Stein Variational Gradient Descent (SVGD), introduced in [LW16], is a deterministic interactingparticle method for sampling from a target probability measure π e V, only requiring access to V. In the mean-field and continuous-time limit, the distribution of particles converges to a flow (ρt) in the space of probability measures that solves a variant of the Fokker-Planck equation with a velocity field smoothed by weighted convolution with a positive definite kernel [LLN19]. This flow can be interpreted as the gradient flow of the relative entropy H( |π) with respect to a "kernelized" Wasserstein metric [Liu17]. The goal of this paper is to investigate the convergence of (ρt) towards π. To this end, we focus on the model case of Riesz kernels of order s on the d-dimensional torus Td. This is a family of translation-invariant kernels whose Fourier coefficients decay as |ξ| 2s. The parameter s hence directly controls the "smoothing strength" of the interaction; in particular, continuous kernels correspond to s > d/2, C1 kernels to s > (d+1)/2, and C2 kernels to s > (d+2)/2. What is known: qualitative weak convergence The starting point of convergence analyses is the energy dissipation formula [Liu17] d dt H(ρt|π) = Is(ρt|π), (1.1) Authors are listed in alphabetical order.

Mediation analysis (Robins & Greenland 1992, Pearl 2001, Imai, Keele & Tingley 2010, Tchetgen Tchetgen & Shpitser 2012) provides a principled framework for investigating causal mechanisms by decomposing the effect of a treatment A on an outcome Y into pathways operating through a mediator of interest M. Classical mediation analysis focuses on the natural indirect effect, corresponding to the pathway from Ato Y through M, and the natural direct effect, corresponding to pathways not through M. These estimands are well understood when a single mediator is present and strong identification assumptions hold. However, in many applications, there exist multiple intermediate variables between treatment and outcome. In such settings, conventional mediation analysis typically requires the absence of treatment-induced mediator-outcome confounders--often referred to as recanting witnesses--as well as the absence of unmeasured confounding. Under these circumstances, commonly used identification assumptions such as sequential ignorability (Imai, Keele & Yamamoto 2010) or nonparametric structural equation models with independent errors (NPSEM-IE) (Pearl 2009) no longer suffice to identify natural indirect effects (Avin et al. 2005, Tchetgen Tchetgen & VanderWeele 2014). Figure 1 illustrates this issue: the recanting witness D is directly affected by A and simultaneously confounds the relationship between M and Y. Such treatment-induced confounding is common in epidemiologic studies, particularly when the mediator of interest occurs long after the treatment initiation (Robins 1999). A motivating example arises in studies of preterm birth. Mediation analysis has been widely used to explore whether adequate prenatal care (A) reduces the risk of preterm birth (Y) through preeclampsia (M) (Vansteelandt & VanderWeele 2012, VanderWeele et al. 2014, Xia & Chan 2023).

Matrix completion is a fundamental problem in machine learning, where the objective is to recover missing entries of a partially observed matrix. A prominent example is the Netflix Prize (Bennett and Lanning, 2007), which involved predicting a matrix of movie ratings by users for recommendation purposes. Beyond recommendation, matrix completion has recently found applications in causal inference for panel data (Athey et al., 2021). A standard assumption in matrix completion is that the underlying matrix is approximately low-rank, reflecting a few latent factors that govern interactions between rows and columns. A substantial body of work has established theoretical guarantees and developed efficient algorithms for matrix completion (Cai, Cand`es and Shen, 2010; Cand`es and Recht, 2008; Keshavan, Montanari, and Oh, 2010; Mazumder, Hastie and Tibshirani, 2010; Recht, 2011), predominantly focusing on cases where the observed entries are continuous-valued. In many applications, however, observations are not continuous-valued but binary.

Many systems in physics, engineering, and biology exhibit multiscale stochastic dynamics, where low-dimensional slow variables evolve under the influence of high-dimensional fast processes. In practice, observations are often limited to a single trajectory of the slow component, while the fast dynamics remain unobserved, making statistical learning challenging. Approaches based on partial differential equations (PDE), such as Fokker-Planck formulations, aim to characterize the evolution of probability densities, typically requiring dense space-time data or grid-based solvers. In contrast, we adopt a trajectory-based perspective and develop a data-driven framework for learning effective stochastic dynamics from a single observed path. We model the dynamics by coupled multiscale stochastic differential equations (SDEs) and first obtain a principled model reduction through stochastic averaging. Unlike generic model reduction techniques such as PCA, this respects the dynamical structure of the original system and explicitly incorporates the interaction between slow and fast scales. A central challenge, however, is that the reduced model depends on the invariant distribution of the fast process, which is a solution to an intractable and often unknown PDE. We introduce a novel learning framework that parameterizes the invariant distribution using normalizing flows, enabling expressive density modeling in the latent fast-variable space. The flow is trained end-to-end by optimizing a penalized likelihood objective induced by the reduced stochastic dynamics. Furthermore, we develop a Bayesian variational inference procedure for uncertainty quantification, employing a second normalizing flow to approximate the posterior distribution over model parameters. This yields a scalable approach to capturing epistemic uncertainty in multiscale systems.

The power method is one of the most fundamental tools for extracting top principal components from data through low-rank matrix approximation. Yet, when the target rank is large, the cost of matrix multiplication associated with this procedure becomes a major bottleneck. We develop an algorithmic and theoretical framework for accelerating the power method using fast sketching, which is a popular paradigm in randomized linear algebra. Our framework leads to simple and provably efficient methods for singular value decomposition, low-rank factorization, and Nyström approximation, which attain strong numerical performance on benchmark problems. The key novelty in our analysis is the use of regularized spectral approximation, a property of fast sketching methods which proves more flexible in generalizing power method guarantees than traditional arguments.

Providing non-conservative uncertainty quantification for function estimates derived from noisy observations remains a fundamental challenge in statistical machine learning, particularly for applications in safety-critical domains. In this work, we propose novel non-asymptotic probabilistic uniform error bounds for kernel-based regression. Compared to related bounds in the literature that are restricted to (conditionally) independent sub-Gaussian noise, our bounds allow to consider a broad class of non-Gaussian distributions, such as sub-Gaussian, bounded, sub-exponential, and variance/moment-bounded noise. Moreover, our results apply to correlated and uncorrelated noise. We compare our proposed error bounds with existing results in terms of the induced uncertainty region and their performance in safe control, demonstrating the tightness of the proposed bounds.

Multicalibration requires predicted scores to agree with label probabilities across rich families of subgroups and score-dependent tests, but existing methods require clean input-label pairs for evaluation and post-processing. This assumption fails in weakly supervised learning (WSL) regimes -- including positive-unlabeled, unlabeled-unlabeled, and positive-confidence learning -- where clean labels are costly or unavailable even though reliable uncertainty estimates may be crucial. We address this gap by developing estimators of multicalibration error and post-hoc correction methods for WSL settings in which clean input-label pairs are unavailable. We propose a unified framework for estimating and correcting multicalibration under weak supervision by combining contamination-matrix risk rewrites with witness-based calibration constraints, yielding corrected multicalibration moments with finite-sample guarantees. We further propose weak-label multicalibration boost (WLMC), a generic post-hoc recalibration algorithm under weak supervision. Finally, we conduct experiments across multiple weak-supervision settings to evaluate multicalibration behavior and offer empirical insight into uncertainty estimation under weak supervision.

Training a language model on data scattered across bandwidth-limited nodes that cannot be centralized is a setting that arises in clinical networks, enterprise knowledge bases, and scientific consortia. We study the regime in which data must remain distributed across nodes, and ask what statistical guarantees are in principle achievable under explicit bandwidth budgets; we aim to characterize what is provably possible, not to demonstrate a deployment-ready system. Existing theory treats either training-time consistency or inference-time calibration in isolation, and none makes bandwidth a first-class statistical parameter. We analyze two protocols, Federated Probe-Logit Distillation (FPLD) for training and Federated Conformal RAG (FC-RAG) for inference, as the analytical vehicles for our results. Our first main result is an explicit high-probability KL-consistency rate for FPLD with simultaneous dependence on node count $K$, per-node sample size $n$, quantization budget $B$, probe-set size $m$, and vocabulary size $V$; bandwidth enters only through an exponentially vanishing quantization term. Our second main result is a distribution-free marginal-coverage bound for FC-RAG, whose novel retrieval-bandwidth slack $Δ_{\mathrm{RAG}} = f_{\max}\sqrt{K^{-2}\sum_i v(B_i)}$ makes per-node retrieval bandwidth a first-class statistical parameter, with arithmetic aggregation across $K$ nodes shrinking the slack as $K^{-1/2}$ in the per-node-uniform regime. A Pinsker-type corollary composes the two bounds into an end-to-end coverage guarantee. Synthetic experiments verify the predicted scaling along the bounds' parameters; small-scale experiments on a GPT-2 testbed illustrate that the qualitative bandwidth-accuracy tradeoff survives on a real language model. A deployment-scale empirical evaluation is out of scope.

In this paper, we study the problem of sampling from a distribution under the constraint of differential privacy (DP). Prior works measure the utility of DP sampling with density ratio-based measures such as KL divergence. However, such formulations suffer from two key limitations: 1) they fail to capture the geometric structure of the support, and 2) they are not applicable when the supports of the distributions differ. To deal with these issues, we develop a novel framework for DP sampling with Wasserstein distance as the utility measure. In this formulation, we propose Wasserstein Projection Mechanism (WPM), a minimax optimal mechanism based on Wasserstein projection. Furthermore, we develop efficient algorithms for computing the proposed mechanisms approximately and provide convergence guarantees.

Generative models trained on finite data face a fundamental tension: their score-matching or next-token objective converges to the empirical training distribution rather than the population distribution we seek to learn. Using rule-valid synthetic tasks, we trace this tension across two training timescales: $τ_{\mathrm{rule}}$, the step at which generations first become rule-valid, and $τ_{\mathrm{mem}}$, the step at which models begin reproducing training samples. Focusing on parity and extending to other binary rules and combinatorial puzzles, we characterize how these two clocks, $τ_{\mathrm{rule}}$ and $τ_{\mathrm{mem}}$, depend on key aspects of the learning setup. Specifically, we show that $τ_{\mathrm{rule}}$ increases with rule complexity and decreases with model capacity, while $τ_{\mathrm{mem}}$ is approximately invariant to the rule and scales nearly linearly with dataset size $N$. We define the \emph{innovation window} as the interval $[τ_{\mathrm{rule}}, τ_{\mathrm{mem}}]$. This window widens with increasing $N$ and narrows with rule complexity, and may vanish entirely when $τ_{\mathrm{rule}} \geq τ_{\mathrm{mem}}$. The same two-clock structure arises in both diffusion (DiT) and autoregressive (GPT) models, with architecture-dependent offsets. Dissecting the learned score of DiT models reveals a corresponding evolution of the optimization landscapes, where rule-valid samples' basins expand substantially around $τ_{\mathrm{rule}}$, while training samples' basins begin to dominate around $τ_{\mathrm{mem}}$. Together, these results yield a unified and predictive account of when and how generative models exhibit genuine innovation.