arXiv subjects

Christian Schmidt

Publications and source records attributed to Christian Schmidt.

At least 19 recordsLinked to original sources

arXiv2026

Application of LLMs to Threat Assessment of Foreign Peacekeeping Missions

We present a novel approach for applying Large Language Models (LLMs) to threat assessment in the context of foreign peacekeeping missions. Building on the PINPOINT project and its use case, the EU Monitoring Mission in Georgia, we combine an interdisciplinary risk-model with OSINT-based media collection and LLM-supported threat extraction. The proposed workflow maps media contents to mission-relevant threats, extracts structured information and applies several additional LLM-based processing steps to improve relevance and grounding. An evaluation of threats extracted from media documents shows high agreement between automatically generated results and human judgment for core aspects such as threat and mission relevance. These results indicate that LLMs provide a promising approach to support analysts in the context of peacekeeping missions.

cs.CR↗

arXiv2026

MERMAID-v1 PET Scanner Prototype: Initial Characterization and First Zebrafish Scans

MERMAID-v1 is a prototype PET scanner designed to support biomedical research involving adult zebrafish and similar species. The current experimental setup has been characterized, and scans of various phantoms, as well as adult zebrafish have been conducted. A dedicated reconstruction software was implemented, including accurate modeling of the parallax effect. The average energy resolution was 21.6% (FWHM at 511keV), with no significant dead-time effects observed for activities up to 18MBq. The absolute sensitivity at the center of the field of view (FOV) ranged from 0.06% to 0.31%, depending on the energy window (from 450-550 to 300-600keV), reflecting the limitations of the current two-head configuration. In the central 12mm of the transaxial FOV, the averaged spatial resolution is approximately 0.77mm (FWHM) transaxially and 0.66mm axially, as evaluated using a point source. Image quality was assessed using a downscaled NEMA-inspired IQ phantom and a 3D-printed Derenzo phantom. The reconstructed images suggest a spatial resolution around 0.7mm - 0.8mm, despite the lack of depth-of-interaction information. The first ex- and in-vivo PET scans of adult zebrafish were successfully performed, showing detectable tracer uptake in organs such as the brain and eyes despite low initial activity levels. These results confirm MERMAID-v1 capability to obtain useful results from the acquired data from living, anesthetized fish in a water-filled imaging chamber. While no scatter, attenuation, or efficiency corrections have yet been implemented, this work establishes a working proof-of-concept for dedicated PET imaging of small aquatic vertebrates. Future developments will focus on developing correction techniques, expanding the detector array, and integrating complementary modalities such as CT.

physics.med-ph↗

arXiv2026

SurGe: Improved Surface Geometry in Point Maps

Recent feedforward 3D reconstruction methods predict point maps and estimate global 3D geometry remarkably well. However, their predictions still exhibit inaccurate local surface geometry, which is clearly visible qualitatively but only weakly reflected in common metrics. To make these errors more explicit in evaluation, we introduce a point map normal metric that evaluates the local surface orientation induced by neighboring 3D predictions. To reduce these errors, we propose two complementary components: a point gradient matching loss that supervises depth-normalized 3D finite differences, and a Neighborhood Attention Decoder (NAD) that progressively upsamples features and uses Neighborhood Attention for local feature mixing. Across eight zero-shot monocular geometry benchmarks, our model, SurGe, achieves the best average rank for global point map AbsRel and consistently improves local point map and point map normal evaluations.

cs.CV↗

arXiv2026

Testing machine-learned distributions against Monte Carlo data for the QCD chiral phase transition

We demonstrate that conditional Masked Autoregressive Flows constitute a flexible interpolation tool for lattice QCD observables, conditioned on bare lattice parameters. As a benchmark, we use the chiral phase structure of QCD with five degenerate light quark flavours, which on coarse lattices exhibits a region of first-order chiral transitions terminating in a critical quark mass. The method successfully reproduces standard reweighting in the gauge coupling, and naturally extends to interpolation in quark mass and spatial volume, for which reweighting is computationally prohibitive or inapplicable, respectively. Once trained, the model generates samples across the full parameter space in minutes, which can be used to obtain consistent first estimates of the critical quark mass without simulating all intermediate parameter values. This offers a concrete reduction in the number of lattice ensembles required. Precision on the critical mass from learned distributions is so far prohibited by the mode-covering effect inherent to maximum-likelihood-based training, which introduces a systematic bias near first-order transitions. At the current stage, the method is well-suited for a range of practical applications: localising phase boundaries, identifying the universal scaling axes at a critical point, and accelerating informed determinations of parameter values ahead of high-precision Monte Carlo campaigns.

hep-lat↗

arXiv2026

Status and perspectives of ILDG

We discuss the status and progress of recent efforts to modernize the International Lattice Data Grid(ILDG).This includes activities of the metadata and middleware workinggroups concerning deployment and operation of crucial services (user management, metadata catalogues, file catalogues) and extensions of the metadata format, which have been tailored according to the needs of the large collaborations. We also report on developments and extensions that are planned to be addressed in the foreseeable future.

hep-lat↗

arXiv2026

Non-perturbative Renormalization of the EMT in Full QCD

The energy-momentum tensor (EMT) is the conserved current corresponding to space-time translation symmetry. Its applications are remarkably diverse, ranging from the thermodynamics to the calculation of transport coefficients. While the EMT is well-defined in the continuum up to a total derivative, with its coefficients fixed by Ward identities, its extension to lattice QCD is not straightforward. The primary challenge arises from the breaking of continuous space-time symmetries by the discrete lattice regulator. Although the EMT can be constructed on the lattice in a way that yields the correct continuum limit, the operators are not uniquely defined. In this proceeding, we construct the EMT for both pure-gauge theory and full QCD, discussing its renormalization in the specific context of determining the coefficients required for shear viscosity. In this context, we present a comparative analysis of the trace anomaly, number density, pressure, energy density and enthalpy density with imaginary chemical potential for multiple $\beta$ values at approximately the same temperature, aimed for the continuum limit.

hep-lat↗

arXiv2025

Block-Sparse Global Attention for Efficient Multi-View Geometry Transformers

Efficient and accurate feed-forward multi-view reconstruction has long been an important task in computer vision. Recent transformer-based models like VGGT, $\pi^3$ and MapAnything have demonstrated remarkable performance with relatively simple architectures. However, their scalability is fundamentally constrained by the quadratic complexity of global attention, which imposes a significant runtime bottleneck when processing large image sets. In this work, we empirically analyze the global attention matrix of these models and observe that the probability mass concentrates on a small subset of patch-patch interactions corresponding to cross-view geometric correspondences. Building on this insight and inspired by recent advances in large language models, we propose a training-free, block-sparse replacement for dense global attention, implemented with highly optimized kernels. Our method accelerates inference by more than $3\times$ while maintaining comparable task performance. Evaluations on a comprehensive suite of multi-view benchmarks demonstrate that our approach seamlessly integrates into existing global attention-based architectures such as VGGT, $\pi^3$ , and MapAnything, while substantially improving scalability to large image collections.

cs.CV↗

arXiv2025

OpenSplat3D: Open-Vocabulary 3D Instance Segmentation using Gaussian Splatting

3D Gaussian Splatting (3DGS) has emerged as a powerful representation for neural scene reconstruction, offering high-quality novel view synthesis while maintaining computational efficiency. In this paper, we extend the capabilities of 3DGS beyond pure scene representation by introducing an approach for open-vocabulary 3D instance segmentation without requiring manual labeling, termed OpenSplat3D. Our method leverages feature-splatting techniques to associate semantic information with individual Gaussians, enabling fine-grained scene understanding. We incorporate Segment Anything Model instance masks with a contrastive loss formulation as guidance for the instance features to achieve accurate instance-level segmentation. Furthermore, we utilize language embeddings of a vision-language model, allowing for flexible, text-driven instance identification. This combination enables our system to identify and segment arbitrary objects in 3D scenes based on natural language descriptions. We show results on LERF-mask and LERF-OVS as well as the full ScanNet++ validation set, demonstrating the effectiveness of our approach.

cs.CV↗

arXiv2025

PCA for Enhanced Cross-Dataset Generalizability in Breast Ultrasound Tumor Segmentation

In medical image segmentation, limited external validity remains a critical obstacle when models are deployed across unseen datasets, an issue particularly pronounced in the ultrasound image domain. Existing solutions-such as domain adaptation and GAN-based style transfer-while promising, often fall short in the medical domain where datasets are typically small and diverse. This paper presents a novel application of principal component analysis (PCA) to address this limitation. PCA preprocessing reduces noise and emphasizes essential features by retaining approximately 90\% of the dataset variance. We evaluate our approach across six diverse breast tumor ultrasound datasets comprising 3,983 B-mode images and corresponding expert tumor segmentation masks. For each dataset, a corresponding dimensionality reduced PCA-dataset is created and U-Net-based segmentation models are trained on each of the twelve datasets. Each model trained on an original dataset was inferenced on the remaining five out-of-domain original datasets (baseline results), while each model trained on a PCA dataset was inferenced on five out-of-domain PCA datasets. Our experimental results indicate that using PCA reconstructed datasets, instead of original images, improves the model's recall and Dice scores, particularly for model-dataset pairs where baseline performance was lowest, achieving statistically significant gains in recall (0.57 $\pm$ 0.07 vs. 0.70 $\pm$ 0.05, $p = 0.0004$) and Dice scores (0.50 $\pm$ 0.06 vs. 0.58 $\pm$ 0.06, $p = 0.03$). Our method reduced the decline in recall values due to external validation by $33\%$. These findings underscore the potential of PCA reconstruction as a safeguard to mitigate declines in segmentation performance, especially in challenging cases, with implications for enhancing external validity in real-world medical applications.

eess.IV↗

arXiv2025

Generalized susceptibilities and the properties of charm degrees of freedom across the QCD crossover temperature

We study the generalized charm susceptibilities in 2+1 flavor QCD on the lattice at several lattice spacings. We show that, below the chiral crossover, these susceptibilities are well described by the hadron resonance gas (HRG) model if charmed hadrons not listed in tables of the Particle Data Group are included. However, the HRG description abruptly breaks down just above the chiral crossover. To understand this, we use a model for the charm pressure in which it is expressed as the sum of partial pressures from charmed baryons, charmed mesons, and charm quarks. We present continuum estimates of these partial pressures and find that, while the partial pressures of charmed mesons and baryons drop below their respective HRG predictions, the charm quark pressure becomes non-zero above the chiral crossover.

hep-lat↗

arXiv2025

Selected topics on the QCD phase diagram at finite temperature and density

We will report recent progress on the QCD phase diagram at finite temperature and density. In particular, we discuss the universal scaling of the chiral transition in the limit of two massless quarks and one strange quark. We also discuss influence of other control parameter as chemical potentials, external magnetic field strength and number of quark flavors on the chiral transition. From calculations of Taylor expansion coefficients of the pressure w.r.t the baryon chemical potential and at imaginary chemical potential, we discuss estimates of the QCD critical point. Those estimates make use of the universal scaling ansatz of the Lee-Yang edge singularity.

hep-lat↗

arXiv2025

Shear viscosity from quenched to full lattice QCD

The shear viscosity of the quark-gluon plasma (QGP) plays a crucial role in interpreting current measurements from heavy-ion collisions and is a key input to hydro-dynamical models. The interest in shear viscosity also lies in the fact that QGP is the most ideal fluid ever observed and has the shear viscosity to entropy ratio ($\eta / s$) close to the theoretical bound $\eta / s \geq 1/ 4 \pi$ in the strong coupling region within AdS/CFT formalism. The lattice determination of $\eta / s$ has been explored for the pure gauge case, but its determination in full QCD remains unexplored, despite its significant importance. In this proceeding, we present updates on extending our quenched findings to full QCD. Specifically, we focus on the renormalization of the energy-momentum tensor with the gradient flow method and provide a progress update on determining the relevant renormalization coefficients for shear viscosity. For this purpose, we have used an imaginary isospin chemical potential.

hep-lat↗

arXiv2025

Lattice gauge ensembles and data management

We summarize the status of lattice QCD ensemble generation efforts and their data management characteristics. Namely, these proceedings combine the contributions to a dedicated parallel session during the 41st International Symposium on Lattice Field Theory (Lattice 2024), during which representatives of 16 lattice QCD collaborations provided details on their simulation program, with focus on plans for publication, data management, and storage requirements. The parallel session was organized by the International Lattice Data Grid (ILDG), following an open call to the lattice QCD community for participation in the session.

hep-lat↗

arXiv2025

The QCD phase diagram, universal scaling, and Lee-Yang zeros

We will report on current progress in the understanding of the QCD phase diagram, including universal scaling in the chiral limit and the vicinity of the QCD critical point. In the latter case we will discuss the universal scaling of Lee-Yang zeros and their determination from multi-point Pad\'e approximations to the baryon number density at imaginary chemical potentials. In particular, reported results include the critical phase transition temperature, the curvature of the critical and pseudo-critical transition temperature with respect to the chemical potential and the location of the QCD critical point.

hep-lat↗

arXiv2024

Look Gauss, No Pose: Novel View Synthesis using Gaussian Splatting without Accurate Pose Initialization

3D Gaussian Splatting has recently emerged as a powerful tool for fast and accurate novel-view synthesis from a set of posed input images. However, like most novel-view synthesis approaches, it relies on accurate camera pose information, limiting its applicability in real-world scenarios where acquiring accurate camera poses can be challenging or even impossible. We propose an extension to the 3D Gaussian Splatting framework by optimizing the extrinsic camera parameters with respect to photometric residuals. We derive the analytical gradients and integrate their computation with the existing high-performance CUDA implementation. This enables downstream tasks such as 6-DoF camera pose estimation as well as joint reconstruction and camera refinement. In particular, we achieve rapid convergence and high accuracy for pose estimation on real-world scenes. Our method enables fast reconstruction of 3D scenes without requiring accurate pose information by jointly optimizing geometry and camera poses, while achieving state-of-the-art results in novel-view synthesis. Our approach is considerably faster to optimize than most competing methods, and several times faster in rendering. We show results on real-world scenes and complex trajectories through simulated environments, achieving state-of-the-art results on LLFF while reducing runtime by two to four times compared to the most efficient competing method. Source code will be available at https://github.com/Schmiddo/noposegs .

cs.CV↗

arXiv2024

Fine-Tuning Image-Conditional Diffusion Models is Easier than You Think

Recent work showed that large diffusion models can be reused as highly precise monocular depth estimators by casting depth estimation as an image-conditional image generation task. While the proposed model achieved state-of-the-art results, high computational demands due to multi-step inference limited its use in many scenarios. In this paper, we show that the perceived inefficiency was caused by a flaw in the inference pipeline that has so far gone unnoticed. The fixed model performs comparably to the best previously reported configuration while being more than 200$\times$ faster. To optimize for downstream task performance, we perform end-to-end fine-tuning on top of the single-step model with task-specific losses and get a deterministic model that outperforms all other diffusion-based depth and normal estimation models on common zero-shot benchmarks. We surprisingly find that this fine-tuning protocol also works directly on Stable Diffusion and achieves comparable performance to current state-of-the-art diffusion-based depth and normal estimation models, calling into question some of the conclusions drawn from prior works.

cs.CV↗

arXiv2024

Searching for the QCD critical endpoint using multi-point Padé approximations

Using the multi-point Padé approach, we locate Lee-Yang edge singularities of the QCD pressure in the complex baryon chemical potential plane. These singularities are extracted from singularities in the net baryon-number density calculated in $N_f=2+1$ lattice QCD at physical quark mass and purely imaginary chemical potential. Taking an appropriate scaling ansatz in the vicinity of the conjectured QCD critical endpoint, we extrapolate the singularities on $N_τ=6$ lattices to pure real baryon chemical potential to estimate the position of the critical endpoint (CEP). We find $T^{\rm CEP}=105^{+8}_{-18}$~ MeV and $μ_B^{\rm CEP} = 422^{+80}_{-35}$~ MeV, which compares well with recent estimates in the literature. For the slope of the transition line at the critical point we find $-0.16(24)$.

hep-lat↗

arXiv2024

SIMULATeQCD: A simple multi-GPU lattice code for QCD calculations

The rise of exascale supercomputers has fueled competition among GPU vendors, driving lattice QCD developers to write code that supports multiple APIs. Moreover, new developments in algorithms and physics research require frequent updates to existing software. These challenges have to be balanced against constantly changing personnel. At the same time, there is a wide range of applications for HISQ fermions in QCD studies. This situation encourages the development of software featuring a HISQ action that is flexible, high-performing, open source, easy to use, and easy to adapt. In this technical paper, we explain the design strategy, provide implementation details, list available algorithms and modules, and show key performance indicators for SIMULATeQCD, a simple multi-GPU lattice code for large-scale QCD calculations, mainly developed and used by the HotQCD collaboration. The code is publicly available on GitHub.

hep-lat↗