Papers

Selective inference (SI) provides statistically valid $p$-values for hypotheses selected by applying an algorithm to the data, correcting for the bias that arises when the same data are used both to select and to test a hypothesis. Developing an SI procedure for a new algorithm, however, has required an expert to derive, and then implement, the selection event, i.e., the conditions under which the hypothesis is selected. Repeating this specialized effort for every new algorithm is why exact SI has so far been available for only a narrow class. We propose AutoSI, a framework that removes this barrier in two ways. First, AutoSI constructs the selection event automatically from the algorithm's individual operations, so the user only writes the algorithm as ordinary NumPy-like code and derives nothing by hand. Second, AutoSI broadens the class of selection events SI can handle: existing exact methods are limited to selection events characterized by linear or quadratic inequalities in the data, whereas AutoSI covers any algorithm expressible through rational functions of the data (ratios of polynomials). We prove that the $p$-values computed by AutoSI are exactly valid in finite samples. We demonstrate AutoSI on three feature-selection methods, each written in a few dozen lines of code. One of these methods, the lasso with its tuning parameter selected by cross-validated $R^2$, cannot be handled within existing exact SI frameworks and is made possible by AutoSI. Experiments on synthetic and real datasets show that the resulting $p$-values control the type I error rate (i.e., the false positive rate) at the nominal level while retaining high power.

Estimating the dynamic effects of economic shocks in short and very short samples is impeded by a lack of degrees of freedom. We offer a solution based on a Bayesian hierarchical framework for estimating local projection (LP) impulse response functions across a panel of related time series. The framework explicitly accommodates unbalanced panels in which some series are substantially shorter than others, allowing the short series to borrow information from longer ones at horizons where the short series carry little or no own data. Since series might exhibit heterogeneous dynamics, we develop a sparse finite mixture pool that clusters units by similarity of their impulse response profiles. We show in simulations that our approach substantially improves LP estimation accuracy relative to the standard approach if the time series are short while producing similar LPs for longer time series. Using a US price dataset, augmented with survey responses, we find that supply-chain and oil shocks trigger heterogeneous reactions of different price measures, with headline price indices responding more sharply than their core counterparts and goods prices changing more than services prices.

Formula One outcomes reflect the joint contributions of drivers and constructors, but these contributions are unobserved and vary over time. We propose a Bayesian state-space model that disentangles dynamic driver and constructor abilities using two observed outcomes: fastest qualifying lap times and race rankings. Both outcomes depend jointly on latent driver and constructor states that evolve at the Grand Prix level, while the race equation additionally accounts for starting-grid position. The decomposition is supported by constraints that center the driver and constructor abilities at zero, together with variation in driver-constructor assignments over time. Bayesian inference is performed using the No-U-Turn sampler under weakly informative priors that treat driver and constructor abilities symmetrically. Applying the model to the Formula One hybrid era from 2014 to 2021, we find substantial heterogeneity in both driver and constructor abilities. Driver abilities are generally more stable over time, whereas constructor abilities exhibit greater variation and, for many driver–constructor combinations, contribute more strongly to observed performance.

An information theory-based framework is proposed in attempt to explain insistence on sameness in autism as an instance of a general behavior pattern in which an individual tries to reduce surprise and uncertainty. It offers a new definition of autism as an impairment in which cognitive functions are restricted to discrimination, memorization and prediction of tangible properties of the environment. An analogy between insistence on sameness and constrained minimization of the entropy metric is observed and examined for a set of assumptions that describe cognitive limitations of a person with autism. The metric is given by the formula $D_H(R, M) = H(R|M) + H(M|R)$, where $R$ represents sequences of random stimuli, $M$ is a memory that stores and retrieves them, and where $H(.|.)$ denotes their conditional entropies interpreted as surprise and uncertainty, respectively. It is first inferred that to minimize the metric an individual can learn about $R$ (and store that knowledge in $M$) or can restrict $R$ to the already known $M$. Then, it is concluded that insistence on sameness is a manifestation of the latter. Moreover, it is shown that the proposed framework: (1) Helps to quantify the concepts of surprise, uncertainty, sensory overload and deprivation, anxiety, comfort zone, disappointment, disorientation, pedantry, rigidness, observance or aberrant precision. (2) Leads to a list of guidelines for learning therapies and daily care routines, and allows them to be defined as optimization algorithms and implemented as programs for robotic live-in caregivers. (3) Can be validated with the help of a Turing test-like approach that requires no experiments involving individuals with autism. The framework-if positively validated-will provide formal foundations and design guidelines for therapies aimed at improving self-reliance of individuals with autism in basic activities of daily living.

In this paper, we propose a novel approach for estimating the minimum of a smooth function and its location from observations corresponding to a multivariate regression function depending a priori on d variables but actually only on r < d active variables and corrupted by some additional noise. Our method consists of two steps: The rst one is a variable selection approach which is used for identifying the r active variables on which f depends and the second one consists in estimating the minimum of the function and its location. The estimation of the minimizers is obtained by using a projected gradient descent where the gradient is estimated using a local polynomial approximation of the regression function limited to its active variables obtained in the rst step. The estimation of the minimum is obtained by evaluating the estimator of the regression function using a local polynomial approach at the estimator of one of the minimizers previously obtained. We establish non asymptotic upper bounds for the quadratic risk of the estimators of the minimizers and of the minimum and prove that they reach the optimal rate that could be expected as if the active variables were known beforehand up to a factor smaller than a power of a logarithmic term.

The mixture of experts (MoE) model provides a flexible approach for modeling heterogeneous regression relationships by allowing covariate-dependent mixing through a gating network, but most existing MoE models rely on parametric assumptions for expert error distributions, typically Gaussian, which can lead to inefficiency and sensitivity to outliers or heavy-tailed behavior when misspecified. We propose a semiparametric MoE model in which each expert error distribution is represented as a nonparametric Gaussian scale mixture estimated via nonparametric maximum likelihood, relaxing parametric assumptions within the Gaussian scale-mixture class while preserving the interpretability and structure of the MoE framework. The resulting model adapts to complex error structures, improves robustness under contamination and heavy tails, and remains competitive under well-specified Gaussian settings, providing a practical and theoretically grounded alternative to parametric MoE formulations.

Functional flow matching is posed on distributions of functions but implemented from finitely many coefficients or point values. Under scattered or adaptive refinement, the resulting conditioning sigma-algebras need not be nested, so martingale convergence does not justify the sensor limit. We prove strong $L^2$ convergence of finite conditional velocity targets for every strongly consistent sequence of finite-rank reconstructions, with quantitative bounds for orthogonal projections and a point-sensor extension through a regularity space. For learned flows, coupling directly to a population superposition path yields an end-to-end Wasserstein bound without assuming uniqueness of the population finite-dimensional ODE. We verify sensor-independent constants for a normalized quadrature neural operator, including globally Lipschitz activations through an explicit magnitude recurrence. A noncommuting trace-class Gaussian example gives boundary multiplier $0$ under projected restriction and $0.72$ under exact conditioning. A spatial regularity–cubature certificate closes the operator-realization term, a Bernstein argument gives a $\widetilde{O}(n^{-1})$ excess-risk term for fixed model dimension and envelopes, and an exactly realizable clipped Gaussian scaling specialization yields an explicit end-to-end rate.

On two-sided content platforms, symmetric two-sided isolation (assigning matched fractions of creators and viewers to isolated treatment and control submarkets) is widely used for creator-side and cold-start experiments because it removes cross-arm marketplace interference. Isolation, however, thins each viewer's candidate catalog, and intuition suggests the resulting engagement cost should fade as the platform grows: a small fraction of a vast catalog is still vast. We show that, in an order-statistics model of engagement, whether this intuition holds depends on the upper tail of match quality. Extreme-value theory yields tail-class loss laws with a sharp dichotomy: for light or bounded tails the loss vanishes as the candidate pool grows, whereas under heavy tails it converges to a size-independent constant, so expanding the candidate pool, even by orders of magnitude, does not asymptotically eliminate the cost. Evidence from two production experiments on a platform with millions of active creators is consistent with this picture: a pure A/A traffic sweep reveals a measurable, depth-graded engagement cost; a one-sided catalog ablation independently shows that per-viewer thinning contributes to the loss; and a tail index calibrated on the small exploration pool predicts an effect consistent with the one observed in the far larger full-catalog ablation. Isolation thus carries a price that experimenters should budget for, like any other cost. We give practitioners a preflight procedure that estimates it before launch, sizes traffic accordingly, and recommends a fallback design when the predicted cost exceeds a chosen tolerance.

Filtering combines model predictions with measurements to estimate the probability density function (PDF) of a system state over time. The PDF often becomes highly asymmetric and even multimodal in nonlinear systems with oscillatory or chaotic dynamics. Such non-Gaussian features violate the single-Gaussian assumption underlying Kalman-type filters. To address this problem, Gaussian mixture filtering has been proposed. However, accurately propagating mixture components and adaptively adjusting their number and weights over time remain open challenges. Here, we develop an adaptive split-combine Gaussian mixture filter (AMF) that estimates the time evolution of asymmetric and multimodal PDFs by adaptively splitting and combining Gaussian particles without auxiliary online numerical optimization. Notably, the proposed splitting method guarantees a reduction in variance along a target level-set-point direction of a Gaussian particle. This enables accurate and efficient propagation of particles. We show that AMF consistently outperforms various baseline filters across diverse benchmarks, including single and coupled slow-fast Van der Pol oscillators and the Lorenz attractor. We also propose a parallel implementation of AMF, allowing high-fidelity PDF estimation with practical computational cost.

On a non-collapsible survival mechanism, longitudinal-matching and target-trial-emulation methods are not competing estimators of one truth but answers to different causal questions, so a benchmark that scores them against a single "true hazard ratio" fabricates bias. We provide the direct comparison of relative efficiency, variance estimation, and model sensitivity that reviews find lacking. On a deliberately non-collapsible continuous-time Cox mechanism with known truth, the dominant families (sequential Cox, sequential stratification, risk-set matching, and inverse-probability-of-treatment-weighted (IPTW) marginal structural models) target numerically distinct causal estimands (marginal, conditional, two average-treatment-effect-on-the-treated, and intention-to-treat versus per-protocol). First, we quantify the phantom bias a shared marginal truth fabricates: 0.32-0.33 log-cumulative-hazard-ratio units for the matching estimators and 0.15 for the conditional method; the associational naive time-dependent Cox sits 0.76 away, a total discrepancy compounding the estimand gap with confounding. Second, a rank reversal: the recommended method flips with the target estimand, and a low-variance off-target estimator can still win on mean-squared error. Third, a cross-family variance result: the cluster-robust sandwich is closer to nominal for the trial-stacking estimator (0.90) but under-covers the matching estimators (0.77-0.82), which a prespecified n=500 bootstrap sub-study brings to 0.95-0.96. Fourth, model sensitivity: omitting a confounder induces 0.45-0.50 log-hazard-ratio bias and undercoverage, and intention-to-treat and per-protocol effects diverge as switching increases; a heart-transplant analysis illustrates these. On a second mechanism three of four findings replicate, the rank reversal attenuating and model sensitivity proving calibration-dependent.

We study asynchronous optimization for finite-sum eigenspace computation in heterogeneous distributed systems. The theoretical foundations for asynchronous eigenspace computation remain scarce, with existing approaches offering limited coverage of dynamics directly on the Grassmannian under stale information. In this paper, we propose a Grassmannian incremental aggregation method that refreshes only arriving components and reuses cached gradients, retaining low per-update cost without global synchronization. The method employs an extrinsic polar update that preserves the intrinsic subspace geometry without requiring parallel transport of stale tangent vectors. Our analysis establishes a tight angle-dependent gradient-dominance characterization of the objective and a basin-invariance property for stale aggregated updates. These yield two-phase linear convergence, comprising an explicit broad-basin regime and a sharper local regime, with constants controlled by component spectral spreads. Experiments on serial and distributed PCA demonstrate improved sample efficiency and wall-clock convergence over representative baselines.

Gradient descent has been of particular interest in modern machine learning beyond sole focus on optimization. Implicit bias emerging from optimization, though not being encoded by the learning objective, often prevents from overfitting to spurious patterns. A typical instance is the max-margin implicit bias of a linear classifier, widely established for exponentially tailed loss functions. Even after having a given dataset separated, the parameter vector continues to evolve towards the max-margin direction asymptotically along the gradient descent dynamics. This phenomenon corroborates a frequent empirical observation of "train longer, generalize better." However, the max-margin convergence is an asymptotic phenomenon, and what is worse, this asymptotic convergence rate is significantly slower than pure convex optimization. Even so, the parameter vector along gradient descent dynamics commonly correlates with the max-margin direction positively (though not exactly) within considerably fewer iterations than the asymptotic rate. By shedding another light on this classical problem, this work aims to understand the mechanism of this early-stage alignment phenomenon. Our theoretical results demonstrate that the parameter vector weakly aligns with the max-margin direction within $O(\exp(\exp(-\delta)))$ iterations, where $\delta>0$ is the permissible alignment error, which is shown to be tight. By tracking the radial and tangential flows, our proof operates on the alignment dynamics directly with dataset geometry and gets rid of the asymptotic expansion, which is a key insight to establishing faster weak alignment.

Multimodal learning of images and tabular data is often impaired by ineffective representations, resulting in redundancy, dispersion, and generalization problems. To tackle this challenge, we introduce Graph-Enhanced Descriptor Sequencing (GEDS), a structured feature sequencing algorithm grounded in principles from the Column Permutation Problem (CPP). GEDS refines statistical descriptors of the features through similarity graph-based computations, systematically determining an effective feature sequencing. We incorporate GEDS within an order-aware efficient transformer framework, utilizing order-aware memory tokens that explicitly adhere to the derived feature sequencing via a dedicated loss function. Experimental results across multimodal benchmarks demonstrate that iStructTab effectively minimizes feature dispersion, improving predictive performance and robustness, and highlighting the significance of structured feature sequencing in multimodal learning.

Large language models are increasingly used for information seeking, yet semantically equivalent questions phrased in different ways can receive answers of considerably different quality. System prompts are widely employed to steer response behavior, but they are typically optimized for average-case quality, so some question phrasings may still receive incomplete or low-quality answers. To address this, we formulate a constrained mixed-strategy GroupDRO framework for system-prompt selection. Instead of optimizing the system-prompt text, the framework assigns weights to system prompts in an existing pool to minimize the worst-case information-quality loss across evaluation metrics and groups, while constraining the mean loss to stay close to that of average-based selection. Because pool generation and selection are decoupled, the method applies to any system-prompt pool and can leverage an ensemble of complementary system prompts rather than a single one. Across five LLMs on two bilingual medical and consumer-finance benchmarks, the constrained method reduces the Overall Mean, Worst 25% Mean, and Worst by 13.1%, 13.2%, and 13.7% on average relative to no mitigation while keeping overall quality close to Average selection. Its multi-prompt weights reveal complementarity across metric-group pairs. Code and data are available at https://github.com/Rainxu09/equitable-system-prompt-selection.

Bayesian Go/No-Go decisions with co-primary endpoints require specifying prior distributions under the Normal-Inverse-Wishart framework; however guidance on how prior hyperparameters influence trial decisions remains limited. We propose a calibrated prior specification framework for bivariate Go/No-Go decisions. Skeptical and enthusiastic priors are calibrated so that each assigns a target probability to a clinically relevant decision region. We prove that for any prior precision $\kappa > 0$, a unique scale parameter $\lambda_0$ achieves the target calibration. Operating characteristics are evaluated across different $\kappa $ via simulation and applied to a phase~3 telitacicept lupus trial.The simulation result indicates $\kappa$ is the primary driver of prior discrimination. At $\kappa = 1$, the go rate difference between priors was 0.07; at $\kappa = 10$ it reached 0.56, with false positive rates below 0.01. Operating characteristics were robust to the degrees of freedom parameter $\nu_0$ and prior correlation $\rho_0$, supporting a default of $\nu_0 = 2$. In the lupus application, prior sensitivity was negligible at $\kappa = 1$ but at $\kappa = 10$ the enthusiastic go rate was three times the skeptical rate at small sample sizes. The framework reduces prior specification to two choices: the prior center and the prior precision $\kappa$. The identification of $\kappa$ as the dominant parameter, together with the cautious choice of $\kappa$ before the trial, motivates adaptive approaches to prior precision.

We present a methodological framework for solving robust inference problems with partially specified distributions over measurable subsets of a parameter space. Partial specifications define a set of admissible distributions. The goal is to determine extremal values (over these admissible distributions) for statistical quantities, where these quantities—these objective functions—are ratios of expectations of analytic functions, continuous functions, piecewise continuous functions, and uniform limits of piecewise continuous functions. We show that extremal values are approached by sequences of admissible distributions, whose limiting extremal distributions are characterised by fixed-point conditions on their support locations. This characterises where extremal distributions place probability mass and yields a practical computational framework for solving the corresponding optimisation problems. We establish convergence and asymptotic properties of the resulting extremal distributions and extremal objective function values. This work extends robust inference methods (e.g. robust Bayesian inference) by combining extremal-distribution reduction, fixed-point characterisation, and approximation-based analysis within a unified framework.

Recent proliferation of data-optimization integration has led to a range of methods that aim to improve the statistical performance of data-driven optimization decisions. However, while many of these methods are motivated intuitively from a robustness or regularization perspective, their resulting statistical benefits are often unclear and, even if available, are established on a case-by-case basis. We provide a systematic dissection of data-driven optimization formulations using the view of "directionally perturbed" empirical optimization (EO). Specifically, this umbrella of formulations, which we call "EO+", covers many existing data-driven optimization methods, including regularization, distributionally robust optimization, transfer learning, and analogous methods for contextual optimization. On the one hand, we argue that without additional, correctly specified, side information, any EO+ method can result in at most second-order improvements. This provides a negative conclusion, namely “no free lunch is possible", on the statistical power of EO+. On the other hand, we show that when leveraging side information that is geometrically effective, achieving first-order improvements is possible by choosing hyperparameters that are significantly larger than what is typically suggested in the literature. Moreover, we construct a principled methodology based on excess risk estimation, via either system knowledge or bootstrap resampling, to maximize the first-order gain. We demonstrate how this gain connects to the control-variate principle, a variance reduction technique in the Monte Carlo simulation literature, which helps explain why geometrically effective side information is necessary.

The Rashomon effect describes the phenomenon that many models can achieve nearly equivalent performance on the same learning task, with significant ramifications for robustness, feature importance, and customizability. These use cases motivate the computation of Rashomon sets: the set of all models whose regularized loss is near-optimal. Decision trees are one of the few model classes for which Rashomon sets can be fully enumerated, but this computation has always been conditional on a binarization of the original data, either restricting which splits each tree is allowed to make or substantially increasing the complexity of an already difficult combinatorial problem. We introduce the first algorithm that exactly enumerates decision-tree Rashomon sets while exploiting the ordered structure of continuous features. We further develop a relaxation for approximate enumeration and an anytime algorithm that progressively refines the set of candidate thresholds, producing increasingly detailed approximations that converge to the continuous-feature Rashomon set. Experiments show that coarse binarization can miss many trees, important features, and predictive multiplicity; our algorithms achieve orders-of-magnitude speedups over existing enumeration methods, with approximations providing further speedups while maintaining near-perfect recall.

We comment on the optimally-transported generalized method of moments (OTGMM) estimator proposed by Schennach & Starck (2026a) and give counterexamples to Theorems 2-6 under their stated assumptions. First, the assumptions used in the small-error analysis are insufficient for consistency in Theorem 2 and asymptotic normality in Theorem 3. Next, we consider the large-error analysis, in which Theorem 4 states that the OTGMM estimator is equivalent to a GMM estimator with modified moments. We show that in a scalar model, Theorem 4 selects a value that differs from the unique OTGMM minimizer and violates the OTGMM sample moment restriction. In an overidentified model satisfying the assumptions used in Theorems 5 and 6, the first component of the Lagrange multiplier has different probability limits under the OTGMM estimator and the GMM estimator with modified moments. Under misspecification, the population value selected by OTGMM depends on the transport metric and on which variables may be adjusted. We give a sufficient condition under which solutions of the modified moment equations also solve the original constrained problem at a given parameter value, and separate conditions for consistency and asymptotic normality of the OTGMM estimator. We also show that Assumption 16 does not imply the matrix bound used in the supplemental proofs and replace Assumption 16 with a matrix condition that yields the bound.

Estimating large and simulation-intensive discrete choice models (DCMs) requires repeated evaluation of utilities, probabilities, derivatives, and simulated likelihoods over many observations, alternatives, and draws. Existing DCM software provides mature econometric workflows, while recent GPU-oriented tools accelerate selected models, leaving a gap between econometric coverage and scalable differentiable computation. We introduce TorchDCM, an open Python package for discrete choice modeling that compiles choice data and model specifications into a unified PyTorch-native likelihood engine for estimation, inference, prediction, and structured reporting on CPU or CUDA devices. The package covers the principal econometric functionality available across Biogeme and Apollo, including multinomial, nested, mixed, ordered, latent-variable, and panel likelihoods. It also supports ragged choice sets, constrained parameters, covariance estimation, willingness-to-pay analysis, elasticities, and extensible likelihood components. We evaluate TorchDCM against seven other estimation packages in aligned synthetic and real-data full-estimation experiments. TorchDCM completes all 45 synthetic cases, runs fastest in every comparable synthetic case, and satisfies the prespecified final-log-likelihood tolerance in every comparison with at least two comparable solutions. More precisely, it reduces median runtime by 89.1%-99.7% relative to Biogeme and Apollo across model-data settings. CUDA provides an additional 12.0-71.0x speedup over single-core TorchDCM. These results establish a scalable and reproducible foundation for econometric estimation and differentiable choice-model development. The open-source package and executed examples are available at https://github.com/mbc96325/torchdcm.

Physics-informed and hybrid models integrate data and physics, and read learned coefficients as physical. Yet a small prediction error guarantees neither that the postulated operator is right, nor that the design identifies the coefficient, nor that training reached its minimum. On one parabolic benchmark where a sensor aggregates several diffusion rates and the truth is known, a correct operator recovers the coefficient to machine precision. A wrong one leaves it $30\%$ off while the in-domain error is 1.36 times the noise, and drifts $27\%$ across windows. At $7\%$ noise the error falls below the noise; the coefficient stays $30\%$ off. Without the truth, a post-training coefficient diagnostic questions the trained model with three classical statistics, without oracle; the first two need no refit. Does it explain the data? Can the design resolve the named physics? Does the coefficient depend on the window? The specification test keeps its pre-registered false-alarm ceiling and rejects every replicate of both wrong operators tested; a network frozen at its minimum returns the same verdicts. Where the design is blind, a wrong operator confined to the unseen rate leaves the test mute; the design reading flags it beforehand. A network stopped short of its minimum mimics a wrong operator; an analytic second fit separates the two. The test has a notch against a missing reaction term, miscalibrates under correlated noise and loses power beyond $7\%$ noise. Each warning names its remedy (operator, experiment or training); silence certifies nothing.

Calibration requires a predictor to be unbiased after conditioning on its own predictions. Multicalibration asks for this guarantee simultaneously across a collection of groups. Many prediction tasks ask for several related features of the same conditional outcome distribution: variance is defined relative to the mean, skewness relative to both mean and variance, and conditional value at risk relative to a quantile. We study multicalibration for a sequence of $k$ properties in which each property is identifiable once the preceding properties are fixed. This framework includes Bayes pairs but does not require the properties to arise from a single loss. For every fixed $k\ge2$, we establish matching upper and lower sample-complexity bounds up to logarithmic factors under regularity conditions. Even with only polylogarithmically many binary groups, achieving multicalibration error $\varepsilon$ requires $\widetilde{\Omega}(\varepsilon^{-(k+2)})$ samples. Conversely, for any finite group family $\mathcal G$, we give a randomized learner using $O(\varepsilon^{-(k+2)}+\varepsilon^{-2}\log|\mathcal G|)$ samples. Thus the sample complexity is $\widetilde{\Theta}(\varepsilon^{-(k+2)})$ for polynomial-size group families. We instantiate the theory for three canonical examples.

We introduce a new class of uniformly ergodic MCMC algorithms, termed Diffeomorphic Contraction Sampler (DCS), and provide fast non-asymptotic mixing guarantees for DCS targeting distributions on $\R^d$ with arbitrarily heavy polynomial tails. DCS provides a solution to a well-known problem for MCMC samplers, which typically struggle with the combination of unbounded high-dimensional state space and vanishing gradients. The DCS pulls back a target on $\R^d$ onto a Euclidean ball $B(R)\subset\R^d$ and then samples from the transformed density on the convex set $B(R)$ via algorithms such as the Ball Walk, Hit-and-Run and others. A radial diffeomorphic contraction is chosen so that the pull-back density on $B(R)$ is bounded, implying uniform ergodicity for all targets with a finite polynomial moment. Non-asymptotic bounds for DCS require stronger assumptions such as log-concavity of the pull-back density. In practice, this is achieved approximately by a preconditioned automorphism of the ball $B(R)$, tuned via Variational Inference. Numerical simulation tests demonstrate that the DCS outperforms significantly the No-U-Turns sampler on multi-dimensional heavy-tailed targets arising as real-world posteriors in PosteriorDB benchmark. DCS also numerically outperforms in high-dimensional examples recently developed spherical projection samplers for heavy-tailed target distributions.

Longitudinal prediction of Alzheimer's disease biomarkers increasingly informs clinical decisions, and a forecast is only useful if it also reports how much to trust it. Conformal prediction supplies this by wrapping any forecaster in a prediction band with a finite-sample coverage guarantee under exchangeability. However, standard population-level conformal prediction guarantees only marginal coverage and may mask substantial under-coverage within clinically important subgroups. We introduce a general mechanism-driven framework for auditing and repairing such subgroup under-coverage. Across two cohorts (ADNI, OASIS-3), two base forecasters, and nine attributes spanning genetic risk, demographics, and clinical severity, we find that population-level bands under-cover high-risk subgroups in 57 of 68 audited combinations, despite achieving nominal marginal coverage. We trace these failures to two mechanisms: (A) rarity, where a group-conditional band calibrated on only $n$ patients covers at most $k/(n+1)$; and (B) tail-heaviness, where a population-wide band is too narrow for a heavy-tailed subgroup and additional data cannot close the gap. Under-coverage falls disproportionately on patients with high genetic risk and disease severity (6.1 pp mean deficit, 95% CI [3.3, 8.9]), while demographic groups remain at the target level on average (0.0 pp, CI [$-1.9$, 1.7]). We pair each mechanism with a corresponding conformal correction: cross-conformal pooling for rarity, per-subgroup calibration for tail-heaviness, and a coverage-safe marginal floor when both arise. Together, these corrections restore target coverage for nearly every high-risk subgroup across both cohorts and forecasters.

We study the problem of aligning data from multiple modalities into a shared representation space, focusing on settings where strong pretrained unimodal encoders are available but cross-modal paired data are scarce. We propose a structure-preserving alignment framework, joint kernel entropic Gromov–Wasserstein Optimal Transport (JK-EGW), which maps multiple modalities into a common latent space by minimizing a quadratic optimal transport objective. JK-EGW leverages fine-grained similarity relationships within and across modalities to construct a global affinity kernel instead of relying on raw feature-space distances. Our framework naturally provides explicit control over the geometry and distribution of the latent embedding. On the theory side, we establish parametric sample complexity rate of $n^{-1/2}$, matching the corresponding rates for standard, entropic and Gromov–Wasserstein optimal transport. On the algorithmic side, we derive a scalable alternating procedure to solve JK-EGW with entropic optimal transport (EOT) updates through a low-rank kernel approximation and a variational lifting. This lifting scheme effectively relieves the burden of a quadratic objective, and allowing us to take the advantage of existing EOT solvers. Empirically, we focus on post-hoc alignment of embeddings from pretrained encoders in data-scarce regimes, and show that our proposed method achieves improved multimodal retrieval performance compared to existing alignment baselines.

Objectives: While causal analysis of travel behavior is an emerging field, estimating heterogeneity in mode choice through causal modeling remains unexplored. This study demonstrates the application of a novel causal method, causal forest, to quantify the heterogeneity in travel mode choice shifts caused by the COVID-19 pandemic. Methods: We applied causal forests, a non-parametric causal machine learning method, to 802,935 trip records from the 2017 and 2022 waves of the National Household Travel Survey. The 2017 wave serves as the pre-pandemic control group, while the 2022 wave represents the treatment condition. Within the potential outcomes framework, we estimate average treatment effects (ATE), heterogeneous treatment effects (HTE), and conditional average treatment effects (CATE) across diverse socio-demographic groups and trip characteristics. Findings: Our results reveal an estimated ATE of a 1.86 percentage point (pp) increase in car-mode share, contrasted with decreases of 0.38 pp and 1.57 pp in public transit and walking, respectively. The largest increases in car use appeared for short-distance trips (one mile or less), households with annual incomes exceeding USD 200,000, and female travelers. Novelty: This is one of the first applications of causal forests to travel mode choice, and the first to use causal machine learning to estimate the pandemic's causal effect on mode choice analysis. Practical Applications: This study discusses methodological advantages, inherent assumptions, and limitations of causal forests within the context of transportation planning. This methodology is applied to COVID-19 travel data to illustrate how causal heterogeneity analysis can offer a deeper understanding of changes in mode choice. These insights are valuable for planners and policymakers in making policies related to mode shifts under an intervention.

Randomized controlled trials with similar protocols may yield conflicting findings when the distribution of relevant effect modifiers differs across study populations. Yet no formal statistical framework exists for defining and assessing whether conflicting trials are reconcilable, despite the importance of this question for evidence synthesis and regulatory decision making. To address this gap, we develop a causal inference framework for evaluating conditional and marginal reconcilability on additive and multiplicative scales in the presence of unmeasured effect modifiers. Within this framework, we use proxy variables for hypothesized unmeasured effect modifiers to develop regression-based tests of conditional reconcilability under parametric structural models. To assess marginal reconcilability, we extend existing transportability methods and develop an equivalence testing framework. We also introduce a reconciliation proportion to quantify the degree of marginal reconciliation. We illustrate these methods using the conflicting Meis and PROLONG trials of 17-alpha-hydroxyprogesterone caproate for preventing recurrent preterm birth. The analyses provided limited evidence that unmeasured effect modifiers such as cervical length, as captured by the selected proxies, were sufficient to marginally reconcile the trials. These findings demonstrate how proximal reconciliation methods may help regulators, researchers, and clinicians evaluate whether differences in study populations explain conflicting trial findings.

The set of $n\times n$ correlation matrices, known as the elliptope, has volume decaying at the super-exponential rate $\exp\{-\tfrac14 n^2\log n\}$. We characterize where this vanishing volume concentrates. A uniform draw is entrywise close to the identity yet globally far from it and nearly singular. Its maximum absolute correlation is of order $\sqrt{\log n/n}$, its Frobenius distance is asymptotic to $\sqrt n$, its empirical spectral distribution converges to the Marchenko-Pastur law with ratio one, and its smallest eigenvalue has the exact $\operatorname{Beta}(1,d)$ distribution, where $d=n(n-1)/2$, and is therefore of order $n^{-2}$. Distinct off-diagonal entries are exactly pairwise independent under every $\operatorname{LKJ}(\eta)$ law, which yields a Chen-Stein proof of the extreme-correlation point-process limit for the whole LKJ family and an $O(n^{-1})$ total-variation bound for finite-dimensional exceedance counts relative to Poisson laws with their exact finite-$n$ means. Finally, for a bounded, centered i.i.d. off-diagonal perturbation of any correlation matrix, the nearest-correlation projection removes a fraction of the squared perturbation that tends to one and recovers the original matrix in average squared error per entry.

Accurate heart-failure prognosis relies on tracking clinical risk over time, yet many machine-learning applications mishandle right-censored survival data by either discarding a patient's observation time or using it as a predictor. Discarding time ignores survival context, while using follow-up time as an input feature introduces severe target leakage that inflates apparent accuracy. We address this by proposing a discrete-time person-period framework for heart-failure mortality classification. Using the UCI Heart Failure Clinical Records cohort ($n=299$, 96 deaths), we transform the data into interval-level binary outcomes and benchmark a Cox proportional hazards baseline against person-period complementary log-log GLM and GAM models, alongside person-period random forest, XGBoost, random survival forest, and DeepSurv classifiers. The person-period GLM reproduces the Cox hazard ratios and concordance, validating the transformation, while the GAM captures significant nonlinear predictor effects and provides the best balance of discrimination and generalization; the flexible classifiers achieve strong raw performance but overfit. Finally, we quantify the leakage effect directly, including observed follow-up duration raises classification AUC from roughly 0.73 to nearly 1.00, confirming that follow-up duration must not be used as a baseline predictor. Overall, these results establish a survival-aware framework that combines flexible classification with valid time-to-event structure.

The 2024 Grants Pass decision newly emboldens US cities to manage unsheltered homelessness through forced displacement. Although literature demonstrates the harmful health and material impacts of this tactic, the longer-term results for housing outcomes and migration patterns remain less clear. In response, this study leverages longitudinal street outreach data to investigate where people move following encampment clearances. We specifically employ relational event models to predict the likelihood of various outcomes post-removal, such as relocating tracts, entering shelter, or obtaining housing. Results suggest that displaced residents do not travel far, yet clearances may still reduce visible homelessness by decreasing the size of camp communities. Furthermore, people appear unlikely to move indoors and instead face high risks of losing contact with service providers. These trends hold regardless of individual demographics, although people with mental health conditions demonstrate stronger attachments to their original sites. Evidence additionally indicates that neighborhood conditions could influence these migration behaviors. Such findings corroborate broader literature on place attachments, residential mobility, and invisibilization of poverty. This paper ultimately addresses the urgent need for a deeper understanding of how forced displacement impacts homelessness and pathways to housing.