Topics

> HPC-Driven Decision Science: Optimization, Simulation, and AI for Economics and Management

Modern global systems, from energy grids and supply chains to volatile financial markets, and techno-economic, as well as environmental dynamics present computational challenges that exceed the limits of traditional desktop computing. This track explores the frontier of high-performance decision intelligence, focusing on the intersection of large-scale data processing, complex simulations, and rigorous optimization. We invite submissions that leverage high-performance computing to solve high-dimensional problems in economics and operations research, including scalable agent-based modeling, discrete-event simulation, and multi-level mathematical optimization. This includes research work on reproducible HPC workflows that integrate simulation with machine learning (simulation-optimization, digital twins) and the use of LLM-augmented pipelines for scenario generation and policy analysis. Submissions may address the technical challenges of efficient, sovereign, and resilient execution on heterogeneous HPC architectures. Key application areas include transportation and logistics, market design, finance, risk management, macro-economics, and public-sector resilience.
This topic is organized by Gesellschaft für Operations Research (GOR) e.V. 

> Advances in Atomistic Simulations

Atomistic simulations are undergoing a profound transformation driven by methodological innovation and the integration of physics-based modeling with data-driven approaches. From electronic structure calculations to large-scale molecular dynamics, the field is evolving toward predictive, automated, and scalable workflows. Advances in HPC and heterogeneous architectures enable access to unprecedented spatial and temporal scales, while novel numerical schemes, adaptive algorithms, and high-throughput strategies systematically explore vast chemical and materials spaces. Within these developments, machine learning-based potentials, surrogate models, and active learning frameworks enhance accuracy, efficiency, and transferability, increasingly complementing first-principles methods and extending simulations to complex and correlated systems.
We invite submissions for talks or posters on algorithmic, methodological, and application-driven advances in atomistic simulations. Topics include scalable electronic structure and molecular dynamics methods, AI-enhanced simulation pipelines, workflow automation, uncertainty quantification, high-throughput frameworks, and hybrid classical–quantum strategies. Contributions that combine physical insight with computational innovation to advance predictive modeling across materials science, chemistry, and related disciplines are particularly encouraged.
This topic is organized by NHR’s Atomistic Simulation Center (ASC) in Berlin/Erlangen/Paderborn. 

> Low and Mixed-Precision Computing for Simulation and AI

Modern computer architecture is undergoing a fundamental shift. Hardware vendors are increasingly designing processors that prioritize reduced-precision formats such as FP16, BF16, FP8, or even FP4, delivering higher throughput at lower energy costs. This transition, originally motivated by AI workloads, is now reshaping the hardware landscape available to the broader HPC and scientific computing community.
As compute centres more and more operate this new generation of hardware, the scientific research community faces a pressing challenge: adapting AI models, numerical methods and simulation codes to effectively leverage reduced-precision capabilities. One now has to explore that transition, addressing e.g. mixed-precision iterative solvers, automatic precision tuning, numerical stability, and reproducibility. This requires to bring together AI experts, HPC practitioners and other researchers to discuss how one can bridge the gap between latest hardware trends and the accuracy demands of real-world applications.