2nd Users' Conference of IT4Innovations

Wednesday 7 Nov 2018, 08:00 → 21:00 Europe/Prague

atrium (IT4Innovations)

Branislav Jansik (IT4Innovations), Paolo Nicolini (Czech Technical University in Prague), Tomáš Kozubek (IT4Innovations), Vít Vondrák (IT4Innovations)

2^nd Users' Conference of IT4Innovations will take place on November 7. All of our users as well as research and project partners from various organisations, research institutions and industry are welcome to attend the Conference.

Registration is open until 31 October 2018.

Attendees will discover more about our future upgrade plans, listen to talks given by our prominent users and can engage in discussions during the Users' Council meeting and poster session.

 

Contribution types: Users' talks/Posters

Users' talks

Selected talks by our prominent users will be presented at scheduled times during the whole conference. Each talk is expected to take max. 15 mins (with discussion included).

Posters

Please note that the required poster size is A1 portrait orientation. 

The poster session will take place on November 7 during the dinner.

Agenda AGENDA

Slides

• Beseda Martin: lib4neuro - a library for artificial neural networks calculations in (not only) molecular physics
• Bohm Stanislav: Snailwatch: Did you make it slow?
• Cerny Miroslav: Interface-induced toughening of nitride superlattices
• Fraile Alberto: Jacob’s Ladder: Prime numbers in 2d
• Grajciar Lukas: Hydrolysis of zeolites: new insights from biased ab initio molecular dynamics
• Holy Daniel: Ion binding at lipid bilayer
• Ingr Marek: Binding of hyaluronan and its neutral analog by TSG-6 Link domain
• Irving Benjamin: Predicting Tribological Properties of Low-Dimensional Materials
• Klimes Jiri: Towards an accurate theoretical description of extended systems
• Kolar Michal H.: Structure and Dynamics of a Ribosome-Regulatory Peptide
• Legut Dominik: Ultrafast demagnetization dynamics at M2,3 edges of hcp Co – first-principles
• Meca Ondrej: Workflow for Parallel Processing of Sequential Mesh Databases
• Mokrejs Martin: Assembling diploid genome of Cobitis taenia using Illumina short reads
• Resler Jaroslav: Radiative transfer model for the microscale urban climate model PALM-4U
• Rezac Jan: Efficient quantum-mechanical calculations for computer-aided drug design
• Rhys Taylor: Optically dark hydrogen clouds in the Virgo galaxy cluster
• Rubes Miroslav: Temperature dependence of CO adsorption in H-FER zeolite
• Skiadopoulou Stella: Magnetoelectric excitations in the polar antiferromagnets Ni-based tellurates
• Wunsch Richard: Origin of globular clusters: star formation in the most extreme conditions

Registration and networking

Welcome and IT4Innovations news

Conveners: Dr Branislav Jansik (IT4Innovations), Prof. Vít Vondrak (IT4Innovations, VSB-Technical University of Ostrava)

Keynote I

Legut Dominik: Ultrafast demagnetization dynamics at M2,3 edges of hcp Co – first-principles calculations

1 Ultrafast demagnetization dynamics at M2,3 edges of hcp Co – first-principles calculations

More information on the evolution of the electronic band structure of a ferro-magnetic material during laser-induced demagnetization is needed to understand properly the underlying mechanisms of femtosecond demagnetization dynamics. These can be obtained by combining experiment and modeling. Using ab initio calculations we investigated the effect of a Stoner-like reduction of the exchange splitting, to simulate the previously proposed fast quenching of the exchange splitting, as well as the effect of Heisenberg-like magnon excitations in hcp Co. In our simulations we employed density functional theory calculations using the WIEN2k code [1]. The exchange splitting was reduced to different values employing the methods described in Ref. [2]. To simulate the effects of these mechanisms on the magneto-optical signal, we computed the quasi-static magneto-optical response at the M2,3 edges using the Kubo formula [3]. Influence of non-equilibrium electron populations was determined and computed data were compared to ultrafast high-harmonics measurements of time-, energy-, and angle-resolved magneto-optical effects spectra obtained at the M$_{2,3}$ edges of Co films after optical pumping to induce ultrafast demagnetization [4]. These Co films were on insulating substrates to exclude possible effects of superdiffusive spin currents. For the examined timescales the contribution of transversal spin excitations (i.e., magnon excitation) has been found to be appreciably higher than that of exchange splitting reduction that would originate from longitudinal spin flips [4]. Interestingly, the ratio between these two was not growing with time in favor of transversal excitations, as would be expected from models based on electron-magnon interactions. In recent paper [5] we identify that the changes in xy are uniform throughout the spectrum, to within our experimental precision. This result suggests that, in the regime of strong demagnetization, the ultrafast demagnetization response is primarily dominated by magnon generation. We estimate the contribution of exchange-splitting reduction to the ultrafast demagnetization process to be no more than 25%. This work was done with experimentalists of Refs. [4,5].

Speaker: Dr Dominik Legut (IT4I)

Slides Legut Dominik: Ultrafast demagnetization dynamics at M2,3 edges of hcp Co – first-principles

10:00 Coffee break

Users Talks I.: 6 contributions each 15 mins

2 Efficient quantum-mechanical calculations for computer-aided drug design

We have improved the description of intermolecular interactions in semi-empirical quantum-mechanical (SQM) methods to the point where large molecular systems can be computed with accuracy allowing reliable predictions in the timescale of minutes.[1,2] This opened the way to applications of these methods to the calculations of biomolecules. In particular, we focus on the calculations of protein-ligand interactions that are applicable in computer-aided drug design (CADD). We've developed a general computational protocol based on SQM methods,[3] which could be simplified further to fit specific applications. The first task in a CADD workflow is finding the geometry of the protein-ligand complex. Docking algorithms can be used to generate suitable candidates (poses), but the accuracy of the scoring functions is not sufficient to identify the correct one. Only rescoring the poses with SQM-based protocol allows unambiguous identification of the native pose. In our study of four different and challenging proteins, the SQM-based scoring based on DFTB3 calculations complemented by a COSMO solvation model was the only approach that yielded no false positives.[4,5] Once the geometries of the protein-ligand complexes are determined, the next step is to predict the binding affinity. This is even more challenging as only small changes in the binding free enrgy translate into order of magnitude differences in dissociation constants. Achieving such accuracy for more challenging systems is clearly out of reach of empirical methods. We've used carbonic anhydrase II, a zinc metalloprotein, as a model for which there exist very detailed experimental data. Here, most of the common scoring functions fail, as does a molecular mechanics forcefields. Describing the active site is challenging even for most SQM methods, and only the DFTB approach is able to predict binding activities that correlate sufficiently well with the experiment.[6] The results obtained so far indicate that the SQM-based scoring is accurate enough to predict protein-ligand binding activities, and we are currently applying it to more diverse set of problems.

Speaker: Dr Jan Řezáč (UOCHB AV ČR)

Slides Rezac Jan: Efficient quantum-mechanical calculations for computer-aided drug design

3 lib4neuro - a library for artificial neural networks calculations in (not only) molecular physics

Machine Learning is a rapidly developing scientific discipline with applications in many other fields. We are interested in the applications of Artificial Neural Networks (ANNs) [1] in physical chemistry and, subsequently, in plasma medicine. Within the scope of the computational chemistry, there are several privileged applications, namely potential energy surfaces representations, solving of differential equations and molecular clusters categorization. For such purposes, it is usually necessary to use very large ANNs which tend to be very computationally demanding, up to the point where they cannot be trained or evaluated on a personal computer in a reasonable time. Also, some applications require a specific architecture of the neural network to work correctly. Unfortunately, those problems are not covered by state-of-the-art software, which either runs only sequentially and thus it is not computationally-efficient enough or it offers only specialized applications, i.e. a very limited network architecture. That led us to the idea of developing our own massively-parallel ANN library, which will allow users to specify their own, arbitrary, architecture, so it might be used in a wide range of applications, overcoming time and memory limitations by utilizing large modern infrastructures to the fullest. Our library *lib4neuro* is implemented in a way that enables us to easily implement and to try new training algorithms, which seem to be one of the most significant aspects considering time-consumption. Today, not only usual back-propagation is being used, but also two global optimization methods are implemented now - particle swarm [2] and simulated annealing [3]. All these methods have a great potential to be efficiently parallelized. lib4neuro is going to utilize not only the common MPI+OpenMP combination but also OpenACC to be able to use GPUs and MICs to off-load computationally-heavy parts. Moreover, MPI will be partly substituted by DASH, an innovative way of distributive-level parallelization developed in HLRS institute [4]. With network flexibility being our priority, lib4neuro is able to construct, train and evaluate very atypical network architectures to the point where an arbitrary number of networks with mutually dependent weights can be constructed and then trained simultaneously. This allows us, for example, to formulate differential equations with initial and boundary conditions using neural networks only and solve them efficiently. lib4neuro is also able to construct ANNs with fully arbitrary connections where no interconnection graph completeness is needed neither for training nor evaluation. With such an architecture, the network can be trained much more efficiently. Also, modified versions of network-processing, inspired by Hopfield networks, Kohonen maps, and convolutional networks are being implemented so it will be possible to use lib4neuro for image-classification or detection of patterns in data. Those features will be further optimized to be used efficiently for the above-mentioned molecular cluster categorization. So far, easy-to-use interface for solving differential equations has been fully implemented and its results will be presented.

Speakers: Mr Martin Beseda (IT4Innovations, VSB-TUO), Mr Martin Mrovec (IT4Innovations, VSB-TUO)

Slides Beseda Martin: lib4neuro - a library for artificial neural networks calculations in (not only) molecular physics

4 Interface-induced toughening of nitride superlattices

The ever more demanding application conditions call for new systems with properties over-performing current materials. Multilayer architecture is one of perspective concept to reach these goals. For example, it has been shown that TiN/VN superlattices with bi-layer period in a range of several nanometers yield hardness values higher that either of the constituent materials [1]. This motivated us to aim on detail exploring the interface impact on the ideal strength of nitride multilayers consisting of various combinations of cubic VN, TiN and AlN. Epitaxial superlattices having a common $(100)$ plane as the interface were loaded along the $[100]$ direction. Brittle cleavage along the $(100)$ planes was simulated by splitting the system into two rigid blocks to obtain cleavage parameters such as cleavage energy, critical stress, and critical length. The total energies of such systems were calculated using Density Functional Theory as implemented in the Vienna Ab initio Simulation Package (VASP) [2]. Values of the cleavage energy and stress were found to oscillate in VN layers [3]. These values oscillate within a wide range without any damping. However, only minor alternations are evident in the TiN part. Moreover, with increasing distance from the VN/TiN interface, the oscillations inside TiN tend to level out to the value of single-phased TiN (the horizontal dotted line). Similar effects were also obtained for other systems, namely AlN/TiN and AlN/VN. We have been able to link the oscillatory behavior to interplanar distances and charge accumulation.

Speaker: Dr Miroslav Černý (CEITEC, Brno University of Technology)

Slides Cerny Miroslav: Interface-induced toughening of nitride superlattices

5 Snailwatch: Did you make it slow?

-

Speaker: Dr Stanislav Böhm (IT4Innovations)

Slides Bohm Stanislav: Snailwatch: Did you make it slow?

6 TEMPERATURE DEPENDENCE OF CO ADSORPTION IN H-FER ZEOLITE

The carbon monoxide is a very sensitive IR probe molecule, and thus is quite often used to characterize adsorption sites in various materials [1]. In zeolites, the calculated CO frequency shifts are in good agreement with experimental observations[2]. However, the CO energetics (e.g. isosteric heats) still presents a challenge for computational chemistry due to the dynamical nature of the adsorption process. Experimentally, we observe about 6-7 kJ/mol difference in isosteric heats at 200 K and 300 K, respectively. This difference cannot be explained by using simple thermodynamics models and molecular dynamics (MD) simulations need to be performed. It is straightforward that the accuracy of the employed potential is of utmost importance. We developed an updated DFT/CC [3] methodology to calculate very accurate interaction energies with siliceous framework and Brønsted acid sites. The DFT/CC model was verified with “golden” standard CCSD(T)/CBS on cluster models (up to 4T) and RPA/RSE calculations on several H-FER and siliceous FER structures. The temperature effects were calculated from up to 80 ps MD trajectories for each T-position in H-FER material. Furthermore, the MD simulations indicate significant differences in isosteric heats for different T-positions in H-FER material at 200 K. This effect is partially diminished at 300 K due to the formation of less stable OC-complexes and desorption from the Brønsted site.

Speaker: Dr Miroslav Rubes (IOCB, AS, CR)

Slides Rubes Miroslav: Temperature dependence of CO adsorption in H-FER zeolite

7 Alloy analogy model: Fully relativistic electrical transport with phonons, magnons, and chemical disorder

We will present implementation and applicability of the alloy analogy model (AAM) within fully relativistic *ab initio* computational framework with the coherent potential approximation (CPA). Reliable treatment of temperature-induced disorder in solids is still a challenging task, especially because of numerical expenses or realisticity of approaches. Our tight-binding linear muffin-tin orbital method numerical (TB-LMTO) numerical codes with the CPA-AAM represent a perfect tool for a study of combined contribution of atomic displacements (phonons), spin fluctuations (magnons) and chemical impurities [1,2]. We will show calculations of electronic structure (Bloch spectral function) and transport properties (electrical resistivity, anomalous Hall conductivity) with primary focus on describing spintronic devices influenced by realist room-temperature disorder. Perfect agreement with experimental data is obtained not only for simple metals and binary alloys but also for complex systems like half-Heusler NiMnSb. Discussion of numerical efficiency of our approach will be included and we will also show prediction of experimentally important but hardly-accessible quantities like spin-polarization of electrical current. Our investigation of finite-temperature material properties is also extremely useful for an identification of chemical impurities in specific samples and the developed methods may be used to study of wide range of metalic materials and to efficiently deal with a combination of chemical, magnetic, and temperature-induced disorder.

Speaker: Mr David Wagenknecht (Department of Condensed Matter Physics, Faculty of Mathematics and Physics, Charles University)

12:00 Lunch

Poster session

Keynote II

Wünsch Richard: Origin of globular clusters: star formation in the most extreme conditions

8 Origin of globular clusters: star formation in the most extreme conditions

Globular clusters (GCs) are very massive spherically symmetric objects consisting of $10^5$ - $10^6$ stars concentrated within a radius of several parsecs. There are approximately 150 GCs in our Galaxy, distributed in a spherical galactic halo. Until recently, it has been believed that they are simple systems consisting of a single generation of stars with the same age and chemical composition. However, approximately 15 years ago it has been found that most if not all GCs consist of two or more stellar populations differing photometrically. Moreover, recent spectroscopic observations show that some populations of stars include products of high temperature hydrogen burning, which takes place in certain types of stars. This chemical feature is universal and unique for GCs. Several hypotheses have been suggested, most of them interpreting the individual stellar populations as subsequent generations with later ones formed out of winds or outflows of the previous generation stars, enriched by the H-burning products. However, all the existing hypotheses include important unsolved problems. For instance, it is not clear how the stellar winds with velocities several thousand km/s (much more than the escape velocity from the cluster) can be captured inside the star cluster and participate on the secondary star formation. Another (so called "mass budget") problem is that the observed enriched generations of stars are substantially more massive than the total amount of mass released by the first generation stars. We develop a model that tries to overcome the two aforementioned problems. The subsequent stellar populations in GCs are formed out of winds of massive stars of the first generation. In dense and massive young star clusters, the hot gas originating from wind-wind collisions becomes thermally unstable and forms dense warm gaseous structures that sink into the cluster centre where they accumulate, cool further due to self shielding of the ionising radiation of stars, and form new stars. This can solve the first problem of the wind capturing, and help with the mass budget: if the second generation is more centrally concentrated, a fraction of the first generation can be lost by galactic tidal forces. We study this model using radiation-hydrodynamic simulations describing the interacting winds of massive stars. The numerical implementation is based on MPI-parallel adaptive mesh refinement code FLASH4 (Fryxell+2000) widely used in many astrophysical applications. Effects of ionising radiation are calculated with the algorithm TreeRay combining the method of reverse ray-tracing with tree (Wunsch+2018). This module is developed in our research group as a part of the wide international collaboration SILCC (Walch+2015) aiming to simulate the life cycle of molecular clouds in galaxies with unprecedented physical complexity. The results confirm that the fast stellar winds can be captured by the suggested mechanism and that the secondary star formation occurs in clusters with parameters expected for young GCs. Additionally, we predict shapes of the spectral line H$\alpha$30 that could be detected in young massive star clusters using the ALMA radiotelescope to test the model observationally.

Speaker: Dr Richard Wunsch (Astronomical Institute, Czech Academy of Sciences)

Slides Wunsch Richard: Origin of globular clusters: star formation in the most extreme conditions

Users' Talks II.: 6 contributions each 15 mins

9 Magnetoelectric excitations in the polar antiferromagnets Ni-based tellurates

Magnetoelectric (ME) multiferroics (MFs) are strong candidates for a wide range of novel hybrid technological applications, such as sensing, energy harvesting, photovoltaics, solid-state refrigeration, data storage, magnonics, and spintronics, to name a few.$^1$ A promising route to design efficient future hybrid devices is the use of the dynamical ME effect, where the order parameters of magnetization and polarization are not static, but oscillatory. Elementary excitations called electromagnons may emerge, as “carriers” of this dynamical ME coupling. These spin excitations can contribute to both the magnetic permeability and dielectric permittivity, allowing to promote the modulation of the index of refraction by both static fields and electromagnetic radiation.$^{2,3}$ The understanding of the quantum-level microscopic mechanisms that lead to ME coupling is essential for the engineering of novel single-phase ME MFs. A solid theoretical approach is essential for further insight in the fundamental physics of magnetoelectric phenomena. Here, we propose a combination of first principles calculations of electronic and magnetic structure, as well as lattice-dynamical (phonon) calculations, in combination with experimental spectroscopic investigation for a series of novel ME MF materials. The polar antiferromagnet Ni3TeO6 exhibits non-hysteretic colossal magnetoelectric effect.$^{4,5}$ A collinear antiferromagnetic order appears below $53$ K, giving rise to spin-induced-polarization. Spin and lattice excitations in Ni3TeO6 were studied, by a combination of infrared, Raman and time-domain THz spectroscopies identifying two spin excitations simultaneously present in Raman and THz spectra corresponding to electromagnons.$^6$ Furthermore, for the first time, we prepared single crystals and polycrystalline samples of Ni2MnTeO6, Ni2CoTeO6 and NiCo2TeO6, showing the same polar structure as Ni3TeO6, from room temperature down to $4$ K, with the $\textit{R}3$ space group symmetry. Specifically for Ni2MnTeO6, magnetic and dielectric measurements have indicated an antiferromagnetic phase transition at $T_N≈70$ K, almost $20$ K higher than that of Ni3TeO6.$^7$ Infrared and Raman spectroscopies revealed all $18$ phonon modes predicted by the factor-group analysis for the $\textit{R}3$ structure. THz spectra below $T_N$ reveal magnons, strongly influenced by external magnetic field. In the compounds with Ni substitution by Co, at least two spin excitations were simultaneously seen in Raman and THz spectroscopy, revealing the polar character of the magnons. Density functional theory calculations by the use of VASP code were employed for the family of Ni-based tellurates Ni$_{3-x}$(Co,Mn)$_x$TeO$_6$, in order to resolve the magnetic ground state and estimate the strength of the exchange interactions. In addition, the lattice vibrations were studied by the use of Phonopy code$^8$ based on the Hellman-Feynman forces obtained within the direct method$^9$ using VASP code. A good agreement between theory and experiment was found. Acknowledgements: “IT4Innovations národní superpočítačové centrum - cesta k exascale, CZ.02.1.01/0.0/0.0/16_013/0001791"

Speaker: Dr Stella Skiadopoulou (IT4Innovations Supercomputing Center - VSB-TUO)

Slides Skiadopoulou Stella: Magnetoelectric excitations in the polar antiferromagnets Ni-based tellurates

10 Structure and Dynamics of a Ribosome-Regulatory Peptide

VemP is a peptide found in a marine bacterium, which can live in fresh or salt water. VemP regulates how other proteins are transported out of the bacterium under different extracellular ionic concentrations. The regulation mechanism involves ribosome stalling, which modulates the protein synthesis. Many physiologic consequences have been identified to be stalling-dependent; apart from the protein excretion also the effect of antibiotics, or protein misfolding. In case of VemP, the stalling is caused by a unique α-helical element formed in the exit tunnel near the catalytic center. The helix interferes with several ribosome residues, which are critical for peptide bond formation. We have probed the structure and dynamics of VemP in the stalled ribosome as well as in the aqueous solution. We have proposed, why the stalling efficiency depends on many more nascent peptide amino acids than in other regulatory nascent peptides. While the amino acids of the VemP helix directly inhibit the peptide synthesis, the more distant VemP parts hold the α-helix in place so it is effective. Also, the simulations suggest that the tunnel walls play a critical role in VemP folding. Our simulations include entire ribosome dissolved in explicit aqueous environment under physiological conditions totaling in more than 2 million particles in the simulation box. Such a challenging setup requires large-scale supercomputer facilities to sample conformational space at relevant time scales. Apart from IT4I resources, we have employed two German supercomputers, namely SUPERMUC in Munich and Hazel Hen in Stuttgart via their respective Open calls.

Speaker: Dr Michal H. Kolář (University of Chemistry and Technology, Prague, Czech Republic)

Slides Kolar Michal H.: Structure and Dynamics of a Ribosome-Regulatory Peptide

11 Optically dark hydrogen clouds in the Virgo galaxy cluster

The "missing satellite problem" is a long-standing discrepancy between cosmological models of galaxy formation and observations. The models (e.g. Moore et al. 1999) have generally only used dark matter, which is computationally cheap and thought to dominate the mass on galaxy scales. While extremely successful at reproducing the large-scale distribution of galaxy structures, e.g. the "cosmic web" of filaments and voids, on smaller scales the missing satellite problem and other issues remain difficult to understand. In particular, models generally predict around a factor of ten more dark matter "halos" than observed galaxies. An increasingly popular suggestion is that most of the dark matter halos found in simulations do exist in reality, but not all of them accumulate enough gas for star formation to occur. The conditions under which gas is converted into stars are poorly understood, but most galaxies are known to possess a gas disc significantly more extended than their stellar component (Broeils & Rhee 1999). Davies et al. 2004 suggested that it might be possible for a dark halo to accumulate enough gas to be detectable with radio surveys without triggering star formation. Whereas gaseous features produced in galaxy-galaxy encounters (tidal debris) generally have low spectral line widths, such "dark galaxies" would have higher widths characteristic of rotation. Surveys such as AGES (Auld et al. 2006) use radio telescopes to search for neutral atomic hydrogen (HI) gas independently of optical emission. Taylor et al. 2012 discovered eight HI clouds with particularly intriguing properties : they have high spectral line widths (180 km/s), no optical counterparts, and are relatively isolated. Taylor et al. 2016,2017 showed that such objects are extremely difficult to produce in galaxy-galaxy encounters. While streaming motions can produce high line widths that could be mistaken for rotation, such features are highly transient - implying that tidal debris should be found close to its parent galaxy. In contrast, if the clouds are indeed rotating, this would imply a high dark matter content which would allow them to remain stable and move freely. Burkhart & Loeb 2016 proposed a new model for the dark HI clouds in which their internal pressure is balanced by the confining pressure of a hot, thin, external medium. The internal pressure would have to be dynamic, not thermal, as the line width of the clouds would imply a temperature (>100,000 K) which is too high to sustain HI. We investigated this model on the Salomon cluster using 3D hydrodynamic simulations in FLASH. The HI is modelled as a Gaussian-density sphere with a turbulent velocity field, with properties that match the AGES clouds, embedded in a hot, thin external medium with properties based on X-ray observations. We found (Taylor, Wünsch & Palouš 2018) that such clouds evolve rapidly and resemble the observed clouds for < 100 Myr. This model would require us to have detected eight clouds in a very brief, unusual stage of their evolution, whereas in the dark galaxy hypothesis they are stable and long-lived.

Speaker: Dr Rhys Taylor (Astronomical Institute of the Czech Academy of Sciences)

Slides Rhys Taylor: Optically dark hydrogen clouds in the Virgo galaxy cluster

12 Binding of hyaluronan and its neutral analog by TSG-6 Link domain.

Hyaluronic acid (HA, hyaluronan), an alternating co-polymer of glucuronic acid and N-acetylglucosamine ([4)-β-D-GlcpA-(1->3)-β-D-GlcpNAc-(1->]n), is a major component of extracellular matrix of animal connective tissues. It plays various roles in signaling cascades and is thus involved in inflammation, progression of various diseases including cancer, and wound healing. HA-binding proteins, hyaladherins, which serve as mediators of these processes, are both membrane-bound (CD44, LYVE-1, RHAMM) or soluble (TSG-6). TSG-6 structure is known from numerous NMR experiments that indicate its interaction with HA, heparin and chondroitinsulfate [1]. We applied the method of molecular-dynamics to study the binding of HA oligosaccharides by TSG-6 Link domain. Two binding sites were identified in a solvent constituted by pure water containing only the neutralizing counterions, one of which is identical with the HA-binding site described previously [2], but the other one, so far unknown, partially overlaps with the binding site of heparin [3] and also for chondroitinsulfate [1]. The specificity of the binding sites for HA and charged oligosaccharides in general was investigated by the comparison of HA with its neutral HA analog containing the glucuronic acid residue instead of glucose [4]. This molecule can be bound by both these sites, but the Helmholtz energies of complex dissociation determined by the umbrella sampling method show a remarkably lower stabilization of the analog in the first site, while in the other site the analog binding is even more stable than that of HA. The second binding site is thus less HA-specific and is able of binding various oligosaccharides independently of their negative charge. When NaCl is added to the solvent in the physiological concentration of 0.15 M, the binding interactions of both the ligands change. For hyaluronan the structue of both the binding modes changes considerably, but the ligand remains bound to the receptor. The neutral analog does not bind the first binding site at all, but its binding to the seco. On the contrary, this liagnd is still bound in the second binding mode is kept with only a small change in the stabilization energy. Thus, TSG-6 Link domain binds both hyaluronan and its neutral analog which indicates the possibility of designing artificial ligands with a tunable activity to hyaladherins with a potential pharmaceutical applications.

Speaker: Dr Marek Ingr (Tomas Bata University in Zlín, Faculty of Technology, Department of Physics and Materials Engineering)

Slides Ingr Marek: Binding of hyaluronan and its neutral analog by TSG-6 Link domain

13 Workflow for Parallel Processing of Sequential Mesh Databases

A wide range of engineers from the industrial practice use numerical methods to simulate physical processes. A process of simulation can be divided into two parts -- creation of a numerical model (finite element mesh, definition of boundary conditions, contact interfaces, etc.) and its solution. Since the solution quality is dependent on the numerical model, it is necessary to use an efficient and robust tool for its creation. A typical example are problems in the structural mechanics, where the creation of a high-quality numerical model is one of the most complicated (and important) parts during the simulation process (especially for complex geometries). In general, high-quality models are usually provided by (commercial) tools that produce sequential mesh databases. Hence, some parallel solvers provide a converter of external (sequential) database files into their (parallel) database structure. However, the converter is usually only sequential. It greatly slows down or inhibits the connection between tools that generate sequential data and parallel solvers. It is one of the factors of a low utilization of HPC by the mainstream engineering community. Our motivation is to create a library enabling the connection between tools for the creation of complex engineering models (such as ANSYS, HyperMesh, ANSA, ABAQUS, etc.) along with open source parallel solvers to allow broader usage of HPC by the engineering community. By the straight connection, we get the robust preprocessing together with possibility to connect various highly parallel solvers that are able to solve non-standard problems. We propose an algorithm that is able to load and decompose a sequentially stored mesh database. Thus, it can be used as a parallel converter for various formats. Moreover, scalability tests show that the algorithm can be also used as a direct loader and preprocessor of massively parallel solvers (e.g ESPRESO). The algorithm is composed of several commonly used approaches that together lead to a robust and fast solution. The workflow is composed from several steps. At the beginning a mesh database is read and parsed. Since the parsed data can be randomly scattered among MPI processes, we apply parallel sorting to get a known data distribution. Then, we are ready to assign regions and make a spatial clusterization. This step assures scalability of the next processing that prepares a mesh for a parallel solver.

Speaker: Ondřej Meca (IT4Innovations)

Slides Meca Ondrej: Workflow for Parallel Processing of Sequential Mesh Databases

14 Predicting Tribological Properties of Low-Dimensional Materials

Low-dimensional materials have recently attracted immense interest due to their fascinating physical properties and potential for application in diverse fields such as (opto)electronics, energy harvesting and dry lubrication. Transition metal dichalcogenides (TMDs), of general form MX2 (M = Mo, W; X = S, Se, Te), are posited as being some of the best solid-state lubricants currently available. They exhibit a lamellar structure in which covalently-bonded MX2 layers are held together by weak van der Waals forces, which, together with very low ideal shear strengths (i.e., the maximum load applied parallel to the face of the material that can be resisted prior to the onset of sliding) render them suitable for use in the mitigation of friction. Our extensive density functional calculations highlight the dependence of important nanomechanical properties of TMDs on their chemical composition and bilayer orientation (sliding direction); in particular, our calculations underscore the intrinsic relationship between incommensurate layers and superlubricity.[2] Our latest calculations have focused on TMD-based van der Waals heterostructures (e.g. WS2 sliding on MoS2), with the aim of formalizing the relationship between fundamental quantum chemical parameters of the constituent elements and the nanomechanical properties of the material. Ultimately, we wish to improve the predictive capabilities of in silico methods during the material design process.

Speaker: Dr Irving Benjamin (Czech Technical University in Prague)

Slides Irving Benjamin: Predicting Tribological Properties of Low-Dimensional Materials

15:00 Coffee break

Users' Council

Users' Talks III.: 6 contributions each 15 mins

15 Jacob’s Ladder: Prime numbers in 2d

The distribution of prime numbers is clearly very complex, and it research history has shown that solid proven conclusions about its structure will probably take a long time to come by. Before that happens, the field is in need of ideas able to provide an intuitive and qualitative understanding. It is beyond the scope of this talk to provide quantitative results. Its main purpose is to offer the Mathematics community a simple model that can be used to explore the ideas we point out, which are interconnected with the most important unsolved problems related to the distribution of prime numbers. In addition, technological implications of this study could be derived, since the Ladder scheme presents an excellent framework for developing pseudo-random number routines and randomness tests and algorithms.

Speaker: Dr Fraile Alberto (Department of Control Engineering. Faculty of Electrical Engineering, Czech Technical University in Prague)

Slides Fraile Alberto: Jacob’s Ladder: Prime numbers in 2d

16 Assembling diploid genome of Cobitis taenia using Illumina short reads

.

Speaker: Dr Martin Mokrejš (IT4Innovations, VŠB – Technical University of Ostrava)

Slides Mokrejs Martin: Assembling diploid genome of Cobitis taenia using Illumina short reads

17 Towards an accurate theoretical description of extended systems

The course of many industrial or natural processes is given by interactions between molecules or between molecules and solids, let us mention the catalytic formation of ammonia or the formation of snowflakes to name but a few. If we want to model and understand such processes we need methods that describe the interactions reliably. This is a difficult task as we often need to use description on the level of quantum mechanics. However, solving the equations of quantum mechanics for extended systems is not possible exactly and we need to use approximations. Methods based on perturbation theory, such as the random phase approximation (RPA) or the second order perturbation theory (MP2) are promising for the treatment of extended systems. However, when applying such methods to extended systems one faces an issue that there are several numerical parameters that affect the results, such as the real space grid density. Converging with these parameters increases substantially the overall cost of the calculations. However, converged properties are needed if we want to understand the accuracy of a given theoretical scheme. We have calculated such highly converged binding energies of molecular solids using the RPA approach. We were able to show that it provides much better and more consistent accuracy compared to the state-of-the-art DFT methods [1,2]. As it is difficult to assess when a convergence has been reached, we have used an alternative scheme to obtain the binding energies of molecular solids. This is called the many-body expansion (MBE) and it assembles the binding energy from contributions of individual dimers, trimers, etc. of molecules in the crystal. However, this doesn't come without issues either, again there are several numerical parameters that need to be controlled and that make the convergence difficult. By calculating binding energies of four molecular crystals of small molecules within periodic boundary conditions and with the MBE we were able to identify some of the issues related to both approaches. Moreover, we were able to identify general guidelines to follow when performing calculations of binding energies of molecular solids [3].

Speaker: Jiri Klimes (Faculty of Mathematics and Physics, Charles University)

Slides Klimes Jiri: Towards an accurate theoretical description of extended systems

18 Hydrolysis of zeolites: new insights from biased ab initio molecular dynamics.

The silicates, based on a tetrahedron-shaped anionic (SiO_4)^(4-) group, is one of the most abundant classes of compounds on Earth being heavily used in industry. However, the silicates, and in particular porous silicates doped with aluminium (zeolites), are often being degraded and eventually destroyed upon prolonged exposure to humidity [1]. Besides the negative effects of prolonged exposure to water, it is possible to direct hydrolysis, which can lead to synthesis of new materials such as hierarchical layered silicates with micro- and meso-porosity [2]. Therefore, with a goal to hinder or direct the silicate hydrolysis, numerous first principle studies have focused on understanding the hydrolysis mechanism. However, they were limited to static calculations not properly accounting for water dynamics in a fully solvated material at realistic conditions [3]. In this study, we employed a (biased) ab initio molecular dynamics to accurately incorporate entropic effects and most importantly to allow for explicit treatment of water molecules in interaction with a zeolites, since water is not just the solvent but also the reactant. New hydrolysis paths, specific for solvated material, have been found, which include Grotthuss proton-hopping mechanism or water splitting on the T-O-T (T=Si, Ge, Al) bonds with protons solvated in the surrounding water (see Figure). These new mechanisms have lower reaction barriers than the most stable mechanism proposed previously using limited solvation model [2]. In addition, the new mechanisms can help understand the feasibility of directed partial hydrolysis on the basis of different accessibility of (TO_4)^(4-) tetrahedra and germanium clustering. Although, the current study considered a specific porous silicate, the UTL zeolite, we expect the new mechanisms to be of importance for other silicates as well.

Speaker: Dr Lukáš Grajciar (Department of Physical and Macromolecular Chemistry, Charles University in Prague)

Slides Grajciar Lukas: Hydrolysis of zeolites: new insights from biased ab initio molecular dynamics

19 Radiative transfer model for the microscale urban climate model PALM-4U

The radiative transfer model represents one of key components of the urban climate model. The scalability of the first version described in the paper [Resler et al 2017] was lower than the scalability of the rest of the PALM-4U model. It allowed to process domains up to extent of a few hundreds of gridpoints in each dimension while the required domains reached the extent of thousands gridpoints. We successively developed new versions 2.0, 2.5 and 3.0 of the RTM model which improved the performance and scalability in orders. It was achieved using different algorithms and the discretization of the problem and by optimization of the MPI communications. The new version is able to process domains in extent of order thousands gridpoints in each direction with very little overhead to the PALM model core and with similar accuracy to the model v. 1.0. We demonstrate the performance and results of the model on pilot testing domains from Prague and Berlin processed inside the project UrbiPragensi.

Speaker: Dr Jaroslav Resler (Institute of Computer Science The Czech Academy of Sciences, Prague)

Slides Resler Jaroslav: Resler Jaroslav: Radiative transfer model for the microscale urban climate model PALM-4U

20 Ion binding at lipid bilayers

Cells interact with the environment through plasma membranes. Because of their complexity simpler lipid bilayers are use as models to study their properties. An important component which modulates plasma membranes is the calcium ion which is for example involved in a number of processes, such as neurotransmitter release and membrane fusion. Calcium is also known to modulate the physical properties of plasma membranes and to accelerate the process of membrane fusion. Previous research suggests that calcium binds more strongly to positively curved bilayers than to flat ones. We build on this finding and extend the investigations by using an improved lipid force field for POPC in our simulations. Using this ECC-POPC force filed[1], which provides a better description of ion-lipid interaction, we characterize the influence that the lipid bilayer shape has on the binding of the calcium ions to the bilayer. To describe the way calcium binds to the bilayer, we first carry out molecular dynamics simulations of flat POPC bilayers at different calcium concentrations and analyze the calcium density distributions in a manner which can be extended to curved systems. After adequately describing the flat bilayer system, we apply the same methodology to bilayers of varying curvatures. We also compare the results of this approach to the calcium density profiles calculated using the instantaneous liquid interface approach. The results of this project will allow us to comment on the mechanisms the cells use to regulate calcium concentration related to curvature of the membrane.

Speaker: Mr Daniel Holý (Institute of Organic Chemsitry and Biochemistry, Czech Academy of Sciences`)

Slides Holy Daniel: Ion binding at lipid bilayer

Conference Dinner & Networking

Poster session