Recently, the fundamental lemma by Willems et. al has been extended towards stochastic LTI systems subject to process disturbances. Using this lemma requires previously recorded data of inputs, outputs, and disturbances. In this paper, we exploit causality concepts of stochastic control to propose a variant of the stochastic fundamental lemma that does not require past disturbance data in the Hankel matrices. Our developments rely on polynomial chaos expansions and on the knowledge of the disturbance distribution. Similar to our previous results, the proposed variant of the fundamental lemma allows to predict future input-output trajectories of stochastic LTI systems. We draw upon a numerical example to illustrate the proposed variant in data-driven control context.
The gradient descent (GD) has been one of the most common optimizer in machine learning. In particular, the loss landscape of a neural network is typically sharpened during the initial phase of training, making the training dynamics hover on the edge of stability. This is beyond our standard understanding of GD convergence in the stable regime where arbitrarily chosen stepsize is sufficiently smaller than the edge of stability. Recently, Wu et al. (COLT2024) have showed that GD converges with arbitrary stepsize under linearly separable logistic regression. Although their analysis hinges on the self-bounding property of the logistic loss, which seems to be a cornerstone to establish a modified descent lemma, our pilot study shows that other loss functions without the self-bounding property can make GD converge with arbitrary stepsize. To further understand what property of a loss function matters in GD, we aim to show arbitrary-stepsize GD convergence for a general loss function based on the framework of \emph{Fenchel--Young losses}. We essentially leverage the classical perceptron argument to derive the convergence rate for achieving $\epsilon$-optimal loss, which is possible for a majority of Fenchel--Young losses. Among typical loss functions, the Tsallis entropy achieves the GD convergence rate $T=\Omega(\epsilon^{-1/2})$, and the R{\'e}nyi entropy achieves the far better rate $T=\Omega(\epsilon^{-1/3})$. We argue that these better rate is possible because of \emph{separation margin} of loss functions, instead of the self-bounding property.
We show how to identify the distributions of the error components in the two-way dyadic model $y_{ij}=c+\alpha_i+\eta_j+\varepsilon_{ij}$. To this end, we extend the lemma of Kotlarski (1967), mimicking the arguments of Evdokimov and White (2012). We allow the characteristic functions of the error components to have real zeros, as long as they do not overlap with zeros of their first derivatives.
This work investigates data-driven prediction and control of Hammerstein-Wiener systems using physics-informed Gaussian process models. Data-driven prediction algorithms have been developed for structured nonlinear systems based on Willems' fundamental lemma. However, existing frameworks cannot treat output nonlinearities and require a dictionary of basis functions for Hammerstein systems. In this work, an implicit predictor structure is considered, leveraging the multi-step-ahead ARX structure for the linear part of the model. This implicit function is learned by Gaussian process regression with kernel functions designed from Gaussian process priors for the nonlinearities. The linear model parameters are estimated as hyperparameters by assuming a stable spline hyperprior. The implicit Gaussian process model provides explicit output prediction by optimizing selected optimality criteria. The model is also applied to receding horizon control with the expected control cost and chance constraint satisfaction guarantee. Numerical results demonstrate that the proposed prediction and control algorithms are superior to black-box Gaussian process models.
In the data-based setting, analysis and control design of dynamical systems using measured data are typically based on overlapping trajectory segments of the input and output variables. This could lead to complex designs because the system internal dynamics, which is typically reflected by the system state variable, is unavailable. In this paper, we will show that the coefficient vector in a modified version of Willems' fundamental lemma is an intrinsic and observable state variable for the system behavior. This argument evolves from the behavioral framework without the requirement of prior knowledge on the causality among system variables or any predefined representation structure (e.g., a state space representation). Such a view allows for the construction of a state map based on the fundamental lemma, bridging the trajectory space and the state space. The state property of the coefficient vector allows for a simple stability design approach using memoryless quadratic functions of it as Lyapunov functions, from which the control action for each step can be explicitly constructed. Using the coefficient vector as a state variable could see wide applications in the analysis and control design of dynamical systems including directions beyond the discussions in this paper.
Recent studies highlight the importance of persistently exciting condition in single signal sequence for model identification and data-driven control methodologies. However, maintaining prolonged excitation in control signals introduces significant challenges, as continuous excitation can reduce the lifetime of mechanical devices. In this paper, we introduce three informativity conditions for various types of multi-signal data, each augmented by weight factors. We explore the interrelations between these conditions and their rank properties in linear time-invariant systems. Furthermore, we introduce open-loop experimental design methods tailored to each of the three conditions, which can synthesize the required excitation conditions either offline or online, even in the presence of limited information within each signal segment. We demonstrate the effectiveness of these informativity conditions in least-squares identification. Additionally, all three conditions can extend Willems' fundamental lemma and are utilized to assess the properties of the system. Illustrative examples confirm that these conditions yield satisfactory outcomes in both least-squares identification and the construction of data-driven controllers.
This work studies linear bandits under a new notion of gap-adjusted misspecification and is an extension of Liu et al. (2023). When the underlying reward function is not linear, existing linear bandits work usually relies on a uniform misspecification parameter $\epsilon$ that measures the sup-norm error of the best linear approximation. This results in an unavoidable linear regret whenever $\epsilon > 0$. We propose a more natural model of misspecification which only requires the approximation error at each input $x$ to be proportional to the suboptimality gap at $x$. It captures the intuition that, for optimization problems, near-optimal regions should matter more and we can tolerate larger approximation errors in suboptimal regions. Quite surprisingly, we show that the classical LinUCB algorithm -- designed for the realizable case -- is automatically robust against such $\rho$-gap-adjusted misspecification with parameter $\rho$ diminishing at $O(1/(d \sqrt{\log T}))$. It achieves a near-optimal $O(\sqrt{T})$ regret for problems that the best-known regret is almost linear in time horizon $T$. We further advance this frontier by presenting a novel phased elimination-based algorithm whose gap-adjusted misspecification parameter $\rho = O(1/\sqrt{d})$ does not scale with $T$. This algorithm attains optimal $O(\sqrt{T})$ regret and is deployment-efficient, requiring only $\log T$ batches of exploration. It also enjoys an adaptive $O(\log T)$ regret when a constant suboptimality gap exists. Technically, our proof relies on a novel self-bounding argument that bounds the part of the regret due to misspecification by the regret itself, and a new inductive lemma that limits the misspecification error within the suboptimality gap for all valid actions in each batch selected by G-optimal design.
We present GliLem -- a novel hybrid lemmatization system for Estonian that enhances the highly accurate rule-based morphological analyzer Vabamorf with an external disambiguation module based on GliNER -- an open vocabulary NER model that is able to match text spans with text labels in natural language. We leverage the flexibility of a pre-trained GliNER model to improve the lemmatization accuracy of Vabamorf by 10% compared to its original disambiguation module and achieve an improvement over the token classification-based baseline. To measure the impact of improvements in lemmatization accuracy on the information retrieval downstream task, we first created an information retrieval dataset for Estonian by automatically translating the DBpedia-Entity dataset from English. We benchmark several token normalization approaches, including lemmatization, on the created dataset using the BM25 algorithm. We observe a substantial improvement in IR metrics when using lemmatization over simplistic stemming. The benefits of improving lemma disambiguation accuracy manifest in small but consistent improvement in the IR recall measure, especially in the setting of high k.
Generalized Schr\"odinger Bridges (GSBs) are a fundamental mathematical framework used to analyze the most likely particle evolution based on the principle of least action including kinetic and potential energy. In parallel to their well-established presence in the theoretical realms of quantum mechanics and optimal transport, this paper focuses on an algorithmic perspective, aiming to enhance practical usage. Our motivated observation is that transportation problems with the optimality structures delineated by GSBs are pervasive across various scientific domains, such as generative modeling in machine learning, mean-field games in stochastic control, and more. Exploring the intrinsic connection between the mathematical modeling of GSBs and the modern algorithmic characterization therefore presents a crucial, yet untapped, avenue. In this paper, we reinterpret GSBs as probabilistic models and demonstrate that, with a delicate mathematical tool known as the nonlinear Feynman-Kac lemma, rich algorithmic concepts, such as likelihoods, variational gaps, and temporal differences, emerge naturally from the optimality structures of GSBs. The resulting computational framework, driven by deep learning and neural networks, operates in a fully continuous state space (i.e., mesh-free) and satisfies distribution constraints, setting it apart from prior numerical solvers relying on spatial discretization or constraint relaxation. We demonstrate the efficacy of our method in generative modeling and mean-field games, highlighting its transformative applications at the intersection of mathematical modeling, stochastic process, control, and machine learning.
Using large language models (LLMs), computers are able to generate a written text in response to a us er request. As this pervasive technology can be applied in numerous contexts, this study analyses the written style of one LLM called GPT by comparing its generated speeches with those of the recent US presidents. To achieve this objective, the State of the Union (SOTU) addresses written by Reagan to Biden are contrasted to those produced by both GPT-3.5 and GPT-4.o versions. Compared to US presidents, GPT tends to overuse the lemma "we" and produce shorter messages with, on average, longer sentences. Moreover, GPT opts for an optimistic tone, opting more often for political (e.g., president, Congress), symbolic (e.g., freedom), and abstract terms (e.g., freedom). Even when imposing an author's style to GPT, the resulting speech remains distinct from addresses written by the target author. Finally, the two GPT versions present distinct characteristics, but both appear overall dissimilar to true presidential messages.
Model Predictive Control (MPC) has demonstrated significant potential in improving energy efficiency in building climate control, outperforming traditional controllers commonly used in modern building management systems. Among MPC variants, data-driven predictive control (DPC) offers the advantage of modeling building dynamics directly from data, thereby substantially reducing commissioning efforts. However, inevitable model uncertainties and measurement noise can result in comfort violations, even with dedicated MPC setups. This work introduces a disturbance-adaptive DPC (DAD-DPC) framework designed to ensure asymptotic satisfaction of predefined violation bounds under sufficient conditions. The framework employs a fully data-driven pipeline based on Willems' Fundamental Lemma and conformal prediction for application in building climate control. The proposed DAD-DPC framework was validated using four building cases, leveraging a high-fidelity simulation platform, BOPTEST, and an actual building testbed, Polydome. For a 5% violation bound, the framework achieved energy consumption increases of 2.82%, 2.24%, 3.75%, and 5.09% while reducing comfort violations by 77.62%, 79.51%, 73.26%, and 39.76%, respectively, compared to other controller configurations. These results demonstrate the framework's ability to effectively balance energy consumption and comfort violations, offering a practical solution for building climate control applications.
This work is concerned with synthesizing safety controllers for discrete-time nonlinear systems beyond polynomials with unknown mathematical models using the notion of k-inductive control barrier certificates (k-CBCs). Conventional CBC conditions (with k=1) for ensuring safety over dynamical systems are often restrictive, as they require the CBCs to be non-increasing at every time step. Inspired by the success of k-induction in software verification, k-CBCs relax this requirement by allowing the barrier function to be non-increasing over k steps, while permitting k-1 (one-step) increases, each up to a threshold epsilon. This relaxation enhances the likelihood of finding feasible k-CBCs while providing safety guarantees across the dynamical systems. Despite showing promise, existing approaches for constructing k-CBCs often rely on precise mathematical knowledge of system dynamics, which is frequently unavailable in practical scenarios. In this work, we address the case where the underlying dynamics are unknown, a common occurrence in real-world applications, and employ the concept of persistency of excitation, grounded in Willems et al.'s fundamental lemma. This result implies that input-output data from a single trajectory can capture the behavior of an unknown system, provided the collected data fulfills a specific rank condition. We employ sum-of-squares (SOS) programming to synthesize the k-CBC as well as the safety controller directly from data while ensuring the safe behavior of the unknown system. The efficacy of our approach is demonstrated through a set of physical benchmarks with unknown dynamics, including a DC motor, an RLC circuit, a nonlinear nonpolynomial car, and a nonlinear polynomial Lorenz attractor.
We derive a general change of variables formula for score functions, showing that for a smooth, invertible transformation $\mathbf{y} = \phi(\mathbf{x})$, the transformed score function $\nabla_{\mathbf{y}} \log q(\mathbf{y})$ can be expressed directly in terms of $\nabla_{\mathbf{x}} \log p(\mathbf{x})$. Using this result, we develop two applications: First, we establish a reverse-time It\^o lemma for score-based diffusion models, allowing the use of $\nabla_{\mathbf{x}} \log p_t(\mathbf{x})$ to reverse an SDE in the transformed space without directly learning $\nabla_{\mathbf{y}} \log q_t(\mathbf{y})$. This approach enables training diffusion models in one space but sampling in another, effectively decoupling the forward and reverse processes. Second, we introduce generalized sliced score matching, extending traditional sliced score matching from linear projections to arbitrary smooth transformations. This provides greater flexibility in high-dimensional density estimation. We demonstrate these theoretical advances through applications to diffusion on the probability simplex and empirically compare our generalized score matching approach against traditional sliced score matching methods.
Incremental input-to-state stability (delta-ISS) offers a robust framework to ensure that small input variations result in proportionally minor deviations in the state of a nonlinear system. This property is essential in practical applications where input precision cannot be guaranteed. However, analyzing delta-ISS demands detailed knowledge of system dynamics to assess the state's incremental response to input changes, posing a challenge in real-world scenarios where mathematical models are unknown. In this work, we develop a data-driven approach to design delta-ISS Lyapunov functions together with their corresponding delta-ISS controllers for continuous-time input-affine nonlinear systems with polynomial dynamics, ensuring the delta-ISS property is achieved without requiring knowledge of the system dynamics. In our data-driven scheme, we collect only two sets of input-state trajectories from sufficiently excited dynamics, as introduced by Willems et al.'s fundamental lemma. By fulfilling a specific rank condition, we design delta-ISS controllers using the collected samples through formulating a sum-of-squares optimization program. The effectiveness of our data-driven approach is evidenced by its application on a physical case study.
Using AI to write formal proofs for mathematical problems is a challenging task that has seen some advancements in recent years. Automated systems such as Lean can verify the correctness of proofs written in formal language, yet writing the proofs in formal language can be challenging for humans and machines. The miniF2F benchmark has 20 IMO problems in its testing set, yet formal proofs are available only for 7 of these problems (3 of which are written only by mathematicians). The model with best accuracy can only prove 4 of these 20 IMO problems, from 1950s and 60s, while its training set is a secret. In this work, we write complete, original formal proofs for the remaining 13 IMO problems in Lean along with 3 extra problems from IMO 2022 and 2023. This effort expands the availability of proof currently in the public domain by creating 5,150 lines of Lean proof. The goal of the paper is to pave the way for developing AI models that can automatically write the formal proofs for all the IMO problems in miniF2F and beyond. In this pursuit, we devise a method to decompose the proof of these problems into their building blocks, constructing a dataset of about 900 lemmas with 25,500 lines of Lean code. These lemmas are not trivial, yet they are approachable, providing the opportunity to evaluate and diagnose the failures and successes of AI models. We then evaluate the ability of GPT-4 in writing formal proofs for these lemmas with zero shot prompting, CoT reasoning and lemma retrieval. In evaluating the responses, we also analyze the confounding factor of LLM's ability to write the proofs in natural language vs Lean language.
Generative AI proposes several large language models (LLMs) to automatically generate a message in response to users' requests. Such scientific breakthroughs promote new writing assistants but with some fears. The main focus of this study is to analyze the written style of one LLM called ChatGPT by comparing its generated messages with those of the recent French presidents. To achieve this, we compare end-of-the-year addresses written by Chirac, Sarkozy, Hollande, and Macron with those automatically produced by ChatGPT. We found that ChatGPT tends to overuse nouns, possessive determiners, and numbers. On the other hand, the generated speeches employ less verbs, pronouns, and adverbs and include, in mean, too standardized sentences. Considering some words, one can observe that ChatGPT tends to overuse "to must" (devoir), "to continue" or the lemma "we" (nous). Moreover, GPT underuses the auxiliary verb "to be" (^etre), or the modal verbs "to will" (vouloir) or "to have to" (falloir). In addition, when a short text is provided as example to ChatGPT, the machine can generate a short message with a style closed to the original wording. Finally, we reveal that ChatGPT style exposes distinct features compared to real presidential speeches.
Recently several large language models (LLMs) have demonstrated their capability to generate a message in response to a user request. Such scientific breakthroughs promote new perspectives but also some fears. The main focus of this study is to analyze the written style of one LLM called ChatGPT 3.5 by comparing its generated messages with those of the recent US presidents. To achieve this objective, we compare the State of the Union addresses written by Reagan to Obama with those automatically produced by ChatGPT. We found that ChatGPT tends to overuse the lemma "we" as well as nouns and commas. On the other hand, the generated speeches employ less verbs and include, in mean, longer sentences. Even when imposing a given style to ChatGPT, the resulting speech remains distinct from messages written by the target author. Moreover, ChatGPT opts for a neutral tone with mainly positive emotional expressions and symbolic terms (e.g., freedom, nation). Finally, we show that the GPT's style exposes distinct features compared to real presidential addresses.
We investigate proximal descent methods, inspired by the minimizing movement scheme introduced by Jordan, Kinderlehrer and Otto, for optimizing entropy-regularized functionals on the Wasserstein space. We establish linear convergence under flat convexity assumptions, thereby relaxing the common reliance on geodesic convexity. Our analysis circumvents the need for discrete-time adaptations of the Evolution Variational Inequality (EVI). Instead, we leverage a uniform logarithmic Sobolev inequality (LSI) and the entropy "sandwich" lemma, extending the analysis from arXiv:2201.10469 and arXiv:2202.01009. The major challenge in the proof via LSI is to show that the relative Fisher information $I(\cdot|\pi)$ is well-defined at every step of the scheme. Since the relative entropy is not Wasserstein differentiable, we prove that along the scheme the iterates belong to a certain class of Sobolev regularity, and hence the relative entropy $\operatorname{KL}(\cdot|\pi)$ has a unique Wasserstein sub-gradient, and that the relative Fisher information is indeed finite.
We give a dimension-independent sparsification result for suprema of centered Gaussian processes: Let $T$ be any (possibly infinite) bounded set of vectors in $\mathbb{R}^n$, and let $\{{\boldsymbol{X}}_t\}_{t\in T}$ be the canonical Gaussian process on $T$. We show that there is an $O_\varepsilon(1)$-size subset $S \subseteq T$ and a set of real values $\{c_s\}_{s \in S}$ such that $\sup_{s \in S} \{{\boldsymbol{X}}_s + c_s\}$ is an $\varepsilon$-approximator of $\sup_{t \in T} {\boldsymbol{X}}_t$. Notably, the size of $S$ is completely independent of both the size of $T$ and of the ambient dimension $n$. We use this to show that every norm is essentially a junta when viewed as a function over Gaussian space: Given any norm $\nu(x)$ on $\mathbb{R}^n$, there is another norm $\psi(x)$ which depends only on the projection of $x$ along $O_\varepsilon(1)$ directions, for which $\psi({\boldsymbol{g}})$ is a multiplicative $(1 \pm \varepsilon)$-approximation of $\nu({\boldsymbol{g}})$ with probability $1-\varepsilon$ for ${\boldsymbol{g}} \sim N(0,I_n)$. We also use our sparsification result for suprema of centered Gaussian processes to give a sparsification lemma for convex sets of bounded geometric width: Any intersection of (possibly infinitely many) halfspaces in $\mathbb{R}^n$ that are at distance $O(1)$ from the origin is $\varepsilon$-close, under $N(0,I_n)$, to an intersection of only $O_\varepsilon(1)$ many halfspaces. We describe applications to agnostic learning and tolerant property testing.
We introduce the new concept of an Arabic Derivational Chain Bank CHAINBANK to leverage the relationship between form and meaning in modeling Arabic derivational morphology. We constructed a knowledge graph network of abstract patterns and their derivational relations and aligned it with the lemmas of the CAMELMORPH morphological analyzer database. This process produced chains of derived words' lemmas linked to their corresponding lemma bases through derivational relations, encompassing 23,333 derivational connections.