Searcharxiv⌕ Search

arXiv subjects

Jia Liu

Publications and source records attributed to Jia Liu.

At least 55 records · Page 3Linked to original sources
arXiv2026

Dark Matter implications from the LZ, PandaX-4T and XENONnT Data

We investigate a possible dark matter origin of the high-energy nuclear-recoil-like events in data from liquid xenon time projection chamber experiments, including LZ, PandaX-4T, and XENONnT, which cannot be explained by standard elastic spin-independent WIMP scattering. Using our unified DIAMX framework, built on openly available data and likelihood models, we perform the first combined profile-likelihood fits to multiple WIMP-search datasets with a total exposure of approximately 8.8 tonne $\times$ year. We consider two broad classes of dark matter-nucleon interactions, involving either velocity-dependent cross sections or inelastic (endo- and exothermic) scattering, which can reproduce the observed high-energy recoil spectrum, reaching local significances up to $3.5σ$. We further quantify the impact of $^{124}$Xe double electron capture (DEC) backgrounds, finding that variations in the poorly known DEC charge yields can shift the inferred significances from a null-like result to $3.5σ$. We further note that extending the same analysis to data from all three experiments with recoil energies up to $300~\mathrm{keV}$, when available, will provide a powerful test of the dark matter interpretation, since the $^{124}$Xe DEC background is expected to be negligible in this high-energy range.

hep-ph↗
arXiv2026

ARGUS: Defending LLM Agents Against Context-Aware Prompt Injection

Large Language Model (LLM) agents are increasingly deployed as task-oriented software systems that use runtime context to decide and act on behalf of users. This delegation model makes prompt injection especially dangerous: an attacker can hide a context-aware instruction inside evidence the agent must use to decide what to do. Existing benchmarks and defenses largely miss this setting. Benchmarks often use context-insensitive tasks where the user prompt already specifies the intended action, together with generic attack payloads independent of context. Existing defenses also do not capture the causal support from runtime evidence to concrete actions, which makes them incomplete and ineffective for context-dependent tasks. We present AgentLure, a benchmark for context-dependent tasks under context-aware prompt injection. AgentLure spans four agentic domains and eight attack vectors across six attack surfaces. To defend this setting, we propose ARGUS, a causal-provenance auditor for LLM agents. Instead of relying only on tool authorization or suspicious-context detection, ARGUS verifies whether each proposed action has a complete benign causal justification. It builds an influence-provenance graph, labels runtime spans, grounds action arguments in supporting evidence, and releases an action only when benign evidence entails it and task invariants hold. On AgentLure, ARGUS reduces attack success rate from 28.8% to 3.8% while preserving 87.5% clean utility, significantly outperforming existing defenses in the security-utility tradeoff.

cs.CR↗
arXiv2026

Optimus: A Generic Operator-Level PyTorch Model Transformation Framework

In large-scale industrial applications, deep learning models that power recommendation and ranking have complex and diverse model architectures. These models are continuously developed and refined by large teams of machine learning engineers, rendering manual optimization infeasible. Consequently, graph-based optimization techniques have become an industry standard for boosting performance, with PyTorch FX transformations leading the charge. These transformations typically rely on a set of human-engineered module-level rewrite rules which are not scalable to diverse model architectures. To address this limitation, we introduce Optimus, a general-purpose model transformation framework built in the PyTorch 2.x (PT2) machine learning compiler. With a concise set of predefined patterns, Optimus applies an efficient greedy search algorithm for pattern matching and replacement, while preserving model semantic. It is designed and implemented as a highly customizable and extensible framework integrated into the PT2 stack. Our evaluation shows that the framework can achieve up to 63% speedup, 6% peak memory reduction, and over 400 second compile time decrease for our industry-scale recommendation models compared to baselines. Optimus is open-sourced together with PyTorch 2.x as a customizable model transformation layer.

cs.PF↗
arXiv2026

Probing Purely Inelastic Scalar Dark Matter Across Colliders and Gravitational Wave Observatories

We propose and study a purely inelastic scalar dark matter model, where two real scalars-dark matter $ϕ_1$ and its excited partner $ϕ_2$ interact with the Standard Model via a Higgs portal. After mass diagonalization, only inelastic couplings remain, allowing the model to evade stringent bounds from direct detection. We show that thermal (co-)annihilation between $ϕ_1$ and $ϕ_2$ naturally yields the observed dark matter relic abundance. The same interaction structure can induce a strongly first-order phase transition in the early universe, generating detectable gravitational waves in upcoming experiments. Meanwhile, the slight mass splitting between $ϕ_1$ and $ϕ_2$, along with the heavy off-shell mediator SM Higgs, leads to long-lived particle signatures of $ϕ_2$ at the HL-LHC via the displaced muon-jets technique. We pinpoint a feasible parameter space where the correct relic abundance, observable gravitational waves, and collider signals can all be achieved concurrently, presenting a valuable chance to validate this scenario through a comprehensive examination encompassing cosmological, astrophysical, and collider investigations.

hep-ph↗
arXiv2026

On strong algebrability and spaceability of continuous functions and fractal dimensions

In this paper, we investigate the strong algebrability and $(α,β)$-lineability/spaceability of continuous functions with prescribed fractal dimensions. For $1< s< r< t\leq2$, we define $$H_s[0,1]=\{f\in C[0,1]:{\dim}_HG_f([0,1])=s\},$$ $$\underline{B}_r[0,1]=\{f\in C[0,1]:\underline{\dim}_BG_f([0,1])=r\}$$ and $$\overline{B}_t[0,1]=\{f\in C[0,1]:\overline{\dim}_BG_f([0,1])=t\}.$$ We prove that $H_s[0,1]\cap\underline{B}_r[0,1]\cap\overline{B}_t[0,1]$ is both strongly $\mathfrak{c}$-algebrable and spaceable. This complements recent findings of Bonilla et al. \cite{BFBS}, Esser et al. \cite{EMVVS}, and Liu et al. \cite{LZS}. We prove that for any $1<s\leq t\leq2$, $H_s[0,1]\cap\overline{B}_t[0,1]$ is $(p,\mathfrak{c})$-spaceable for $p=1,2$. We also prove that $H_s[0,1]\cap\overline{B}_t[0,1]$ is $(n,m+n)$-lineable for any $m,n\in\mathbb{N}$, thus complementing the recent work of Liu et al. \cite{LS}.

math.FA↗
arXiv2026

Streaming-dLLM: Accelerating Diffusion LLMs via Suffix Pruning and Dynamic Decoding

Diffusion Large Language Models (dLLMs) offer a compelling paradigm for natural language generation, leveraging parallel decoding and bidirectional attention to achieve superior global coherence compared to autoregressive models. While recent works have accelerated inference via KV cache reuse or heuristic decoding, they overlook the intrinsic inefficiencies within the block-wise diffusion process. Specifically, they suffer from spatial redundancy by modeling informative-sparse suffix regions uniformly and temporal inefficiency by applying fixed denoising schedules across all the decoding process. To address this, we propose Streaming-dLLM, a training-free framework that streamlines inference across both spatial and temporal dimensions. Spatially, we introduce attenuation guided suffix modeling to approximate the full context by pruning redundant mask tokens. Temporally, we employ a dynamic confidence aware strategy with an early exit mechanism, allowing the model to skip unnecessary iterations for converged tokens. Extensive experiments show that Streaming-dLLM achieves up to 68.2X speedup while maintaining generation quality, highlighting its effectiveness in diffusion decoding. The code is available at https://github.com/xiaoshideta/Streaming-dLLM.

cs.LG↗
arXiv2026

FISHER: A Foundation Model for Multi-Modal Industrial Signal Comprehensive Representation

Industrial signal analysis is hindered by severe data heterogeneity, which we characterize as the M5 problem. Existing solutions rely on specialized models that lack robustness and scalability, while large-scale pre-training has rarely been investigated in this area. In this work, we derive a prioritized roadmap for the M5 problem and propose FISHER, a Foundation model for multi-modal Industrial Signal compreHEnsive Representation. To address the foremost multi-sampling-rate problem, FISHER utilizes a novel sub-band modeling approach that treats sampling rate increments as concatenated sub-band information, enabling the adaptive usage of full signal bandwidth without resampling. FISHER is pre-trained by teacher-student self-distillation over external audio and music data. We also establish the RMIS benchmark, comprising 19 datasets across four modalities. In the experiment, FISHER outperforms 24 state-of-the-art series encoders (up to 2B) with much smaller sizes (up to 16x), showcasing groundbreaking diagnostic accuracy and remarkable versatility. We further demonstrate that 1) seamless adaptation to variable sampling rates is the key to generalization 2) audio and music data provide better temporal variability, which is essential for pre-training. Both FISHER and RMIS are open-sourced.

cs.LG↗
arXiv2026

Ferromagnetic broadband sensing of axionlike dark matter

Levitated particles have demonstrated ultrahigh sensitivity to magnetic fields and accelerations owing to their extremely low dissipation. Such systems have strong potential for fundamental physics research, particularly for the detection of axions and axionlike particles, well-motivated dark matter candidates spanning a broad mass range. In this context, both high sensitivity and large bandwidth are essential. Here, we demonstrate a levitated magnet magnetometer based on an engineered double-resonance mode, achieving an effective linewidth at its optimal sensitivity that is approximately three orders of magnitude broader than those of previous approaches. Together with a hard-magnet array that enhances the axion-induced signal and soft-ferromagnetic shielding that suppresses environmental magnetic noise, this system constitutes a hybrid ferromagnetic platform for axionlike dark matter searches. We search for axionlike dark matter through its photon coupling $g_{aγ}$ over the $40$-$3000\,\mathrm{Hz}$ frequency range and establish new direct limits in this frequency band. The best sensitivity is achieved near the upper resonance around $276\,\mathrm{Hz}$, where the magnetometer reaches a magnetic-field resolution of $0.7\,\mathrm{fT}$, corresponding to a limit of $g_{aγ}\sim10^{-7}\,\mathrm{GeV}^{-1}$. At this frequency, this result improves upon previous direct limits by more than four orders of magnitude. The demonstrated high-bandwidth levitated sensor may also enable a broad range of applications, including biological sensing and precision measurements.

hep-ex↗
arXiv2026

DUET: Decentralized Bilevel Optimization without Lower-Level Strong Convexity

Decentralized bilevel optimization (DBO) provides a powerful framework for multi-agent systems to solve local bilevel tasks in a decentralized fashion without the need for a central server. However, most existing DBO methods rely on lower-level strong convexity (LLSC) to guarantee unique solutions and a well-defined hypergradient for stationarity measure, hindering their applicability in many practical scenarios not satisfying LLSC. To overcome this limitation, we introduce a new single-loop DBO algorithm called diminishing quadratically-regularized bilevel decentralized optimization (DUET), which eliminates the need for LLSC by introducing a diminishing quadratic regularization to the lower-level (LL) objective. We show that DUET achieves an iteration complexity of $O(1/T^{1-5p-\frac{11}{4}τ})$ for approximate KKT-stationary point convergence under relaxed assumptions, where $p$ and $τ$ are control parameters for LL learning rate and averaging, respectively. In addition, our DUET algorithm incorporates gradient tracking to address data heterogeneity, a key challenge in DBO settings. To the best of our knowledge, this is the first work to tackle DBO without LLSC under decentralized settings with data heterogeneity. Numerical experiments validate the theoretical findings and demonstrate the practical effectiveness of our proposed algorithms.

math.OC↗
arXiv2026

Rethinking Shrinkage Bias in LLM FP4 Pretraining: Geometric Origin, Systemic Impact, and UFP4 Recipe

FP4 training promises substantial reductions in memory and computation cost for LLM pretraining, yet current FP4 hardware paths and recipes, including NVIDIA Blackwell/Rubin-class systems and AMD MI350-series GPUs, remain centered on E2M1 data elements. In this study, we identify a fundamental limitation of that choice: non-uniform formats such as E2M1 inherently suffer from Shrinkage Bias, a systematic negative rounding error caused by the geometric asymmetry of their representable bins. We show that this bias accumulates multiplicatively across layers and is amplified by the Random Hadamard Transform (RHT), providing a unified explanation for the training instability observed in existing E2M1-based FP4 recipes. In contrast, uniform grids (E1M2/INT4) bypass this grid-geometry error and better convert the improved bucket utilization from RHT into higher quantization quality. Based on this finding, we propose UFP4, a uniform 4-bit training recipe that applies RHT to all three training GEMMs while restricting stochastic rounding to dY alone. On Dense 1.5B, MoE 7.9B, and MoE 124B long-run pretraining, UFP4 consistently achieves lower BF16-relative loss degradation than strong E2M1-based baselines, supported by scaling-law analysis and ablation studies. Our results suggest that future accelerators should support E1M2/INT4-style uniform 4-bit grids as first-class training primitives alongside E2M1.

cs.AI↗
arXiv2026

How to Train a Real-World Silicon Concierge? Internalizing Complex Business Workflow to Only OneModel

Traditional industrial agents rely on modular pipelines, including Router, Retriever, Planner, Executor, Responder, Reviewer, and other components. These systems often fracture into a labyrinth of ad-hoc patches, leading to cascading errors and high latency. We propose OneModel, an applicable paradigm shift from external workflows to internalized knowledge representation. Unlike modular systems that slice fluid user intents into static steps, OneModel consolidates complex business logic and SOPs directly into the model parameters. Through Continual Pre-training (CPT) and logic-compilation SFT, we transform fragmented business rules into intuitive model reasoning within a unified attention space. Deployed in our global financial service system, OneModel effectively breaks the trade-off between latency, accuracy, and complexity. Online A/B testing demonstrates an end-to-end latency reduction of more than 50 percent, from 18.7 seconds to 8.0 seconds, while the Intelligent Resolution Rate (IRR) increases from 64.3 percent to 83.3 percent. The results show that OneModel can replace brittle engineering logic with internalized cognitive intuition, offering a scalable blueprint for transitioning industrial agents from complex, error-prone workflows to unified model architectures.

cs.CL↗
arXiv2026

Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale

Efficient and scalable agentic intelligence requires models that can deliver both low-latency responses and strong reasoning capabilities while remaining practical to train, serve, and deploy. In this report, we present Ling-2.6 and Ring-2.6, a family of models designed to address this challenge at scale. Ling-2.6 is optimized for instant response generation and high capability per output token, whereas Ring-2.6 is tailored for deeper reasoning and more advanced agentic workflows. Instead of training from scratch, we upgrade the Ling-2.0 base model through architectural migration pre-training and large-scale post-training. This upgrade is guided by a unified co-design of model architecture, optimization objectives, serving systems, and agent training environments, enabling improvements in both model capability and deployment efficiency. At the architectural level, we introduce a hybrid linear attention design that integrates Lightning Attention with MLA, improving the efficiency of long-context training and decoding. To further enhance token efficiency, we optimize capability per output token through Evolutionary Chain-of-Thought, Linguistic Unit Policy Optimization, bidirectional preference alignment, and shortest-correct-response distillation. For agentic capabilities, we propose KPop, a reinforcement learning framework designed to support stable training of Ring-2.6-1T on large-scale environment-grounded data. KPop improves training efficiency through asynchronous scheduling across coding, search, tool use, and workflow execution, enabling scalable learning from complex agent-environment interactions. Together, Ling-2.6 and Ring-2.6 provide a practical pathway toward efficient, scalable, and open agentic systems. We open-source all checkpoints in the 2.6 family to support further research and development in practical agentic intelligence.

cs.CL↗
arXiv2026

Maximum Utility Split Method for Utility Preference Elicitation

In this paper, we propose a new approach, called maximum utility split (MUS) scheme, which is built on random utility split (RUS) scheme but with a notable difference: one lottery is designed with two fixed outcomes but with varying probability, and the other has a deterministic outcome specifically chosen at the point where the range between the largest and smallest possible utility values is maximized. Consequently, the probability of random lottery is set such that the range of the ambiguity set of utility functions is reduced by half at the point. Under moderate conditions, we show that MUS can successively generate a sequence of such questionnaires and effectively reduce the ambiguity set, eventually converging to the true utility function as the number of questionnaires increases. The main challenge is to effectively identify the point with the largest utility range for a given ambiguity set constructed from preference information. Based on the structure of the ambiguity set, we propose an interval-based algorithm which identifies each certain-outcome lottery by solving a sequence of linear programs. Moreover, to deal with the case where elicitation terminates before the ambiguity set reduces to a singleton, we demonstrate how to figure out a nominal utility function by solving optimization programs. These identify the smallest and largest utility functions under the Kantorovich metric within the ambiguity set, after which we identify a nominal utility function located in the middle of them. Finally, numerical results demonstrate the efficiency of the MUS method and the performance of a robo-advisor system based on MUS-type queries and the nominal utility elicitation. While the main discussions focus on concave utility functions, we also demonstrate how the MUS approach can be extended to accommodate general non-concave utility functions, particularly S-shaped ones.

math.OC↗
arXiv2026

Coordinate-wise Polyhedral Method for Eliciting Multivariate Linear Utility and Univariate Nonlinear Utility Functions

In this paper, we propose a coordinate-wise polyhedral method (CPM) for cutting polyhedra with theoretical guarantees of convergence. Unlike the existing polyhedral method, which designs pairwise comparison queries by solving coupled optimization problems and performs cuts subsequently, CPM specifies coordinate-wise cuts in advance and then designs corresponding pairwise comparison queries by solving a linear system of equations. Under this framework, we show that CPM reduces the diameter of the polyhedron at a linear convergence rate. Moreover, we extend CPM to a univariate piecewise-linear utility function by representing it with its increment vector over consecutive breakpoints. We show that the Kantorovich distance between the true utility function and the estimated one obtained by CPM decreases at a linear convergence rate. Further, we extend the CPM to general nondecreasing Lipschitz continuous utility functions by piecewise-linear approximation (PLA). We introduce an adaptive-breakpoint strategy to avoid direction errors caused by PLA. We prove that the set of piecewise-linear utility functions corresponding to the ambiguity set of increment vectors converges to the true utility function as the number of queries increases, and derive an explicit bound for the approximation. Finally, to evaluate the performance of CPM, we conduct a series of numerical experiments. The results demonstrate comparable convergence to the standard polyhedral method in the case of linear multivariate utility functions. For nonlinear univariate utility functions, CPM achieves stable convergence to the true utility function, in line with the theoretical findings.

math.OC↗
arXiv2026

Centrifugal-corrected harmonic oscillator model for spherical proton emitters

In the present work, we propose an improved harmonic oscillator model to systematically evaluate the proton radioactivity half-lives in spherical nuclei, incorporating centrifugal potential effects. By fitting the experimental data, the centrifugal parameter $d = 0.143$ for the correction term $dl(l+1)$ and nuclear potential depth $V_0 = 62.4$ MeV are obtained. The model integrates the relativistic mean field (RMF) theory with the BCS method based on the DD-ME2 force to determine spectroscopic factors $S_p$. Moreover, by verifying the linear relationship between the logarithm of the normalized width $\log_{10}{γ^2}$ and fragmentation potential $V_{frag}$, the connection between nuclear structure and tunneling dynamics is confirmed, and an analytical expression for the adjustable parameter $d$ corresponding to the centrifugal potential is derived as $d^{\rm{Ae}}$ $\approx$ 0.167. Compared with $d^{\rm{Ae}}$, the modified model based on $d$ yields results in better agreement with experimental half-lives, and is able to control the error of the experimental data within a factor of 2.4. Furthermore, the extended improved model is used to predict the half-lives of some possible proton radioactivity candidates in NUBASE2020 that are energetically allowed or have been observed but not yet quantified. This work improves the accuracy of proton radioactivity studies and provides a robust theoretical framework for future nuclear structure research.

nucl-th↗
arXiv2026

Mitigating the Curse of Dimensionality in Uniform Convergence of Deep Neural Networks via Smooth Activations

This paper establishes a theoretical framework for the uniform convergence of smoothly activated deep neural network (DNN) estimators. While standard ReLU networks achieve minimax-optimal rates in the $L^2(P)$ norm for various nonparametric regression tasks, we establish a theoretical lower bound demonstrating that least-squares ReLU estimators can suffer from the curse of dimensionality in their uniform convergence behavior. Motivated by the need for reliable uniform guarantees in downstream tasks requiring worst-case reliability, we address this limitation by analyzing smoothly activated DNNs (smooth DNNs), encompassing both feedforward and residual structures. We establish novel pseudo-dimension bounds, non-asymptotic approximation guarantees, and Hölder-norm bounds for the approximators of these models. Leveraging these results, we derive non-asymptotic uniform convergence rates for smooth DNN estimators across multiple statistical contexts, including Huber, least-squares, quantile, and logistic regression. We prove that smooth DNNs can mitigate the {curse of dimensionality} in uniform convergence by adaptively exploiting the low-dimensional hierarchical composition structure of the target function. Supported by both simulation studies and a real-world application, our results position smooth DNNs as a theoretically grounded and practically viable alternative to ReLU networks for statistical learning tasks requiring uniform guarantees.

cs.LG↗
arXiv2026

From Live to Recording: Consumer Demand and Response to Price Across the Livestreaming Lifecycle

Livestreaming has evolved into a thriving industry where creators can directly monetize and engage with their audiences and followers. In practice, creators and platforms typically concentrate their marketing efforts on the period leading up to the livestream. However, livestreaming events naturally transition into recorded formats once the event concludes, creating potential "residual" opportunities for monetization. This study systematically examines consumer demand for live events throughout the entire livestream life-cycle, using data from a large livestreaming platform that allows consumers to purchase the recorded version of a paid live event after the livestream ends. We find that the demand is surprisingly more price-sensitive during the pre-livestream period compared to the post-period. This is partly driven by two mechanisms: consumer self-selection (infrequent consumers who may have missed the live events exhibit a higher willingness to pay for recorded versions) and quality uncertainty (consumers face higher uncertainty in event quality during the pre-period than in the post-period). Our findings generate implications for the pricing and targeting strategies in livestreaming markets.

stat.ML↗
arXiv2026

On strong spaceability of continuous functions and fractal dimensions

Given $s\in(1,2]$, define $$H_s[0,1]=\{f\in C[0,1]:{\dim}_HG_f([0,1])=s\}$$ and $$\overline{B}_s[0,1]=\{f\in C[0,1]:\overline{\dim}_BG_f([0,1])=s\}.$$ The main goal of this paper is to study the $(α,β)$-lineability/spaceability of the sets $H_s[0,1]$ and $\overline{B}_s[0,1]$. As a principal result, we prove that $H_s[0,1]$ is $(p,\mathfrak{c})$-spaceable for $p=1,2$ and also $(n,n+m)$-lineable for any $m,n\in\mathbb{N}$. This partially answers a question raised by Liu et al. concerning the Hausdorff dimension of graphs of continuous functions. Furthermore, for a cardinal number $α$, we prove that $\overline{B}_s[0,1]$ is $(α,\mathfrak{c})$-spaceable if and only if $α<\aleph_0$. This completely resolves an open question raised by Liu et al. concerning the upper box dimension of graphs of continuous functions.

math.FA↗