Czech researchers turn to LUMI for new advances in biomedicine, AI, smart cities, and particle physics

In the 35th round of the IT4Innovations Open Access Grant Competition, 22 researchers and their teams from Czech institutions have been granted access to the LUMI supercomputer. The awarded projects originate from IT4Innovations, the Czech Technical University in Prague, Charles University, the Czech Academy of Sciences, Masaryk University, Brno University of Technology, the University of West Bohemia in Pilsen, CEITEC, and Palacký University Olomouc.

The projects span a broad scientific spectrum – from advanced materials and magnetism, through artificial intelligence and trustworthy data processing, to biomedical research, sustainable energy, and fundamental physics, including support for CERN experiments.

Examples of the awarded projects:

Design of polysaccharide-binding peptides using molecular simulations and evolutionary algorithms

Call: 35th Open Access Grant Competition, OPEN-35-3

Researcher: Denys Biriukov

Institution: Masaryk University

Field: Biosciences

Denys Biriukov from Masaryk University will use the LUMI supercomputer to design new short peptides capable of recognising and binding polysaccharides on the surfaces of mammalian and bacterial cells. By combining evolutionary algorithms with molecular simulations, he will search for peptide motifs with high affinity for selected saccharide structures, such as hyaluronan or polysaccharides typical of E. coli and S. aureus. The results may contribute to the development of more precise biosensors, targeted therapeutics, and new diagnostic tools, thereby supporting progress in biomedicine and biotechnology.

ILTIR: Instance-level text-to-image retrieval

Call: 35th Open Access Grant Competition, OPEN-35-7

Researcher: Vladan Stojnić

Institution: Czech Technical University in Prague

Field: Informatics

Imagine trying to find a photo of your childhood stuffed toy monkey in a huge collection of images. Not just any toy monkey, but your orangutan-like one with shaggy reddish-brown fur, big round eyes, a light beige face, velcro on its hands, and a stitched smile. Most image search tools today would simply return generic pictures of toy monkeys. They often miss the specific details that matter most. Vladan Stojnić from the Czech Technical University in Prague received LUMI computational resources to develop a method that can find images based on detailed, object-level descriptions. Instead of searching for broad categories, like “stuffed toy,” the goal of our team is to help computers recognize and retrieve images of specific items that match a unique description. To achieve this, we plan to build on top of recent models that connect images and text and train them with data designed for this task. They’ll also adjust how these models learn, so they better understand the fine details that set one object apart from another.

Synthetic environments for next-gen safety and efficiency in cities

Call: 35th Open Access Grant Competition, OPEN-35-29

Researcher: Petr Strakoš

Institution: IT4Innovations

Field: Informatics

The IT4Innovations team will use the Karolina, LUMI, and Barbora NG supercomputers to develop intelligent systems for safe transport and efficient energy management in cities and smart buildings. They will focus on object recognition, human activity detection and real-time anomaly identification using artificial intelligence. To train the algorithms, they will also use synthetically generated visual scenarios from 3D environments, enabling them to simulate complex or rare situations that cannot be easily captured in the real world. The outcome will be a new generation of systems capable of responding swiftly and reliably in critical situations. This research is being carried out as part of international cooperation and the InnovAIte project.

Image: The image shows synthetically generated visual data from a virtual 3D environment designed for the development of AI algorithms in the field of transport.

Computational support for CERN experiments

Call: 35th Open Access Grant Competition, OPEN-35-22

Researcher: Michal Svatoš

Institution: Czech Academy of Sciences

Field: Physics

Experiments measuring particle collisions in the Large Hadron Collider (LHC) at CERN require enormous computing capacity for data analysis and Monte Carlo simulations. To meet these demands, the Worldwide LHC Computing Grid (WLCG) was established as a distributed computing environment, supported by both pledged resources, and unpledged resources provided by HPC centers, and occasional cloud resources and resources from volunteers via BOINC.

Among the significant contributors of computing capacity are the resources provided by IT4Innovations, including the Czech supercomputers Karolina and Barbora, and LUMI. A team from the Institute of Physics of the Czech Academy of Sciences has already developed an efficient environment for the automatic submission of workloads to IT4Innovations supercomputers. Processor utilisation is optimised using the HyperQueue tool developed by IT4Innovations, ensuring maximum efficiency of the available resources.

The overall aim is to increase the computing capacity for processing LHC experimental data, thereby strengthening the computational resources available to the Czech Republic and supporting advanced analyses and precise measurements in research into the fundamental properties of matter.