Searcharxiv⌕ Search

arXiv subjects

Zixuan Qiu

Publications and source records attributed to Zixuan Qiu.

6 recordsLinked to original sources

Dirac eigenvalue comparison

We study how curvature bounds control the Dirac eigenvalues of closed spin manifolds, comparing the positive and negative eigenvalues separately. Within a fixed conformal class and spin structure, pointwise ordering of scalar curvature relative to a metric of negative scalar curvature implies comparison at every nonzero index. On the two-sphere, a holomorphic ladder on the half-canonical bundle proves the corresponding comparison in both directions under a Gaussian curvature bound. On surfaces of higher genus, the same bundle identities propagate comparison from the scalar and spinor levels to higher canonical powers. In the round conformal class on the higher dimensional sphere, a prescribed scalar curvature bound yields an upper bound in absolute value for the first $2^{[n/2]}$ Dirac eigenvalues. When the conformal class varies, we can show that the corresponding comparisons on hyperbolic surfaces and odd-dimensional spheres no longer hold; while on even-dimensional spheres, the first gap and all sufficiently high-index eigenvalues of either sign satisfy a one-direction comparison.

math.DG↗

Weighted eigenvalues of Dirac operators: complete continuity and comparison

We give a min-max characterization of all of the weighted Dirac eigenvalues, and show that the weighted eigenvalues and eigenspaces of Dirac operators are continuous with respect to weak $L^p$ convergence of the weight, for any $p>n$. Moreover, we establish a comparison result for such weighted eigenvalue problems including the case of nontrivial harmonic spinors.

math.SP↗

Glasseek: A Transparent Recipe for Training and Trustworthy Evaluation of DeepSearch Agents

Deep search enables language agents to solve complex information-seeking tasks through iterative web exploration. However, developing robust open-source agents remains challenging for three reasons. First, opaque trajectory-construction pipelines make training data difficult to audit and reproduce. Second, context compression can create a substantial mismatch between training and inference. Third, agents may retrieve benchmark content from the open web during evaluation, contaminating the results and overstating their performance. To overcome these bottlenecks, we introduce Glasseek, a fully transparent deep search framework spanning the complete development lifecycle, releasing all data, model weights, training recipes, and code. Specifically, for data construction, we establish a verifiable trajectory curation pipeline that systematically generates high-quality exploration paths. For training, we propose Compaction-Managed GSPO (CM-GSPO), which uses the same policy for search and summarization while optimizing only execution sub-trajectories, enabling end-to-end search-policy training across context-compaction boundaries. For evaluation, we conduct the first systematic audit of test-set leakage in open-web agents and release a modular anti-leakage plugin that seamlessly integrates into existing evaluation harnesses. Under this rigorous evaluation protocol, our 9B agent outperforms frontier proprietary agents across almost all deep research benchmarks spanning diverse task types, and achieves the best overall performance among recent open-weight agents, setting a new state of the art at the 9B scale.

cs.AI↗

Continuity of Weighted Dirac Spectra

For the weighted Dirac eigenvalue problem, we show that the two-sided weighted spectrum depends continuously on the weight under continuous deformations within a uniformly elliptic class. Moreover, for differentiable families of weights we obtain a quantitative Lipschitz estimate for the full spectrum in the arsinh--metric, based on a weighted Hellmann--Feynman variational identity.

math.SP↗

Windowed Fourier Propagator: A Frequency-Local Neural Operator for Wave Equations in Inhomogeneous Media

Wave equations are fundamental to describing a vast array of physical phenomena, yet their simulation in inhomogeneous media poses a computational challenge due to the highly oscillatory nature of the solutions. To overcome the high costs of traditional solvers, we propose the Windowed Fourier Propagator (WFP), a novel neural operator that efficiently learns the solution operator. The WFP's design is rooted in the physical principle of frequency locality, where wave energy scatters primarily to adjacent frequencies. By learning a set of compact, localized propagators, each mapping an input frequency to a small window of outputs, our method avoids the complexity of dense interaction models and achieves computational efficiency. Another key feature is the explicit preservation of superposition, which enables remarkable generalization from simple training data (e.g., plane waves) to arbitrary, complex wave states. We demonstrate that the WFP provides an explainable, efficient and accurate framework for data-driven wave modeling in complex media.

cs.LG↗

INTEGRALBENCH: Benchmarking LLMs with Definite Integral Problems

We present INTEGRALBENCH, a focused benchmark designed to evaluate Large Language Model (LLM) performance on definite integral problems. INTEGRALBENCH provides both symbolic and numerical ground truth solutions with manual difficulty annotations. Our evaluation of nine state-of-the-art LLMs reveals significant performance gaps and strong correlations between problem difficulty and model accuracy, establishing baseline metrics for this challenging domain. INTEGRALBENCH aims to advance automated mathematical reasoning by providing a rigorous evaluation framework specifically tailored for definite integral computation.

cs.AI↗