arXiv subjects

Fabian Schmidt

Publications and source records attributed to Fabian Schmidt.

At least 19 recordsLinked to original sources

arXiv2026

Sensor Configuration Matters: A Systematic Evaluation of Multimodal SLAM on Quadruped Robots

Autonomous navigation of quadrupedal robots in diverse environments fundamentally relies on resilient Simultaneous Localization and Mapping (SLAM). While visual-inertial SLAM has matured across wheeled, handheld, and aerial platforms, a critical evaluation gap remains regarding how hardware-level sensor configurations affect performance under the aggressive dynamics of legged locomotion. Quadrupeds introduce distinct embodiment-induced sensory challenges, including foot-impact shocks, high-frequency mechanical vibrations, and rapid angular rotations, which degrade standard perception pipelines. To address this gap, we present a systematic evaluation of state-of-the-art visual, visual-inertial, and LiDAR-visual-inertial SLAM methods using the GrandTour dataset recorded on an ANYmal D quadruped. We isolate and quantify the impacts of camera modalities, shutter techniques, and inertial sensor tiers, analyzing their trade-offs across localization accuracy, algorithmic robustness, and computational resource utilization. Our empirical findings demonstrate that hardware selection has substantial influence on system resilience: stereo configurations consistently outperform monocular and RGB-D modalities, global shutter cameras significantly mitigate motion-induced tracking failures compared to rolling shutter cameras, and, crucially, standard inertial integration can degrade the performance of primarily vision-based frameworks under harsh legged locomotion. These insights additionally offer concrete design guidelines for tailoring custom sensor payloads to achieve dependable perception on agile legged systems.

cs.RO↗

arXiv2026

Field-level vs summaries: convergence of information in non-Gaussian density fields

We elucidate the sources of information gain in weakly non-Gaussian cosmological fields at the field- vs. summary-statistic-level in a controlled setting. Specifically, we compare field-level inference (FLI) with the standard power spectrum plus bispectrum (P${+}$B), and a family of composite-operator correlators (OCs) built from auto- and cross-spectra of local powers of the galaxy density field. The forward model is a linear density field with a single local quadratic coupling $\lambda$ and Gaussian noise; this minimal nonlinear setup interpolates between a purely Gaussian dataset ($\lambda=0$) and a non-Gaussian one ($\lambda\sim 1$), while keeping the analytical structure tractable. FLI is performed by jointly sampling the initial conditions, bias and noise parameters via MCMC; the summary posteriors are obtained with simulation-based inference (SBI) as well as Fisher estimates. In the Gaussian limit, the P${+}$B, OCs and FLI yield equivalent constraints, in agreement with the perturbative expectation. As the nonlinear coupling $\lambda$ increases, the summary-based uncertainties on the model parameters grow faster than the FLI ones, leading to an increasing information loss for a fixed set of summaries. This loss is largely, but not completely, recovered by adding OCs corresponding to up to the 6-point function. The information loss over FLI becomes even more pronounced for lower-noise data, where summaries corresponding to up to the 6-point function still capture significantly less information than the field.

astro-ph.CO↗

arXiv2026

A differentiable forward model for weakly perturbed stellar streams: substructure forecasts from density and kinematics spectra

Stellar streams are a promising way to gravitationally detect low-mass substructure, since their low dynamical temperature makes them retain the imprint of weak gravitational perturbations. We develop a fast, differentiable forward model for perturbed stellar streams in the diffusion regime, where the stream is heated by many small velocity kicks rather than by a few strong encounters. The substructure population enters only through its power spectrum, so the computational cost is insensitive to the number of perturbers, and alternative dark matter models and/or baryonic perturbers can be explored by changing this single input. We validate the simulations against analytical predictions, then forecast the sensitivity of a GD-1-like stream to the substructure power spectrum, adding to the stream density the full kinematics, both proper motions and the radial velocity. Kinematic information tightens the constraints by a factor of $\sim 3$-$5$ relative to density alone, improving the precision on the dark matter free-streaming cutoff scale from $\sim 1.2$ dex to $\sim 0.25$ dex at a fiducial value of $M_{\rm hm} = 10^6 M_\odot$ for a $5$ Gyr stream. A single well-measured stream could thus constrain dark matter competitively with current limits from strong lensing and satellite counts.

astro-ph.GA↗

arXiv2026

Revisiting Vul-RAG: Reproducibility and Replicability of RAG-based Vulnerability Detection with Open-Weight Models

Large language models (LLMs) have shown strong potential for automated software vulnerability detection, particularly in retrieval-augmented generation (RAG) settings. However, for approaches relying on proprietary models and APIs, reproducibility and replicability remain largely unexplored, raising the question of whether reported results generalize or depend primarily on specific model choices. In this work, we present a reproducibility study of Vul-RAG, a RAG-based framework for source code vulnerability detection that enhances LLMs with high-level vulnerability knowledge. We first replicate the results in a fully local and open-weights setting using the reported open-weight baseline models. We then extend the evaluation to a diverse set of recent open-weight LLMs, including code-specialized, general-purpose, and reasoning models of varying parameter sizes. The results confirm that the findings of Vul-RAG are reproducible under local deployment, but with minor deviations. Across all evaluated models, we observe a performance plateau at approximately 0.30 pairwise accuracy (code pairs for which both the vulnerable and the patched function are correctly classified). Notably, this plateau persists even for more recent and advanced models, indicating that improvements in model capacity alone do not substantially enhance performance. Finally, we discuss practical implications and trade-offs between detection effectiveness, model capabilities, and model scale. Implementation and evaluation artifacts are publicly available at https://github.com/hs-esslingen-it-security/revisiting-Vul-RAG.

cs.SE↗

arXiv2026

On the Relation Between Field-Level Posteriors, Correlators, and their Likelihoods

We develop a field-level posterior for cosmological data by marginalizing over initial conditions and noise in a general forward model. While our focus is on large-scale structure data, the results generalize to any weakly non-Gaussian observable. Moreover, the construction is non-perturbative with respect to the forward model and applies equally well to perturbative calculations, simulation-based predictions, and more general effective descriptions. Expanding the FLP around its Gaussian limit, we derive a general expression for the Fisher matrix and reorganize the field-level information into contributions associated with the connected correlators of the evolved field. This makes explicit which terms are captured by likelihood analyses based on the power spectrum, the bispectrum, or finite sets of summary statistics, and which are lost under compression. We recover the standard Gaussian-covariance result for the power spectrum, show that the Gaussian bispectrum likelihood reproduces the corresponding field-level contribution, and show how cross-covariances among summaries progressively reconstruct more of the full field-level information. As an application to the BAO scale, we show how the field contains all the information required for its optimal reconstruction in the presence of noise, and identify the contributions in the FLP needed to attain this limit. We also show that the reconstruction of the initial field arises naturally as a byproduct of our approach, yielding the optimal estimate of the initial conditions given the data and the noise. Our results provide a unified framework to compare field-level and correlator-based inference, to quantify the information loss induced by compression, and to explore the role of stochasticity.

astro-ph.CO↗

arXiv2026

Cosmological hydrodynamical simulations of clustering dark energy with Nefertiti

We present the first cosmological simulations that consistently include nonlinear clustering dark energy evolved as a fluid with the numerical hydrodynamics code Nefertiti. Dark energy perturbations become fully nonlinear on small scales, developing significant density fluctuations without exhibiting the catastrophic instabilities previously reported. We show results for the density distribution, power spectrum, and halo profiles of dark energy. Clustering dark energy contributes to the total density perturbation at the $\sim 10\%$ level inside and around massive halos in our simulations with constant $w=-0.9$, a significant potential signal for lensing and dynamical probes. These simulations pave the way to robust constraints on the speed of sound of dark energy perturbations from large-scale structure data.

astro-ph.CO↗

arXiv2026

LAD-Drive: Bridging Language and Trajectory with Action-Aware Diffusion Transformers

While multimodal large language models (MLLMs) provide advanced reasoning for autonomous driving, translating their discrete semantic knowledge into continuous trajectories remains a fundamental challenge. Existing methods often rely on unimodal planning heads that inherently limit their ability to represent multimodal driving behavior. Furthermore, most generative approaches frequently condition on one-hot encoded actions, discarding the nuanced navigational uncertainty critical for complex scenarios. To resolve these limitations, we introduce LAD-Drive, a generative framework that structurally disentangles high-level intention from low-level spatial planning. LAD-Drive employs an action decoder to infer a probabilistic meta-action distribution, establishing an explicit belief state that preserves the nuanced intent typically lost by one-hot encodings. This distribution, fused with the vehicle's kinematic state, conditions an action-aware diffusion decoder that utilizes a truncated denoising process to refine learned motion anchors into safe, kinematically feasible trajectories. Extensive evaluations on the LangAuto benchmark demonstrate that LAD-Drive achieves state-of-the-art results, outperforming competitive baselines by up to 59% in Driving Score while significantly reducing route deviations and collisions. We will publicly release the code and models on https://github.com/iis-esslingen/lad-drive.

cs.RO↗

arXiv2025

Enhancing LLM-based Autonomous Driving with Modular Traffic Light and Sign Recognition

Large Language Models (LLMs) are increasingly used for decision-making and planning in autonomous driving, showing promising reasoning capabilities and potential to generalize across diverse traffic situations. However, current LLM-based driving agents lack explicit mechanisms to enforce traffic rules and often struggle to reliably detect small, safety-critical objects such as traffic lights and signs. To address this limitation, we introduce TLS-Assist, a modular redundancy layer that augments LLM-based autonomous driving agents with explicit traffic light and sign recognition. TLS-Assist converts detections into structured natural language messages that are injected into the LLM input, enforcing explicit attention to safety-critical cues. The framework is plug-and-play, model-agnostic, and supports both single-view and multi-view camera setups. We evaluate TLS-Assist in a closed-loop setup on the LangAuto benchmark in CARLA. The results demonstrate relative driving performance improvements of up to 14% over LMDrive and 7% over BEVDriver, while consistently reducing traffic light and sign infractions. We publicly release the code and models on https://github.com/iis-esslingen/TLS-Assist.

cs.CV↗

arXiv2025

GraphPilot: Grounded Scene Graph Conditioning for Language-Based Autonomous Driving

Vision-language models have recently emerged as promising planners for autonomous driving, where success hinges on topology-aware reasoning over spatial structure and dynamic interactions from multimodal input. However, existing models are typically trained without supervision that explicitly encodes these relational dependencies, limiting their ability to infer how agents and other traffic entities influence one another from raw sensor data. In this work, we bridge this gap with a novel model-agnostic method that conditions language-based driving models on structured relational context in the form of traffic scene graphs. We serialize scene graphs at various abstraction levels and formats, and incorporate them into models via structured prompt templates, enabling systematic analysis of when and how relational supervision is most beneficial and computationally efficient. Extensive evaluations on the LangAuto and Bench2Drive benchmarks show that scene graph conditioning yields large and persistent improvements. We observe a substantial performance increase in the Driving Score of our proposed approach versus competitive LMDrive, BEVDriver, and SimLingo baselines. These results indicate that diverse architectures can effectively internalize and ground relational priors through scene graph-conditioned training, even without requiring scene graph input at test-time. Code, fine-tuned models, and our scene graph dataset are publicly available at https://github.com/iis-esslingen/GraphPilot.

cs.CV↗

arXiv2025

Non-Gaussian Galaxy Stochasticity and the Noise-Field Formulation

We revisit the stochastic, or noise, contributions to the galaxy density field within the effective field theory (EFT) of large-scale structure. Starting from the general, all-order expression of the EFT partition function, we elucidate how the stochastic contributions can be described by local nonlinear couplings of a single Gaussian noise field. We introduce an alternative formulation of the partition function in terms of such a noise field, and derive the corresponding field-level likelihood for biased tracers. This noise-field formulation can capture the complete set of stochastic contributions to the galaxy density at the field level in a normalized, positive-definite probability density which is suitable for numerical sampling. We illustrate this by presenting the first results of EFT-based field-level inference with non-Gaussian and density-dependent stochasticity on dark matter halos using LEFTfield.

astro-ph.CO↗

arXiv2025

Probabilistic Textual Time Series Depression Detection

Accurate and interpretable predictions of depression severity are essential for clinical decision support, yet existing models often lack uncertainty estimates and temporal interpretability. We propose PTTSD, a Probabilistic framework for Depression Detection from clinical interview utterance sequences that predicts PHQ-8 scores while modeling calibrated uncertainty. PTTSD includes sequence-to-sequence and sequence-to-one variants, both combining LSTMs, self-attention, and residual connections with Gaussian or Student's-t output heads trained via negative log-likelihood. The sequence-to-sequence variant enables temporal analysis of how predictive confidence evolves over an interview, despite the target being a single session-level score. Evaluated on E-DAIC and DAIC-WOZ, PTTSD achieves competitive performance among text-only systems (e.g., MAE = 3.85 on E-DAIC, 3.55 on DAIC) and produces well-calibrated prediction intervals. Ablations confirm the value of attention and probabilistic modeling, while a three-part calibration analysis and qualitative case studies highlight the clinical relevance of uncertainty-aware prediction.

cs.CL↗

arXiv2025

Parity Violation in Galaxy Shapes: Primordial Non-Gaussianity

We present a comprehensive study of galaxy intrinsic alignment (IA) as a probe of parity-violating primordial non-Gaussianity (PNG). Within the effective field theory (EFT) framework, we show that the parity-odd IA power spectrum is sensitive to the collapsed limit of the parity-odd primordial trispectrum. For a $U(1)$-gauge inflationary model, the IA power spectrum is proportional to the power spectrum of the curvature perturbation, $P_\zeta(k) \propto k^{-3}$. However, the proportionality constants contain not only the PNG amplitude but also undetermined EFT bias parameters. We use $N$-body simulations to determine the bias parameters for dark matter halos. Using these bias parameters, we forecast IA's constraining power, assuming data from the Dark Energy Spectroscopic Instrument (DESI) and the Rubin Observatory Legacy Survey of Space and Time (LSST). We find that the IA power spectrum can improve the current limits on the amplitude of parity-violating PNG derived from galaxy four-point correlation and CMB trispectrum analyses. Moreover, galaxy shapes are complementary to these probes as they are sensitive to different scales and trispectrum configurations. Beyond galaxy shapes, we develop a new method to generate initial conditions for simulations and forward models from the parity-odd trispectrum with an enhanced collapsed limit.

astro-ph.CO↗

arXiv2025

Mixed Dark Matter and Galaxy Clustering: The Importance of Relative Perturbations

We develop a perturbative model to describe large-scale structure in cosmologies where dark matter consists of a mixture of cold (CDM) and warm (WDM) components. In such mixed dark matter (MDM) scenarios, even a subdominant warm component can introduce distinctive signatures via its free-streaming effects, altering the evolution of density and velocity perturbations. We present linear-order solutions for both total and relative perturbations in the two-fluid system, identifying novel contributions to galaxy bias caused by the relative density and velocity modes between the components. Incorporating these effects into the galaxy bias expansion, we compute the linear galaxy power spectrum in both real and redshift space. Using Fisher matrix forecasts, we assess the sensitivity of upcoming surveys such as DESI and PFS to MDM scenarios. Our results demonstrate that neglecting relative perturbations can lead to significant biases in inferred constraints on the warm dark matter fraction, particularly for lighter WDM masses ($\lesssim 150~\mathrm{eV}$ and $\lesssim 80~\mathrm{eV}$) for PFS and DESI, respectively. This framework provides a consistent and generalizable approach for incorporating multi-component dark matter dynamics into galaxy clustering analyses.

astro-ph.CO↗

arXiv2025

A Systematic Literature Review on Detecting Software Vulnerabilities with Large Language Models

The increasing adoption of Large Language Models (LLMs) in software engineering has sparked interest in their use for software vulnerability detection. However, the rapid development of this field has resulted in a fragmented research landscape, with diverse studies that are difficult to compare due to differences in, e.g., system designs and dataset usage. This fragmentation makes it difficult to obtain a clear overview of the state-of-the-art or compare and categorize studies meaningfully. In this work, we present a comprehensive systematic literature review (SLR) of LLM-based software vulnerability detection. We analyze 263 studies published between January 2020 and November 2025, categorizing them by task formulation, input representation, system architecture, and techniques. Further, we analyze the datasets used, including their characteristics, vulnerability coverage, and diversity. We present a fine-grained taxonomy of vulnerability detection approaches, identify key limitations, and outline actionable future research opportunities. By providing a structured overview of the field, this review improves transparency and serves as a practical guide for researchers and practitioners aiming to conduct more comparable and reproducible research. We publicly release all artifacts and maintain a living repository of LLM-based software vulnerability detection studies at https://github.com/hs-esslingen-it-security/Awesome-LLM4SVD.

cs.SE↗

arXiv2025

Monodromic Dark Energy and DESI

Recent baryon acoustic oscillation (BAO) measurements from the Dark Energy Spectroscopic Instrument (DESI) collaboration have renewed interest in dynamical dark energy models, particularly those that cross the "phantom divide" ($w_{\rm DE} = -1$). We present the first observational constraints on monodromic k-essence, a physically motivated scalar field dark energy scenario capable of realizing rapid oscillations about the phantom divide. Using cosmic microwave background (CMB) information, DESI DR2 BAO measurements, and Type Ia supernovae observations, we constrain the amplitude, frequency, phase, and power-law index describing the monodromic k-essence scenario at the background level. We find that the monodromic dark energy scenario can fit these datasets with a $\chi^2$ that is comparable to the phenomenological $w_0$-$w_a$ parametrization. While the CMB and BAO data alone are consistent with the standard $\Lambda$CDM model, the inclusion of DESY5 supernovae shows a preference for a non-zero amplitude, $A=0.44^{+0.16}_{-0.12}$ (fully marginalized 68% C.L.). Conversely, inclusion of the Pantheon-Plus supernovae provides no evidence for monodromic k-essence, with $A<0.43$ (95% C.L.). We show that constraints on both monodromic dark energy and $w_0$-$w_a$ models are sensitive to the DESI DR2 LRG2 BAO distance, especially in the absence of supernovae data.

astro-ph.CO↗

arXiv2025

Large-Scale-Structure Observables in General Relativity Validated at Second Order

We present a second-order calculation of relativistic large-scale-structure observables in cosmological perturbation theory, specifically the "cosmic rulers and clock", which are the building-blocks of any other large-scale-structure observable, including galaxy number counts, on large scales. We calculate the scalar rulers (longitudinal perturbation and magnification) and the cosmic clock to second order, using a fully non-linear covariant definition of the observables. We validate our formulae on three non-trivial space-time metrics: two of them are null tests on metrics which are obtained by applying a gauge transformation to the background space-time, while the third is the "separate universe" curved background, for which we can also compute the observables exactly. We then illustrate the results by evaluating the second-order observables in a simplified symmetric setup. On large scales, they are suppressed over the linear contributions by $\sim 10^{-4}$, while they become comparable to the linear contributions on mildly non-linear scales. The results of this paper form a significant (and the most complicated) part of the relativistic galaxy number density at second order.

astro-ph.CO↗

arXiv2025

Forward vs Backward: Improving BAO Constraints with Field-Level Inference

We present results of field-level inference of the baryon acoustic oscillation (BAO) scale $r_s$ on rest-frame dark matter halo catalogs. Our field-level constraint on $r_s$ is obtained by explicitly sampling the initial conditions along with the bias and noise parameters via the LEFTfield EFT-based forward model. Comparing with a standard reconstruction pipeline applied to the same data and over the same scales, the field-level constraint on the BAO scale improves by a factor of $\sim 1.2-1.4$ over standard BAO reconstruction. We point to a surprisingly simple source of the additional information.

astro-ph.CO↗

arXiv2025

On the Connection between Field-Level Inference and $n$-point Correlation Functions

Bayesian field-level inference of galaxy clustering guarantees optimal extraction of all cosmological information, provided that the data are correctly described by the forward model employed. The latter is unfortunately never strictly the case. A key question for field-level inference approaches then is where the cosmological information is coming from, and how to ensure that it is robust. In the context of perturbative approaches such as effective field theory, some progress on this question can be made analytically. We derive the parameter posterior given the data for the field-level likelihood given in the effective field theory, marginalized over initial conditions in the zero-noise limit. Particular attention is paid to cutoffs in the theory, the generalization to higher orders, and the error made by an incomplete forward model at a given order. The main finding is that, broadly speaking, an $m$-th order forward model captures the information in $n$-point correlation functions with $n \leqslant m+1$. Thus, by adding more terms to the forward model, field-level inference is made to automatically incorporate higher-order $n$-point functions. Also shown is how the effect of an incomplete forward model (at a given order) on the parameter inference can be estimated.

astro-ph.CO↗