arXiv subjects

Jiajia Huang

Publications and source records attributed to Jiajia Huang.

9 recordsLinked to original sources

arXiv2026

Van der Waals interactions in supercritical water under Earth's mantle conditions

The properties of water under high pressure and high temperature (HP-HT) are critical in multiple geochemical processes in deep Earth. Ab initio molecular dynamics (AIMD) is a promising approach to study water under extreme conditions without any empirical parameters. However, the accuracy of AIMD simulations is determined by the exchange-correlation (XC) functional including the dispersion correction used in density functional theory calculations. While van der Waals (vdW) interactions are well known to be critically important for water under ambient conditions, the influence of the dispersion correction on the properties of HP-HT water as found in Earth's mantle remains largely unexplored. To address this, we carried out AIMD simulations for supercritical water at 1, 5, and 10 GPa, and 1000 K. We compared PBE, PBE-D3, RPBE-D3, and SCAN functionals, where D3 means Grimme's D3 dispersion correction. We compared the structural, diffusion, and vibrational properties of water as computed with these XC functionals. Overall, the discrepancies between the functionals are reduced under extreme P-T conditions relative to ambient conditions. PBE and PBE-D3 exhibit higher proton-transfer rates than RPBE-D3 and SCAN, suggesting that while vdW interactions do not significantly affect the water structure under extreme conditions, the oxygen-hydrogen bond strength does influence proton transfer. Our results provide molecular-level insight into water in Earth's mantle and offer valuable guidance for selecting appropriate XC functionals in AIMD simulations of aqueous solutions under extreme conditions.

physics.comp-ph↗

arXiv2026

TaxPraBen: A Scalable Benchmark for Structured Evaluation of LLMs in Chinese Real-World Tax Practice

While Large Language Models (LLMs) excel in various general domains, they exhibit notable gaps in the highly specialized, knowledge-intensive, and legally regulated Chinese tax domain. Consequently, while tax-related benchmarks are gaining attention, many focus on isolated NLP tasks, neglecting real-world practical capabilities. To address this issue, we introduce TaxPraBen, the first dedicated benchmark for Chinese taxation practice. It combines 10 traditional application tasks, along with 3 pioneering real-world scenarios: tax risk prevention, tax inspection analysis, and tax strategy planning, sourced from 14 datasets totaling 7.3K instances. TaxPraBen features a scalable structured evaluation paradigm designed through process of "structured parsing-field alignment extraction-numerical and textual matching", enabling end-to-end tax practice assessment while being extensible to other domains. We evaluate 19 LLMs based on Bloom's taxonomy. The results indicate significant performance disparities: all closed-source large-parameter LLMs excel, and Chinese LLMs like Qwen2.5 generally exceed multilingual LLMs, while the YaYi2 LLM, fine-tuned with some tax data, shows only limited improvement. TaxPraBen serves as a vital resource for advancing evaluations of LLMs in practical applications.

cs.CL↗

arXiv2025

From Scores to Skills: A Cognitive Diagnosis Framework for Evaluating Financial Large Language Models

Large Language Models (LLMs) have shown promise for financial applications, yet their suitability for this high-stakes domain remains largely unproven due to inadequacies in existing benchmarks. Existing benchmarks solely rely on score-level evaluation, summarizing performance with a single score that obscures the nuanced understanding of what models truly know and their precise limitations. They also rely on datasets that cover only a narrow subset of financial concepts, while overlooking other essentials for real-world applications. To address these gaps, we introduce FinCDM, the first cognitive diagnosis evaluation framework tailored for financial LLMs, enabling the evaluation of LLMs at the knowledge-skill level, identifying what financial skills and knowledge they have or lack based on their response patterns across skill-tagged tasks, rather than a single aggregated number. We construct CPA-KQA, the first cognitively informed financial evaluation dataset derived from the Certified Public Accountant (CPA) examination, with comprehensive coverage of real-world accounting and financial skills. It is rigorously annotated by domain experts, who author, validate, and annotate questions with high inter-annotator agreement and fine-grained knowledge labels. Our extensive experiments on 30 proprietary, open-source, and domain-specific LLMs show that FinCDM reveals hidden knowledge gaps, identifies under-tested areas such as tax and regulatory reasoning overlooked by traditional benchmarks, and uncovers behavioral clusters among models. FinCDM introduces a new paradigm for financial LLM evaluation by enabling interpretable, skill-aware diagnosis that supports more trustworthy and targeted model development, and all datasets and evaluation scripts will be publicly released to support further research.

cs.CE↗

arXiv2025

Comment on "Mineral-water reactions in Earth's mantle: Predictions from Born theory and ab initio molecular dynamics" by Fowler et al. 2024 (Geochim. Cosmochim. Acta 372, 111-123)

This comment addresses discrepancies in dielectric constant calculations of water under extreme conditions (~10 GPa and 1000 K) between Fowler et al.'s recent study [Geochim. Cosmochim. Acta 372, 111-123 (2024)] and the earlier work by Pan et al. [Proc. Natl. Acad. Sci. 110, 6646-6650 (2013)]. Through reproduced ab initio molecular dynamics (AIMD) simulations using the CP2K code with extended duration and identical system size, we rigorously validate that Pan et al.'s original results (39.4) are well-converged, contrasting with Fowler et al.'s reported value of 51. The observed discrepancy cannot be attributed to simulation duration limitations, but rather to methodological differences in dipole moment calculation. Our analysis highlights critical issues in the treatment of dipole moment fluctuations in periodic systems within the framework of modern theory of polarization. This clarification has significant implications for modeling mineral-water interactions in Earth's mantle using Born theory.

physics.geo-ph↗

arXiv2024

AuditWen:An Open-Source Large Language Model for Audit

Intelligent auditing represents a crucial advancement in modern audit practices, enhancing both the quality and efficiency of audits within the realm of artificial intelligence. With the rise of large language model (LLM), there is enormous potential for intelligent models to contribute to audit domain. However, general LLMs applied in audit domain face the challenges of lacking specialized knowledge and the presence of data biases. To overcome these challenges, this study introduces AuditWen, an open-source audit LLM by fine-tuning Qwen with constructing instruction data from audit domain. We first outline the application scenarios for LLMs in the audit and extract requirements that shape the development of LLMs tailored for audit purposes. We then propose an audit LLM, called AuditWen, by fine-tuning Qwen with constructing 28k instruction dataset from 15 audit tasks and 3 layers. In evaluation stage, we proposed a benchmark with 3k instructions that covers a set of critical audit tasks derived from the application scenarios. With the benchmark, we compare AuditWen with other existing LLMs from information extraction, question answering and document generation. The experimental results demonstrate superior performance of AuditWen both in question understanding and answer generation, making it an immediately valuable tool for audit.

cs.CL↗

arXiv2024

FinBen: A Holistic Financial Benchmark for Large Language Models

LLMs have transformed NLP and shown promise in various fields, yet their potential in finance is underexplored due to a lack of comprehensive evaluation benchmarks, the rapid development of LLMs, and the complexity of financial tasks. In this paper, we introduce FinBen, the first extensive open-source evaluation benchmark, including 36 datasets spanning 24 financial tasks, covering seven critical aspects: information extraction (IE), textual analysis, question answering (QA), text generation, risk management, forecasting, and decision-making. FinBen offers several key innovations: a broader range of tasks and datasets, the first evaluation of stock trading, novel agent and Retrieval-Augmented Generation (RAG) evaluation, and three novel open-source evaluation datasets for text summarization, question answering, and stock trading. Our evaluation of 15 representative LLMs, including GPT-4, ChatGPT, and the latest Gemini, reveals several key findings: While LLMs excel in IE and textual analysis, they struggle with advanced reasoning and complex tasks like text generation and forecasting. GPT-4 excels in IE and stock trading, while Gemini is better at text generation and forecasting. Instruction-tuned LLMs improve textual analysis but offer limited benefits for complex tasks such as QA. FinBen has been used to host the first financial LLMs shared task at the FinNLP-AgentScen workshop during IJCAI-2024, attracting 12 teams. Their novel solutions outperformed GPT-4, showcasing FinBen's potential to drive innovation in financial LLMs. All datasets, results, and codes are released for the research community: https://github.com/The-FinAI/PIXIU.

cs.CL↗

arXiv2024

Super-concentrated alkali hydroxide electrolytes for rechargeable Zn batteries

Rechargeable Zn batteries offer safe, inexpensive energy storage, but when deeply discharged to compete with lithium-ion batteries, they are plagued by parasitic reactions at the Zn anodes. We apply super-concentrated alkaline electrolytes to suppress two key parasitic reactions, hydrogen evolution and ZnO passivation. An electrolyte with 15 M KOH displays a broad electrochemical window (>2.5 V on Au), a high ZnO solubility (>1.5 M), and an exceptionally high ionic conductivity (>0.27 S/cm at 25 C). Spectroscopies and ab-initio molecular dynamics simulation suggest K+-OH- pairs and a tightened water network to underpin the stability. The simulation further reveals unique triggered proton hopping that offsets the lack of water wires to sustain the conductivity. Low hydrogen evolution, confirmed via online mass spectroscopy, and slow passivation enable a NiOOH||Zn battery to deliver a cumulative capacity of 8.4 Ah cm-2 and a Zn-air battery to last for over 110 hours.

cond-mat.mtrl-sci↗

arXiv2024

Superconducting Klein and anti-Klein tunneling in Weyl junctions

Klein tunneling is an old topic in relativistic quantum physics, and has been observed recently in graphene where massless particles reside. Here, we propose a new heterostructure platform for Klein tunneling to occur, which consists of a Weyl-semimetal-based normal state/superconductor (NS) junction. By developing a Blonder-Tinkham-Klapwijk-like theory, we find that Klein tunneling occurs at normal incidence, which can lead to differential conductance doubling. If the (single) Weyl semimeltals are replaced by double Weyl semimetals, anti-Klein tunneling will take place of Klein tunneling. Our work provides a theoretical guide for the detection of (anti-)Klein tunneling in three-dimensional chiral NS junctions.

cond-mat.mes-hall↗

arXiv2022

A semiclassical approach to surface Fermi arcs in Weyl semimetals

We present a semiclassical explanation for the morphology of the surface Fermi arcs of Weyl semimetals. Viewing the surface states as a two-dimensional Fermi gas subject to band bending and Berry curvatures, we show that it is the non-parallelism between the velocity and the momentum that gives rise to the spiral structure of Fermi arcs. We map out the Fermi arcs from the velocity field for a single Weyl point and a lattice with two Weyl points. We also investigate the surface magnetoplasma of Dirac semimetals in a magnetic field, and find that the drift motion, the chiral magnetic effect and the Imbert-Fedorov shift are all involved in the formation of surface Fermi arcs. Our work not only provides an insightful perspective on the surface Fermi arcs and a practical way to find the surface dispersion, but also paves the way for the study of other physical properties of the surface states of topological semimetals, such as transport properties and orbital magnetization, using semiclassical methods.

cond-mat.mes-hall↗