstrate it in three scenarios, including one where a real LLM judge validated at "substantial agreement" would have led a researcher to publish a spurious finding. evalstats is publicly available at https://github.com/ianarawjo/evalstats.

As browser-use agents improve, benchmarks keep pace by adding new tasks, websites, and applications. Doing so is costly, and because many factors change at once, it is unclear what makes the new tasks harder. In this work, we instead make existing tasks harder by changing the environment in which they are performed, so that difficulty becomes a programmable property of the environment. We introduce BreakingWeb, in which each task is run twice: once on an unmodified website and once under an intervention, a controlled change to the website's content, server state, network responses, or rendered page. Each intervention is reproducible, keeps the user instruction and success criterion fixed, can be detected and worked around by a competent user, and is labeled with the behavior that recovery from it mainly requires, such as verification. BreakingWeb contains $519$ task pairs on seven self-hosted websites, and each run is scored on the website's final backend state. We evaluate six strong browser-use agents, three agents that observe only screenshots, and human participants. Interventions lower the pass rates of the six browser-use agents by $22.9$ percentage points on average, and each agent fails on nearly half of the tasks it solves on the unmodified website. Humans are affected much less: their pass rate drops by $10.0$ points on the first attempt and by $5.7$ points when they repeat the task. Most failures of the six agents ($75\%$) are belief failures, in which the agent declares the run finished although a required change was never made. Our code, data, and environments are available at https://www.breakingweb.app.

Compact surrogates could reduce the cost of simulating large language model societies, but must reproduce collective behavior. We compare individual predictions and collective forecasts using 9,455 published trajectories and new experiments on opinion dynamics. Neighbor information improves individual prediction in all 16 public-data settings and pooled collective forecasts on held-out questions, although collective gains depend on transfer conditions. Tests on 24 new statements do not confirm earlier contrasting history effects in forecasts from the initial state. Qwen benefits from history after three observed rounds. These findings motivate direct collective validation, explicit limits on available observations, and comparisons with simple baselines.

In-car conversational assistants (ICAs) are increasingly integrated into vehicles to support route planning, vehicle control, and information access. Ensuring their reliability is challenging due to multi-turn interactions, the absence of explicit ground truth, and strict safety constraints. Existing evaluation techniques fall short, as they target single-turn settings and fail to capture constraint handling, context retention, and safety-critical behavior across turns. We propose an automated framework for testing the multi-turn conversational capabilities of ICAs. The system is treated as a black box and evaluated via closed-loop simulation with a strategy-guided user simulator, an adversarial strategy manager, and a two-tier LLM judge assessing turn-level failures and conversation-level quality. We evaluate the approach on an industrial ICA with six LLM backends and twelve human annotators. The automated judge shows substantial agreement with humans, and strategy guidance uncovers 2.96 times more unique failure types per conversation and more than doubles the number of unique failing conversations compared to unguided simulation.

Tool retrieval is a critical bottleneck for LLM-based agents operating over large, heterogeneous API ecosystems. Existing approaches face an inherent trade-off: semantic retrievers are fast but suffer from the semantic-functional gap, while execution-based validation improves precision at the cost of prohibitive latency. We propose Lookahead-R, a planning-based framework that reformulates tool retrieval as a resource-constrained sequential decision-making problem. At its core, Lookahead-R introduces a lightweight execution-aware surrogate world model that jointly predicts tool execution success, latency cost, and semantic utility—without invoking real APIs. This world model drives a cost-sensitive, uncertainty-guided Monte Carlo Tree Search that navigates the tool space under strict budget constraints. Evaluated on the large-scale ToolBench benchmark, Lookahead-R achieves a superior accuracy-efficiency trade-off across all test scenarios. On the most challenging I3 split, it attains an NDCG@5 of 91.40%, outperforming the state-of-the-art ToolGen (90.16%) by 1.24%. Ablation studies confirm that explicit latency modeling is the key discriminative signal for identifying high-quality tools under resource constraints.

Large language models are increasingly used in oncology applications, but their predictions are often weakly grounded in explicit medical structure. We present TRACE, a deployable tree-relational enhancement framework for oncology LLMs. TRACE separates expensive offline structure learning from lightweight online inference: oncology concepts and relations are organized into an updatable tree-relational structure, refined using LM-loss-derived evidence, and retrieved at inference time as compact prompt evidence. This design supports task-adaptive evidence selection without requiring supervised labels in the zero-shot setting. Across ten oncology classification tasks and one MedQuAD CancerGov QA benchmark, TRACE improves both label-free evaluation and supervised fine-tuning. Additional analyses show that TRACE improves over vanilla RAG and generic GraphRAG, remains useful under leakage-controlled METABRIC inputs, and produces interpretable evidence paths aligned with clinical reasoning. These results suggest that explicit, updatable medical structure is a practical path toward more accurate and auditable oncology LLM deployment.

The rapid advancement of generative AI raises concerns about the misuse of Multimodal LLMs (MLLMs) for large-scale disinformation campaigns on social media. Despite existing research on textual disinformation, a fundamental question remains unanswered: can MLLMs be exploited to fabricate realistic multimodal fake news, and can they reliably detect it? We introduce a multi-agent framework in which a story agent, an image agent, and a critic agent collaborate to produce fake social media posts that plausibly counter true news. We apply the framework to generate over 9,000 paired multimodal news posts across science, health, and entertainment domains, and benchmark 16 open- and closed-source MLLMs for automated detection. We find that most models fall substantially short of human-level accuracy and fail critically on identifying image authenticity. Our research provides a foundation for developing robust defenses against social media fake news. Code and data are available at https: //github.com/xiuzhenzhang/Multimodal.

Experience reuse can reduce repeated exploration in embodied agents, but a trajectory that succeeded previously may be unsuitable for the current execution context. Existing memory systems pri marily optimize construction and retrieval; semantic relevance and historical success therefore remain insufficient when retrieved ex perience contains incompatible actions or an inappropriate level of structure. We introduce Memory Adaptation for Task-Conditioned Execution (MATE), a deterministic post-retrieval procedure that converts trajectories into execution-oriented memory. MATE re moves obsolete control context, extracts condition-action-effect transitions, applies verified action normalization, selects a task dependent representation, and serializes the result under a fixed budget without additional LLM inference. On 134 ALFWorld tasks, MATE achieves task success rates of 81.3% and 93.3% with Qwen2.5-14B and 72B while using approximately one-tenth of the tokens required by raw trajectories. Controlled comparisons show that verified action normalization is the principal mechanism by which MATE restores the utility of retrieved experience, support ing memory adaptation as a distinct stage between retrieval and embodied execution.

Over perfect fields of characteristic $p\ge5$, the unipotent radicals of the height-one special, Witt-Jacobson, and Hamiltonian automorphism groups in at least three, two, and four variables, respectively, attain the largest derived length allowed by their nilpotency classes. More strongly, every term of each derived series equals the corresponding power-of-two term of the lower central series. These equalities hold on finite-field points, with explicit Frattini quotients and lower $p$-central series. For the special family, a Cartier quotient measures the difference between the congruence and lower central filtrations. The Hamiltonian correction comes from flux and the socle line; the Witt-Jacobson family has none. We also identify the contact automorphism group scheme and, at arbitrary divided-power height, compute the Cartier-Verschiebung maps on restricted derivations and the Verschiebung filtration and distribution algebra of the Frobenius thickening of the special radical.

The shifted congruence series of finite special Nottingham jet groups is realized by an integral augmentation filtration. For volume-preserving formal substitutions in characteristic $p\ge5$ and $n\ge3$ variables, the filtration induced by the action on truncated power series has the congruence subgroups as its dimension subgroups. The canonical weighted augmentation filtration has the same dimension subgroups. The group symbols form a degree-truncated Lie algebra of divergence-free polynomial fields. Since brackets with quadratic fields give the subspace of exact fields in every coefficient degree at least three, the Lie indecomposables occur only in degrees $(p-1)(n-1)+pr$, with $r\ge0$, where they are determinant-weighted Frobenius twists of tensor products of symmetric powers with the defining module.

Cross-dataset transfer in rice leaf disease classification remains a significant challenge, with models trained on one image collection performing substantially worse when deployed on another. We conduct a systematic benchmark across three Bangladeshi rice leaf disease datasets (5,419 images, 6 transfer pairs, 3 CNN backbones, 3 random seeds) to characterize and diagnose this failure. Strong augmentation recovers a mean cross-dataset macro-F1 improvement of +0.070 (Wilcoxon p < 0.001, 15 of 18 transfer pairs positive). Removing non-leaf image content via segmentation shows directional benefit (mean +0.066, p = 0.062, n = 36 paired observations) that is consistent across two independent segmentation methods but does not reach conventional significance. A self-supervised ViT control (DINOv2 linear probe) exhibits equivalent cross-dataset collapse to CNNs, ruling out architecture inductive bias as the primary driver and pointing to acquisition-condition shift. Adaptive batch normalization uniformly harms transfer performance, with harm magnitude correlating with source-target label-prior divergence and model depth (Spearman rho = 0.621, p = 0.009). Grad-CAM attribution analysis on 12 sampled predictions does not distinguish correct from incorrect cross-domain predictions (p = 0.462), indicating that common attribution proxies are insufficient for diagnosing shift at practical sample sizes. We document all frozen results, prespecified analysis criteria, and reproducibility artifacts in a public repository with SHA-256 integrity verification. This work establishes a rigorous empirical baseline for understanding cross-dataset generalization in agricultural computer vision and identifies both effective (augmentation) and ineffective (AdaBN) adaptation strategies.

In embodied voice interaction, cocktail-party speech perception requires both speech separation and speaker attribution across machine-generated and human speech sources. However, conventional audio-only blind source separation remains permutation ambiguous, making the correspondence between separated streams and physical speakers unclear. This paper presents RadarVox, a radar-audio multimodal benchmark for identity-aware cocktail-party speech separation. RadarVox provides acoustic mixtures and source-level radar displacement signals from loudspeaker-emitted and human speech, enabling source-aware speaker assignment. FMCW radar captures laryngeal mechanical motion, providing speaker-specific spatial-motion cues unavailable to a single-channel microphone. We inject radar-derived priors into a DPRNN separator via gated fusion and learn a speaker-aware cross-modal matcher to associate unordered speech streams with radar-observed speakers. Experiments on multi-speaker mixtures show that radar cues provide complementary benefits, achieving scale-invariant signal-to-distortion ratio (SI-SDR) values of 9.75 dB and 7.13 dB in two- and three-speaker scenarios, respectively. More importantly, the proposed method improves ordered SI-SDR by more than 13 dB over audio-only methods while achieving over 80% speaker assignment accuracy.

We study a two-dimensional Gaussian Rouse chain divided into two contiguous blocks with different local friction coefficients but identical Hookean elasticity. In quiescent conditions, friction heterogeneity leaves the continuous-time equilibrium Gaussian conformation unchanged while producing distinct block mobilities and a modified generalized relaxation spectrum. The exact friction-weighted translational diffusion coefficient retains the scaling $D_\gamma\propto N^{-1}$, with a friction-dependent prefactor; the corresponding diffusion law and generalized-mode formulation extend beyond the diblock to more general local friction profiles. We then place the chain in an incompressible Batchelor-Kraichnan random flow. Under a common external flow, the high-friction block becomes more extended than the low-friction block. The deformation asymmetry strengthens with friction contrast and develops across the weak- to moderate-flow regime, remaining clearly visible at the strongest finite-time conditions. It also persists when the chain-specific longest-mode Weissenberg number is matched across friction ratios, showing that, within the same finite-time protocol, the effect is not accounted for solely by a shift of the global relaxation timescale. In generalized-mode coordinates, the smooth random strain does not directly mix mode indices and preferentially amplifies the slowest mode. At larger flow strengths, extension distributions broaden markedly and exhibit pronounced temporal drift over the simulated time window. Since all simulated chain-specific longest-mode Weissenberg numbers remain below unity, this drift is interpreted as slow convergence toward the broad, power-law-tailed stationary Hookean regime near the coil-stretch threshold. Overall, friction heterogeneity alone is sufficient to redistribute deformation along an otherwise mechanically uniform polymer.

Visual Language Models (VLMs) have achieved remarkable success across diverse tasks, yet they struggle with high-resolution inputs where critical information resides in small regions or detailed, cluttered scenes. While several approaches address this limitation, a systematic understanding of why models fail at high resolutions is lacking. We introduce a controlled evaluation framework that disentangles resolution-related performance degradation from task difficulty through semantics-preserving transformations. We propose two simple metrics: Area Under the Scaling Curve (AUSC), which quantifies scaling robustness independent of baseline accuracy, and Prediction Variance Score (PVS), which measures resolution-induced prediction instability. Through comprehensive experiments across 5 model families and 5 benchmarks, we identify three primary failure modes: (1) information loss from downsampling at vision token limits, (2) tokenization artifacts from patch boundary shifts and positional encoding fragility under non-standard aspect ratios, and (3) attention dilution as token counts increase. Our analysis reveals that even state-of-the-art models suffer from performance drops when processing high-resolution images, with degradation patterns varying systematically by architectural family. We provide actionable insights for model architecture design and data augmentation strategies to mitigate these limitations.

Byte-Pair Encoding (BPE) constructs subword vocabularies through greedy pair merging, but conventional BPE requires the number of merges or target vocabulary size to be specified externally. Significance-Gain Pair Encoding (SG-BPE) replaces frequency-only selection with a statistical criterion based on how strongly an observed pair exceeds its expected co-occurrence under an independence model. This paper introduces MDL-Calibrated Significance-Gain Pair Encoding (MDL-SG), a three-stage procedure separating discovery, replication, and utility. Candidate pairs are ranked by Significance-Gain on a discovery partition, tested for replication on a separate partition using an exact one-sided hypergeometric test with per-iteration Benjamini-Hochberg correction, and then evaluated on a utility partition using a Minimum Description Length (MDL) criterion. Merging stops automatically when no replicated candidate yields positive held-out MDL gain. On WikiText-103, MDL-SG stops at 209, 433, and 847 merges for 120K, 250K, and 500K-character tokenizer-training samples, respectively. At 500K characters, it selects a stored vocabulary of 1,017 tokens without prescribing the vocabulary size in advance. In a compute-matched TinyGPT experiment with identical 2,024,448-parameter models and 500 optimizer updates per language model, MDL-SG achieves validation/test BPC of 3.2612/3.2436, compared with test BPC of 3.2894 for SG-BPE and 3.3493 for frequency BPE. Frequency BPE achieves stronger raw compression, while MDL-SG achieves lower BPC, showing that compression-oriented merge selection and language-model utility need not coincide.

We describe the Lean 4 formalization of Langlands's Second Main Lemma for local epsilon factors over nonarchimedean local fields, in mixed and equal characteristic. The lemma compares the induction expressions attached to compatible characters of the intermediate fields of a bicyclic extension of prime-square degree. We explain the mathematical constructions and comparisons that constitute the proof, and identify their Lean declarations.

We investigate holographic entanglement entropy and its generalization to pseudo-entropy in a boosted BTZ black hole geometry, which is holographically dual to a moving strongly coupled plasma. By solving the extremal surface equation, we derive an exact closed-form expression for the holographic entanglement entropy as a function of the boundary interval length, boost velocity, temperature, and conserved quantities. The resulting exact analytic formula correctly reproduces the standard thermal BTZ entanglement entropy in the vanishing-boost limit and smoothly reduces to the vacuum conformal result at zero temperature, thereby providing robust consistency checks. Motivated by the off-diagonal structure of the boosted metric arising from the Lorentz transformation, we then study two families of analytic continuations to Euclidean signature. Both require a simultaneous continuation of the time coordinate and the boost parameter so that the Euclidean metric remains real. If only these two parameters are continued, the boundary interval becomes imaginary and the analytically continued holographic entanglement entropy is complex-valued; we interpret this complex entropy as the holographic dual of pseudo-entropy for the boosted thermal state. If, in addition, a conserved charge is also continued, the boundary interval remains real and the resulting entropy is real. Our construction establishes a direct connection between boosted HEE and holographic pseudo-entropy, extending previous analytic frameworks developed for rotating BTZ geometries to Lorentz-boosted spacetimes.

We conjecture a positive four-shift identity for rank-two Nahm sums with denominator steps $(1,31)$, giving a mixed-base realization of the Monster McKay–Thompson series $T_{31A}$ through the classical Rogers–Ramanujan function $U(1,31)$. A theta dissection directly derives the modular target. For the underlying quadratic exponent $3r^2+31rs+93s^2$, we prove nonmodularity of every individual sum with rational linear terms and normalization, together with an exclusion from finite Fricke systems with meromorphic finite-width expansions at infinity. Exact agreement through $q^{2000}$ and six vanishing radial corrections support the proposed combination; a five-term dilogarithm certificate and an exact amplitude calculation prove its leading asymptotic behavior. We further propose a three-component Fricke law for a dual combination, with an explicit icosahedral matrix involving fifth-root constants, and a quadratic identity linking its components to $T_{31A}$. The dual translation law and matrix algebra are proved; its Fricke transformation and moonshine identity remain conjectural, supported respectively by numerical evaluations and exact finite coefficient checks.

Generative adversarial networks (GANs) are increasingly used to augment medical imaging datasets, but synthetic images do not necessarily provide downstream classification benefits. This study investigates whether class-specific Deep Convolutional Generative Adversarial Network (DCGAN) augmentation improves brain tumor classification when the classifier and evaluation set are held constant. Experiments were conducted on 7,200 brain magnetic resonance imaging (MRI) scans across four classes: glioma, meningioma, pituitary tumor, and no tumor. For each class, 1,400 real images were used for training and 400 were reserved for testing. A baseline Swin Transformer classifier was trained using only the real training images and compared with a second model trained using the same real images augmented with 500 DCGAN-generated images per class. Both conditions were evaluated on the identical held-out test set. The two models achieved the same overall accuracy of 96%, while macro F1 remained effectively unchanged and ROC-AUC decreased slightly from 0.987 to 0.982 after augmentation. Class-level analysis showed small redistributions in errors rather than a consistent performance gain. FID values between real and synthetic images ranged from 209.15 to 314.27, indicating substantial distributional differences under the adopted evaluation setup. These results suggest that synthetic augmentation should not be assumed to improve medical image classification and should instead be evaluated for both distributional fidelity and downstream task utility.

NAND-backed storage offers the capacity needed to scale LLM prefix caching, but its block I/O path incurs CPU cache contention and host-DRAM staging in addition to NAND latency. Our characterization shows that these interface costs persist even with DRAM as the storage medium, motivating CXL-SSDs for byte-addressable access to NAND-backed capacity. Surprisingly, however, a stock CXL-SSD remains about 3$\times$ slower than local DRAM and no faster than an NVMe SSD, while generic prefetching provides little benefit. We present LM-CXD, a CXL-SSD specialized for LLM prefix caching. LM-CXD bridges the semantic gap between the serving engine, which knows which KV chunks will be consumed, and the device, which controls their placement and movement. It makes KV chunks device-visible I/O units, exposes NAND-to-DRAM progress to the serving engine, and uses device DRAM as a GPU-accessible buffer. LM-CXD further coordinates request scheduling with windowed prefetching and pipelines layerwise KV movement with GPU computation to hide NAND latency under limited device DRAM. Across five LLM models, LM-CXD reduces average TTFT over a stock CXL-SSD by up to 2.6$\times$ with compute asynchronous prefetching and 4.03$\times$ with layerwise prefetching, achieving TTFT within 1.5$\times$ of local DRAM on average.

We examine the round Euclidean no-boundary saddle proposed by Liu, Quintin, and Afshordi in "Ultraviolet Completion of the Big Bang in Quadratic Gravity." For a free, massless, conformal realization of their large matter sector, the omitted type-A trace anomaly contributes at leading order in the matter multiplicity. Within the one-loop renormalization-group-improved pure quadratic truncation, both the sphere effective action and the local trace equation show that a stationary point of the running $R^2$ coupling supports a round sphere only if the total Euler anomaly coefficient vanishes. Using the proposed beta function, we further find that no real stationary coupling ratio exists when this coefficient exceeds $79/420$, a bound violated parametrically by the assumed matter sector when the remaining gravitational contributions are perturbatively controlled and of order unity. The local obstruction also applies to a round half-sphere and cannot be removed by shifting its radius. Under these assumptions, the proposed round starting geometry and its stated continuation to Lorentzian inflation are excluded, without ruling out inflation from other initial conditions.