Searcharxiv⌕ Search

arXiv subjects

Quan Cheng

Publications and source records attributed to Quan Cheng.

At least 19 recordsLinked to original sources
arXiv2026

PlaceReasoner-Beta: Reasoning-Driven Macro Placement and Benchmarking

Automated macro placement remains a fundamental challenge in VLSI physical design. Despite decades of research, existing approaches predominantly optimize hand-crafted proxy objectives, such as estimated wirelength, and typically produce placements through one-shot numerical optimization, limiting their ability to incorporate visual layout context, codified design expertise, and downstream physical-design feedback in a unified loop. We present PlaceReasoner-Beta, a verifier-guided multi-agent framework that reformulates macro placement as a closed-loop reasoning problem rather than black-box optimization. A vision-language model (VLM) planner generates candidate placements from the floorplan image, macro specifications, and connectivity structure; a geometric verifier enforces physical legality and expert placement principles; a physical verifier refines candidates using early implementation feedback; and a post-route optimizer further improves promising layouts using final PPA. To enable reproducible evaluation, we introduce PlaceReasoner-Bench, a fully open end-to-end benchmark built from open RTL designs, EDA tools, and technology libraries. It comprises 8 designs at two aspect ratios, yielding 16 tasks with fixed floorplans and I/O assignments, so methods differ only in macro positions and orientations and are evaluated using routed PPA and DRC rather than pre-route proxies. Across the benchmark, PlaceReasoner-Beta achieves the best timing among DRC-clean methods on all square tasks, reducing post-route TNS by 61.2% at 1:1 and 53.0% at 2:1 relative to the classical baseline field. It also shortens routed wirelength on most designs despite never explicitly optimizing it, demonstrating that reasoning over spatial structure under physical-design feedback can improve end-to-end layout quality beyond proxy-objective optimization.

cs.AI↗
arXiv2026

Group Distributionally Robust Optimization over Decentralized Networks

We study group distributionally robust optimization (GDRO) over decentralized networks, where workers collaborate to learn robust models by minimizing the maximum local risk. Existing methods optimize only dual-regularized surrogates and establish guarantees for a globally averaged model that is not available to workers. In this paper, we directly optimize the original GDRO objective and establish guarantees for local models in semi-decentralized (SemiDec) and fully decentralized (Dec) settings. A key difficulty is that applying the primal-dual decomposition from centralized GDRO yields worker-dependent losses whose evaluation requires unavailable cross-worker information. To address this challenge, we develop a novel error decomposition that recasts decentralized GDRO as (i) a decentralized online convex optimization (D-OCO) problem and a prediction-with-expert-advice problem in SemiDec, and as (ii) two coupled D-OCO problems in Dec, with discrepancy terms capturing inconsistencies among local decisions. By isolating consensus errors in analysis, we develop modular frameworks that support different consensus routines. We then establish the \emph{first} per-worker convergence guarantees for optimizing the original GDRO objective over decentralized networks. Instantiations with accelerated gossip and Choco-gossip achieve an $O(\sqrt{(\log m)/T})$ convergence rate over $T$ update rounds, while experiments demonstrate improved worst-worker performance over baselines.

math.OC↗
arXiv2026

PDAGENT-BENCH: Characterizing, Grounding, and Architecting LLM/VLM Agents for VLSI Physical Design

Large Language Models and vision-language models have shown remarkable success in the front-end design of Very Large-Scale Integrated Circuits, yet their capabilities for VLSI physical design remain significantly underexplored. The primary cause is the lack of standardized benchmarks for evaluating agentic physical design workflows that require high-dimensional, multi-stage optimization under strict design constraints, coordinated interaction with diverse Electronic Design Automation tools, and iterative refinement. This work introduces PDAGENT-BENCH, a comprehensive and multi-dimensional benchmark for evaluating LLM/VLM-based agents across the physical design stack. PDAGENT-BENCH integrates both task-level assessment and workflow-level execution. The benchmark suite contains 353 curated problems that combine conceptual questions with real-world industrial artifacts, with expert-validated references and executable solutions. In addition, the benchmark provides a unified, human-aligned agentic physical design workflow framework that enables closed-loop evaluation of holistic physical design in realistic EDA environments. Experiments on 11 state-of-the-art models reveal that while modern LLMs/VLMs perform competitively on conceptual tasks, they remain substantially limited in tool-centric execution (e.g., 42.2% on Innovus script generation) and long-horizon, multi-stage reasoning. Our studies further show that human-skill-enhanced agentic workflows significantly improve end-to-end physical design performance. PDAGENT-BENCH establishes a standardized, reproducible, and realistic evaluation framework for advancing LLM/VLM-driven holistic physical design automation. To ensure full reproducibility, we release \textsc{PDAgent-Bench} at https://github.com/fenggev56/PDAGENT-BENCH.

cs.AR↗
arXiv2026

SILK: Closing the Time-of-Check-to-Time-of-Use Gap in RoT-Protected AI Systems

Root-of-trust (RoT) authentication verifies a DNN model at load time, but weights may subsequently traverse DRAM, DMA, interconnect, and prefetch paths before reaching the compute engine. Post-verification tampering along this path can therefore alter the weights actually consumed while leaving the authenticated model image unchanged, creating a time-of-check-to-time-of-use (TOCTOU) integrity gap. We present SILK (Streaming Inline Lightweight Keying), an in-place integrity mechanism that verifies the weight stream at the final pre-compute boundary. SILK repurposes quantized-weight LSBs as secret-keyed integrity bits and chains dependencies across weight bytes, so a local modification perturbs multiple integrity checks. A lightweight streaming checker recomputes these checks without separate authentication tags and uses commit gating to prevent unverified weights from reaching computation. Under a secure pseudorandom function (PRF), the forgery probability decreases exponentially with the number of affected checks, and measured miss rates closely track the analytical bound. SILK detects every stream-modifying instance in our functional attack suite. For INT8, it limits quality loss to at most 0.76 pp across evaluated CNNs and 0.17 perplexity across eight LLMs, while INT4 and MXFP4 provide a configurable security-quality tradeoff through check sparsity. On a Xilinx ZCU102, the synthesized reference pipelined implementation sustains 756 MB/s at only 1.00% of the equivalent area cost of a Caliptra 2.x RoT, while a configuration with a conservative per-attempt forgery bound of 2^-128 still sustains 678 MB/s at 6.15% of the RoT cost.

cs.CR↗
arXiv2026

CircuitsDNA: Discovering Unconventional Multi-Accuracy Arithmetic Circuits via Evolutionary Synthesis

Emerging edge AI workloads increasingly require arithmetic units that can trade computational accuracy for efficiency on demand. However, existing approximate arithmetic circuits are typically fixed-accuracy or rely on predefined structures for runtime configurability. This work introduces CircuitsDNA, an evolutionary framework that automatically evolves accuracy-configurable arithmetic circuits supporting multiple accuracy modes within a single circuit. It integrates three key features: 1) multi-threshold verifiability miter to enforce mode-specific accuracy requirements, 2) resource-limited verifiability-driven search to reduce verification overhead without sacrificing correctness, enabling efficient exploration of large circuit design, and 3) feedback-driven adaptive mutation to prioritize effective structural modifications and accelerate search convergence. Experimental results show that the 8-bit multiplier variants synthesized in 28-nm CMOS reduce the area-power product by up to 56% on INT8 DNN workload and 93% under exhaustive activity, compared with an exact 8-bit multiplier. Across CNNs and DeiTs, the accuracy loss relative to FP32 remains below 2% after fine-tuning under worst-case error (WCE) budgets of at most 1%. CircuitsDNA eliminates all search stalls observed in conventional methods across 8/12/16-bit multipliers, while adaptive mutation provides up to 1.33 times faster convergence than its non-adaptive counterpart.

cs.NE↗
arXiv2026

Heterogeneity-Aware Microscaling for Efficient Low-Bit LLM Inference

Microscaling (MX) is now the standard for low-bit large language model (LLM) inference. Its 4-bit form MXFP4 still loses substantial accuracy, because existing MX formats fix either the element format or the precision-recovery scheme across blocks, and thus capture only limited quantization heterogeneity. Quantization heterogeneity appears at two levels: 1) across blocks, the preferred element format and precision-recovery scheme vary; 2) across operands, weights and activations require different encoding. We introduce AdaMX (Adaptive Microscaling), a heterogeneity-aware format and accelerator. It selects the precision-recovery scheme per block and the representation per operand, at no increase in equivalent bit width (EBW). One design covers two block sizes, giving a higher-accuracy operating point and a lower-EBW operating point that saves storage. We implement a 22nm FD-SOI AI accelerator prototype with the proposed decoder, computing unit, and quantization logic. Against an otherwise identical MXFP4 accelerator with FP4-only multipliers, AdaMX adds about 1% system energy. At the lower-EBW point, AdaMX stays more accurate than the baseline while lowering both memory footprint and energy. Across LLMs from 3B to 70B, AdaMX removes 83% of the MXFP4 accuracy loss on commonsense and 82% on MMLU, and 43% and 27% of the NVFP4 loss. AdaMX also generalizes to multimodal models. On Gemma-4 12B, it leads MXFP4 on all four vision-language benchmarks and keeps up to 96% of FP16 accuracy.

cs.AR↗
arXiv2026

Superfluid Angular Momentum Reservoir Effect in Pulsar Glitches and Crab Pulsar Glitch Time Prediction

Pulsar glitches are usually regarded as stochastic, independent events triggered by sudden angular momentum transfer from the neutron star's superfluid interior to its crust. However, dense glitching episodes in the Crab pulsar suggest that some temporally proximate small glitches may instead form parts of broader dynamical episodes. Here we reanalyse more than five decades of Crab timing data by grouping nearby glitches into glitch clusters. In this clustered sequence, adjacent waiting times are consistent with preferred temporal organization around $\sim 3.5$ yr, and every-other cluster intervals indicate a longer-timescale component near $\sim 7$ yr. Cluster size correlates more strongly with preceding than with subsequent waiting times, with the clearest signal arising from the longer pre-history of the system. These results suggest that clustering primarily regularizes the temporal structure of the Crab glitch record and support a picture in which Crab glitches are better interpreted as temporally coupled, history-dependent collective events rather than as fully independent stochastic occurrences.

astro-ph.HE↗
arXiv2026

Exploring Superfluid Angular Momentum Reservoir Effect on Pulsar Glitches and Forecasting Next Glitches of the Crab Pulsar

Pulsar glitches are generally viewed as stochastic events driven by sudden angular momentum transfer from the neutron star's superfluid interior to its crust. Except two peculiar pulsars with quasi-periodic glitches, this stochastic view has prevailed. Here, by clustering temporally proximate small glitches of the Crab pulsar, we uncover clear evidence of an underlying quasi-periodic modulation, challenging the paradigm of purely random behavior. Furthermore, our correlation analyses reveal a strong positive relationship between glitch cluster size and waiting time since the preceding clusters. These findings demonstrate the effect of angular momentum reservoir operating over long-term scales and enable the predictions of next glitching window. Remarkably, two minor glitches detected in July and August 2025, which align with our initial prediction made in June, should be confirmed as the onset of this predicted activity. Inspired by the initial success, we forecast the occurrence of a major glitch from now until August 2026, with possible glitch size up to a relative change in rotational frequency of $697.2 \times 10^{-9}$. Physically, the observed long-term quasi-periodicity and cluster size-waiting time correlations imply that each glitch event releases only a fraction of the stored superfluid angular momentum. This partial-release mechanism provides a unified framework for both stochastic and quasi-periodic glitch behaviors across different pulsars, underscoring the universality of the superfluid angular momentum reservoir effect. As the most intensively monitored object, the Crab pulsar serves as a natural laboratory for studying angular momentum inside neutron stars.

astro-ph.HE↗
arXiv2026

Stable Routing for Mixture-of-Experts in Class-Incremental Learning

Class-incremental learning (CIL) requires models to learn new classes sequentially while preserving prior knowledge. Recently, approaches that combine pre-trained models with mixture-of-experts (MoE) have received increasing attention in CIL: they typically expand experts during learning and employ a router to assign weights across experts. However, existing MoE methods often overlook routing drift induced by expert expansion. Once new experts are introduced, the router may reassign samples from earlier classes to newly added experts, thereby perturbing previously established expert compositions and causing interference even when old experts remain frozen. We argue that expandable MoE in CIL requires two complementary properties: stable old-class routing for knowledge preservation and sufficient capacity utilization for new-class adaptation. To this end, we propose Stable Routing for MoE (StaR-MoE), a routing-level framework for expandable MoE in CIL. By incorporating sensitivity-aware routing alignment, StaR-MoE aligns current old-class routing behavior with historical routing distributions through sensitivity-guided constraints. Complementarily, StaR-MoE introduces asymmetric capacity regularization to encourage effective utilization of the expanded expert pool without compromising class-specific routing specialization. Extensive experiments across four standard CIL benchmarks demonstrate that StaR-MoE consistently improves both average and last accuracy over state-of-the-art methods, highlighting the importance of stable routing.

cs.CV↗
arXiv2026

VeRA+: Vector-Based Lightweight Digital Compensation for Drift-Resilient RRAM In-Memory Computing

RRAM-based in-memory computing (IMC) offers high energy efficiency but suffers from conductance drift that severely degrades long-term accuracy. Existing approaches including retraining, noise-aware training, and Batch Normalization (BN)-based calibration either require RRAM rewriting, demand large storage overhead, or rely on online correction. We propose VeRA+, a lightweight drift compensation framework that reuses shared projection matrices and introduces only two compact drift-specific vectors per drift level. A drift-aware scheduling algorithm offline-trains a small set of VeRA+ parameters and selects the appropriate set over time without any on-chip retraining or data replay. VeRA+ preserves up to 99.77% of the drift-free accuracy after ten years of simulated drift and reduces storage overhead by more than three orders of magnitude compared with BN-based calibration. To validate VeRA+ under realistic device behavior, we extract one-week drift statistics from measurements on our fabricated 1T1R RRAM devices and use them to simulate realistic drifted weights. Under these measured drift conditions, VeRA+ achieves accuracy close to the drift-free baseline, providing an efficient and practical solution for long-term drift resilience in RRAM-IMC.

cs.AR↗
arXiv2026

Why the Valuable Capabilities of LLMs Are Precisely the Unexplainable Ones

This paper proposes and argues for a counterintuitive thesis: the truly valuable capabilities of large language models (LLMs) reside precisely in the part that cannot be fully captured by human-readable discrete rules. The core argument is a proof by contradiction via expert system equivalence: if the full capabilities of an LLM could be described by a complete set of human-readable rules, then that rule set would be functionally equivalent to an expert system; but expert systems have been historically and empirically demonstrated to be strictly weaker than LLMs; therefore, a contradiction arises -- the capabilities of LLMs that exceed those of expert systems are exactly the capabilities that cannot be rule-encoded. This thesis is further supported by the Chinese philosophical concept of Wu (sudden insight through practice), the historical failure of expert systems, and a structural mismatch between human cognitive tools and complex systems. The paper discusses implications for interpretability research, AI safety, and scientific epistemology.

cs.AI↗
arXiv2026

Via Negativa for AI Alignment: Why Negative Constraints Are Structurally Superior to Positive Preferences

Recent empirical results have demonstrated that training large language models (LLMs) with negative-only feedback can match or exceed standard reinforcement learning from human feedback (RLHF). Negative Sample Reinforcement achieves parity with PPO on mathematical reasoning; Distributional Dispreference Optimization trains effectively using only dispreferred samples; and Constitutional AI outperforms pure RLHF on harmlessness benchmarks. Yet no unified theoretical account explains why negative signals are so effective. This paper proposes such an account: positive preferences and negative constraints are structurally asymmetric. Positive preferences ("which is better") encode continuously coupled, context-dependent human values that cannot be exhaustively specified -- leading models to learn surface correlates such as agreement with the user (sycophancy). Negative constraints ("what is wrong") encode discrete, finite, independently verifiable prohibitions that can converge to a stable boundary. This asymmetry -- rooted in Popper's falsification logic and the epistemology of negative knowledge -- explains both the sycophancy failure of preference-based RLHF and the surprising effectiveness of negative-signal methods. We argue that alignment research should shift its center of gravity from "learning what humans prefer" to "learning what humans reject," and offer testable predictions for this framework.

cs.AI↗
arXiv2026

Studies on the spin and magnetic inclination evolution of magnetars Swift J1834.9-0846 under wind braking

The magnetar Swift J1834.9-0846 presents a significant challenge to neutron star spin-down models. It exhibits two key anomalies: an insufficient rotational energy loss rate to power its observed X-ray luminosity, and a braking index of $ = 1.08\pm 0.04$, which starkly contradicts the canonical magnetic dipole value of $n=3$. To explain these anomalies, we develop a unified spin-evolution model that self-consistently integrates magnetic dipole radiation, gravitational wave emission, and wind braking. Within this framework, we constrain the wind braking parameter to $κ\in [13, 37]$ from the nebular properties, finding it contributes substantially (17%-51%) to the current spin-down torque. Bayesian inference reveals that the birth period is poorly constrained by present data and is prior-dependent, indicating a millisecond birth is allowed but not required. Furthermore, we constrain the number of precession cycles to $ξ\sim 10^{4}$--$10^{5}$, and our analysis favors a toroidally-dominated internal magnetic field configuration as the most self-consistent explanation for the low braking index. Finally, we assess the continuous gravitational-wave detectability. The present-day signal is undetectable. However, the early-time signal might have reached the projected sensitivity of next-generation gravitational-wave observatories, such as the Advanced Laser Interferometer Gravitational-Wave Observatory (aLIGO) and the Einstein Telescope (ET), although a confident detection would require exceptionally stable rotation, an assumption considered highly optimistic for a young magnetar. This work establishes a unified framework that links magnetar spin-down with their interior physics and multi-messenger observables, providing a physically consistent interpretation for Swift J1834.9-0846 and a new tool for understanding similar extreme neutron stars.

astro-ph.HE↗
arXiv2025

Continuous Subspace Optimization for Continual Learning

Continual learning aims to learn multiple tasks sequentially while preserving prior knowledge, but faces the challenge of catastrophic forgetting when adapting to new tasks. Recently, approaches leveraging pre-trained models have gained increasing popularity in mitigating this issue, due to the strong generalization ability of foundation models. To adjust pre-trained models for new tasks, existing methods usually employ low-rank adaptation, which restricts parameter updates to a fixed low-rank subspace. However, constraining the optimization space inherently compromises the model's learning capacity, resulting in inferior performance. To address this limitation, we propose Continuous Subspace Optimization for Continual Learning (CoSO) to fine-tune the model in a series of subspaces rather than a single one. These sequential subspaces are dynamically determined through the singular value decomposition of the gradients. CoSO updates the model by projecting gradients onto these subspaces, ensuring memory-efficient optimization. To mitigate forgetting, the optimization subspace of each task is constrained to be orthogonal to the historical task subspace. During task learning, CoSO maintains a task-specific component that captures the critical update directions for the current task. Upon completing a task, this component is used to update the historical task subspace, laying the groundwork for subsequent learning. Extensive experiments on multiple datasets demonstrate that CoSO significantly outperforms state-of-the-art methods, especially in challenging scenarios with long task sequences.

cs.CV↗
arXiv2025

Could the stochastic gravitational wave background from newborn magnetars be detected by the advanced LIGO and Einstein Telescope?

Newborn magnetars are important gravitational wave sources due to their ultra-strong magnetic fields and fast spins, and the entire population in the Universe may significantly contribute to the stochastic gravitational wave background (SGWB). In this work, we investigate the SGWB from newborn magnetars and assess its detectability by the advanced LIGO (aLIGO) and Einstein Telescope (ET) based on three typical formation mechanisms of magnetars, i.e., the $α-Ω$ dynamo, convective dynamo, and magnetic flux conservation. For the two dynamo scenarios, when calculating the SGWB, we creatively incorporate the anti-correlations between the magnetic fields and initial spin periods $P_{\rm i}$ with the initial dipole-field distribution of newborn magnetars. For the flux-conservation scenario, a bimodal lognormal form is adopted to describe the distribution of initial dipole fields, and all magnetars are assumed to have the same $P_{\rm i}$. Our results show that the SGWB from newborn magnetars may be undetectable by the aLIGO and ET if the magnetars are formed due to these mechanisms since the signal-to-noise ratio of the SGWB with respect to the ET for an observation time of one year is only 0.37 for the $α-Ω$ dynamo, $3\times10^{-4}$ for the convective dynamo, and at most 0.21 for the flux conservation.

astro-ph.HE↗
arXiv2025

Physics of Strong Magnetism with eXTP

In this paper we present the science potential of the enhanced X-ray Timing and Polarimetry (eXTP) mission, in its new configuration, for studies of strongly magnetized compact objects. We discuss the scientific potential of eXTP for quantum electrodynamic (QED) studies, especially leveraging on the recent observations made with the NASA IXPE mission. Given eXTP's unique combination of timing, spectroscopy, and polarimetry, we focus on the perspectives for physics and astrophysics studies of strongly magnetized compact objects, such as magnetars and accreting X-ray pulsars. Developed by an international Consortium led by the Institute of High Energy Physics of the Chinese Academy of Sciences, the eXTP mission is expected to launch in early 2030.

astro-ph.HE↗
arXiv2025

Dense Matter in Neutron Stars with eXTP

In this White Paper, we present the potential of the enhanced X-ray Timing and Polarimetry (eXTP) mission to constrain the equation of state of dense matter in neutron stars, exploring regimes not directly accessible to terrestrial experiments. By observing a diverse population of neutron stars - including isolated objects, X-ray bursters, and accreting systems - eXTP's unique combination of timing, spectroscopy, and polarimetry enables high-precision measurements of compactness, spin, surface temperature, polarimetric signals, and timing irregularity. These multifaceted observations, combined with advances in theoretical modeling, pave the way toward a comprehensive description of the properties and phases of dense matter from the crust to the core of neutron stars. Under development by an international Consortium led by the Institute of High Energy Physics of the Chinese Academy of Sciences, the eXTP mission is planned to be launched in early 2030.

astro-ph.HE↗
arXiv2025

Efficient Calibration for RRAM-based In-Memory Computing using DoRA

Resistive In-Memory Computing (RIMC) offers ultra-efficient computation for edge AI but faces accuracy degradation due to RRAM conductance drift over time. Traditional retraining methods are limited by RRAM's high energy consumption, write latency, and endurance constraints. We propose a DoRA-based calibration framework that restores accuracy by compensating influential weights with minimal calibration parameters stored in SRAM, leaving RRAM weights untouched. This eliminates in-field RRAM writes, ensuring energy-efficient, fast, and reliable calibration. Experiments on RIMC-based ResNet50 (ImageNet-1K) demonstrate 69.53% accuracy restoration using just 10 calibration samples while updating only 2.34% of parameters.

cs.AR↗