Scientific Computing Core

SCC supports CASUS researchers in all aspects of scientific computing and data-driven research. It also conducts research on specific topics of its own. The members of core come from various scientific disciplines and complement the subject-specific expertise of the scientific staff with methods, tools and extensive computer science knowledge.

Our department supports the application area scientists in all groups in CASUS with all aspects of Computational Science and data-driven research. Our team from diverse scientific backgrounds complements the application field scientific expertise with strong computer science expertise and latest methods and tools. Together, we achieve not only faster but better scientific results. Furthermore, we are active in our own research topics.

Together, we achieve not only faster but better scientific results. Furthermore, we are active in our own research topics.

Dr Andreas Knüpfer

Head of Scientific Computing Core

Contact

a.knuepfer@hzdr.de‎

+49 3581 375 23 123

Center for Advanced Systems Understanding

Conrad-Schiedt-Straße 20

D-02826 Görlitz

CP2K development

Some research groups at CASUS employ condensed matter simulations. For that purpose, CASUS hosts a development group for CP2K. CP2K is a quantum-chemistry and solid-state physics software package capable of performing large-scale massively parallel calculations on different levels of theory.

Research Data Management

Research Data Management (RDM) is a crucial step for all fields of science relying on data, no matter if medium size, large or huge amounts of data. SCC provides know-how and infrastructure for proper data management that helps with every-day data handling as well as large-scale data workflows according to the F.A.I.R. principles and scientific community standards.

Platform portability with Alpaka

The Alpaka library is a performance-portability programming model designed for various computing architectures, making it easier to develop and optimize C++ applications that run on different hardware. It abstracts the hardware specifics, so developers can write code that is portable across various platforms such as CPUs, GPUs, and other accelerators. Alpaka provides a uniform and flexible interface that helps in managing the complexities associated with different architectures, focusing on parallel computing paradigms. This allows developers to maximize performance without having to tailor their code for specific hardware platforms, enhancing productivity and scalability across diverse computational environments.

Support - Where2Test online platform on COVID19 pandemic analysis

During the COVID19 pandemic CASUS developed and ran the Where2Test platform. It combined data collection about COVID infections with demographics as well as prediction and simulation of hypothetical scenarios. For this purpose it employed High-Performance Computing (HPC), epidemiological modelling, control theory, and numeric optimization. In addition to traditional scientific papers, the CASUS effort has produced several web applications that made the HPC forecasts available to general public, and provided useful tools for dealing with the practical aspects of life under pandemic such as workplace risk calculator, optimizer for office occupancy, or testing strategy optimizer for medical care facilities.

Public–private partnerships on digital health

CASUS is a partner in several data driven, public–private partnership research projects. Among them are three notable digital health projects: PIONEER (big-data platform for prostate cancer patient data in Europe), OPTIMA (first interoperable and GDPR-compliant European real-world oncology data and evidence generation platform based on the needs of clinicians and patients with prostate, lung and breast cancer) and the UroEvidenceHub, an analytics platform for urology and related medical disciplines. CASUS hosts these big-data platforms and provides expertise in setting up a federated infrastructure to conduct statistical and artificial-intelligence-driven data analysis for developing decision-making tools for clinical usage.

AI models and surrogate models

The CASUS Scientific Computing Core aims to facilitate collaboration between the CASUS research teams. Currently, we are involved in pairing the award-winning the Materials Learning Algorithms (MALA) software package with the multivariate polynomial interpolation package minterpy.

Scientific visualizations with Scenery and Sciview

Scenery is a scenegraphing and rendering library. It allows you to quickly create high-quality 3D visualizations based on mesh data. Scenery is using Vulkan to render and its pipelines are configurable using YAML files, so it’s easy to switch between e.g. forward shading and deferred shading, as well as stereo rendering. Rendering pipelines can be switched on-the-fly. Sciview is an ImageJ/FIJI plugin for 3D visualization of images and meshes and uses Scenery as a rendering back-end. Both Scenery and Sciview support rendering to head-mounted virtual reality (VR) goggles like the HTC Vive or Oculus Rift via OpenVR/SteamVR.

Publications

Prediction of Steady and Unsteady Flow Quantities Using Multiscale Graph Neural Networks

Sebastian Strönisch, Maximilian Sander, Marcus Meyer, Andreas Knüpfer - ASME August 28, 2024

Analysis, optimization and uncertainty quantification of the aerodynamic behaviour of turbomachinery components is a fundamental part of the current industrial design process and requires the extensive use of compute-intensive CFD simulations. In this paper we investigate whether graph neural networks can be useful as surrogate models to accelerate the design process, for example in a multi-fidelity framework…

The impact of NGH on cancer cell death

H. Tahmasbi, K. Ramakrishna, M. Lokamani, A. Cangi - Phys. Rev. Mater. 8, 033803 (2024)

We created a computational workflow to analyze the potential energy surface (PES) of materials using machine-learned interatomic potentials in conjunction with the minima hopping algorithm. We demonstrate this method by producing a versatile machine-learned interatomic potential for iron hydride via a neural network using an iterative training process to explore its energy landscape under different pressures…

Inverting the Kohn-Sham equations with physics-informed machine learning

V. Martinetto, K. Shah, A. Cangi, A. Pribram-Jones - Mach. Learn.: Sci. Technol. 5 015050 (2024)

Electronic structure theory calculations offer an understanding of matter at the quantum level, complementing experimental studies in materials science and chemistry. One of the most widely used methods, density functional theory, maps a set of real interacting electrons to a set of fictitious non-interacting electrons that share the same probability density…