arXiv subjects

Sadegh Soudjani

Publications and source records attributed to Sadegh Soudjani.

At least 19 recordsLinked to original sources

arXiv2026

Verified Linear Programming through Tolerance-Aware Precision Boosting

Linear programming plays a fundamental role in computer science, with applications in optimization, formal verification, SMT solving, and numerous other domains. When exactness guarantees are required, numerical inaccuracies arising from floating-point arithmetic can compromise soundness, whereas exact rational arithmetic often incurs significant computational overhead. In this paper, we investigate the numerical stability of the simplex algorithm and establish conditions under which a precision-boosting floating-point implementation provably produces the same pivot decisions and final basis as an exact rational implementation, from which the final result is reconstructed and certified exactly. Our analysis shows that correctness depends on both arithmetic precision and a careful handling of numerical tolerances. Based on these results, we develop a tolerance-aware, precision-boosting simplex algorithm with formal correctness guarantees. Finally, we introduce a delta-complete termination criterion that allows the algorithm to terminate once certified upper and lower bounds on the optimal objective differ by at most a user-specified threshold delta, providing a certified, user-controlled optimality gap.

math.OC↗

arXiv2026

Non-parametric Formal Synthesis of Unknown Stochastic Systems: Asymptotic Convergence Guarantees

Data-driven techniques have shown promising potential for checking behavior of complex systems operating in safety-critical domains against safety and other temporal requirements. This paper studies a class of data-driven techniques that are based on learning a representation of the system from data using non-parametric estimation. The proposed approach is able to formally verify discrete-time stochastic dynamical systems against temporal logic specifications only using observation samples and without the knowledge of the model, and provides a probabilistic guarantee on the satisfaction of the specification. We first consider finite abstract representations of the system in the form of Markov decision processes (MDPs) and derive asymptotic convergence guarantees between the transition probabilities of the abstract MDP and their estimation using Bernstein's inequality and statistical properties of non-parametric estimators. We then propose theoretical results for estimating the asymptotic upper bound of the \emph{Lipschitz constant} (LC) of the stochastic system, which can determine the size of the finite abstract MDP for a given precision error. Under appropriate assumptions, our results prove that the asymptotic convergence rate of the estimations is $O(n^{-1/(3+d)})$ for both the transition probabilities and the LC, where $\mathsf d$ is the dimension of the system and $n$ is the data scale. By integrating these results, we can guarantee the asymptotic closeness in formal verification and policy synthesis performed on the original system and its finite abstraction based on the size of the dataset. Multiple case studies are presented to validate the effectiveness of the proposed method.

eess.SY↗

arXiv2026

Sound Probabilistic Safety Bounds for Large Language Models

We propose a novel framework for computing rigorous bounds on the probability that a large language model (LLM) generates harmful output to a given prompt. We study a new application of the Clopper-Pearson confidence intervals to obtain probably approximately correct (PAC) bounds for this problem. As our main technical contribution, we propose an algorithm that leverages features in the latent space to prioritize exploring branches in the auto-regressive generation tree that are more likely to produce harmful outputs. Our approach in particular enables the efficient computation of useful lower bounds, even in scenarios where the true harm probability is extremely small, and crucially, the obtained lower bounds are sound, i.e., formally proven to be less than the actual harmfulness probability: our experimental results demonstrate the effectiveness of our method by computing non-trivial lower bounds on state-of-the-art LLMs. This study newly enables the evaluation and statistical certification of LLMs.

cs.CL↗

arXiv2026

Context-Triggered Robust MPC for Temporal Logic Specifications

We consider the problem of synthesizing robust feedback controllers for discrete-time linear systems that ensure the satisfaction of context-dependent linear temporal logic specifications in the presence of additive bounded disturbances. Building on existing results that reduce context-triggered temporal logic synthesis to the realization of context-dependent reach-avoid-stay (cRAS) objectives, we focus on the corresponding low-level control synthesis problem. We first employ certificate-based conditions for the almost-sure satisfaction of RAS specifications. Based on these conditions, we propose a switching control architecture that combines robust model predictive control (MPC) with a local invariant controller, and show that the resulting MPC value function serves as a reachability certificate while avoidance is enforced through robust constraints and the stay is enforced via the local controller. To obtain computationally tractable formulations for the resulting robust optimizations, we employ convex duality to reformulate the robust constraints into equivalent deterministic optimization problems, yielding convex quadratic and second-order cone programs for relevant geometric settings. The proposed framework is demonstrated on a robot navigation problem with context-triggered logical switches in both static and moving environments. The results show significantly larger feasible sets than Lyapunov-based approaches, while naturally accommodating dynamic environments and online task reconfiguration.

eess.SY↗

arXiv2026

Kernel-Based Safe Exploration in Deep Reinforcement Learning

Safety has been a major concern when deploying deep reinforcement learning algorithms in the real world. A promising direction that ensures that the learned policy does not visit unsafe regions is to learn a \emph{barrier function} along with the policy. A barrier is a function from states to reals that assigns low values to the initial states, high values to the unsafe states, and decreases in expectation on each transition; such a function can be used to bound the probability of reaching unsafe states. Previous attempts learned a barrier function directly from exploration data, but this required either large amounts of data or restrictions on the system dynamics. In this paper, we show how kernel embeddings can be used to learn barrier functions during deep reinforcement learning for stochastic systems with unknown dynamics. Our algorithm, \emph{kernel-based safe exploration (KBSE)}, learns an optimal policy and a barrier simultaneously during exploration. The barriers are computed iteratively, represented as conditional mean embeddings, and provide better probabilistic safety guarantees with more exploration. The exploration algorithm uses the learned barrier functions to identify safety violations. In the case of violation, it intervenes to modify the unsafe action to a safe action, thereby ensuring that the exploration is restricted to actions that bound the probability of reaching unsafe states. We evaluate KBSE on several complex continuous control benchmarks. Experimental results establish our new algorithm to be suitable for synthesizing control policies that are probabilistically safe without degradation in reward accumulation.

eess.SY↗

arXiv2026

Barrier Certificates for Uncertain Temporal Specifications

This paper studies satisfying temporal logic specifications on stochastic dynamical systems, where the predicates evolve randomly over time. Such randomness may arise from uncertain environment models or external stochastic processes causing the sets associated with predicate satisfaction to vary in a non-deterministic manner. As a result, verifying whether a stochastic dynamical system satisfies a temporal specification depends also on the uncertainty in the predicates. We develop a certificate-based framework to bound the probability of satisfying temporal logic specifications with randomly evolving predicates. We first show that temporal logic specifications with stochastic predicates can be transformed to specifications with deterministic predicates on an augmented space which is extended to include the stochastic space of predicate's uncertainty. We then utilize barrier certificates on an augmented space to provide tractable optimization-based conditions and to avoid the computational burden of dynamic programming. Focusing on linear dynamics and safety-type specifications, we derive analytical conditions under which barrier certificates guarantee bounds on the probability of violating the stochastic safety predicates. The approach is demonstrated on numerical case studies.

math.OC↗

arXiv2026

Safety Certification is Classification

The goal of this paper is certifying safety of dynamical systems subject to uncertainty. Existing approaches use trajectory data to estimate transition probabilities, and compute safety probabilities recursively via dynamic programming (DP). This recursion may lead to compounding errors in the certified safety probability, thus collapsing to a vacuous lower bound for growing horizons $T$. We propose a kernel embedding framework that treats safety certification as a classification problem on trajectory data, directly estimating the $T$-step safety probability without recursion. We show that the framework subsumes well-established approaches from the literature (e.g., barrier certificates, robust Markov models) as special cases, and allows us to go beyond their limitations. As the main consequence, it bypasses compounding error across the horizon and enables certification for systems with non-Markovian dynamics. We demonstrate that direct estimators remain stable independent of the certification horizon and in the non-Markovian setting, whilst DP-based certificates silently go unsound -- confirmed in simulation on a neural-controlled quadrotor.

cs.AI↗

arXiv2026

Almost Sure Reachability in Continuous-time Stochastic Systems

We provide certificates for almost sure reachability of continuous-time stochastic systems governed by stochastic differential equations (SDEs). We first show that a standard Euler-Maruyama discretization may fail to preserve almost sure reachability property of the system using a double-well Langevin system. This observation motivates us to develop certificates for almost sure reachability directly on the continuous-time system. We introduce a pair of certificates, a drift function and a variant function, and prove necessity and sufficiency for almost sure reachability of an open bounded target set. Using these certificates, for linear SDEs, we give a characterization of almost sure reachability in terms of the spectral structure of the system matrices. For polynomial SDEs, we fix a polynomial template for the drift function and choose the variant function template as an exponential function composed with a polynomial. This allows us to translate the conditions in the certificates into sum-of-squares (SOS) constraints. We then propose an alternating scheme to resolve bilinearities. We illustrate the approach on the double-well Langevin example, showing that continuous-time SOS certificates recover almost sure reachability that is lost under time discretization. Moreover, we verify the SOS approach on a polynomial system.

eess.SY↗

arXiv2026

BarrierBench: Evaluating Large Language Models for Safety Verification in Dynamical Systems

Safety verification of dynamical systems via barrier certificates is essential for ensuring correctness in autonomous applications. Synthesizing these certificates involves discovering mathematical functions with current methods suffering from poor scalability, dependence on carefully designed templates, and exhaustive or incremental function-space searches. They also demand substantial manual expertise--selecting templates, solvers, and hyperparameters, and designing sampling strategies--requiring both theoretical and practical knowledge traditionally shared through linguistic reasoning rather than formalized methods. This motivates a key question: can such expert reasoning be captured and operationalized by language models? We address this by introducing an LLM-based agentic framework for barrier certificate synthesis. The framework uses natural language reasoning to propose, refine, and validate candidate certificates, integrating LLM-driven template discovery with SMT-based verification, and supporting barrier-controller co-synthesis to ensure consistency between safety certificates and controllers. To evaluate this capability, we introduce BarrierBench, a benchmark of 100 dynamical systems spanning linear, nonlinear, discrete-time, and continuous-time settings. Our experiments assess not only the effectiveness of LLM-guided barrier synthesis but also the utility of retrieval-augmented generation and agentic coordination strategies in improving its reliability and performance. Across these tasks, the framework achieves more than 90% success in generating valid certificates. By releasing BarrierBench and the accompanying toolchain, we aim to establish a community testbed for advancing the integration of language-based reasoning with formal verification in dynamical systems. The benchmark is publicly available at https://hycodev.com/dataset/barrierbench

cs.AI↗

arXiv2026

Temporal Logic Resilience for Continuous-time Systems

In this paper, we present a novel framework for quantifying a lower bound on resilience in continuous-time (non)linear systems subject to external disturbances while ensuring satisfaction of signal temporal logic specifications. Unlike robustness, which evaluates how well a system satisfies a specification under a given disturbance, resilience measures the maximum disturbance a system can tolerate from a given initial state while maintaining specification satisfaction. We first derive bounds on the perturbed trajectories and then use them to formulate a computational method based on scenario optimization to efficiently compute the maximum admissible disturbance. We validate our approach through case studies, including dc motor, temperature regulation, a nonlinear numerical example, and a vehicle collision avoidance case.

eess.SY↗

arXiv2026

Incremental Data-Driven Policy Synthesis via Game Abstractions

We address the synthesis of control policies for unknown discrete-time stochastic dynamical systems to satisfy temporal logic objectives. We present a data-driven, abstraction-based control framework that integrates online learning with novel incremental game-solving. Under appropriate continuity assumptions, our method abstracts the system dynamics into a finite stochastic (2.5-player) game graph derived from data. Given a requirement over time on this graph, we compute the winning region -- i.e., the set of initial states from which the objective is satisfiable -- in the resulting game, together with a corresponding control policy. Our main contribution is the construction of abstractions, winning regions and control policies \emph{incrementally}, as data about the system dynamics accumulates. Concretely, our algorithm refines under- and over-approximations of reachable sets for each state-action pair as new data samples arrive. These refinements induce structural modifications in the game graph abstraction -- such as the addition or removal of nodes and edges -- which in turn modify the winning region. Crucially, we show that these updates are inherently monotonic: under-approximations only grow, over-approximations only shrink, and the winning region only expands. We exploit this monotonicity by defining an objective-induced ranking function on the nodes of the abstract game that increases monotonically as new data samples are incorporated. These ranks underpin our novel incremental game-solving algorithm, which employs customized gadgets (DAG-like subgames) within a rank-lifting algorithm to efficiently update the winning region. Numerical case studies demonstrate significant computational savings compared to the baseline approach, which re-solves the entire game from scratch whenever new data samples arrive.

cs.GT↗

arXiv2026

Resilient and Effort-Optimal Controller Synthesis under Temporal Logic Specifications

In this paper, we consider the notions of effort and resilience of a dynamical control system defined by the maximum disturbance the system can withstand while satisfying given finite temporal logic specifications. Given a dynamical system and a specification, the objective is to synthesize the controller such that the system satisfies the specification while maximizing its resilience, taking into account input constraints. In addition, we introduce a new metric, called the effort metric, which characterizes the minimal input bound necessary to satisfy a given specification for a perturbed system. The problem for both metrics is formulated as a robust optimization program where the objective is to compute the maximum resilience for the system with input constraints or the minimal effort while simultaneously synthesizing the corresponding controller parameters. Moreover, we study the trade-off between resilience and effort, where we seek to maximize resilience and minimize the control effort. For linear systems and linear controllers, exact solutions are provided for the class of time-varying polytopic specifications for the closed-loop and open-loop systems. For the case of nonlinear systems, nonlinear controllers, and more general specifications, we leverage tools from the scenario optimization approach, offering a probabilistic guarantee of the solution as well as computational feasibility. Different case studies are presented to illustrate the theoretical results.

eess.SY↗

arXiv2026

Average Reward Reinforcement Learning for Omega-Regular and Mean-Payoff Objectives

Recent advances in reinforcement learning (RL) have renewed interest in reward design for shaping agent behavior, but manually crafting reward functions is tedious and error-prone. A principled alternative is to specify behavioral requirements in a formal, unambiguous language and automatically compile them into learning objectives. $ω$-regular languages are a natural fit, given their role in formal verification and synthesis. However, most existing $ω$-regular RL approaches operate in an episodic, discounted setting with periodic resets, which is misaligned with $ω$-regular semantics over infinite traces. For continuing tasks, where the agent interacts with the environment over a single uninterrupted lifetime, the average-reward criterion is more appropriate. We focus on absolute liveness specifications, a subclass of $ω$-regular languages that cannot be violated by any finite prefix and thus aligns naturally with continuing interaction. We present the first model-free RL framework that translates absolute liveness specifications into average-reward objectives and enables learning in unknown communicating Markov decision processes (MDPs) without episodic resetting. We also introduce a reward structure for lexicographic multi-objective optimization: among policies that maximize the satisfaction probability of an absolute liveness specification, the agent maximizes an external average-reward objective. Our method guarantees convergence in unknown communicating MDPs and supports on-the-fly reductions that do not require full environment knowledge, enabling model-free learning. Experiments across several benchmarks show that the continuing, average-reward approach outperforms competing discount-based methods.

cs.AI↗

arXiv2026

Multi-Agent Temporal Logic Planning via Penalty Functions and Block-Coordinate Optimization

Multi-agent planning under Signal Temporal Logic (STL) is often hindered by collaborative tasks that lead to computational challenges due to the inherent high dimensionality of the problem, preventing scalable synthesis with satisfaction guarantees. To address this, we formulate STL planning as an optimization program under multi-agent STL constraints and introduce a penalty-based unconstrained relaxation that can be efficiently solved via a Block-Coordinate Gradient Descent (BCGD) method, where each block corresponds to a single agent's decision variables, thereby mitigating complexity. By utilizing a quadratic penalty function defined via smooth STL semantics, we show that BCGD iterations converge to a stationary point of the penalized problem under standard regularity assumptions. To enforce feasibility, the BCGD solver is embedded within a two-layer optimization scheme: inner BCGD updates are performed for a fixed penalty parameter, which is then increased in an outer loop to progressively improve multi-agent STL robustness. The proposed framework enables scalable computations and is validated through various complex multi-robot planning scenarios.

eess.SY↗

arXiv2026

Stochastic Generalized Dynamic Games with Coupled Chance Constraints

This paper investigates stochastic generalized dynamic games with coupling chance constraints, where agents have incomplete information about uncertainties satisfying a concentration of measure property. This problem, in general, is non-convex and NP-hard. To address this, we propose a convex under-approximation by replacing chance constraints with tightened expected-value constraints, yielding a tractable game. We prove the existence of a stochastic generalized Nash equilibrium (SGNE) in this new game and show that its variational SGNE is an $\boldsymbol{\varepsilon}$-SGNE for the original game, with $\boldsymbol{\varepsilon}$ expressed via the approximation errors and Lagrange multipliers. A semi-decentralized, sampling-based algorithm with time-varying step sizes is developed, requiring no prior knowledge of the uncertainty distribution or expectation evaluations. Unlike existing methods, it avoids step-size tuning based on Lipschitz constants or adaptive rules. Under standard assumptions on the pseudo-gradient, the algorithm converges almost surely to an SGNE.

eess.SY↗

arXiv2026

Kernel-Based Learning of Safety Barriers

The rapid integration of AI algorithms in safety-critical applications such as autonomous driving and healthcare is raising significant concerns about the ability to meet stringent safety standards. Traditional tools for formal safety verification struggle with the black-box nature of AI-driven systems and lack the flexibility needed to scale to the complexity of real-world applications. In this paper, we present a data-driven approach for safety verification and synthesis of black-box systems with discrete-time stochastic dynamics. We employ the concept of control barrier certificates, which can guarantee safety of the system, and learn the certificate directly from a set of system trajectories. We use conditional mean embeddings to embed data from the system into a reproducing kernel Hilbert space (RKHS) and construct an RKHS ambiguity set that can be inflated to robustify the result to out-of-distribution behavior. We provide the theoretical results on how to apply the approach to general classes of temporal logic specifications beyond safety. For the data-driven computation of safety barriers, we leverage a finite Fourier expansion to cast a typically intractable semi-infinite optimization problem as a linear program. The resulting spectral barrier allows us to leverage the fast Fourier transform to generate the relaxed problem efficiently, offering a scalable yet distributionally robust framework for verifying safety. Our work moves beyond restrictive assumptions on system dynamics and uncertainty, as demonstrated on two case studies including a black-box system with a neural network controller.

cs.AI↗

arXiv2026

Policy Synthesis for Interval MDPs via Polyhedral Lyapunov Functions

Decision-making under uncertainty is central to many safety-critical applications, where decisions must be guided by probabilistic modeling formalisms. This paper introduces a novel approach to policy synthesis in multi-objective interval Markov decision processes using polyhedral Lyapunov functions. Unlike previous Lyapunov-based methods that mainly rely on quadratic functions, our method utilizes polyhedral functions to enhance accuracy in managing uncertainties within value iteration of dynamic programming. We reformulate the value iteration algorithm as a switched affine system with interval uncertainties and apply control-theoretic stability principles to synthesize policies that guide the system toward a desired target set. By constructing an invariant set of attraction, we ensure that the synthesized policies provide convergence guarantees while minimizing the impact of transition uncertainty in the underlying model. Our methodology removes the need for computationally intensive Pareto curve computations by directly determining a policy that brings objectives within a specified range of their target values. We validate our approach through numerical case studies, including a recycling robot and an electric vehicle battery, demonstrating its effectiveness in achieving policy synthesis under uncertainty.

eess.SY↗

arXiv2025

LUCID: Learning-Enabled Uncertainty-Aware Certification of Stochastic Dynamical Systems

Ensuring the safety of AI-enabled systems, particularly in high-stakes domains such as autonomous driving and healthcare, has become increasingly critical. Traditional formal verification tools fall short when faced with systems that embed both opaque, black-box AI components and complex stochastic dynamics. To address these challenges, we introduce LUCID (Learning-enabled Uncertainty-aware Certification of stochastIc Dynamical systems), a verification engine for certifying safety of black-box stochastic dynamical systems from a finite dataset of random state transitions. As such, LUCID is the first known tool capable of establishing quantified safety guarantees for such systems. Thanks to its modular architecture and extensive documentation, LUCID is designed for easy extensibility. LUCID employs a data-driven methodology rooted in control barrier certificates, which are learned directly from system transition data, to ensure formal safety guarantees. We use conditional mean embeddings to embed data into a reproducing kernel Hilbert space (RKHS), where an RKHS ambiguity set is constructed that can be inflated to robustify the result to out-of-distribution behavior. A key innovation within LUCID is its use of a finite Fourier kernel expansion to reformulate a semi-infinite non-convex optimization problem into a tractable linear program. The resulting spectral barrier allows us to leverage the fast Fourier transform to generate the relaxed problem efficiently, offering a scalable yet distributionally robust framework for verifying safety. LUCID thus offers a robust and efficient verification framework, able to handle the complexities of modern black-box systems while providing formal guarantees of safety. These unique capabilities are demonstrated on challenging benchmarks.

eess.SY↗