Protecting Floating-Point Computation for DNN Binaries with MBA Obfuscation
This submission was made prematurely and has been withdrawn by the authors for substantial revision before further dissemination.
cs.CR↗
arXiv subjects
Wensheng Tang
Publications and source records attributed to Wensheng Tang.
Push Your Agent: Measuring and Enforcing Quantitative Goal Persistence in Long-Horizon LLM Agents
Long-horizon language agents can make many plausible local tool calls yet fail to persist until a requested count is actually complete. We study this gap as Quantitative Goal Persistence (QGP): whether an agent keeps working until an external verifier confirms enough distinct valid items. PushBench turns this into a benchmark for repository-artifact collection and verifier-backed work units, so repeated work, duplicate submissions, false completion, and progress drift are measured directly rather than hidden behind a final success flag. In matched controller comparisons, a state-tracking retrieval controller reaches 69-78% success while eliminating duplicate submissions, and a backlog-tracking work-unit controller reaches 25-50% success in settings where standard and completion-gated controllers complete no task instances. Black-box frontier-agent evaluations with Claude Code (Sonnet 4.6) and Codex CLI (gpt-5.4) solve many 50-artifact tasks but drop to 3 out of 9 successes per condition at 100 artifacts. The results show that quantitative goals stress a different reliability requirement from local task competence: agents must maintain verified progress and stop only when the requested work is complete.
cs.LG↗
ConCovUp: Effective Agent-Based Test Driver Generation for Concurrency Testing
Concurrency testing is essential to improve the reliability and security of multi-threaded programs. Dynamic analysis tools, such as TSan, depend on high-quality test drivers that reach critical shared-memory interactions at runtime. However, current testing practices predominantly focus on sequential logic, leaving a gap in automated concurrent test generation. Recently, large language models (LLMs) have shown promise in generating sequential tests, but they struggle to produce effective concurrent tests without a deep understanding of concurrency semantics. This paper presents ConCovUp, a multi-agent framework that combines LLMs with program analysis. ConCovUp grounds test generation in static analysis to extract shared memory accesses and their calling contexts. To trigger hard-to-reach accesses, it introduces an LLM-driven backward tracing approach, leveraging the model's semantic reasoning to deduce concrete inputs that satisfy complex path constraints, and iteratively refines the generated tests via dynamic execution feedback. Our evaluation on nine real-world C/C++ libraries shows that ConCovUp improves average Shared Memory Access Pair Coverage (SMAP Coverage) from 36.6% to 68.1% over the general Claude Code agent baseline.
cs.SE↗
Ghost in the Agent: Redefining Information Flow Tracking for LLM Agents
Autonomous Large Language Model (LLM) agents are increasingly deployed to conduct complex tasks by interacting with external tools, APIs, and memory stores. However, processing untrusted external data exposes these agents to severe security threats, such as indirect prompt injection and unauthorized tool execution. Securing these systems requires effective information flow tracking. Yet, traditional taint analysis that is designed for program memory states fundamentally fails when applied to LLMs, where data propagation is governed by probabilistic natural language reasoning. In this paper, we present NeuroTaint, the first comprehensive taint tracking framework tailored for the unique information flow characteristics of LLM agents. Our key insight is that taint propagation in LLM agents must be understood not only as explicit content transfer, but also as semantic transformation, causal influence on decisions, and cross-session persistence through memory. NeuroTaint therefore audits execution traces offline to reconstruct provenance from untrusted sources to privileged sinks using semantic evidence, causal reasoning, and persistent context tracking, rather than relying on exact string matches or pre-defined source-sink paths alone. Extensive evaluation using TaintBench, our 400-scenario benchmark spanning 20 real-world agent frameworks, shows that NeuroTaint substantially outperforms FIDES, an information-flow-control (IFC)-style baseline for LLM agents, in source-sink propagation detection. We further show that NeuroTaint remains effective on established agent-security benchmarks, including InjecAgent and ToolEmu, while operating offline with modest additional auditing cost.
cs.CR↗
Energy-preserving continuous-stage partitioned Runge-Kutta methods
In this paper, we present continuous-stage partitioned Runge-Kutta (csPRK) methods for energy-preserving integration of Hamiltonian systems. A sufficient condition for the energy preservation of the csPRK methods is derived. It is shown that the presented condition contains the existing condition for energy-preserving continuous-stage Runge-Kutta methods as a special case. A noticeable and interesting result is that when we use the simplifying assumptions of order conditions and the normalized shifted Legendre polynomials for constructing high-order energy-preserving csPRK methods, both the Butcher "weight" coefficients $B_τ$ and $\widehat{B}_τ$ must be equal to $1$. As illustrative examples, new energy-preserving integrators are acquired by virtue of the presented condition, and for the sake of verifying our theoretical results, some numerical experiments are reported.
math.NA↗
Energy-preserving continuous-stage Runge-Kutta-Nyström methods
Many practical problems can be described by second-order system $\ddot{q}=-M\nabla U(q)$, in which people give special emphasis to some invariants with explicit physical meaning, such as energy, momentum, angular momentum, etc. However, conventional numerical integrators for such systems will fail to preserve any of these quantities which may lead to qualitatively incorrect numerical solutions. This paper is concerned with the development of energy-preserving continuous-stage Runge-Kutta-Nyström (csRKN) methods for solving second-order systems. Sufficient conditions for csRKN methods to be energy-preserving are presented and it is proved that all the energy-preserving csRKN methods satisfying these sufficient conditions can be essentially induced by energy-preserving continuous-stage partitioned Runge-Kutta methods. Some illustrative examples are given and relevant numerical results are reported.
math.NA↗
Continuous-stage Runge-Kutta methods based on weighted orthogonal polynomials
We develop continuous-stage Runge-Kutta methods based on weighted orthogonal polynomials in this paper. There are two main highlighted merits for developing such methods: Firstly, we do not need to study the tedious solution of multi-variable nonlinear algebraic equations associated with order conditions; Secondly, the well-known weighted interpolatory quadrature theory appeared in every numerical analysis textbook can be directly and conveniently used. By introducing weight function, various orthogonal polynomials can be used in the construction of Runge-Kutta-type methods. It turns out that new families of Runge-Kutta-type methods with special properties (e.g., symplectic, symmetric etc.) can be constructed in batches, and hopefully it may produce new applications in numerical ordinary differential equations.
math.NA↗
Chebyshev symplectic methods based on continuous-stage Runge-Kutta methods
We develop Chebyshev symplectic methods based on Chebyshev orthogonal polynomials of the first and second kind separately in this paper. Such type of symplectic methods can be conveniently constructed with the newly-built theory of weighted continuous-stage Runge-Kutta methods. A few numerical experiments are well performed to verify the efficiency of our new methods.
math.NA↗
Symplectic integration with Jacobi polynomials
In this paper, we study symplectic integration of canonical Hamiltonian systems with Jacobi polynomials. The relevant theoretical results of continuous-stage Runge-Kutta methods are revisited firstly and then symplectic methods with Jacobi polynomials will be established. A few numerical experiments are well performed to verify the efficiency of our new methods.
math.NA↗
An extended framework of continuous-stage Runge-Kutta methods
We propose an extended framework for continuous-stage Runge-Kutta methods which enables us to treat more complicated cases especially for the case weighting on infinite intervals. By doing this, various types of weighted orthogonal polynomials (e.g., Jacobi polynomials, Laguerre polynomials, Hermite polynomials etc.) can be used in the construction of Runge-Kutta-type methods. Particularly, families of Runge-Kutta-type methods with geometric properties can be constructed in this new framework. As examples, some new symplectic integrators by using Legendre polynomials, Laguerre polynomials and Hermite polynomials are constructed.
math.NA↗
Energy-preserving integration of non-canonical Hamiltonian systems by continuous-stage methods
As is well known, energy is generally deemed as one of the most important physical invariants in many conservative problems and hence it is of remarkable interest to consider numerical methods which are able to preserve it. In this paper, we are concerned with the energy-preserving integration of non-canonical Hamiltonian systems by continuous-stage methods. Algebraic conditions in terms of the Butcher coefficients for ensuring the energy preservation, symmetry and quadratic-Casimir preservation respectively are presented. With the presented condition and in use of orthogonal expansion techniques, the construction of energy-preserving integrators is examined. A new class of energy-preserving integrators which is symmetric and of order $2m$ is constructed. Some numerical results are reported to verify our theoretical analysis and show the effectiveness of our new methods.
math.NA↗
Two types of variational integrators and their equivalence
In this paper, we introduce two types of variational integrators, one originating from the discrete Hamilton's principle while the other from Galerkin variational approach. It turns out that these variational integrators are equivalent to each other when they are used for integrating the classical mechanical system with Lagrangian function $L(q,\dot{q})=\frac{1}{2}\dot{q}^TM\dot{q}-U(q)$ ($M$ is an invertible symmetric constant matrix). They are symplectic, symmetric, possess super-convergence order $2s$ (which depends on the degree of the approximation polynomials), and can be related to continuous-stage partitioned Runge-Kutta methods.
math.NA↗
Synthesizing Conjunctive Queries for Code Search
This paper presents Squid, a new conjunctive query synthesis algorithm for searching code with target patterns. Given positive and negative examples along with a natural language description, Squid analyzes the relations derived from the examples by a Datalog-based program analyzer and synthesizes a conjunctive query expressing the search intent. The synthesized query can be further used to search for desired grammatical constructs in the editor. To achieve high efficiency, we prune the huge search space by removing unnecessary relations and enumerating query candidates via refinement. We also introduce two quantitative metrics for query prioritization to select the queries from multiple candidates, yielding desired queries for code search. We have evaluated Squid on over thirty code search tasks. It is shown that Squid successfully synthesizes the conjunctive queries for all the tasks, taking only 2.56 seconds on average.
cs.PL↗
High order symplectic integrators based on continuous-stage Runge-Kutta Nystrom methods
On the basis of the previous work by Tang \& Zhang (Appl. Math. Comput. 323, 2018, p. 204--219), in this paper we present a more effective way to construct high-order symplectic integrators for solving second order Hamiltonian equations. Instead of analyzing order conditions step by step as shown in the previous work, the new technique of this paper is using Legendre expansions to deal with the simplifying assumptions for order conditions. With the new technique, high-order symplectic integrators can be conveniently devised by truncating an orthogonal series.
math.NA↗
Symmetric integrators based on continuous-stage Runge-Kutta-Nystrom methods for reversible systems
In this paper, we study symmetric integrators for solving second-order ordinary differential equations on the basis of the notion of continuous-stage Runge-Kutta-Nystrom methods. The construction of such methods heavily relies on the Legendre expansion technique in conjunction with the symmetric conditions and simplifying assumptions for order conditions. New families of symmetric integrators as illustrative examples are presented. For comparing the numerical behaviors of the presented methods, some numerical experiments are also reported.
math.NA↗
Continuous-stage Runge-Kutta-NystrÖm methods
We develop continuous-stage Runge-Kutta-NystrÖm (csRKN) methods in this paper. By leading weight function into the formalism of csRKN methods and modifying the original pattern of continuous-stage methods, we establish a new and larger framework for csRKN methods and it enables us to derive more effective RKN-type methods. Particularly, a variety of classical weighted orthogonal polynomials can be used in the construction of RKN-type methods. As an important application, new families of symmetric and symplectic integrators can be easily acquired in such framework. Numerical experiments have verified the effectiveness of the new integrators presented in this paper.
math.NA↗
A note on continuous-stage Runge-Kutta methods
We provide a note on continuous-stage Runge-Kutta methods (csRK) for solving initial value problems of first-order ordinary differential equations. Such methods, as an interesting and creative extension of traditional Runge-Kutta (RK) methods, can give us a new perspective on RK discretization and it may enlarge the application of RK approximation theory in modern mathematics and engineering fields. A highlighted advantage of investigation of csRK methods is that we do not need to study the tedious solution of multi-variable nonlinear algebraic equations stemming from order conditions. In this note, we will discuss and promote the recently-developed csRK theory. In particular, we will place emphasis on structure-preserving algorithms including symplectic methods, symmetric methods and energy-preserving methods which play a central role in the field of geometric numerical integration.
math.NA↗
Symplecticity-preserving continuous-stage Runge-Kutta-Nyström methods
We develop continuous-stage Runge-Kutta-Nyström (csRKN) methods for solving second order ordinary differential equations (ODEs) in this paper. The second order ODEs are commonly encountered in various fields and some of them can be reduced to the first order ODEs with the form of separable Hamiltonian systems. The symplecticity-preserving numerical algorithm is of interest for solving such special systems. We present a sufficient condition for a csRKN method to be symplecticity-preserving, and by using Legendre polynomial expansion we show a simple way to construct such symplectic RKN type method.
math.NA↗