Xiaofan Li

Ting Gao, Jinqiao Duan, Xiaofan Li et al.

The mean first exit time and escape probability are utilized to quantify dynamical behaviors of stochastic differential equations with non-Gaussian alpha-stable type Levy motions. Both deterministic quantities are characterized by differential-integral equations(i.e.,differential equations with non local terms) but with different exterior conditions. The non-Gaussianity of noises manifests as nonlocality at the level of mean exit time and escape probability. An objective of this paper is to make mean exit time and escape probability as efficient computational tools, to the applied probability community, for quantifying stochastic dynamics. An accurate numerical scheme is developed and validated for computing the mean exit time and escape probability. Asymptotic solution for the mean exit time is given when the pure jump measure in the Levy motion is small. From both the analytical and numerical results, it is observed that the mean exit time depends strongly on the domain size and the value of alpha in the alpha-stable Levy jump measure. The mean exit time can measure which of the two competing factors in alpha-stable Levy motion, i.e. the jump frequency or the jump size, is dominant in helping a process exit a bounded domain. The escape probability is shown to vary with the underlying vector field(i.e.,drift). The mean exit time and escape probability could become discontinuous at the boundary of the domain, when the process is subject to certain deterministic potential and the value of alpha is in (0,1).

Bruce Yang, Xinfeng He, Huan Gao et al.

Effective prompt design is essential for improving the planning capabilities of large language model (LLM)-driven agents. However, existing structured prompting strategies are typically limited to single-agent, plan-only settings, and often evaluate performance solely based on task accuracy - overlooking critical factors such as token efficiency, modularity, and scalability in multi-agent environments. To address these limitations, we introduce CodeAgents, a prompting framework that codifies multi-agent reasoning and enables structured, token-efficient planning in multi-agent systems. In CodeAgents, all components of agent interaction - Task, Plan, Feedback, system roles, and external tool invocations - are codified into modular pseudocode enriched with control structures (e.g., loops, conditionals), boolean logic, and typed variables. This design transforms loosely connected agent plans into cohesive, interpretable, and verifiable multi-agent reasoning programs. We evaluate the proposed framework across three diverse benchmarks - GAIA, HotpotQA, and VirtualHome - using a range of representative LLMs. Results show consistent improvements in planning performance, with absolute gains of 3-36 percentage points over natural language prompting baselines. On VirtualHome, our method achieves a new state-of-the-art success rate of 56%. In addition, our approach reduces input and output token usage by 55-87% and 41-70%, respectively, underscoring the importance of token-aware evaluation metrics in the development of scalable multi-agent LLM systems. The code and resources are available at: https://anonymous.4open.science/r/CodifyingAgent-5A86

Xiaofan Li, Xing Gao

In recent years, various software supply chain (SSC) attacks have posed significant risks to the global community. Severe consequences may arise if developers integrate insecure code snippets that are vulnerable to SSC attacks into their products. Particularly, code generation techniques, such as large language models (LLMs), have been widely utilized in the developer community. However, LLMs are known to suffer from inherent issues when generating code, including fabrication, misinformation, and reliance on outdated training data, all of which can result in serious software supply chain threats. In this paper, we investigate the security threats to the SSC that arise from these inherent issues. We examine three categories of threats, including eleven potential SSC-related threats, related to external components in source code, and continuous integration configuration files. We find some threats in LLM-generated code could enable attackers to hijack software and workflows, while some others might cause potential hidden threats that compromise the security of the software over time. To understand these security impacts and severity, we design a tool, SSCGuard, to generate 439,138 prompts based on SSC-related questions collected online, and analyze the responses of four popular LLMs from GPT and Llama. Our results show that all identified SSC-related threats persistently exist. To mitigate these risks, we propose a novel prompt-based defense mechanism, namely Chain-of-Confirmation, to reduce fabrication, and a middleware-based defense that informs users of various SSC threats.