arXiv subjects

Jia Li

Publications and source records attributed to Jia Li.

At least 19 recordsLinked to original sources

arXiv2026

TV-Regulated OPD: Direction Matters in On-Policy Distillation

On-Policy Distillation (OPD) facilitates the transfer of knowledge from domain expert to student in the post-training phase of Large Language Models (LLMs). However, the supervision signals in mainstream OPD methods suffer from high variance and noise which is generally instable during training. In this work, we systematically investigated what really matters to the performance and the fundamental mechanisms behind the instability during training. We found that retaining only the sign of token-level advantages is sufficient to achieve the performance comparable to standard OPD. Meanwhile, smoother and bounded advantages can stabilize the training process without sacrificing its performance. These motivated us to shape the advantages using the Total Variation (TV) and propose a robust TV regulated On-Policy Distillation (TV-OPD) method. Benefiting from the bounded and diminished advantages, TV-OPD exhibits stable training dynamics and steady late-stage performance. We conducted comprehensive experiments and found that, across various settings, TV-OPD consistently achieved better performance and lower variance in the late-stage of training.

cs.LG↗

arXiv2026

Single Image to Textured 3D Object Generation in Frequency Domain: From Theory to Pipeline

Single-view 3D reconstruction, also known as image-to-3D, is a persistently challenging task due to the extreme lack of information. Recently, diffusion models pre-trained on large-scale datasets served as 2D priors are used to solve the ill-posed task but suffer from color deviation and view inconsistency, which can be curbed by using diffusion models fine-tuned with 3D annotated data served as 3D priors. However, 3D priors lack high-frequency details, which cannot be solved by direct complementation with 2D priors in spatial domain for introducing erroneous low-frequency 2D prior guidance. In this paper, we revisit the characteristics of different diffusion priors from the frequency perspective. Based on our observations, we theoretically present a unified framework of hybrid optimization using multiple diffusion priors in frequency domain. Under this framework, we further propose Morpheus3D, a pipeline of 3D object generation from any single unposed image in the wild. Morpheus3D enhances 3D prior with high-pass image-prompt 2D prior guidance to reconstruct high-quality 3D objects while effectively suppressing view inconsistency, low-frequency color deviation, and high-frequency lacking problems. Both quantitative and qualitative experiments on the public and our collected datasets with complex textures show that our method exhibits significant improvements in generation quality.

cs.CV↗

arXiv2026

Glivenko--Cantelli Theorems for Integrated Volatility Functionals in Pure-Jump Semimartingales with an Application to Cryptocurrency Markets

We develop a two-step procedure for estimating integrated volatility functionals, defined through the occupation measure of the latent spot volatility process, when the asset price is a pure-jump semimartingale. In the first step, block-based estimators formed from absolute powers of high-frequency increments uniformly approximate local averages of powers of volatility. In the second step, these estimates are aggregated into an empirical occupation measure. Since price increments have infinite variance in this setting, arguments based on local Gaussianity are unavailable, and the uniform theory instead rests on maximal inequalities tailored to the stable regime. We establish Glivenko--Cantelli-type uniform consistency over classes of bounded monotone, Lipschitz-in-parameter, and locally H\"older test functions. These results deliver consistent estimation of volatility occupation times and quantiles, together with an argmax-consistency theory for $M$-estimators built on nonparametrically recovered latent processes. We further propose a stability-based rule for selecting the power index of the volatility estimator, which tracks an infeasible ex ante optimal choice closely in Monte Carlo experiments. An application to high-frequency cryptocurrency markets illustrates the framework in a jump-dominated, heavy-tailed environment.

math.ST↗

arXiv2026

Prototype-guided transfer of sparse literature knowledge for electrolyte additive discovery

Electrolyte additive discovery remains challenging because experimentally validated molecules are sparse, whereas accessible chemical spaces are vast and largely unlabeled. This challenge is amplified in lithium-ion batteries, where additive performance arises from coupled interfacial reactions rather than a single molecular property. Here, we develop a prototype-guided molecular intelligence, ProtoMI, a literature-driven framework that learns transferable structural priors from reported electrolyte additives and uses them to prioritize candidates in unlabeled chemical space. For boron-containing additives, ProtoMI combines 126 literature-reported molecules with 179,977 unlabeled candidates. Graph contrastive learning identifies seven chemically interpretable prototypes from the reported additives, and prototype guided semi-supervised contrastive learning adapts these prototypes to the candidate space under source-target distribution mismatch. In retrospective temporal validation, ProtoMI achieves enrichment factors of 9.2-45.6 while screening less than 2% of the candidate space. A subsequent translation step identifies four commercially accessible candidates. One representative candidate, 4,4,5,5-Tetramethyl-2-[10-(1naphthyl)anthracen-9-yl]-1,3,2-dioxaborolane (TNDB), improves high-temperature LiFePO4||graphite cycling at 55 {\deg}C by 34.93% relative to the baseline electrolyte. An arsenal of characterizations and operando optical fiber Fourier transform infrared spectroscopy suggest that TNDB forms B-containing, F/P/O-modified inorganic interphases, suppresses solvent decomposition and reduces Fe deposition on graphite. This case study shows how sparse literature knowledge can guide experimentally efficient molecular discovery in data-scarce battery-additive spaces.

physics.chem-ph↗

arXiv2026

Agentic Quantitative Trading: A Survey of Workflows, Systems, and Evaluation

Quantitative trading is moving from isolated predictive models toward agentic workflows that combine reasoning, tool use, memory, and feedback. This survey reviews agentic quantitative trading across five stages: factor mining, signal discovery, portfolio construction, order execution, and risk management. We further examine agentic quant trading systems through architecture, coordination, and adaptation, while comparing benchmarks across strategy construction, offline trading, live market evaluation, and reliability assessment. Our review finds that current systems remain concentrated on signal discovery, while complete integration with portfolio construction, execution, and risk control is still uncommon. Multi-agent systems also rely heavily on aggregation despite increasingly diverse workflow structures. Benchmark evidence further shows that strong model or forecasting capability does not reliably translate into trading performance under live market conditions and reliability controls. We conclude with future directions for more complete trading workflows, stronger coordination, and evaluation matched to the capability being assessed.

q-fin.CP↗

arXiv2026

Adapting Dense Vision-Language Relationships for Multi-label Classification with Partial Label

Learning multi-label image classification with incomplete annotations is a challenging task that has been widely studied for its superior trade-off between high efficiency and less labor consumption on large-scale datasets. Predominant methods rely on strong prior assumptions to recover the missing semantics from partial annotations. However, these statistic priors suffer from unstable semantic mistakes and thus lead to catastrophic overfitting. Toward this end, we propose a Language-driven Dense Semantic Adaptor (LDSA) that excavates prior-adaptive relationships from multimodal pretrained CLIP models. In our approach, the densely contrastive adaptor is first proposed to construct dense visual contrastive constraints, transferring the task-specific knowledge to visual domains. We then propose a language-driven interactive decoder with the help of class-specific prompt tuning, which adapts language proxies with visual domains. With the collaborative learning of proposed modules, experimental results demonstrate our proposed LDSA achieves a new state of the art on public multi-label classification benchmarks, and interpretable analyses reveal that our LDSA discovers implicit semantic relationships with the prior-adaptive learning scheme.

cs.CV↗

arXiv2026

LiST: Local-Simplex Test-Time LoRA Fusion

Task-specific LoRA adapters offer a modular way to specialize large language and vision-language models. However, existing adapter composition methods are mostly static and cannot adapt to individual test inputs. To address these issues, we propose \textbf{LiST}, a label-free test-time LoRA fusion framework that converts an existing LoRA bank into a target-conditioned local simplex and searches sample-specific fusion weights at inference time. LiST builds joint task representations from LoRA parameter anchors and prompt-level behavior vectors, retrieves neighboring adapters as a local search space, and performs branch-preserving fusion without updating the backbone or adapters. Candidate weights are selected by a prompt-level energy with prior, geometric, and stochastic-consistency constraints, and are deployed only when they pass a safe acceptance rule. Otherwise, LiST falls back to a target-conditioned prior. Experiments on multimodal and language benchmarks show that LiST outperforms static LoRA merging and conventional test-time adaptation baselines, while preserving task-specific adapter utility and improving robustness on unseen tasks.

cs.CV↗

arXiv2026

State-Space Model-Enabled Reinforcement Learning for Magnetic Configuration Controlon EXL-50U

Accurate feedback control of the plasma current ($I_p$) and centroid position $(R_c,Z_c)$ is essential for the stable operation of spherical torus (ST) plasmas. Conventional proportional-integral-derivative (PID) controllers require extensive manual tuning and struggle with the fast, strongly coupled dynamics that arise as plasma performance improves. Reinforcement learning (RL) has recently emerged as a promising alternative to such complex magnetic control problems, yet its practical deployment on ST devices remains challenging. This paper presents a practical RL controller for the EXL-50U ST, trained within a rigid RZIP state-space model (SSM) that enables efficient offline policy learning. A lightweight plasma position reconstructor is developed to estimate $(R_c,Z_c)$ from magnetic probe signals within the real-time control cycle. The trained policy is seamlessly deployed on the EXL-50U plasma control system, achieving stable regulation of $I_p$ and $(R_c,Z_c)$ and sustaining discharges up to 650 ms under RL control. These results demonstrate the feasibility and practical potential of model-informed RL for magnetic configuration control in ST devices, offering a promising direction beyond conventional PID-based schemes.

physics.plasm-ph↗

arXiv2026

Amorphous and Nanocrystalline Topological Semimetal YPtBi/W/CoFeB Heterostructures for BEOL-Compatible Spin-Orbit Torque Devices

Spin-orbit torque (SOT) devices require spin-source materials that combine efficient charge-to-spin conversion with back-end-of-line (BEOL) thermal compatibility. Here, we show that YPtBi/W/CoFeB heterostructures deposited directly on Si/SiOx remain predominantly amorphous or weakly nanocrystalline from room temperature to 400 {\deg}C while preserving a large effective damping-like SOT response. Anomalous Hall and harmonic Hall measurements, together with X-ray diffraction, cross-sectional transmission electron microscopy, X-ray reflectivity, and electron energy-loss spectroscopy, show that the response does not correlate with bulk crystallization of YPtBi. Instead, the interfacial analysis indicates that the strongest trend of the spin Hall angle is associated with the chemistry of the upper YPtBi/W boundary: the effective SOT response tracks the integrated W concentration at that YPtBi surface. Meanwhile, a two-spin source analysis shows that the Pt-W-rich interlayer provides only a small positive correction, insufficient to explain the large negative effective spin Hall angle by itself. The dominant control variable is therefore inferred to be the incorporation of W into the upper YPtBi interface, which plausibly modifies the local electronic structure of YPtBi and amplifies the stack-level response. These results provide a more physically constrained interpretation of the stack behavior and identify a BEOL-compatible route to disordered topological spin-source layers for scaled SOT memory and compute-in-memory hardware.

cond-mat.mtrl-sci↗

arXiv2026

Sharp Sobolev Approximation on General Domains by Linearized Shallow Networks with Analytic Activations

We study Sobolev approximation on bounded domains by linearized shallow neural networks whose inner parameters are prescribed independently of the target function. Our main step is a one-dimensional construction for analytic activations. We prove that quasi-Chebyshev parameter sets with univariate resolution $m$ generate fixed feature spaces attaining the sharp $H^r$-to-$H^s$ approximation order $m^{-(r-s)}$ for a class of analytic activations satisfying a quantitative non-cancellation condition on their Taylor coefficients. Combining this result with the ridge-function lifting theorem in [SIAM J. Math. Anal. 30 (1998), pp. 155-189] and its extension to arbitrary quasi-uniform direction sets established in this work, we construct tensor-product-type parameter sets that attain the sharp rate $$\|f-f_n\|_{L^2(\Omega)}\lesssim n^{-\frac rd}\|f\|_{H^r(\Omega)},\quad f\in H^r(\Omega)$$ for all $r>0$. In contrast to the finite-difference construction in [Neural Comput. 8 (1996), pp. 164-177], whose explicit admissibility condition may require an extremely small parameter scale, the proposed parameter sets remain distributed over fixed intervals and are therefore more amenable to practical computation.

math.NA↗

arXiv2026

$R^3$-Bench: LLMs Struggle with Resource-Rational Reasoning under Shared Budgets

In cognitive science, resource rationality asks how an agent should allocate limited computation to maximize expected value. Most reasoning and agent benchmarks use independent per-task budgets; existing shared-budget studies do not calibrate suite performance against the same model's demonstrated single-problem competence. We introduce $R^3$-Bench, which evaluates six-problem suites under shared budgets across mathematics, competitive programming, and abstract reasoning in tool-free and agentic settings. Matched single-problem response curves define an offline empirical oracle over observed successes. Across 72 main-table cells for six models, the oracle mean matches or exceeds the contest mean in all cells and is strictly higher in 71. Under moderate tool-free pressure, equal-allocation replay also exceeds contest performance for four of six models. Trajectory diagnostics reveal limited strategy updating and pressure-dependent failure patterns. In a three-model diagnostic under strong agentic pressure, at least one fixed scheduler exceeds the contest mean in six of nine cells, but no policy dominates across domains. These results expose a persistent gap between demonstrated competence and shared-budget realization.

cs.CL↗

arXiv2026

SubZero+: Efficient Zeroth-Order LLM Fine-Tuning via Large Learning Rates

Zeroth-order (ZO) optimization enables backpropagation-free fine-tuning of large language models, but existing ZO methods suffer from high-variance gradient estimators, making convergence unstable and highly sensitive to learning rates. We propose SubZero+, an improved SubZero framework that improves stability in three complementary ways: (i) multi-query gradient estimation within layer-specific low-rank subspaces to reduce variance without exhibiting the multi-query paradox; (ii) a subspace Adam optimizer that performs adaptive updates using in-subspace multi-query gradient statistics; and (iii) a sign correction for QR-based subspace construction to ensure Haar-distributed projection matrices, eliminating implementation-dependent orientation ambiguity. Experiments on models from 1.3B to 32B across SuperGLUE, under both full-parameter tuning and LoRA, show that SubZero+ consistently outperforms prior ZO baselines, enlarges the stable learning-rate range, and narrows the gap to first-order methods with minimal extra memory overhead.

cs.LG↗

arXiv2026

CETalk: Continuous Valence-Arousal Control for Audio-Driven 3D Talking Head Generation

Emotional 3D talking head generation aims to synthesize expressive facial animations with accurate lip synchronization. However, existing methods often rely on discrete emotion categories, which fail to capture the continuous evolution of affect. They also overlook the temporal frequency mismatch between audio articulation and emotional expression. In this paper, we propose CETalk, an audio-driven 3D facial animation framework conditioned on continuous Valence--Arousal (VA) representations for fine-grained emotion control. CETalk predicts a sequence of FLAME parameters through three key components: a Dynamic Emotion Modulation Module that adaptively scales emotional intensity using audio-derived cues; a Multi-Scale Temporal Modeling mechanism that employs parallel branches to decouple high-frequency articulatory movements from low-frequency emotional dynamics; and a Dynamic Fusion Mechanism that integrates these multi-scale features via an adaptive gating network. To support training and evaluation, we construct 3D-VA-MEAD, a large-scale dataset with automatically estimated VA annotations and reconstructed 3D facial motions. Extensive experiments demonstrate that CETalk outperforms state-of-the-art methods in both lip-sync accuracy and emotional expressiveness, while enabling smooth and controllable emotion transitions.

cs.CV↗

arXiv2026

Envs-FORGE: Frontier-Optimized Reward-Grounded Environment Synthesis for Agent RL

Reinforcement learning (RL) for terminal agents needs executable training environments with reliable rewards and useful difficulty. Fixed recipes such as few-shot, Self-Instruct, and Evol-Instruct apply the same prompting policy to every seed, even when the current policy would benefit from a harder, easier, or simply different task. We present Envs-FORGE, a prompting policy that converts verifier rewards into per-seed environment-synthesis actions. Envs-FORGE estimates seed pass rates, scores six projection--direction actions around a target learning frontier, and solves a per-seed mixed-integer linear program (MILP) to choose the action that conditions generation. The selected action drives synchronized rewriting of the instruction, fixtures, oracle solution, tests, and Docker environment; only gold-verified bundles enter RL training. The indexed MILP form also supports optional soft skill coverage for portfolio planning. On Qwen 3.5 35B, Envs-FORGE improves Pass@1 over Base by 9.2 percentage points on tb-core (40.0% to 49.2%) and 6.4 points on tb-2.0 (23.0% to 29.4%), exceeding the strongest fixed-recipe baseline by 2.4 and 2.1 points. It reaches 77.1% on SWE-bench Verified versus 73.4% for Base, and improves tb-core by 6.8--9.2 points across the evaluated 4B--35B models. All synthesis methods export 100 verified environments and use 2.27M--2.88M synthesis tokens, placing the comparison at the same downstream training-set size and the same operational scale. The source code is available at https://github.com/DataArcTech/DataArc-SynData-Toolkit/.

cs.CL↗

arXiv2026

Finiteness of Equidistant Affine Subspaces

Let $0\leqslant k<n$, and let $\mathcal N(k,n)$ denote the supremum of the cardinalities of finite families of affine $k$-planes in $\R^n$ whose pairwise Euclidean separation distances are all equal to one. For $k=1$, the finiteness of $\mathcal N(1,n)$ was recently established by Solymosi and Zahl. We extend this result to affine subspaces of arbitrary dimension and prove the fully explicit bound \[ \mathcal N(k,n) \leqslant 2^{\,2^{\,10(k+1)(n-k)}}. \] The main difficulty in higher dimensions is that the rational distance formula degenerates along several possible rank strata of $\dim(U+V)$, rather than only along the parallel locus. For each fixed rank $r$, we replace the choice of a nonvanishing Gram minor by a global sum-of-squares expression involving all minors. This produces a polynomial relation $H_r(x,y)=0$ which encodes both the unit-distance equation and an upper-rank condition, while also vanishing on the diagonal. We propagate this relation by means of the restricted Zariski closure and decompose the resulting algebraic set into an explicitly bounded number of connected Nash submanifolds. An analytic diagonal approximation argument shows that, within each Nash piece, the maximal possible value of $\dim(U_x+U_y)$ must decrease. Iterating this rank descent over all possible ranks and combining the finitely many standard affine Grassmann charts yields the stated double-exponential upper bound.

math.MG↗

arXiv2026

LazyTrain: Limited-resource Allocation toward Zero-waste Yield Optimization in Large Language Model Training

Training large language models on limited hardware is increasingly a scheduling problem across GPU compute, host memory, PCIe transfer, and storage bandwidth. Existing offloading systems reduce GPU residency, and MegaTrain shows that a CPU-master layer-streaming executor can train large models on a single GPU, but fixed checkpointing and placement heuristics still leave communication exposed on the critical path. We propose LazyTrain, an optimization layer over a layer-streaming executor. LazyTrain formulates checkpoint selection, activation placement, recomputation, and CPU-GPU-NVMe communication overlap as a mixed-integer scheduling problem, then executes the solved policy during training. It further couples 8-bit optimizer states with fast gradient clipping as a single Hybrid 8-bit operator: state compression reduces optimizer-state memory, while fast clipping counteracts the additional CPU-side update overhead. Across H800 experiments from Qwen2.5-3B to Qwen3.6-27B, LazyTrain improves sustained TFLOPS over matched baselines runs by approximately 1.24$\times$; RTX 3090 experiments likewise increase the maximum feasible batch size by one at each model scale. In the primary Qwen3.6-27B H800 MetaMathQA run, LazyTrain reaches 219.95 TFLOPS and 1361 tokens/s at batch size 72, peaks at 68.84\,GB of GPU memory, and obtains 95.42\% exact-match accuracy on the full evaluation split. The source code is available at https://github.com/DataArcTech/LazyTrain.

cs.CL↗

arXiv2026

Regularized Multitangent Functions and Reduction Theorem

We develop a direct analytic theory of stuffle-regularized multitangent functions and prove their reduction to finite linear combinations of monotangent functions, without using mould calculus. We first establish an asymptotic comparison between one-sided truncated multiple Hurwitz zeta functions and their stuffle regularizations at the natural parameter $T_N-H(s)$, where $T_N$ is the harmonic truncation and $H(s)$ is the harmonic-number function. Applying this comparison to symmetric multitangent truncations yields meromorphic, $1$-periodic regularized multitangent functions. An explicit partial-fraction decomposition of each summand, combined with asymptotic estimates for moving truncation ranges, gives formulas for the reduction coefficients in terms of stuffle-regularized multiple zeta values. The constant term and the coefficient of the monotangent $\mathcal T(1;s)$ are determined from the limits as $\operatorname{Im}s\to\pm\infty$: both vanish whenever the index contains an entry greater than $1$, whereas the exceptional indices $\{1\}^r$ are evaluated through a sine-quotient generating function. As a consequence, we obtain a family of relations among regularized multiple zeta values.

math.NT↗

arXiv2026

Efficient Grammar-Constrained Decoding via Parser Stack Classification

LLMs are widely used to generate structured output like source code or JSON. Grammar-constrained decoding (GCD) can guarantee the syntactic validity of the generated output, by masking out tokens that violate rules specified by a context-free grammar. However, the online computational overhead of existing GCD methods, with latency typically scaling linearly with vocabulary size, limits the throughput of LLMs, especially for models with large vocabularies. To address this issue, we propose PSC, a novel grammar-constrained decoding method. By combining acceptance conditions of all vocabulary tokens into a single classifier of the parser stack during preprocessing, PSC can compute the complete vocabulary mask by checking the parser stack exactly once per decoding step, with time complexity independent of the vocabulary size. Experiments show that PSC computes masks up to 700$\times$ faster than baselines on complex programming language grammars, and up to 30$\times$ faster for schema-conformant JSON; end-to-end LLM throughput with PSC approaches that of unconstrained decoding. We analyze the preprocessing overhead for preprocessing providers and decoding users, and provide a break-even point analysis to help users decide whether to do preprocessing by themselves.

cs.SE↗