Publication Server of Zuse Institute Berlin (ZIB)
Not a member yet
6648 research outputs found
Sort by
Physics informed Bayesian optimization for inverse design of diffractive optical elements
No full textSequential Controlled Langevin Diffusions
No full textAn effective approach for sampling from unnormalized densities is based on the idea of gradually transporting samples from an easy prior to the complicated target distribution. Two popular methods are (1) Sequential Monte Carlo (SMC), where the transport is performed through successive annealed densities via prescribed Markov chains and resampling steps, and (2) recently developed diffusion-based sampling methods, where a learned dynamical transport is used. Despite the common goal, both approaches have different, often complementary, advantages and drawbacks. The resampling steps in SMC allow focusing on promising regions of the space, often leading to robust performance. While the algorithm enjoys asymptotic guarantees, the lack of flexible, learnable transitions can lead to slow convergence. On the other hand, diffusion-based samplers are learned and can potentially better adapt themselves to the target at hand, yet often suffer from training instabilities. In this work, we present a principled framework for combining SMC with diffusion-based samplers by viewing both methods in continuous time and considering measures on path space. This culminates in the new Sequential Controlled Langevin Diffusion (SCLD) sampling method, which is able to utilize the benefits of both methods and reaches improved performance on multiple benchmark problems, in many cases using only 10% of the training budget of previous diffusion-based samplers
Melting Transitions in Small Aluminum Clusters Simulated with Energies Approaching DFT Accuracy.
No full textWe describe a computational framework for modelling melting-like transitions in atomic clusters that combines first-principles energy calculations, global optimization, and machine-learned interatomic potentials. A diverse set of configurations is generated by global optimization, and energies are calculated by Density Functional Theory. The energies are fitted to an accuracy of 10 meV/atom or better with an Allegro E(3)-equivariant neural network potential. The model allows efficient parallel tempering Monte Carlo simulations with near DFT-level accuracy. This methodology was validated by simulating Na_20 and comparing it to earlier experimental and computational results. We used it to study melting-like transitions in Al_n+ clusters (n=9 to 16), and Al_n and Al_n^- (n=12, 13,14). The simulated heat capacity of these clusters, in particular Al_16+, are in qualitative agreement with experiments.
The melting point of Al_n+ clusters with n=11-16 are well above the bulk melting point (934 K). The closed-shell Al_13- species has an exceptionally high melting point, close to 2100 K
Incremental Heuristics for Periodic Timetabling
No full textWe present incremental heuristics for the Periodic Event Scheduling Problem (PESP), the standard mathematical tool to optimize periodic timetables in public transport. The core of our method is to solve successively larger subinstances making use of previously found solutions. Introducing the technical notion of free stratifications, we formulate a general scheme for incremental heuristics for PESP. More practically, we use line and station information to create heuristics that add lines or stations one by one, and we evaluate these heuristics on instances of the benchmarking library PESPlib. This approach is indeed viable, and leads to new incumbent solutions for six PESPlib instances
Uncovering Hidden Resonances in Non-Hermitian Systems with Scattering Thresholds (Laser Photonics Rev. 19(21)/2025)
No full text
From Vision to Implementation: Current Practices and Challenges of Software Management Plans in Research
No full textImproving the Identification of Layers in 3D Images of Ancient Papyrus using Artificial Neural Networks
No full textThe process of digitally unfolding ancient documents, such as folded papyrus packages, from 3D image data aims to be a non-invasive means to make previously hidden writing visible without risking to damage the precious documents. One of the main tasks necessary to digitally unfold a document is the geometric reconstruction of the writing substrate, which is a prerequisite for its subsequent unfolding. All current reconstruction methods require the existence of an interspace between different layers of the document to ensure a correct topology. Layers that appear merged together in the 3D image often result in wrong connections between layers and thus also in a wrong topology of the reconstructed geometry, which hinders the successful unfolding. Here, we propose to use a neural network to facilitate the discrimination of the layers. Using papyrus documents as an example of a particularly difficult writing material, we show that this significantly reduces the number of wrong connections and improves the overall identification of the layers. This in turn enables fully automatic digital unfolding of large areas of highly complex papyrus packages. Utilizing explainable AI (XAI) further allows us to explore the results of the applied neural network
Bi-invariant Geodesic Regression with Data from the Osteoarthritis Initiative
No full textMany phenomena are naturally characterized by measuring continuous transformations such as shape changes in medicine or articulated systems in robotics. Modeling the variability in such datasets requires performing statistics on Lie groups, that is, manifolds carrying an additional group structure. As the Lie group captures the symmetries in the data, it is essential from a theoretical and practical perspective to ask for statistical methods that respect these symmetries; this way they are insensitive to confounding effects, e.g., due to the choice of reference coordinate systems. In this work, we investigate geodesic regression---a generalization of linear regression originally derived for Riemannian manifolds. While Lie groups can be endowed with Riemannian metrics, these are generally incompatible with the group structure. We develop a non-metric estimator using an affine connection setting. It captures geodesic relationships respecting the symmetries given by left and right translations. For its computation, we propose an efficient fixed point algorithm requiring simple differential expressions that can be calculated through automatic differentiation. We perform experiments on a synthetic example and evaluate our method on an open-access, clinical dataset studying knee joint configurations under the progression of osteoarthritis