31ST open access competition RESULTS



Researcher: Rafael Dolezal  


Exploring the influence of phospholipid bilayer composition and surface tension on the key macroswitches in the human orexin receptor 2 using molecular dynamics and machine learning         

Karolina CPU  Alloc=250;  LUMI-C  Alloc=300;  LUMI-G  Alloc=5160  

In biochemistry, it takes a few minutes to reliably distinguish agonists from antagonists of the human orexin receptor 2 (OX2R), a member of the GPCR family. Current state-of-the-art computational chemistry methods are however experiencing serious troubles to achieve such a classification of OX2R modulators based only on receptor-based drug design methods (i.e. molecular dynamics). For these reasons, only a massive wet-chemistry high-throughput screening has been able so far to discover some new hits of OX2R agonists that are crucial for developing a drug against narcolepsy. In this project, I will focus on designing all-atom molecular dynamics simulation protocol for OX2R (i.e. unmodulated, agonized, antagonized) embedded in several types of phospholipid bilayers (i.e. POPC, POPE), enriched with cholesterol at various concentrations, employing isobaric-isothermal-isosurface-tension ensemble (NPγT). After finding the optimal setting that provides the membrane parameters comparable with experimental measurements, mobility of key transmembrane helices of OX2R (i.e. macroswitches) will be determined by the method of potential of mean force (PMF). Finally, the PMF energies as well as the ligand-residue and residue-residue contact maps in OX2R will be processed by machine learning tools do derive a predictive model to digitally assess modulating capacity of novel ligands on OX2R. This objective might bring a breakthrough step in the current in silico design of OX2R modulators.


Researcher: Agnieszka Stańczak      


Understanding structure/function correlations over the CBC enzymes           

Barbora CPU  Alloc=25000;  Karolina CPU  Alloc=60000         

Coupled-binuclear copper (CBC) metalloenzymes are a broad class of enzymes that are activated by dioxygen and contain a unique [Cu2O2] catalytic core. This core is responsible for catalyzing challenging biochemical transformations, such as regioselective monooxygenations/oxidations of substituted phenols. Despite almost four decades of intense experimental and theoretical research, factors governing the diverse reactivity of CBC enzymes remain only partially understood.  In this project we will correlate spectroscopy and kinetic experimental data with the state-of-the-art computations (including hybrid quantum and molecular mechanical, QM/MM, quantum mechanics, QM methods, as well as conformational sampling approaches) to explore internal architecture of different CBC family members and understand its connection with exhibited diverse reactivity. The complete mechanistic understanding across known CBC enzymes will facilitate discovery of new chemistries within the protein family, inspire development of small-molecule catalysts by implementing second sphere ligands interactions to the [Cu2O2] ‘catalytic core’, and open new directions in the design of small-molecule drugs and development of protein therapeutics tools.


Researcher: Luigi Cigarini     


Thermoelectric effects in twisted van der Waals materials     

Barbora CPU  Alloc=11800;  Barbora FAT  Alloc=200;  Barbora GPU  Alloc=700;  DGX-2  Alloc=100;  Karolina CPU  Alloc=19300;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=5000;  LUMI-C  Alloc=5400        

Two-dimensional systems are receiving increasing attention for their extraordinary properties in various fields of physics, such as photonics and electronics. In particular, recent technological advancements have led to the realization of precision microrotatory devices capable of creating few-layered twisted systems, where two adjacent planes are artificially maintained at a twisted angle relative to each other. Recent studies have reported important observations of properties of these twisted nanostructures at precise twisting angles, particularly in fields such as superconductivity and nonlinear optics, where the interaction between electronic structure and atomic nuclei vibrations plays a fundamental role in determining the nature of phenomena. This leads us to hypothesize analogous effects, which have not yet been explored, in the field of thermoelectric conversion, which exhibits similar interplays. Improving the performance of thermoelectric devices could have significant implications in the energy sector, such as developing devices for waste heat recovery in industrial systems or wearable devices powered by body heat.


Researcher: Sergiu Arapan    


Increasing the Curie temperature in uranium hydrides             

Barbora CPU  Alloc=5300;  Karolina CPU  Alloc=5800;  Karolina GPU  Alloc=5500;  LUMI-C  Alloc=4900 

From the fundamental point of view, UH3 represents a reference material for the 5f magnetism in light actinides, having been discovered as the first 5f ferromagnet. With the increased demand for high performance permanent magnets for green energy production and use, the uranium hydrides could become a less expensive and environmentally more friendly solution for replacing Rare Earth based permanent magnets. One obstacle is the relatively low Curie temperature, and, in this work, we apply electronic structure calculations technique to predict the effect of adding transition metals in UH3 in order to increase the Curie temperature.



Researcher: Mariia Savenko 


MD Simulations of hyaluronan as a target for anticancer drugs           

LUMI-C  Alloc=8800;  LUMI-G  Alloc=8600      

The extracellular matrix (ECM) is no longer considered exclusively a passive protective layer around the cell. The heterogeneity of ECM structure, and its changes with cell type and age, diseases or medical conditions, such as cancer, give a potential for developing cell specific molecules to prevent or slow down the development of ECM-dependent illnesses. The goal of the proposed project is to investigate the potential of hyaluronan (HA) – a polysaccharide found in the ECM - as a target to anticancer drugs. Particularly, we aim at investigating the interactions of hyaluronans of various length with toll-like receptors (TLRs), as TLRs are activating pathways linked to tumor growth. Using extensive molecular dynamics (MD) simulations based on docking results we aim at unraveling the full cooperation between HA and TLR4-MD-2 complex involved in TLR4-MD-2 receptor activation. We also aim at extending these simulations to a potential inhibitor, the HA-binding peptide Pep-1 that was developed to hamper TLR-associated cancer growth. Understanding how Pep-1 potentially hampers TLR associated cancer grow via HA binding opens novel path for developing anticancer drugs targeting hyaluronans.


Researcher: Dominik Legut  


Development and validation of OstravaJ: a novel computational method for calculating exchange interactions in Heisenberg magnets             

Barbora CPU  Alloc=20000;  Barbora FAT  Alloc=300;  Barbora GPU  Alloc=2000;  DGX-2  Alloc=300;  Karolina CPU  Alloc=20000;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=5000;  Karolina VIZ  Alloc=100;  LUMI-C  Alloc=9300    

Magnetic exchange interactions (Jij) are the primary reason why materials are magnetic. Their strength and structure determines most of the properties which make magnetic materials interesting and are often measured by the experimentalists, from simple things like the Curie temperature and hysteresis curves to the appearance of skyrmions. Such intrinsic property as the Jij can therefore be explicitly tracked into many of real world (extrinsic) magnetic properties. The exchange interactions themselves are quite difficult to determine experimentally, but they can be calculated using ab initio. However, even this approach is not foolproof and can yield wrong results. Most of available codes for calculating exchange interactions are based on the LKAG method, deeming the energy difference method too impractical. In this project, we want to further develop and validate OstravaJ, a code for calculating exchange interactions via the energy difference method. The main challenge in this method is to generate a suitable set of magnetic configurations, since the magnetic configuration space is exponential and the number of suitable configurations is small. In the proof-of-concept of OstravaJ, we have already demonstrated that this obstacle can be overcome. Now, we need the computational time to validate OstravaJ on an initial set of magnetic materials. The result of the project will be a reliable version of OstravaJ that can be a serious alternative to LKAG formalism used codes and therefore can be utilized for materials where the LKAG method fails.


Researcher: Katerina Ruzickova        


Analysis of terrain slope in relation to geological regions in a large area         

LUMI-C  Alloc=100       

This project extends the project Slope analysis from geologist view (Marschalko et al., 2024). Its goal is to analyze the relationship between terrain slopes and geological regions. There was found relationship in the average slope gradient in the engineering geological zones which gradually increases towards the individual soil engineering geological zones (Quaternary) to the rocks and semi-rocks (pre-Quaternary). Rock and semi-rock engineering geological zones always have a higher slope in the statistical characteristics than the soil engineering geological zones. The analysis was carried out for the entire territory of the Czech Republic, and Poland was also partially processed. The next stage will be the completion of the analysis for Poland and also the calculation for the Slovak Republic. The relationships derived for the Czech Republic will be verified on the data of neighboring countries and the specifics of other locations will be evaluated. Terrain characteristics in relation to geological characteristics in this scope and sense of processing have not yet been evaluated. The geological classification of sub-sites is expert and advanced.


Researcher: Simona Kocour 


Adjusting Privacy Levels 3D Maps        

Karolina CPU  Alloc=500;  Karolina GPU  Alloc=2500 

This project introduces a novel approach to 3D maps that balance precision and privacy. 3D maps are 3D representations of a scene that allow intelligent systems to perceive their surroundings. These maps serve a multitude of purposes, including localizing and guiding users in Augmented and Virtual realities, autonomous navigation of robots and cars, documenting historical and cultural heritage, and content creation. Typically, the more accurate the map is, the better it is for the target applications that can perceive even fine geometric and texture details. However, there are instances in which a user might not want to share all the details present in a map. An example is cloud-based localization where the 3D map is stored on servers, which exposes the content of the 3D map to the server provider. The level of detail that a user is willing to share will depend on the application and the content of the map and can range from removing personal details, objects, textures, or even whole rooms. The goal of this project is to develop a novel 3D map representation that gives users control over the level of detail stored and remains compatible with end-user applications.


Researcher: Debora LANČOVÁ          


3D non-ideal MHD simulations of accretion on magnetized star        

Karolina CPU  Alloc=13100     

The project aims to simulate the astrophysical plasma near a magnetized star. While real stars usually have complex geometry of the magnetic field, simulations are mostly done with the most straightforward field geometry: stellar dipole. We will introduce multipole stellar fields with mixed dipole, quadrupole and octupole geometry in our simulations. We will simulate the cases when the magnetic field is skewed concerning the stellar rotation axis, as in the observed stars. The geometry of such fields is complex and non-axisymmetric, demanding full 3D simulations, for which we will use the well-tested resistive magnetohydrodynamics modules in the PLUTO code.


Researcher: Michal Novotny


Thermal properties of Sc2C, Ti2C and V2C MXenes with mixed terminations

Barbora CPU  Alloc=20000;  Barbora GPU  Alloc=5000;  Karolina CPU  Alloc=5000   

MXenes, emerging as promising 2D materials, naturally occur with diverse surface terminations influencing their properties. While metallic MXenes show excellent electrical conductivity but poor Seebeck coefficient, semiconducting counterparts display the opposite trend. This project intends to  investigate the thermal conductivity and capacity of MXenes with mixed terminations, focusing on single metal carbide MXenes, terminated by F, O, OH. Using density functional theory and phonon dispersion spectra analysis, the project aims to understand how terminal group concentration and patterns affect thermal properties. Insight into these fundamental characteristics accelerates MXene development for various applications, including sensors and thermoelectric devices.


Researcher: Martin PYKAL     


Nanostructuring in Supercapacitors: Insights from Molecular Dynamics Simulations

Karolina CPU  Alloc=20000     

Recent advancements in the field of graphene-based materials offer many novel promising materials for energy storage and energy storage applications. This study proposes the use of molecular dynamics (MD) simulations to investigate the electrode-electrolyte interface in supercapacitors, focusing on the diverse structure of graphene derivatives as the primary electrode materials and the related supercapacitor’s performance. It is largely determined by the interaction between the electrode and the electrolyte. The project aims to employ an advanced form of MD, the constant potential method (CPM), to facilitate precise modeling of atomistic electrode charges while examining the charging mechanism at the nanoscale. The research hopes to elucidate key aspects of supercapacitor behavior, offering valuable atomistic insights for the design and optimization of high-performance energy storage devices.


Researcher: Jaroslav Hron    


Patient specific geometry flow simulations in aneurysms       

Karolina CPU  Alloc=2100;  Karolina GPU  Alloc=200 

Rupture of intracranial aneurysms (IA) leads to one of the most severe types of stroke with high morbidity and mortality. Deviations in hemodynamics inside an IA are associated with inflammation reaction in the vessels including activation of macrophages resulting in endothelial damage. Structural changes of the vessel wall lining lead to decreased strength of the wall increasing the risk of IA growth or rupture. We study the association between hemodynamics via computational fluid dynamics (CFD) and inflammation-related vessel wall damage from samples of IA sacs harvested during aneurysm surgery (clipping).


Researcher: Petr Šesták         


Revealing new kwinking mechanics in NiTi shape memory alloys by atomistic simulations based on machine learning          

Karolina CPU  Alloc=15650;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=50;  LUMI-C  Alloc=2800;  LUMI-G  Alloc=1800        

The aim of this project is to develop a machine-learned interatomic potential (ML) for the martensitic B19' phase of NiTi shape memory alloy and use it to study the recently discovered and established \kwinking\" mechanics in NiTi alloy, which is undoubtedly the most widely used shape memory material in applications. The kwinking mechanism represents cooperative twinning and formation and propagation of kink bands by cooperative plastic slip, and is a plastic forming mechanism that provides the high ductility of the martensitic phase that is essential for applications of this alloy. This project will allow to develop the mechanics of kwinking in NiTi B19' martensite on an atomistic scale, which includes the study of the atomic structure of this defect and the mechanism of its formation and propagation. At the same time, the analysis of the evolved ML atomic potential will allow us to understand the conditions that lead to this unusual mechanism of plastic deformation, thus providing a clue for the search for new alloys with similar suitable properties as NiTi. This work is part of a large-scale OP JAK institutional project FerrMion (Ferroic Multifunctionalities CZ.02.01.01/00/22_008/0004591) dedicated to ferroic materials in general, where atomistic calculations will be combined with experiments at the atomistic scale in the new Laboratory of Atom Probe Tomography (the first APT laboratory in Central Europe).


Researcher: Libor Dostál       


Computational investigations on the reactivity of complexes containing group 15 elements           

Karolina CPU  Alloc=49500;  Karolina FAT  Alloc=100 

The transformation of organic substrates belongs to the most important fields of chemistry. While transition metal complexes are commonly applied to catalyze reactions such as hydrogenation, oxidation, or additions of other small molecules, the necessary metals are usually rare, expensive or highly toxic. A solution to these problems may be the usage of more abundant main group elements, such as in group 15 (pnictogens) in the periodic table. The present project aims to gain a deep understanding of the reactivity of complexes containing arsenic, antimony or bismuth with various organic substrates. To complement the experimental studies, this research utilizes DFT and ab initio methods to find the reaction mechanisms and the thorough analysis thereof.


Researcher: Ales Prachar      



Barbora CPU  Alloc=1000;  Karolina CPU  Alloc=600  

Shape optimization of aircraft engine nacelles takes on added significance with the integration of electric propulsion systems, emphasizing cooling and thermal management alongside aerodynamic efficiency. Electric propulsion systems generate substantial heat during operation, particularly in components such as motors, power electronics, and batteries. Efficient cooling mechanisms are crucial to maintain optimal operating temperatures and ensure the longevity and reliability of these systems. Nacelle shape optimization involves integrating advanced cooling channels and heat exchangers to manage heat effectively while minimizing aerodynamic drag. The outer shape optimization is based on an adjoint based CFD optimization.


Researcher: Miroslav Kolos  


MXenes for groundwater remediation 

Barbora CPU  Alloc=30000;  Karolina CPU  Alloc=28500         

Chlorinated hydrocarbons, such as trichloroethylene (TCE), are persistent and toxic pollutants in groundwater, posing global environmental challenges. While conventional remediation technologies often fall short in efficacy and cost-effectiveness, advanced nanomaterials offer promising avenues for innovation. Building upon our previous research, which demonstrated the potential of nitridated iron nanoparticles (N-nZVI) in TCE dechlorination, we propose to extend our investigation to MXenes, a class of two-dimensional transition metal carbides and nitrides. MXenes, known for their versatile properties and reactivity, have yet to be fully explored in the context of groundwater remediation. This project aims to employ density functional theory (DFT) calculations to elucidate the interactions between MXenes and various chlorinated hydrocarbons, seeking to uncover the mechanisms underlying their potential remediation capabilities. By advancing our understanding of these fundamental interactions, we aim to pave the way for developing more efficient and stable nanomaterials for groundwater decontamination. This research not only contributes to the scientific understanding of MXenes in environmental applications but also holds the promise of enhancing the design and implementation of nanomaterial-based remediation strategies for chlorinated hydrocarbons.


Researcher: Jan Kubíček       


Automatic Detection and Classification based on Deep Learning of Significant Retinal Areas for Retinopathy of Prematurity Screening           

Karolina CPU  Alloc=50;  LUMI-G  Alloc=900  

Retinopathy of prematurity (ROP) represents a significant and severe disease, which significantly affect premature born infant’s vision as it affects their developing retinal blood vessels and cause retinal lesions (hemorrhages). In the case of presence of ROP, early diagnostic plays a crucial role in mitigation of ROP symptoms. The current trends of neonatal ophthalmology lead to developing intelligent methods, which are able to automatically detect early symptoms of ROP and ensure a reliable treatment. The main aim of our research is developing autonomous segmentation systems for automatic recognition and classification of ROP symptoms severity based on selected deep learning architectures. This research uses retrospective retinal images of premature born infants from the imaging systems: Clarity RetCam 3, Phoenix ICON, and RetCam Envision. In total, we use approximately 9000 images, labeled by neonatal ophthalmological expert, who classified individual stages of ROP and healthy infants. This research proposes various deep-learning architectures for automatic detection of optical disc, retinal blood vessels, retinal lesions, and consequently classify ROP symptoms to predict ROP severity. In this context, the proposed system represents a hybrid segmentation-classification scheme for ROP screening with significant potential in neonatal clinical practice. This research is a part of project LERCO (The Life Environment Research Center Ostrava) with the aim of bringing new trends of intelligent methods for ROP screening as a prevention of vision loss in premature born infants.


Researcher: Radim Uhlar      


Monte Carlo simulations of gamma fields and responses of the detectors inside the research reactor LR-0 (Research Centre Řež) for the improvement of nuclear data libraries             

Karolina CPU  Alloc=44900     

Many of the current power reactors are in operation for several decades, thus attention is paid to the possibility of their long-term operation. One of the limits is the condition of the reactor pressure vessel and reactor internals. Replacing them is very complicated and expensive. Gamma radiation in a reactor is produced in fission reactions, during the decay, and also during neutron interactions. This prompt component of gamma radiation has relatively high yields in the high energy region (several MeV) and carries significant energy. This radiation then causes significant heating of the material in the reactor parts. This, together with the production of gases due to the neutron interaction can lead to significant void swelling and affects the reactor long-term operation. For that reason, precise experiments focused on prompt gamma validation are demanded, because the description of the prompt gamma radiation production in current nuclear data libraries is significantly inaccurate. Prompt gammas were measured inside the concrete shielding simulator of the VVER-1000 reactor at the LR-0 reactor at the Research Centre Řež and time-consuming Monte Carlo simulations of the experimental setup are necessary to obtain improved prompt gamma radiation production rates for nuclear data libraries. This would result in significant financial savings because precise description of prompt gammas may allow for longer operation of nuclear power plants (more precise simulations, lower safety margins, etc.).


Researcher: Martin Srejber   


In-silico insight into endosomal escape           

Karolina CPU  Alloc=28800;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=800;  LUMI-C  Alloc=3400;  LUMI-G  Alloc=4800        

In the past few decades, liposomal delivery systems (such as lipid nanoparticles, LNPs) became one of the most promising delivery options in fields like cancer chemotherapy, gene therapy, liposome-entrapped drug delivery or vaccines. Recent advances in RNA-based medicine have provided new opportunities for facilitating genetically engineered messenger RNA (mRNA) as a potent information carrier. In the best-known recent example, Covid-19 vaccines, the immunogenic messenger RNA coding SARS-CoV-2 spike protein is enveloped by a liposome consisting of a mixture of natural and engineered lipids – ionizable lipids (ILs) PEGylated lipids, phosphatidylcholines, and cholesterol. The main aim of this project is to investigate and rationalize the complex and multi-step mechanism underlying RNA delivery upon LNP administration. We plan to investigate the step-by-step route of LNP after its administration including the interaction of LNP with plasma membrane model, the maturation of endosome after endocytosis up to pH dependent changes in LNP structure and RNA release.  We anticipate that this work will aid in the development of a framework to be utilized in future studies concerning liposomal-based drug delivery.


Researcher: Jakub Hromádka             


Computational study of the edge localized modes in the COMPASS-U tokamak        

Karolina CPU  Alloc=25000     

Magnetically Confined Fusion (MCF) plasma devices represent a promising way to a practically unlimited and environmentally friendly energy sources.  At present, number of MCF test devices, tokamaks, are under operation and some are under construction. One of such devices is the COMPASS-U, which will start operation in 2026 at the Institute of Plasma Physics of the Czech Academy of Sciences. COMPASS-U is a middle size, high magnetic field tokamak, where among others one of the main problems of MCF will be studied – plasma and power exhaust. The aim of the proposed project is to estimate power and plasma loads to the plasma-facing components in the COMPASS-U during transient events, so called Edge Localized Modes (ELM), when highest loads are expected. The outcomes of the project will be history of power and plasma loads to the so called divertor plates and the related erosion rates of the divertor material. The latter will be used to estimate the possible contamination of the COMPASS-U plasma during the ELM activity. For simulations we will use electrostatic PIC MC code BIT1.


Researcher: Petr Hellinger    


Plasma turbulence vs. fire hose instabilities: Kármán-Howarth-Monin analysis         

Karolina CPU  Alloc=40000     

Turbulence is a ubiquitous phenomenon in space and laboratory plasmas. A nonlinear coupling between scales leads to a spread of energy over a wide range of scales, spectral energy densities exhibit a power-law behavior, and the energy flows/cascades from large to small scales. At small scales the energy is dissipated and plasma particles are heated. In usual fluids the heating is connected with the irreversible particle-particle collisions; however, in many space and astrophysical plasmas the collisions are too rare and the heating proceeds via reversible channels. The turbulent motion in weakly collisional plasma and other effects such as changes in volume (expansion) lead to a formation of complex particle distribution functions, which are far from thermal equilibrium. These particle distributions may become unstable and generate waves; this is a process, which acts oppositely with respect to the dissipation: it transforms part of the internal energy to a combination of the kinetic and magnetic energies. To have a better insight to this complicated problem numerical simulation are necessary. We propose to study plasma turbulence in the context of one well-known example of weakly collisional turbulent system, in the solar wind (a flow of magnetized plasma from the Sun). We will use a three dimensional hybrid code (where electrons are assumed as a fluid but ions are treated as particles) to study properties of turbulent cascade towards ion scales, their heating, and the opposite process of instability generations of waves.


Researcher: Petr Vrchota       



Karolina CPU  Alloc=800          

With the recent rapid evolution of the UAM class concepts in past years, the reemerged interest in VTOL concepts and their aerodynamics arises. The tilt-wing is one of the possible solutions for such a vehicle. Recent decades of tilt-wing aerodynamic research point to the importance of the understanding transition regime from cruise to hover. Especially the wing stall phenomena which affect descent capabilities and control during the landing phase. Understanding such a regime is crucial for optimal wing design. High-lift devices, namely slot, reducing the speed and increasing the critical angle of attack in transition mode will allow a more efficient transition from cruise mode to hover. At the same time, it will reduce energy demands and enable better use of arrival and departure corridors in urban areas.


Researcher: Jakub Šebesta  


Magnetoelasticity of non-cubic materials       

Barbora CPU  Alloc=10100;  Barbora FAT  Alloc=200;  Barbora GPU  Alloc=800;  DGX-2  Alloc=100;  Karolina CPU  Alloc=13800;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=1400;  LUMI-C  Alloc=1600;  LUMI-G  Alloc=250             

Elastic material properties are widely treated in computational physics using direct calculations or simulations as the obtained data are crucial to investigate mechanical properties. However, for magnetic materials, the physical picture becomes more complicated since the presence of magnetism brings enhanced complexity. Unlike non-magnetic material, where the description relies only on the collective excitations of the lattice(phonons), regarding magnetic material excitations in the spin structure–magnons have to be considered as well. Further, phonon and magnon excitations exist as separate features, but they interact together. Meanwhile, phonon quasiparticles can interact with the electron subsystems forming polarons. It means that a complex theoretical description including all the subsystems and their mutual interactions is desirable. It has been shown for simple materials that phonon-magnon interaction significantly re-scales phonon group velocities or other phonon-related processes as thermal conductivity. Moreover, the effects strongly depend on the selected temperature. It appears that an effective description might be an approach based on the spin-lattice (SL) dynamic simulation possessing reliable inter-atomic potentials and magnetic interactions. This project aims on the magnetoelastic behavior beyond commonly considered cubic systems to cover a wider range of material with interesting properties.


Researcher: Mikulas Matousek         


Accurate potential energy surfaces for the Fe(II)-Porphyrin     

Karolina CPU  Alloc=28000     

Porphyrins are one of the most important metal complexes in biochemistry. While mostly known for carrying oxygen in blood, they also are responsible for performing many reactions in the human body.  Despite that, there still remain many open questions → one being the ground spin state of the Fe(II) porphyrin molecule. There are some suggestions that the geometry used might play a crucial role in determining the ground state. Therefore, we would like to calculate the full potential energy surface, to demonstrate the dependency on geometry, which will require us to perform many expensive high level calculations, which cannot be done without the use of a supercomputing centre.


Researcher: Jiri Brabec           


Theoretical study of the CO2 capture using metal organic framework-based materials

Karolina CPU  Alloc=35000     

Over the past few decades our improved understanding of the risks of climate change have pushed most of the world to commit to a reduction of CO2 emissions. Meeting this target will be increasingly difficult given that most models predict that, to limit warming, CO2 emissions would have to stop increasing by the 2nd half of the 21st century. The development of the CO2 capturing materials and detailed understanding of the process is very important for achieving this task. Recently, we started to study the electronic structure of the metal organic frameworks (MOFs) based on copper paddle-wheel building unit and its interaction with CO2. In this work, we will extend our project by using more advanced methods as AC0-DMRG and later DMRG-AC-in-DFT, in order to properly describe different type of interactions including strong correlation effects, dynamical correlations and dispersion.


Researcher: Jakub Benda      


Attosecond streaking in polar and non-polar linear molecules            

Karolina CPU  Alloc=20000     

The possibility of experimental investigation of the so-called “photoionization delays”, enabled by invention of ultrashort pulses, has opened a whole new window into not only the electronic, but also nuclear dynamics of the photoeffect. Photoionization delays are manifestation of prologed dwelling of the photoelectron in the vicinity of the molecule and have been shown to bear imprint of electronic shape resonances, below-threshold resonances, nuclear motion, interaction of the permanent and the transition molecular dipoles with the transient laser fields and other phenomena. So far their measurement has been predominantly indirect, typically aided by additional long-wavelength fields. One of such methods is the “attosecond streaking”, which uses moderately strong femtosecond IR pulses to probe the photoelectron wavepacket. Theoretical simulations of streaking needed to interpret the experiments cannot be perturbative and require solution of the fully coupled time-dependent Schrödinger equation, often very complex if the molecular electron correlation has to be accounted for.


Researcher: Tomas Blejchar


Optimization of design and production procedures of vertical pumps using modern technologies (Optimalizace návrhových a výrobních postupů vertikálních čerpadel s využitím moderních technologií)          

Karolina CPU  Alloc=47900;  Karolina VIZ  Alloc=500;  LUMI-G  Alloc=5000   

Pumps are generally the most used machines in the industry and everyday life. Waterworks and sewerage are based on applying pumps of all types and parameters. The pump's design is relatively complicated and influenced by many input parameters. A well-designed pump reaches maximum efficiency at the operating point, which defines the pump parameters when it is most often operated. Especially for pumps being operated continuously, e.g., 24 hours per day and seven days per week, the issue of effectiveness is crucial. The process of pump optimisation is relatively complicated and time-consuming. The first step consists of a manual elementary rough calculation, which is then compared with a numerical simulation. Then, the initial design is often modified to achieve the required parameters. The project aims to apply a neural network and AI approach to simplify pump design. The well-trained neural network will correct elementary rough calculations, and it will be able to find the optimum pump design in one step. That saves energy and computational sources necessary for repeated numerical simulations and the time required for the best design of the pump.


Researcher: Urszula Wdowik             


Exploring properties of novel densified Mg- and Zr-based metallic glasses by ab initio molecular dynamics    

Barbora CPU  Alloc=15000;  Barbora FAT  Alloc=100;  Barbora GPU  Alloc=4200;  DGX-2  Alloc=600;  Karolina CPU  Alloc=11300;  Karolina GPU  Alloc=11000

Metallic glasses are amorphous solids, usually multicomponent metal alloy systems, with liquid-like atomic structure involving short-to-medium-range order. They represent a relatively new class of structural and functional materials with unique physical and mechanical properties, which remain, however, very sensitive to preparation conditions and thermal treatment. Ultra-high pressure opens a new route to produce a high-density amorphous state of bulk metallic glasses from glass-forming multicomponent alloys. The present theoretical research addresses ab initio molecular dynamics simulations of densified ternary and quaternary Mg- and Zr- based glass-forming alloys. The main purpose of this project is to support and explain results of the ongoing experimental studies performed within the research project Tailoring properties and structure of Mg- and Zr- based metallic glasses by ultra-high pressure, Grant No. 2022/47/B/ST8/03153 National Science Center (NCN) Poland.


Researcher: Ridha Eddhib     



Karolina CPU  Alloc=91000     

Angle-Resolved Photoemission Spectroscopy (ARPES) offers deep insights into studies of solid state physics, allowing to visualize electronic structure and quantum states of a material. However, its precision critically depends on the exact calibration of the photon beam’s polarization. This project aims to apply Machine Learning (ML) techniques to enable more precise calibration and tuning of photon beam’s polarization. Currently experimentalists do not possess any tool to verify the exact polarization of their beam at the nano-level. Our approach will offer them a convenient tool to guide adjustments of the ARPES setups, aiming to achieve 100% polarization. This will be achieved by training an equivariant neural network, based on pytorch [1] and e3nn library [2], which are state of the art software packages to experiment with Machine Learning methods.  The training will be conducted on numerical results produced for monolayer graphene as a test material, making use of high-performance computing (HPC) resources to calculate one-step model photoemission [3] results as a function of various parameters. This dataset will be generated by means of the SPRKKR (Spin-Polarized Relativistic Korringa-Kohn-Rostoker) package [4].    [1]          Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., ... & Chintala, S. (2019). Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32. [2]              Geiger, M., & Smidt, T. (2022). e3nn: Euclidean neural networks. arXiv preprint arXiv:2207.09453. [3]              J. Minár, J. Braun, S. Mankovsky, and H. Ebert, “Calculation of angle-resolved photo emission spectra within the one-step model of photo emission - Recent developments,” J. Electron Spectros. Relat. Phenomena, vol. 184, no. 3–6, pp. 91–99, Apr. 2011, doi: 10.1016/j.elspec.2011.01.009. [4]          H. Ebert, D. Ködderitzsch, and J. Minár, “Calculating condensed matter properties using the KKR-Green’s function method—recent developments and applications,” Reports Prog. Phys., vol. 74, no. 9, p. 096501, Sep. 2011, doi: 10.1088/0034-4885/74/9/096501.


Researcher: Radim Špetlík


Improving Performance of Computer Vision Methods by Score Distillation Sampling from Large Pretrained Text-To-Image Models  

LUMI-G  Alloc=8000    

Score distillation sampling plays a pivotal role in various domains where efficient data representation and model optimization are required. The process involves distilling essential information from complex data distributions to facilitate model training and inference.   Thanks to availability of large pre-trained text-to-image models such as StableDiffusion, boosting the performance of computer vision methods is now available.  In our project, we are interested in score distillation sampling from large pretrained text-to-image models as introduced in [1] to improve performance of machine learning methods in: (i) image style transfer, (ii) temporal super-resolution, and (iii) optical flow estimation.


Researcher: Theodorus Petrus Cornelis Klaver       


Atomistic simulation of carbon hyperbolic pseudospheres    

Barbora CPU  Alloc=8900;  Barbora FAT  Alloc=200;  Barbora GPU  Alloc=1000;  Karolina CPU  Alloc=7900;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=200;  LUMI-C  Alloc=3800;  LUMI-G  Alloc=1000        

Graphene, the single atomic layer sheet of carbon that holds many records for materials properties, can be bent into tiny trumpet-like surfaces with special curvatures. Such surfaces are well less than 1/1000th the width of a human hair and they include 'double trumpet-like' hyperbolic surfaces that may be relevant in very different areas of research. They may act as desalination filters while they may also be relevant in the study of black holes. The creation and study of curved single layer carbon surfaces at such a small scale is difficult to achieve in experiments, but here computer simulations provide a solution. Atomistic simulations, i.e. simulations in which individual atoms are the distinguishable building blocks of larger structures, have become very powerful over past decades. This is thanks to the relentless increase in computing power and theoretical and algorithmic developments. As a result, simulated experiments can in some cases relatively easily give insights that would be quite difficult to obtain from experiments. The project aims to use atomistic simulations of different accuracy and computational cost to virtually create curved nano-scale carbon surfaces that unavoidably contain crystal defects. The impact of these defects on the overall stability, exact shape and conductivity of the surface structures will subsequently be studied. The most promising simulation results are then good candidates for experimental study.


Researcher: Petra KÜHROVÁ


Tuning AMBER All-Atom Force Field: Investigating the Thermal Stability of Nucleic Acid Duplexes

Karolina CPU  Alloc=5000;  Karolina GPU  Alloc=2000;  LUMI-G  Alloc=17000             

The thermal stability of nucleic acid duplexes is a crucial factor in various biological processes and technological applications. In this study, we will investigate the thermal stability of nucleic acid duplexes using enhanced sampling molecular dynamics methods, specifically Replica Exchange Molecular Dynamics or Replica Exchange with Solute Tempering. Our aim is to understand the influence of key parameters of nucleic acids on the melting temperature of duplexes. We will explore factors such as sequence composition, length of duplex, and solvent environment to elucidate their effects on duplex stability. Additionally, we will be tuning current all-atom AMBER nucleic acids force fields, utilizing not only general hydrogen bond fix but also quantum dynamics simulations. By integrating these methodologies, we aim to provide deeper insights into the thermodynamic properties of nucleic acid duplexes and enhance the accuracy of force field parameters for improved simulations of nucleic acid systems.


Researcher: Zdeněk Futera  


Solvation and Electric Field Effects to Electronic Transport on Protein Metal Junctions

Barbora CPU  Alloc=50000;  Karolina CPU  Alloc=25000         

Proteins, the building blocks of life, exhibit remarkable charge transfer properties that are essential for various biological processes and have potential applications in bioelectronics. Recent experimental techniques, such as electrochemical scanning tunneling microscopy (EC-STM), have revealed that proteins adsorbed on metal electrodes can conduct electricity via quantum tunneling, even in solution environments. However, the underlying mechanisms and factors influencing this quantum tunneling phenomenon in solvated protein-metal junctions remain poorly understood. This project aims to investigate the solvation and non-trivial electric field effects on electronic transport properties of protein-metal junctions using advanced computational methods based on density functional theory (DFT). By studying the adsorption geometries, electronic structure, and charge transport characteristics of proteins like cytochrome b562 and on gold electrodes in aqueous environments, we seek to elucidate the role of solvation, protein-electrode binding, and redox cofactors in mediating the tunneling.


Researcher: Martin Friak       


Fine-tuning thermodynamic stability of hydrogen-storage materials 

Barbora CPU  Alloc=31680;  Karolina CPU  Alloc=53760         

A greener future of our civilization critically depends on our ability to store the energy. Focusing specifically on the hydrogen technologies, one of the most promising materials for hydrogen storage is LaNi5. Unfortunately, its thermodynamic stability needs to be optimized so that the hydrogen is not bound too strongly (as that prevents de-charging via release of hydrogen atoms from the material), but the hydrogen atoms should not be bound too weakly either as that would render the material unsuitable for storage applications. An intensively investigated way of fine-tuning the thermodynamic stability of LaNi5 is the use of alloying elements, i.e. impurities incorporated into the crystal lattice. Tin atoms bear promise of proving the desirable modifications of the thermodynamic stability of LaNi5. We suggest to use advanced quantum-mechanical calculations to compute the impact of Sn atoms on materials properties of LaNi5. As these calculations include phonon calculations and as well as different magnetic states, they are very computer-time a memory demanding and supercomputers are needed.


Researcher: Jana Precechtelova       


Advancing IDP Analysis: Integrating Voronoi Tessellation with Neural Networks for Precision Modeling      

Barbora CPU  Alloc=4000;  Barbora GPU  Alloc=2100;  Karolina CPU  Alloc=1800;  LUMI-G  Alloc=4000                

This interdisciplinary project aims to revolutionize our understanding and analysis of intrinsically disordered proteins (IDPs) by developing advanced computational methods that integrate mathematical and deep learning approaches. The core objective is to generate ideal ensembles of IDPs, providing unprecedented insights into their diverse conformations and interactions, which are crucial for understanding their roles in various diseases. Utilizing techniques such as molecular dynamics simulations, cluster analysis, dimensionality reduction, and Voronoi tessellation, the project seeks to develop a novel python software package, StructGenie, for accurate ensemble generation and validation against experimental data.  Through its innovative approach to studying protein-protein interactions and protein dynamics, the project has the potential to significantly impact drug discovery, pharmacokinetics, and the broader field of protein research and dysfunction.


Researcher: Mohamed Bensalem     


Mechanical FEM simulations of nanoindentation and micropillar compression          

Barbora CPU  Alloc=30000;  Karolina CPU  Alloc=30000         

The characterization of mechanical properties in single and polycrystalline materials is critical for the  development and optimization of new materials for advanced engineering applications. In this context,  this work aims to presents a comprehensive approach through combining nanoindentation, capillary  compression experiments, and Finite Element Method (FEM) simulations to estimate the elastic  modulus, hardness, yield strength, and plastic deformation behaviors of both single and polycrystalline  materials. The experimental results will be compared and validated through FEM simulations, which  incorporate material constitutive models developed from the gathered data. This integrated method  enables the investigation of size effects, anisotropy, and the influence of microstructural features such  as grain boundaries and defects on the material's mechanical properties. This work will provide a robust  framework for predicting the mechanical behavior of materials under various loading conditions. The  outcomes of this research offer valuable guidelines for the design and selection of materials in industries  where mechanical performance is crucial, including aerospace, automotive, and microelectronics.


Researcher: Shivprasad Shivaram Shastri  


Ab-initio study of the photocatalytic activity of van der Waals heterostructures         

Barbora CPU  Alloc=25200;  Karolina CPU  Alloc=17125

Photocatalytic solar-to-hydrogen conversion by water splitting is regarded as an important sustainable energy source.  Van der Waals heterostructures can be suitable candidates to this aim, thanks to  favourable carrier migration paths and large surface area favouring catalytic activity. In this project, we consider PtXY and WXY (X,Y=S, Te, Se, X≠Y)  heterostructures and study the band alignment, charge transfer and optical absorption in order to assess the photocatalytic activity. Moreover, vacancies are known to occur in the as prepared materials or due to aging; such vacancies can enhance or degrade the efficiency. So, we will also study the effect of vacancies on the photocatalytic performance in these materials. The results will provide useful physical insights to design new materials for photocatalytic applications.


Researcher: Ievgeniia Korniienko     


Magnetization dynamics induced by ultrashort terahertz radiation: Toward designing spin-based terahertz sensors 

Barbora CPU  Alloc=20000;  Barbora FAT  Alloc=300;  Barbora GPU  Alloc=1000;  DGX-2  Alloc=300;  Karolina CPU  Alloc=40000;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=10000;  LUMI-C  Alloc=8200      

Terahertz (THz) radiation is widely used in fields from medicine to security, high-speed telecommunications and fundamental researches. However, despite the wide variety of THz detectors operating on the basis of diverse phenomena, there is still room for their improvement. Thus, incoherent quantum THz detectors allow only signal amplitude detection but not its phase, and coherent detection systems, which allow detecting the phase of the signal, face the problem of quantum noise. Moreover, when the interaction of THz radiation with matter is analyzed (sensors), it is common practice to perform this analysis using the phenomena sensitive to the electrical component of electromagnetic THz radiation. However, in this project the aim is to utilize the magnetic component, which is smaller in magnitude and therefore much challenging to create a detector sensitive to it. The detection of the magnetic component of the electromagnetic THz pulse has a significant advantage, which consists in the possibility of relatively simple detection not only the signal’s amplitude, but also its phase. Such approach could  provide information about the entire pulse by itself or supplement the information obtained in another way from the electrical component, expanding the possibilities or improving the accuracy of detection. The recent works show the possibility of creation a magnetic field sensitive THz detector. In our research, we further develop this idea and analyze what useful or additional data can be obtained from the analysis of the magnetic dynamics under the THz pulse magnetic field influence in a ferromagnet. We focus on the practical aspects of creating such THz sensor, such as investigating the material properties limits  under which a spin-based sensor can be most effective, and explore ferromagnetic candidates potentially most suitable for creating it.


Researcher: Tomas Brzobohaty         


FALCON - foreseeing the next generation of aircraft   

Karolina CPU  Alloc=2900;  Karolina GPU  Alloc=300;  LUMI-C  Alloc=700;  LUMI-G  Alloc=10000    

Direct aviation emissions accounted for 3.8% of total CO2 emissions and 13.9% of the emissions from transport in the EU in 2017, making it the second biggest source of greenhouse gas emissions after road transport. In addition, the growing amount of air traffic means that many EU citizens are still exposed to high noise levels. Intensified research and innovation activities are therefore needed to reduce all aviation impacts and emissions (CO2 and non-CO2, noise, manufacturing) for the EU to reach its policy goals towards a net-zero greenhouse gas emissions by 2050. One of the main levers to decrease CO2 emissions is to reduce the airframe structural weight. As an answer, FALCON’s ambition is to enhance the design capabilities of the European industrial aircraft sector, focusing on fluid-structure interaction (FSI) phenomena to improve the aerodynamic performances of aircraft (unsteady loads). Specifically, FALCON aims to develop high-performance, predictive and multi-disciplinary tools for FSI in aeronautics, to reduce the aeroacoustics and aeroelastic instabilities using multi-fidelity optimization. This will also benefit specific noise emissions generated by flexible and mobile airframe structures when exposed to both low and high-speed fluid flows.


Researcher: Jiří Pittner           


Relativistic coupled cluster methods augmented by higher excitations from a DMRG wave function  

Karolina CPU  Alloc=17400     

The density matrix renormalization group (DMRG) method has proven itself as a powerful method providing qualitatively correct description of strongly correlated molecular systems. Nevertheless, its lack of the dynamical correlation is a serious drawback, which in general prevents to achieve the goal of obtaining chemical accuracy. Various post-DMRG approaches have thus been suggested; in our group we have focused on the DMRG-tailored coupled cluster method, including a relativistic version. The tailored CC method was able to yield satisfactory numerical results, but it still has the disadvantage that the correlation in the active space is \"frozen\" at the DMRG level and does not reflect interaction with the outer orbital space. In this project we plan to address this issue and explore another way to externally correct the single- and multireference CC methods using selected higher excitations obtained from the MPS wave function provided by the DMRG method and to extend the relativistic treatment to CC methods externally corrected by higher excitations. The methods developed in course of this project should be applicable to accurate calculations of heavy-element containing molecules, like e.g. actinide compounds important in the nuclear research and industry.


Researcher: Georgios Tolias 


SLIIR: Super-Large-scale Instance-level Image Retrieval         

LUMI-G  Alloc=5000    

The proliferation of visual content across personal and online photo collections has underscored the need for effective browsing mechanisms. Specifically, the need for searching of particular objects within photo repositories is prevalent across various real-world applications, such as reverse image search, i.e. the visual-based counterpart of text-based Google search. However, current benchmarks are hampered by their limited scale and the presence of noisy image annotations, often restricted to photos of popular sightseeing attractions. Consequently, the insights drawn fail to reflect the complexities of real-world scenarios.  This project aims to address these limitations by establishing a novel instance-level image retrieval benchmark characterized by accurate annotations and unparalleled scale, surpassing existing benchmarks by two orders of magnitude. In contrast to existing benchmarks, this benchmark will include a diverse range of objects to provide a comprehensive evaluation platform. A wide spectrum of existing models and methodologies will be evaluated, with a particular emphasis on leveraging global and local image representation additionally to geometric information.  By revisiting prior conclusions within this enhanced framework, the project endeavors to establish the de-facto benchmark for image retrieval research in the coming decade. Furthermore, it aspires to serve as a benchmark for evaluating the efficacy of foundational models in visual representation learning, thus contributing to the advancement of the field.


Researcher: Assia Benbihi    


3D Reconstruction and Feature Matching Benchmarks (CPU resources completion)              

Karolina CPU  Alloc=29000;  LUMI-G  Alloc=200          

This project proposes two computer vision benchmarks related to 3D reconstruction and feature matching in urban environments under a “walking pedestrian” scenario. The availability of well-defined benchmarks facilitates progress on a wide array of computer vision, machine learning, and robotic problems by allowing researchers to evaluate and compare their contributions under a unified setup. These benchmarks aim to summarize real-world challenges in carefully curated datasets and provide an analysis of the state-of-the-art that simplifies experimental evaluations, encourages open-source and repeatable research, highlights new research directions relevant to the community, and spurs research progress. Benchmarks are all the more appreciated as they require considerable effort to collect data in a way meaningful for the task at hand. The data must exhibit new challenges yet to be tackled, provide annotations to enable the evaluation, and be memory and runtime-efficient. Also, the evaluation of existing methods is computationally heavy and the analysis of the results, while extremely insightful to the community, requires a serious time investment. With the release of two benchmarks, we propose a significant contribution that will benefit both the research community and the industrial applications related to 3D reconstruction and feature matching.


Researcher: Raul Chametla 


Magnetic coupling triggered by heat release of planetary embryos and its impact on planet migration            

Karolina CPU  Alloc=32400;  Karolina GPU  Alloc=19200        

Recent studies attempting to solve the problem of fast inward migration show that the planet’s orbital dynamics can be governed by torques due to dust-hole and filamentary structure of dust around of the planet and heating torques arising from pebble accretion. However, both mechanisms have been analyzed separately under approach that does not include important physical effects such as: the back-reaction of the dust on gas, the dust turbulent diffusion and the gas thermal ionisation around of the planet where is an increase of temperature. Latter, can produce magnetic coupling in the gas disk which can trigger turbulence around of the planet. Then, the total torque on the planet and, therefore, the migration can change substantially.  Our main aim is to investigate how is affected the planetary migration when we consider the magnetic field of the planet, the heat release of the planetary embryo (thermal torques) and the drag forces between the dust and gas (dust torques).   We will perform three-dimensional multifluid magnetohydrodynamic simulations using the FARGO3D code. Here, we will include ionisation models dependent on gas density and temperature allowing for the first time to self-consistently calculate the coupling of the magnetic field to the gas. Therefore, our models can give a new direction towards the solution of fast inward migration.


Researcher: Aleš Horák          


Slama var - Slavonic Large Foundational Language Model for AI variants       

LUMI-G  Alloc=5000    

The proposed Slama project focuses on building a new foundational language model concentrated on main Slavonic languages (Czech, Slovak, Polish, ...). The project’s primary goal is to explore the performance differences between state-of-the-art pre-trained multilingual models (where English texts represent the majority of training data) and a model tailored specifically to the Slavonic language group. The research will focus on developing generative models whose training data are more balanced in favor of the Slavonic language group rather than English. Therefore it should provide better results when used in AI tools processing mainly Slavonic languages. The resulting foundational model can then be easily applied in a range of AI tasks. This LUMI-G extension of the main Slama project on Karolina extends the training for two other model combinations with varying parameters regarding the number of languages the model is trained on ranging from Czech and Slovak only to a mix of ten European languages.


Researcher: Jana Pavlů           


LaNiSn as energy storage material      

Karolina CPU  Alloc=17153;  Karolina GPU  Alloc=300;  LUMI-C  Alloc=600   

Our modern society based on the utilisation of advanced materials requires thedevelopment of new structural materials that could be used at higher operationaltemperatures or reveal unique properties. Nevertheless, the properties of thosematerials are significantly affected by structural defects such as interfaces and theprocesses taking place on them. Hence, the proposed project aims to understand howthe atoms and structural defects, such as phase boundaries, antisites or vacancies,behave in the SiC-TiSi2 nanocomposites and affect their properties. Unfortunately,these relations very often touch the experimentally unreachable areas. However, theycan be studied through theoretical approaches such as computational modelling; here,the new on-the-fly machine-learned force fields based on the ab initio moleculardynamics. This new approach overcomes some limitations of the present atomisticsimulations (small computational cells and 0 K temperature) and classical moleculardynamics (low precision of interatomic potentials). Hence, we will be able to includemore realistic conditions in our study and save CPU time.


Researcher: Martin Dračínský            


Investigation of nuclear quantum effects in multicomponent solids 

Karolina CPU  Alloc=40000     

Multicomponent molecular solids have numerous applications ranging from drug delivery systems to materials science. For example, many efforts have been recently devoted to the design and investigation of multicomponent pharmaceutical solids, such as salts and cocrystals. The understanding and correct description of intermolecular hydrogen bonds are crucial in these systems. However, experimental distinction between these solid forms is often challenging and common computational approaches may also fail. Nuclear quantum effects, such as hydrogen atom delocalization and tunneling, are very important for accurate predictions of the structure and properties of these systems. This project aims to apply a recently developed computational protocol for accurate predictions of hydrogen-atom positions. The approach will be based on advanced path-integral molecular dynamics simulations that account for the nuclear quantum effects. These computational studies will be complemented with solid-state NMR and X-ray diffraction experiments. The results of this project will help to understand intermolecular interactions in multicomponent molecular solids.


Researcher: Pavel Hobza       


Covalent Dative Bonding, H-Bonding, and Charge Transfer Complexes: Surprising Stability/Instability Trends with Increasing Solvent Polarity    

Barbora CPU  Alloc=4200;  Barbora FAT  Alloc=100;  Karolina CPU  Alloc=45900;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=10100;  LUMI-C  Alloc=3800;  LUMI-G  Alloc=20000    

The project aims to improve our understanding of how solvent polarity influences the stability of covalent dative, and non-covalent complexes using computational methods. The computational analyses will be closely integrated with experimental work conducted by experimentalists utilizing state-of-the-art techniques. This collaboration has the potential to significantly contribute to our comprehension of how solvents impact the stability of complexes, with broad practical applications across diverse fields. Utilizing specific DFT functionals, the project will assess the electronic and optical properties of different complexes, considering both implicit and explicit solvent environments.


Researcher: Petr Marek          


Foundational Czech Language Model Karolina

GPU  Alloc=5900;  LUMI-G  Alloc=6000            

Our aim is to develop a foundational language model tailored for the Czech language, inspired by the advanced capabilities of the Phi models. The initiative aims to transcend the prevalent linguistic limitations by utilizing a robust dataset comprising 350 billion Czech tokens, marking a significant leap towards inclusivity in language technologies. This novel project contemplates the creation of a purely Czech model or a bilingual Czech-English version, with further considerations for integrating other Central-European languages to enhance linguistic diversity.


Researcher: Petr Kurfürst      


Passages of the star through the Galactic jet 

Karolina CPU  Alloc=3600        

For a long time, there has been a broad debate about the reason for the relative lack in the number of red giants/supergiants around the center of our Galaxy. Many theories attempt to explain this phenomenon, for example, by their tidal disruption near the center, collisions with the galactic accretion disk, stellar collisions etc. A promising analytical theory was recently published (Zajacek et al. 2020), the main idea of which is the long-term ablation of the outer envelopes of these stars during many passages through the highly relativistic Galactic jet. We are currently working on a detailed numerical verification of this theory. We determined the initial parameters of the red giant star in detail either using the MESA stellar evolution code or by the radial polytropic structure. Using the advanced and standardized astrophysical code CASTRO, we calculate the 3D models of ablation rate of red giants’ outer layers as they pass through variously parameterized jets. Our goal is also to calculate star-jet interactions at several different distances from the central black hole. Due to the considerable number of stars in the Nuclear Stellar Cluster, a change in the structure of such individual stars can likely cause an overall change in the observable characteristics of the entire central region. We also anticipate the impact of multiple repeated star-jet interactions on an active galactic nucleus (AGN) jet density structure and morphology, the process that can cause its observable variability.


Researcher: Ales Vitek           


Big water clusters III    

Karolina CPU  Alloc=400          

Water clusters change their properties with the increase of its size.  Properties of very small clusters up to 10 molecules are strongly influenced by the geometry of the most stable isomers. At size close to twenty molecules intermolecular distances decreases because of the polarization forces and clusters starts to create cage structures, when not all molecules are on the cluster surface. When the size of clusters grows, up to 50 molecules, the number of stable isomers is so high that particular isomers cease to play an important role and thermodynamic properties of clusters are an average of properties of high number of structural isomers which coexists together. Our work is focused on properties of clusters of 50 water molecules at width intervals of temperature and pressure. This size of water cluster is enough big to be accessible to the real experiment, on the other side, is still accessible for current supercomputers. In the past, a lot of thermodynamics simulations have been performed for water clusters up to 20 molecules at constant-volume conditions, this work brings data of greater sizes and especially investigate pressure-induced structural changes.


Researcher: Frantisek Karlicky          


Many-body physics of van der Waals heterostructures from 2D materials     

Barbora CPU  Alloc=90000;  Barbora FAT  Alloc=1000;  Karolina CPU  Alloc=25000;  Karolina GPU  Alloc=500;  LUMI-C  Alloc=8200

Current needs of applied research on two-dimensional (2D) materials for flexible and ultrathin functional devices require computational predictions and computer-aided design. Stacking of two or more 2D materials leads to van der Waals (vdW) heterostructures, which are promising for engineering of its electronic and optical properties. Delicate physical effects in heterostructures will be studied by costly many-body methods because the usual density functional theory cannot describe the corresponding physics correctly. The project will contribute to the fundamental understanding of vdW heterostructures and boost their experimental research as well as technological applications.


Researcher: Dominik Legut  


Giant negative thermal expansion - Pyrophosphates  

Barbora CPU  Alloc=17900;  Barbora FAT  Alloc=100;  Barbora GPU  Alloc=2000;  Karolina CPU  Alloc=19600;  Karolina FAT  Alloc=100;  Karolina GPU  Alloc=9400;  LUMI-C  Alloc=8300

Pyrophosphates exhibit remarkable versatility across various material science and chemical applications, especially when incorporating transition metal ions into nanocrystalline or amorphous structures. Metal (II) pyrophosphates, in particular, possess diverse biological, magnetic, and chemical properties, rendering them valuable in catalysts, nanobiomaterials, and other applications. This study is part of a broader effort aimed at comprehensively understanding the crystal properties of pyrophosphate compounds, with a focus on divalent ion pyrophosphates due to their significant range of behaviors and crystallographic similarities. The project will systematically investigate the structural, electronic, magnetic, and lattice dynamics properties of iron pyrophosphate, as well as the negative thermal expansion phenomenon in copper pyrophosphate. These theoretical analyses will utilize density functional theory (DFT) and ab initio codes such as VASP and ALAMODE, incorporating anharmonic corrections. The project is structured into three main tasks, with two focusing on the crystal properties of iron pyrophosphate and one on the negative thermal expansion of copper pyrophosphate. The first task involves determining the crystal structure, electronic properties, lattice dynamics, and potential phases of iron pyrophosphate through first-principles theory. The second task centers on investigating the negative thermal expansion phenomenon in copper pyrophosphate. Overall, this research aims to advance our understanding of pyrophosphate compounds and their potential applications in various fields.