arXiv subjects

Tobias Schröder

Publications and source records attributed to Tobias Schröder.

16 recordsLinked to original sources

arXiv2026

TabPFN-3.5: Technical Report

We introduce TabPFN-3.5, our new flagship Tabular Foundation Model. It significantly outperforms its predecessor, TabPFN-3, and all existing baselines across a broad range of tabular problems. TabPFN-3.5 sets a new state of the art on standard tabular prediction in TabArena, and extends it to the data practitioners encounter in practice: non-i.i.d. data with temporal or grouped splits, tables with strings, text and images, high-cardinality categorical features, and wide tables with many features. These gains carry over to our task-specific harnesses: state of the art on relational data and stronger time-series forecasting. For faster inference, our variant TabPFN-3.5-Fast runs up to 3x faster than TabPFN-3 while keeping most of the accuracy gains. In addition, we upgrade TabPFN-3.5-Plus, expanding our multimodal capabilities with advanced text and date handling alongside proprietary inference optimizations. Finally, we release a new version of our Thinking mode, TabPFN-3.5-Thinking, which scales inference-time computation to push the state of the art further. It benefits from our stronger base model and from inference-time improvements that make it up to 12x faster than TabPFN-3-Thinking.

cs.LG↗

arXiv2026

WildCat: Near-Linear Attention in Theory and Practice

We introduce WildCat, a high-accuracy, low-cost approach to compressing the attention mechanism in neural networks. While attention is a staple of modern network architectures, it is also notoriously expensive to deploy due to resource requirements that scale quadratically with the input sequence length $n$. WildCat avoids these quadratic costs by only attending over a small weighted coreset. Crucially, we select the coreset using a fast but spectrally-accurate subsampling algorithm -- randomly pivoted Cholesky -- and weight the elements optimally to minimise reconstruction error. Remarkably, given bounded inputs, WildCat approximates exact attention with super-polynomial $O(n^{-\sqrt{\log(\log(n))}})$ error decay while running in near-linear $O(n^{1+o(1)})$ time. In contrast, prior practical approximations either lack error guarantees or require quadratic runtime to guarantee such high fidelity. We couple this advance with a GPU-optimized PyTorch implementation and a suite of benchmark experiments demonstrating the benefits of WildCat for image generation, image classification, and language model KV cache compression.

cs.LG↗

arXiv2026

Kinematic Anisotropies in PTA Observations: Analytical Toolkit

The reported evidence for an isotropic gravitational-wave background (GWB) from pulsar timing array (PTA) collaborations has motivated searches for extrinsic and intrinsic anisotropies. Kinematic anisotropies may arise as a consequence of a boosted observer moving with respect to the frame in which the GWB appears isotropic. In this work, we present an analytical toolbox to describe the effects of kinematic anisotropies on the overlap reduction function. Our analytical results differ from previous findings at the quadrupole order and are detailed in three appendices. For the first time, we also derive the corresponding auto-correlation using two approaches, taking the pulsar distances to be infinite or finite, respectively. Our formulas can be used in forecasts or Bayesian analysis pipelines.

gr-qc↗

arXiv2025

Gravitational waves from low-scale cosmic strings without scaling

Cosmic strings are predicted in many extensions of the Standard Model and constitute a plausible source of gravitational waves (GWs) from the early Universe. In a previous article arXiv:2405.10937v2, we pointed out that the GW spectrum from a population of string loops in the scaling regime can exhibit a sharp cutoff frequency associated with the fundamental oscillation mode of string loops. In this paper, we study the effect of particle decay due to kink-kink collisions and cusps on the GW spectrum in the nonscaling scenario introduced in Ref. arXiv:1911.12066. We find analytical conditions for the existence of a cutoff frequency in the fundamental spectrum and provide expressions for this frequency. In large regions of parameter space, our results in the nonscaling model turn out to be identical to those in the scaling model. Finally, we demonstrate how the spectrum changes when transitioning from the regime with a cutoff frequency to the regime without a cutoff frequency. Our analytical estimates are validated at qualitatively different benchmark points by comparing them with numerical spectra.

astro-ph.CO↗

arXiv2025

Deep Optimal Sensor Placement for Black Box Stochastic Simulations

Selecting cost-effective optimal sensor configurations for subsequent inference of parameters in black-box stochastic systems faces significant computational barriers. We propose a novel and robust approach, modelling the joint distribution over input parameters and solution with a joint energy-based model, trained on simulation data. Unlike existing simulation-based inference approaches, which must be tied to a specific set of point evaluations, we learn a functional representation of parameters and solution. This is used as a resolution-independent plug-and-play surrogate for the joint distribution, which can be conditioned over any set of points, permitting an efficient approach to sensor placement. We demonstrate the validity of our framework on a variety of stochastic problems, showing that our method provides highly informative sensor locations at a lower computational cost compared to conventional approaches.

stat.ML↗

arXiv2025

On the overlap reduction function of pulsar timing array searches for gravitational waves in modified gravity

Pulsar Timing Array (PTA) searches for gravitational waves (GWs) aim to detect a characteristic correlation pattern in the timing residuals of galactic millisecond pulsars. This pattern is described by the PTA overlap reduction function (ORF) Γ_ab(ξ_ab), which is known as the Hellings--Downs (HD) curve in general relativity (GR). In theories of modified gravity, the HD curve often receives corrections. Assuming, e.g., a subluminal GW phase velocity, one finds a drastically enhanced ORF in the limit of small angular separations between pulsar a and pulsar b in the sky, ξ_ab --> 0. In particular, working in harmonic space and performing an approximate resummation of all multipole contributions, the auto correlation coefficient Γ_aa seems to diverge. In this paper, we confirm that this divergence is unphysical and provide an exact and analytical expression for Γ_aa in dependence of the pulsar distance L_a and the GW phase velocity v_ph. In the GR limit and assuming a large pulsar distance, our expression reduces to Γ_aa = 1. In the case of subluminal phase velocity, we show that the regularization of the naive divergent result is a finite-distance effect, meaning that Γ_aa scales linearly with fL_a, where f is the GW frequency. For superluminal phase velocity (subluminal group velocity), which is relevant in the case of massive gravity, we correct an earlier analytical result for Γ_ab. Our results pave the way for fitting modified-gravity theories with nonstandard phase velocity to PTA data, which requires a proper understanding of the auto correlation coefficient Γ_aa.

gr-qc↗

arXiv2025

The NANOGrav 15 yr Data Set: Running of the Spectral Index

The NANOGrav 15-year data provides compelling evidence for a stochastic gravitational-wave (GW) background at nanohertz frequencies. The simplest model-independent approach to characterizing the frequency spectrum of this signal consists in a simple power-law fit involving two parameters: an amplitude A and a spectral index γ. In this paper, we consider the next logical step beyond this minimal spectral model, allowing for a running (i.e., logarithmic frequency dependence) of the spectral index, γ_run(f) = γ+ β\ln(f/f_ref). We fit this running-power-law (RPL) model to the NANOGrav 15-year data and perform a Bayesian model comparison with the minimal constant-power-law (CPL) model, which results in a 95% credible interval for the parameter βconsistent with no running, β\in [-0.80,2.96], and an inconclusive Bayes factor, B(RPL vs. CPL) = 0.69 +- 0.01. We thus conclude that, at present, the minimal CPL model still suffices to adequately describe the NANOGrav signal; however, future data sets may well lead to a measurement of nonzero β. Finally, we interpret the RPL model as a description of primordial GWs generated during cosmic inflation, which allows us to combine our results with upper limits from big-bang nucleosynthesis, the cosmic microwave background, and LIGO-Virgo-KAGRA.

astro-ph.HE↗

arXiv2024

Energy-Based Modelling for Discrete and Mixed Data via Heat Equations on Structured Spaces

Energy-based models (EBMs) offer a flexible framework for probabilistic modelling across various data domains. However, training EBMs on data in discrete or mixed state spaces poses significant challenges due to the lack of robust and fast sampling methods. In this work, we propose to train discrete EBMs with Energy Discrepancy, a loss function which only requires the evaluation of the energy function at data points and their perturbed counterparts, thus eliminating the need for Markov chain Monte Carlo. We introduce perturbations of the data distribution by simulating a diffusion process on the discrete state space endowed with a graph structure. This allows us to inform the choice of perturbation from the structure of the modelled discrete variable, while the continuous time parameter enables fine-grained control of the perturbation. Empirically, we demonstrate the efficacy of the proposed approaches in a wide range of applications, including the estimation of discrete densities with non-binary vocabulary and binary image modelling. Finally, we train EBMs on tabular data sets with applications in synthetic data generation and calibrated classification.

stat.ML↗

arXiv2024

Gravitational waves from low-scale cosmic strings

Cosmic strings are a common prediction in many grand unified theories and a promising source of stochastic gravitational waves (GWs) from the early Universe. In this paper, we point out that the GW signal from cosmic strings produced at a comparatively low energy scale, $v \lesssim 10^9 \textrm{GeV}$, exhibits several novel features that are not present in the case of high-scale cosmic strings. Our findings notably include (i) a sharp cutoff frequency $f_{\rm cut}$ in the GW spectrum from the fundamental oscillation mode on closed string loops and (ii) an oscillating pattern in the total GW spectrum from all oscillation modes whose local minima are located at integer multiples of $f_{\rm cut}$. These features reflect the fact that string loops produced in the early Universe fail to shrink to zero size because of GW emission within the age of the Universe, if their tension is low enough. In addition, they offer an exciting opportunity to directly probe the discrete spectrum of oscillation modes on closed string loops in GW observations. For strings produced at a scale $v \sim 10^9\textrm{GeV}$, the novel features in the GW spectrum are within the sensitivity reach of future experiments such as BBO and DECIGO.

astro-ph.CO↗

arXiv2024

Embedded Domain Walls and Electroweak Baryogenesis

Embedded walls are domain wall solutions which are unstable in the vacuum but stabilized in a plasma of the early Universe. We show how embedded walls in which the electroweak symmetry is restored can lead to an efficient scenario of electroweak baryogenesis. We construct an extension of the Standard Model of particle physics in which embedded walls exist and are stabilized in an electromagnetic plasma.

hep-ph↗

arXiv2023

Energy Discrepancies: A Score-Independent Loss for Energy-Based Models

Energy-based models are a simple yet powerful class of probabilistic models, but their widespread adoption has been limited by the computational burden of training them. We propose a novel loss function called Energy Discrepancy (ED) which does not rely on the computation of scores or expensive Markov chain Monte Carlo. We show that ED approaches the explicit score matching and negative log-likelihood loss under different limits, effectively interpolating between both. Consequently, minimum ED estimation overcomes the problem of nearsightedness encountered in score-based estimation methods, while also enjoying theoretical guarantees. Through numerical experiments, we demonstrate that ED learns low-dimensional data distributions faster and more accurately than explicit score matching or contrastive divergence. For high-dimensional image data, we describe how the manifold hypothesis puts limitations on our approach and demonstrate the effectiveness of energy discrepancy by training the energy-based model as a prior of a variational decoder model.

stat.ML↗

arXiv2023

Training Discrete Energy-Based Models with Energy Discrepancy

Training energy-based models (EBMs) on discrete spaces is challenging because sampling over such spaces can be difficult. We propose to train discrete EBMs with energy discrepancy (ED), a novel type of contrastive loss functional which only requires the evaluation of the energy function at data points and their perturbed counter parts, thus not relying on sampling strategies like Markov chain Monte Carlo (MCMC). Energy discrepancy offers theoretical guarantees for a broad class of perturbation processes of which we investigate three types: perturbations based on Bernoulli noise, based on deterministic transforms, and based on neighbourhood structures. We demonstrate their relative performance on lattice Ising models, binary synthetic data, and discrete image data sets.

stat.ML↗

arXiv2023

PTArcade

This is a lightweight manual for PTArcade, a wrapper of ENTERPRISE and ceffyl that allows for easy implementation of new-physics searches in PTA data. In this manual, we describe how to get PTArcade installed (either on your local machine or an HPC cluster). We discuss how to define a stochastic or deterministic signal and how PTArcade implements these signals in PTA-analysis pipelines. Finally, we show how to handle and analyze the PTArcade output using a series of utility functions that come together with PTArcade.

hep-ph↗

arXiv2023

The NANOGrav 15-year Data Set: Search for Signals from New Physics

The 15-year pulsar timing data set collected by the North American Nanohertz Observatory for Gravitational Waves (NANOGrav) shows positive evidence for the presence of a low-frequency gravitational-wave (GW) background. In this paper, we investigate potential cosmological interpretations of this signal, specifically cosmic inflation, scalar-induced GWs, first-order phase transitions, cosmic strings, and domain walls. We find that, with the exception of stable cosmic strings of field theory origin, all these models can reproduce the observed signal. When compared to the standard interpretation in terms of inspiraling supermassive black hole binaries (SMBHBs), many cosmological models seem to provide a better fit resulting in Bayes factors in the range from 10 to 100. However, these results strongly depend on modeling assumptions about the cosmic SMBHB population and, at this stage, should not be regarded as evidence for new physics. Furthermore, we identify excluded parameter regions where the predicted GW signal from cosmological sources significantly exceeds the NANOGrav signal. These parameter constraints are independent of the origin of the NANOGrav signal and illustrate how pulsar timing data provide a new way to constrain the parameter space of these models. Finally, we search for deterministic signals produced by models of ultralight dark matter (ULDM) and dark matter substructures in the Milky Way. We find no evidence for either of these signals and thus report updated constraints on these models. In the case of ULDM, these constraints outperform torsion balance and atomic clock constraints for ULDM coupled to electrons, muons, or gluons.

astro-ph.HE↗

arXiv2021

Acoustic Leak Detection in Water Networks

In this work, we present a general procedure for acoustic leak detection in water networks that satisfies multiple real-world constraints such as energy efficiency and ease of deployment. Based on recordings from seven contact microphones attached to the water supply network of a municipal suburb, we trained several shallow and deep anomaly detection models. Inspired by how human experts detect leaks using electronic sounding-sticks, we use these models to repeatedly listen for leaks over a predefined decision horizon. This way we avoid constant monitoring of the system. While we found the detection of leaks in close proximity to be a trivial task for almost all models, neural network based approaches achieve better results at the detection of distant leaks.

cs.LG↗

arXiv2020

Emosaic: Visualizing Affective Content of Text at Varying Granularity

This paper presents Emosaic, a tool for visualizing the emotional tone of text documents, considering multiple dimensions of emotion and varying levels of semantic granularity. Emosaic is grounded in psychological research on the relationship between language, affect, and color perception. We capitalize on an established three-dimensional model of human emotion: valence (good, nice vs. bad, awful), arousal (calm, passive vs. exciting, active) and dominance (weak, controlled vs. strong, in control). Previously, multi-dimensional models of emotion have been used rarely in visualizations of textual data, due to the perceptual challenges involved. Furthermore, until recently most text visualizations remained at a high level, precluding closer engagement with the deep semantic content of the text. Informed by empirical studies, we introduce a color mapping that translates any point in three-dimensional affective space into a unique color. Emosaic uses affective dictionaries of words annotated with the three emotional parameters of the valence-arousal-dominance model to extract emotional meanings from texts and then assigns to them corresponding color parameters of the hue-saturation-brightness color space. This approach of mapping emotion to color is aimed at helping readers to more easily grasp the emotional tone of the text. Several features of Emosaic allow readers to interactively explore the affective content of the text in more detail; e.g., in aggregated form as histograms, in sequential form following the order of text, and in detail embedded into the text display itself. Interaction techniques have been included to allow for filtering and navigating of text and visualizations.

cs.HC↗