Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf

Short talk about alpaka, SIMD in C++, Vc, VecCore, LLAMA and std::simd.

Wissenschaftliche Experimente nutzen eine große Bandbreite an verschiedenen Software-Werkzeugen in den verschiedenen Phasen des Projektes von der Proposal-Einreichung über die Datennahme bis zur finalen Publikation. Eine große Herausforderung für Wissenschaftseinrichtungen ist es, WissenschaftlerInnen für die Dokumentation der genutzten Werkzeuge in allen Phasen des Forschungsprojektes zusätzliche Metadaten gemäß der FAIR-Prinzipien zur Verfügung zu stellen. Das Ziel der HELmholtz ScIentific Project WORkflow PlaTform (HELIPORT) ist es daher den kompletten Lebenszyklus eines wissenschaftlichen Projekts zu registrieren und die zugehörigen Programme und Systeme miteinander zu verknüpfen. Die maschinenlesbare Dokumentation aller im jeweiligen Forschungsprojekt durchgeführten Arbeitsschritte gemeinsam mit den dazugehörigen Metadaten macht jeden Arbeitsschritt transparent, verständlich und zitierbar und trägt somit zur Einhaltung guter wissenschaftlicher Praxis bei.

The dynamic properties of a warm dense matter (WDM) system is strongly influenced by electron dynamics. Therefore, there is a need for computational methods which explicitly treat the electron dynamics, while still scalable and independent on any equilibrium assumptions, for the ability to model ongoing experimental efforts. We present an extension of the wave packet molecular dynamic scheme, where the electron state has an arbitrary Gaussian form, increasing the degree of freedom in the model and allowing for anisotropy in collisional dynamics and molecular bonds. The model employs a generalised scheme for the computation of short-range interactions, Ewald summation and Pauli interactions. The highly parallelised molecular dynamics framework LAMMPS has been utilized for the implementation of the model, with the intention to allow investigations of larger-scale systems. We present both ground state as well as the thermodynamic properties of the model, specialising in hydrogen systems.

We investigate the emergence of electronic excitations from the inhomogeneous electronic structure at warm dense matter parameters based on first-principles calculations. The emerging modes are controlled by the imposed perturbation amplitude. They include satellite signals around the standard plasmon feature, transformation of plasmons to optical modes, and double-plasmon modes. These modes exhibit a pronounced dependence on the temperature. This makes them potentially invaluable for the diagnostics of plasma parameters in the warm dense matter regime. We demonstrate that these modes can be probed with present experimental techniques.

Warm dense matter is a highly active research area both at the frontier and interface of material science and plasma physics. We assess the performance of commonly used exchange-correlation (XC) approximation (LDA, PBE, PBEsol, and AM05) in the spin-polarized inhomogeneous electron gas under warm dense conditions based on exact path-integral quantum Monte-Carlo calculations. This extends our recent analysis on the relevance of inhomogeneities in the spin-unpolarized warm dense electron gas [Z.~Moldabekov et al., J. Chem. Phys. 155, 124116 (2021)]. We demonstrate that the predictive accuracy of these XC functionals deteriorates with (1) a decrease in density (corresponding to an increase in the inter-electronic correlation strength) and (2) an increase of the characteristic wave number of the density perturbation. We provide recommendations for the applicability of the considered XC functionals at conditions typical for warm dense matter. Furthermore, we hint at future possibilities for constructing more accurate XC functionals under these conditions.

Correlated many-fermion systems emerge in a broad range of phenomena in warm dense matter, plasmonics, and ultracold atoms. Quantum hydrodynamics (QHD) complements first-principles methods for many-fermion systems at larger scales. We illustrate the failure of the standard Bohm potential central to QHD for strong perturbations when the density perturbation is larger than about 10−3 of the mean density. We then extend QHD to this regime via the many-fermion Bohm potential from first-principles. This may lead to more accurate QHD simulations beyond their common application domain in the presence of strong perturbations at scales unattainable with first-principles methods.

Warm dense matter (WDM)–an exotic, highly compressed state of matter between solid and plasma phases is of high current interest, in particular for astrophysics and inertial confinement fusion. For the latter, in particular the propagation of compression shocks is crucial. The main unknown in the shock propagation in WDM is the behavior of the electrons since they are governed by correlations, quantum and spin effects that need to be accounted for simultaneously. Here we describe the shock dynamics of the warm dense electron gas using a quantum hydrodynamic model. From the numerical hydrodynamic simulations we observe that the quantum Bohm pressure induces shear force which weakens the formation and strength of the shock.
In addition, the Bohm pressure induces an electron density response which takes the form of oscillations. This is confirmed by the theoretical analysis of the early stage of the shock formation. Our theoretical and numerical analysis allows us to identify characteristic dimensionless shock propagation parameters at which the effect of the Bohm force is important.

The extent of data collection in computational history is often delimited by the specific
hypotheses that drive the research in question. Such a parsimonious approach is completely
logical and in many cases sufficient; moreover, there is no such thing as “total” data
collection, because the data is to a degree in the eye of the beholder. At the same time,
however, historical research has a tried and tested tradition of more source-driven research,
where the close reading of sources often drives the direction of study more than the testing
of hypotheses. In this paper, we present our experience of developing a thorough data model
and user interface for the collection of structured data from medieval inquisitorial registers,
focusing mainly on the social, spatial, and semantic relations between historical actors,
groups, places, physical objects, concepts, and events. We undertook this as part of a
project that seeks to provide a networked perspective on religious dissent and its repression
in medieval Europe (Dissident Networks Project / DISSINET, https://dissinet.cz). In this
paper, we would like to discuss our data model and data collection practices as well as to
open the data model to suggestions on how it can be mapped upon existing standards in
order to enhance its interoperability.
From our experience, we derive several proposals which could be of interest to historians
who, on the continuous scale between hypothesis-driven and source-driven data collection,
lean somewhat more towards the latter. Our point of departure is that a data model for
source-driven data collection should allow as much relational complexity as the natural
language of our sources does. Our approach is not completely new from a conceptual or
technical point of view; it is based on statements composed of predicates and actants
(subjects and objects), and therefore close to the idea of “semantic triples”. However, we dig
quite deeply into the language of our sources to propose a way of recording its minutiae,
allowing for fuzziness and uncertainty, modifiers (e.g., adjectives, adverbs), temporal and
spatial relations (incl. relative chronology), modality (negative, question, possibility,
desirability etc.), and give specific meaning to the different actant positions (subject, objects)
of each verb for analytical purposes. We thus join the minority strand in current
computational history which departs from the idea of factoids, decided upon at the moment of
data collection, and rather translate the source into structured data quite extensively to later
adapt the specific data projections to the needs of particular research questions and
hypotheses we want to test.
This approach allows us to preserve the semantic structure and detail of the sources, while
also producing highly structured data suited to various kinds of computational analyses such
as social network analysis, socio-semantic network analysis, geospatial data analysis,
various regression models, etc. Data collection (or rather, production), in our view, thus
already amounts to computational modelling – i.e., in the first instance, we model the source
itself, preserving its original language, its vagaries, and its complexity, and only at a later
stage various research problems in our focus.
The talk does not focus so much on technical solutions (e.g., review of available data
collection environments) or standards. Rather, we explore conceptual issues and a practical
workflow that we believe can be inspirational for computational history more generally.
Nevertheless, the paper includes a brief demonstration of the inkVisitor software, a webbased open-source data collection environment currently under development in our project.

Background and purpose

Major differences exist among proton therapy (PT) centres regarding PT delivery in adult cancer patient. To obtain insight into current practice in Europe, we performed a survey among European PT centres.
Materials and methods

We designed electronic questionnaires for eight tumour sites, focusing on four main topics: 1) indications and patient selection methods; 2) reimbursement; 3) on-going or planned studies, 4) annual number of patients treated with PT.
Results

Of 22 centres, 19 (86%) responded. In total, 4,233 adult patients are currently treated across Europe annually, of which 46% consists of patients with central nervous system tumours (CNS), 15% head and neck cancer (HNC), 15% prostate, 9% breast, 5% lung, 5% gastrointestinal, 4% lymphoma, 0.3% gynaecological cancers. CNS are treated in all participating centres (n=19) using PT, HNC in 16 centres, lymphoma in 10 centres, gastrointestinal in 10 centres, breast in 7 centres, prostate in 6 centres, lung in 6 centres, and gynaecological cancers in 3 centres. Reimbursement is provided by national health care systems for the majority of commonly treated tumour sites. Approximately 74% of centres enrol patients for prospective data registration programs. Phase II-III trials are less frequent, due to reimbursement and funding problems. Reasons for not treating certain tumour types with PT are lack of evidence (30%), reimbursement issues (29%) and/or technical limitations (20%).
Conclusion

Across European PT centres, CNS tumours and HNC are the most frequently treated tumour types. Most centres use indication protocols. Lack of evidence for PT and reimbursement issues are the most reported reasons for not treating specific tumour types with PT.

For the best possible limit of detection of any thin film‐based magnetic field sensor, the functional magnetic film properties are an essential parameter. For sensors based on magnetostrictive layers, the chemical composition, morphology and intrinsic stresses of the layer have to be controlled during film deposition to further control magnetic influences such as crystallographic effects, pinning effects and stress anisotropies. For the application in magnetic surface acoustic wave sensors, the magnetostrictive layers are deposited on rotated piezoelectric single crystal substrates. The thermomechanical properties of quartz can lead to undesirable layer stresses and associated magnetic anisotropies if the temperature increases during deposition. With this in mind, we compare amorphous, magnetostrictive FeCoSiB films prepared by RF and DC magnetron sputter deposition. The chemical, structural and magnetic properties determined by elastic recoil detection, X‐ray diffraction, and magneto‐optical magnetometry and magnetic domain analysis are correlated with the resulting surface acoustic wave sensor properties such as phase noise level and limit of detection. To confirm the material properties, SAW sensors with magnetostrictive layers deposited with RF and DC deposition have been prepared and characterized, showing comparable detection limits below 200 pT/Hz(^1/2) at 10 Hz. The main benefit of the DC deposition is achieving higher deposition rates while maintaining similar low substrate temperatures.

Leaching using deep eutectic solvent (DES) can be an alternative solution for environmentally friendly leaching. Two choline chloride (ChCl) based DESs are investigated using two different oxidants for each DES, which are ChCl-ethylene glycol and ChCl-lactic acid with iodine (I2) or hydrogen peroxide (H2O2) as the oxidants. This study attempts a new method of determining leaching dissolution rate by using 3D image analysis from single particle leaching experiment to optimize leaching process through time. X-ray computed tomography (CT) is used to scan the particle to get 3D images, volume and surface area of the particle before and after the leaching. Gold, silver and telluride minerals are used as the target minerals, whereas galena, chalcopyrite and pyrite are used as the gangue minerals. The dissolution rate of these minerals calculated from the 3D image processing are then combined with mineral liberation analysis (MLA) data of a sulphide concentrate. This MLA data provides mineralogical characteristic along with scanning electron microscopy (SEM)-based image to be used in leaching simulation to get mineral recoveries. This method shows that selectivity of each target and gangue mineral dissolution in different solution can be observed and compared. The dissolution rate can also be applied in predicting leaching using different characteristic of ores or concentrates. Batch leaching experiments using all the leaching solutions are also done to compare the leaching simulation result.

Air quality regulations have reduced emissions of pollutants in the U.S., but many prognostic studies suggest that future air quality might be degraded by global climate change. The simulated climate by various climate models shows a large variation in the future decades, and it is important to account for such variations to study future air quality. A typical approach to study future air quality projections uses three-dimensional (3D) Eulerian models, but these models are computationally too expensive to perform an ensemble of long-term simulations for various climate projections. Therefore, we have developed a machine learning (ML) based air quality model to study, in an efficient way, how future air quality might be influenced by climate change. Our ML model uses two-phase random forest to predict the O3 and PM2.5 concentrations with training datasets of key meteorological information and air quality pollutant emissions. To evaluate the model performance, we used the input datasets for the U.S. Environmental Protection Agent (EPA) the Community Multiscale Air Quality Modeling System (CMAQ) simulations and compared our model predictions against the CMAQ output as a benchmark. The 1995 – 1997 data were used to train the ML model; 2025 – 2035 data were used to evaluate it. The ML model is well performed for hourly O3 predictions over the whole domain in four selected months (January, February, July, and August), and the R2 values are in 0.5 – 0.7, the normalized mean bias (NMB) values are within ±3%, the overall normalized mean error (NME) values are below 20%. Compared to CMAQ, our ML model tends to overpredict the O3 in the Southeast U.S and California, and underpredict in the Central U.S, and the NMB values computed for each grid are generally within ±10%. Predicting PM2.5 is more challenging than predicting O3, but our ML model performance is still acceptable. The overall R2 values of PM2.5 predictions are in 0.4 – 0.6, and the NMB values are within ±6%, but the NME can be up to 60%. The NMB in each grid is within ±30%. There is no clear trend for the regional variation of ML model performance for PM2.5. Our ML model performs better for summer PM2.5 (July and August) than winter (January and February): NME is 10% - 20% lower in summer. While the model performs better in winter than summer with about 10% lower NME for O3. Our ML model with GPU acceleration runs less than one hour using a single GPU processor to predict 11-year one-month (total 11 months) simulations. It uses significantly less computing resources compared to the 3D models, like CMAQ, while it results in comparable predictability to CMAQ. It shows that our ML model a reliable and efficient tool to assess the air quality under various climate change scenarios.

The so called down cycling effect denotes the unwanted mixing of different materials in the recycling stream resulting in materials with lower quality and functionality than the single parts that where introduced to the recycling process. In a project dealing with different analytical methods a sensor strategy for a desired sorting technique was tested. The methods that were applied on different alloys including steel samples, brass and aluminium included: X-ray fluorescence techniques (handheld XRF, WDXRF), laser induced breakdown spectroscopy (LIBS), and spark-OES but also neutron activation.
Spark-OES was used as benchmark, which was verified by using certified reference materials, suitable for the alloys we investigated. The actual tests were done on nine different, commercially available test alloys with known compositions (given ranges of elemental concentrations, though). Initially, they were tested for homogeneity using spatially resolved EDXRF measurements. Deviations to the mean composition were generally lower than 0.2 wt.%. Nevertheless few inhomogeneities were found including impurities and general inhomogenous areas which are attributed to quality issues during manufacturing (probably with no effect on the desired function).
Summarizing the results of this comparative study, it turned out that for specific elements-matrix combinations, like e.g. Si in steels and Cu in Al alloys, LIBS show more accurate results, whereas the XRF techniques exhibit better performances for the main metal quantification due to its direct quantification of the main element (Seidel et al. 2021). Besides the main elements, we tested for Al, Fe, Cu, Si, Mn, Mg, Cr, Ni, Zn depending on the alloy type.
The results were used to simulate a first approximation on a possible sorting result of real scrap metal samples derived from a magnetic separated fraction, on which we applied LIBS, pXRF and WDXRF. These preliminary calculation considered the analytical deviation, different chemical compositions and the elements of interest vs. analytical method.

References:

Seidel, P., Ebert, D., Schinke, R., Möckel, R., Raatz, S., Chao, M., Niederschlag, E., Kreschel, T., Gloaguen, R., Renno, A. D. (2021). Comparison of Elemental Analysis Techniques for the Characterization of Commercial Alloys. Metals, 11(5), 736.

To close the loop of the respective material demands, the role of battery recycling is steadily increasing, especially in the light of changing mobility. The analytical requirements for the battery material are challenging, since they are complex and heterogeneous secondary materials. They contain Li in different compounds, carbon in the form of graphite (“black mass”), but also metals like Mn, Fe, Cu, Al, Co, Ni etc. partly in several oxidation states (pure metal foils vs. different compounds). All these components are also highly affected by the recycling stages (e.g. pyrolysis, leaching, ect.). On the other hand, with the main aim of an effective recycling the demand for as detailed as possible analytical data is likewise high. Generally, there is no single method available for tackling all the analytical issues and a combination of methods is inevitable. In different recycling projects, we developed methods for scanning electron microscope (SEM) techniques (in our case MLA mineral liberation analyser and TIMA Tescan integrated mineral analyzer), quantitative XRD (X-ray diffraction) and XRF (X-ray fluorescence) with support of ICP-OES.
While SEM techniques are essential to provide information on a particle base (incl. chemical composition, grain size, liberation etc.) the other techniques provide pure bulk chemical analytics. An approach – with slight amendments – known from the so called automated mineralogy applied to the materials revealed very useful results for understanding processing parameters, i.e. for the beneficiation of black mass in high quality (Vanderbruggen et al. 2021). On the other hand by simply adding an internal standard (we chose ZrO2) it is possible to develop a quick and easy method for XRF analysis where the sum of “invisible” elements (e.g. C, Li, O, F) can be determined easily with accurate quantification of the other elements, mainly focusing on the metal contents. Lithium containing compounds can be detected by XRD, but the methods have some drawbacks when it comes e.g. to the pure metal contents.
Therefore, a combination of these different approaches helps to increase the analytical precision in order to develop efficient recycling strategies.
References:
Vanderbruggen, A., Gugala, E., Blannin, R., Bachmann, K., Serna-Guerrero, R., & Rudolph, M. (2021). Automated mineralogy as a novel approach for the compositional and textural characterization of spent lithium-ion batteries. Minerals Engineering, 169, 106924.

Chemical transport models (CTMs) are widely used for air quality forecasts but require heavy computational burden and often suffer from a systematic bias that leads to missed poor air pollution events. In this research, we developed a machine learning (ML) modeling framework to provide O3 and PM2.5 forecasts at the monitoring sites throughout the Pacific Northwest (PNW). We used the historical archives from the Weather Research and Forecasting (WRF) meteorological model forecasts, and Air Quality System (AQS) observation datasets of O3 and PM2.5 to build a reliable forecasting system that consists of two ML models, ML1 and ML2: ML1 uses the random forest (RF) classifier and multiple linear regression (MLR) models, and ML2 uses a two-phase RF regression model with best-fit weighting factors. The 10-time, 10-fold cross-validation analysis is used to evaluate our ML forecasting system. Compared to the air quality forecasts based on a CTM, ML1 improves forecast skill for mid-to-high O3 events, which captures 77% more unhealthy events, while ML2 improves forecast skill for low-to-mid O3 events (R2 = 0.78). For PM2.5, our ML model performs well regardless of a season and captures 70% more summer and 30% more winter high-PM2.5 events while the CTM shows systematic biases during summer and winter: 31% underprediction during summer and 5.1% overprediction during winter. The ML modeling framework is now used as an operational forecast of O3 and PM2.5 at the monitoring sites in the PNW.

openPMD is a community meta-data standard for self-describing, FAIR and compatible exchange of scientific data on top of modern file formats.

The openPMD-api has successfully established an ADIOS2 implementation for efficient and scalable I/O and recently stabilized support for streaming features of ADIOS2. This talk evaluates two use cases for the SST engine in scientific simulation workflows with an I/O stack built by openPMD-api and ADIOS2. The first use case aims for an ad-hoc implementation of asynchronous IO, while the second use case discusses a loosely coupled simulation-analysis workflow with massive data exchange, both evaluated on Summit using the Infiniband network (RDMA).

Digitalisation is an important tool to develop and improve industrial processes. The collection and storage of detailed information have to be especially implemented with regard to modern recycling technologies in the field of lithium-ion battery (LIB) processing. Here, real-time recording should enable quick adjustments of the processing parameters and thus improve the process results and product qualities.
The DIGISORT project is part of the greenBatt research cluster which examines heuristically LIB recycling. The project itself focusses on the digitalisation of the mechanical separation via air flow sorting by implementing new sensor technology. Through separation, the components and valuables of lithium-ion batteries have to be transferred into concentrates of high qualities and high recovery rates. In detail, the feed material in air flow sorting is a mixture of electrode foils which is contaminated with other material fractions. Depending on the types of batteries processed and the previous processing steps and parameters, the properties of the mixture vary in size, mass, degree of liberation and composition. As a result, stationary conditions rarely apply in battery recycling and challenge the following separation processes.
A unique sensor technology is developed in DIGISORT recording the most important properties of the material in-line and on-line considering individually each fragment (particle) discretely. Image data with spectroscopic or hyperspectral information will be combined on particle level. Due to the enormous depth of information, these data sets will be structured as information vector for each particle. The control of the air flow sorting using the information vector is intended to increase the separation efficiency and thus the product quality and yield of the aluminium and copper concentrates.
The present contribution introduces the broadness of material characteristics in LIB recycling. Different model mixtures as well as real recycling materials are characterised with regard to their size, shape and composition using a SOPAT camera. These off-line generated data serve as a comparison for the sensor technology to be developed in DIGISORT. In addition to characterisation of the feed material, the camera will be used to describe the separation products and process efficiency.

The Electron Linac for beams with high Brilliance and low Emittance (ELBE) at Helmholtz-Zentrum Dresden-Rossendorf (HZDR) is in operation since 2001. It is operated using the SCADA system WinCC by Siemens. The majority of ELBE systems is connected to WinCC via industrial Ethernet and proprietary S7 communication. However, in recent years new subsystems had to be integrated into the existing infrastructure, which do not provide S7 communication interfaces. Instead, OPC UA has been chosen for system integration. We will show how we use OPC UA as a common communication layer between industrial and scientific instruments as well as proprietary and open source control system software. For example, OPC UA support has been implemented for the ChimeraTK framework developed at DESY. ChimeraTK is used at ELBE e.g. for integrating MicroTCA.4 based subsystems like the digital LLRF system. Furthermore, we are developing a machine data interface for ELBE users. In combination with a certification authority, which hands out user certificates for data access, external users can gain read and write access to different ELBE subsystem data provided by a single OPC UA server.

This paper aims to create a transition path from file-based IO to streaming-based workflows for scientific applications in an HPC environment. By using the openPMP-api, traditional workflows limited by filesystem bottlenecks can be overcome and flexibly extended for in situ analysis.
The openPMD-api is a library for the description of scientific data according to the Open Standard for Particle-Mesh Data (openPMD).
Its approach towards recent challenges posed by hardware heterogeneity lies in the decoupling of data description in domain sciences, such as plasma physics simulations, from concrete implementations in hardware and IO.
The streaming backend is provided by the ADIOS2 framework, developed at Oak Ridge National Laboratory.
This paper surveys two openPMD-based loosely-coupled setups to demonstrate flexible applicability and to evaluate performance.
In loose coupling, as opposed to tight coupling, two (or more) applications are executed separately, e.g. in individual MPI contexts, yet cooperate by exchanging data. This way, a streaming-based workflow allows for standalone codes instead of tightly-coupled plugins, using a unified streaming-aware API and leveraging high-speed communication infrastructure available in modern compute clusters for massive data exchange.
We determine new challenges in resource allocation and in the need of strategies for a flexible data distribution, demonstrating their influence on efficiency and scaling on the Summit compute system.
The presented setups show the potential for a more flexible use of compute resources brought by streaming IO as well as the ability to increase throughput by avoiding filesystem bottlenecks.

The use and development of lithium-ion batteries (LIBs) are promoting the technological transformation of individual mobility, consumer electronics and electric energy storage. At their end of life, the complex compounds are disposed by different recycling technologies with defined secondary raw material production. The applied depollution temperatures of the process routes influence not only the recycling efficiency but also the process expenditure, design, medium and costs. Different pretreatment strategies in terms of dismantling depth and depollution temperature are existing. Furthermore, manual and mechanical methods for cell opening are distinguished, which together with the depollution leads to a respective organic solvent evaporation. In this contribution to LIB recycling, the influence of different dismantling depths, achieved by manual cell opening, on the thermal depollution of the LIB cells regarding the mass difference originating by organic solvent evaporation are quantified, in order to determine cell and equipment properties for a safe cell opening. As a result, combinations of thermal depollution and manual cell opening are discussed regarding technical and economic feasibility. The process medium and equipment properties for a safe cell opening are determined. Furthermore, recommendations for future industrial LIB waste management are presented.

In light of the flow regime transition (bubbly flow to stratified) phenomenon occurring in the vane-type gas-liquid separator, a morphology-adaptive hybrid model featuring with multi fields is applied to predict the complex dynamic behaviors of interfacial structures in it. By means of this model, the coexistence of dispersed bubbles and continuous gas core is able to be described simultaneously with the same set of equations. One key issue related to the simulation of the multi-scale interfacial structure in the separator is the morphology transition criterion, more specifically, the formation of continuous gas out from dispersed gas. A preliminary transition criterion combining coalescence transfer and absorption transfer based on local volume fraction of the dispersed gas and the continuous gas has been implemented in the multi-field Eulerian two-fluid framework. Several simple swirling flow cases are adopted to test its performance and evaluate the effects of key model parameters. The capability of the new hybrid model representing both bubbly flow and continuous gas core as well as their transition is verified for a vane-type separator under two circumstances, namely, the stable gas core and the unstable gas core. For both cases, the numerical results present high similarities with the experimental results, which suggest that the hybrid model is capable of capturing the complex gas-core behavior in the vane-type gas liquid separator and serving as a reliable numerical predicting tool.

Invertible neural networks are a recent technique in machine learning promising neural network architectures that can be run in forward and reverse mode. In this paper, we will be introducing invertible surrogate models that approximate complex forward simulation of the physics involved in laser plasma accelerators: iLWFA. The bijective design of the surrogate model also provides all means for reconstruction of experimentally acquired diagnostics. The quality of our invertible laser wakefield acceleration network will be verified on a large set of numerical LWFA simulations.

Chemical transport models (CTMs) are widely used for air quality forecasts, but these models require large computational resources and often suffer from a systematic bias that leads to missed poor air pollution events. For example, a CTM-based operational forecasting system for air quality over the Pacific Northwest, called AIRPACT, uses over 100 processors for several hours to provide 48-h forecasts daily, but struggles to capture unhealthy O3 episodes during the summer and early fall, especially over Kennewick, WA. This research developed machine learning (ML) based O3 forecasts for Kennewick, WA to demonstrate an improved forecast capability. We used the 2017–2020 simulated meteorology and O3 observation data from Kennewick as training datasets. The meteorology datasets are from the Weather Research and Forecasting (WRF) meteorological model forecasts produced daily by the University of Washington. Our ozone forecasting system consists of two ML models, ML1 and ML2, to improve predictability: ML1 uses the random forest (RF) classifier and multiple linear regression (MLR) models, and ML2 uses a two-phase RF regression model with best-fit weighting factors. To avoid overfitting, we evaluate the ML forecasting system with the 10-time, 10-fold, and walk-forward cross-validation analysis. Compared to AIRPACT, ML1 improved forecast skill for high-O3 events and captured 5 out of 10 unhealthy O3 events, while AIRPACT and ML2 missed all the unhealthy events. ML2 showed better forecast skill for less elevated-O3 events. Based on this result, we set up our ML modeling framework to use ML1 for high-O3 events and ML2 for less elevated O3 events. Since May 2019, the ML modeling framework has been used to produce daily 72-h O3 forecasts and has provided forecasts via the web for clean air agency and public use: http://ozonematters.com/. Compared to the testing period, the operational forecasting period has not had unhealthy O3 events. Nevertheless, the ML modeling framework demonstrated a reliable forecasting capability at a selected location with much less computational resources. The ML system uses a single processor for minutes compared to the CTM-based forecasting system using more than 100 processors for hours.

Transglutaminase 2 (TG2) is a protein expressed in many tissues that exerts numerous, some-times contradictory intra- and extracellular functions under both physiological and pathophysi-ological conditions. In the context of tumor progression, it has been found to be involved in cell adhesion, DNA repair mechanisms, the induction of apoptosis, and mesenchymal transdifferen-tiation, among others. Here, we hypothesized that TG2 also contributes to the radioresistance of two human melanoma cell lines A375 and MeWo which differ in their basal TG2 biosynthesis by examining their proliferation and clonal expansion after irradiation. For this purpose, cellular TG2 biosynthesis and TG2 activity were modulated by transfection-induced overexpression or TG2 knock-out and application of TG2-selective inhibitors. Proliferation and clonal expansion of TG2-overexpressing cells was not enhanced over wildtype cells, suggesting that increased TG2 biosynthesis does not further enhance the radioresistance of melanoma cells. Conversely, TG2 knock-out in A375 cells reduced their clonal and spheroidal expansion after irradiation, which indicates a contribution of TG2 to the radioresistance of melanoma cells. Since TG1, TG3, and partly also TG6 biosynthesis was detectable in A375 and MeWo cells, it can be assumed that these other members of the TG family may exert a partially compensatory effect.

The depletion of ore deposits, the increasing demand for raw materials, the need to process low-grade, complex and finely disseminated ores, and the reprocessing of tailings are challenges especially for froth flotation separation technologies. Even though they are capable of handling relatively fine grain sizes, the flotation separation of very fine and ultrafine particles faces many problems still. Further, the flotation of low-contrast semi-soluble salt-type minerals with very similar surface properties, many complex interactions between minerals, reagents and dissolved species often result in poor selectivity. This study investigates the flotation beneficiation of ultrafine magnesite rich in dolomite from desliming, currently reported to the tailings. The paper especially focuses on the impact of the depressant sodium hexametaphosphate (SHMP) on the following: (i) the froth properties using dynamic froth analysis (DFA), (ii) the separation between magnesite and dolomite/calcite, and (iii) its effect on the entrainment. As a depressant/dispersant, SHMP has a beneficial impact on the flotation separation between magnesite and dolomite. However, there is a trade-off between grade and recovery, and as well as the dewatering process which needs to be considered. When the SHMP increases from 200 g/t to 700 g/t, the magnesite grade increases from 67% to 77%, while recovery decreases massively, from 80% to 40%. The open circuit with four cleaning stages obtained a concentrate assaying 77.5% magnesite at a recovery of 45.5%. The dolomite content in the concentrate is about 20%, where 80% of dolomite was removed and importantly 98% of the quartz was removed, with only 0.3% of the quartz in the final concentrate. Furthermore, the application of 1-hydroxyethylene-1,1-diphosphonic acid (HEDP) as a more environmentally friendly and low-cost alternative to SHMP is presented and discussed. Using only 350 g/t of HEDP can achieve a similar grade (76.3%), like 700 g/t of SHMP (76.9%), while obtaining a 17% higher magnesite recovery as compared to 700 g/t of SHMP. Interestingly, the proportion of hydrophilic quartz minerals ending up in the concentrate is lower for HEDP, with only 1.9% quartz at a recovery of 21.5% compared to the 2.7% of quartz at a recovery of 24.9% when using SHMP. The paper contributes in general to understanding the complexity of the depressant responses in froth flotation.

Vietnamese phosphate deposits are of marine sedimentary origin and one of the largest phosphate rock deposits of Southeast Asia. Lao Cai phosphate reserves have been estimated at about 526 million tonnes with hypothetical reserves of about 2.6 billion tonnes. Phosphate rock has been considered as a secondary rare earth elements (REEs) source due to REEs often found as tracers of geochemical processes in the crystal structure of phosphates via substitution of Ca. Although the content of REEs in apatite is rather low, it is still relevant by total mass due to large tonnages of phosphate rock mined annually.
The impact of milling conditions on particle properties, pulp/ froth properties (Ca2+/ Mg2+ ion concentrations, pulp rheology, froth structure, etc.), which in turn influences on the flotation responses (grade, recovery, flotation kinetics and selectivity between apatite, REEs-bearing apatite and carbonate minerals) are given. With a rougher flotation feed containing about 12 % P2O5 and 200 g/t REEs, the obtained recoveries were 88-93 % for apatite and 75-84 % for REEs. This indicated that REEs can be enriched by froth flotation via co-flotation of the REEs-bearing apatite or true flotation of rare earth bearing minerals. However, there was only 24-29 % of dolomite removed from a feed containing about 7.3 % MgO. Thus, the separation of carbonate from finely disseminated siliceous carbonaceous apatite ores is given to be challenging due to fine intergrowth and a significant amount of dolomite with similar flotation properties like apatite.

Milling and flotation operations are typically studied separately in the literature, although it is well-established that the grinding environment is a determining factor on the flotation performance. This study aims at evaluating the effect of grinding conditions on the discharge particle properties and the subsequent flotation responses in the specific case of semi-soluble salt-type minerals. Three different ores were used as case studies: a scheelite (low-grade), an apatite (medium-grade) and a fluorite (high grade) ore. All three ores were dry-milled, wet-milled, and wet-milled in presence of depressants down to a particle size distribution specific to each ore, and the mill discharge was then floated under pre-established conditions in an Outotec mechanical cell. The concentrates and tailings of flotation tests were characterized with a Mineral Liberation Analyzer (MLA). This article presents the comparative results of the flotation tests conducted on the scheelite ore based on particle properties including size, composition, liberation, association, entrainment and shape.

With the great demand for high radiation tolerant materials for advanced nuclear energy technologies, Fe-Cr alloys are at the forefront with long standing validated performance. Yet, the real mechanism behind their high radiation resistance is in question and understanding the effect of varying Cr percentage is a grand challenge limiting further improvements. Here we applied depth resolved atomic scale probe of defects to uncover the real mechanism on how Cr improves radiation resistance and explain the controversial impact of increasing Cr percentage. By combining depth-resolved positron annihilation lifetime spectroscopy and Doppler broadening spectroscopy we investigated the effect of Cr alloying on the formation and evolution of atomic size clusters induced by ion irradiation in Fe. We also used atom probe tomography to investigate the possible presence of Cr clusters or alpha' phase with high Cr composition. The study reveals that the well-known resistance to radiation in Fe-Cr alloys arises from the stabilization of vacancy clusters around Cr atoms which act as sinks for radiation-induced defects. Thus, Cr atoms do not provide a direct sink for interstitials; rather defect complexes for that consist of Cr atoms and vacancies in turn act as sinks for irradiation-induced vacancies and interstitials. Most importantly, we find that lower amounts of Cr create smaller, uniformly distributed defect clusters that act as efficient sinks for radiation damage, but larger quantities of Cr form a defect structure that is less homogenous in size and spatial distribution, resulting in less efficient damage recombination. No evidence of phase alpha’ was found before or after irradiation, which indicates that it does not contribute to the observed radiation tolerance.

UC and UMeC2 (Me=Fe, Zr, Mo) carbides were studied by high-energy-resolution fuorescence detected x-ray absorption (HERFD-XAS) technique at the U M4 and L3 edges. Both U M4 and L3 HERFD-XAS of UMeC2 reveal some differences from that of UC and between each other in terms of the spectral width and energy position.
The observed differences are attributed to the consequences of the U 5f; 6d-4d(3d) hybridization in UMeC2. Calculations of the U M4 HERFD-XAS spectra were also performed using the Anderson impurity model (AIM). Based on the analysis of the data, the 5f occupancy in the ground state of UC was estimated to be 3.05 electrons. This fnding is also supported by the analysis of U N4;5 XAS of UC and by the results of the AIM calculations of the U 4f x-ray photoelectron spectrum of UC.

The structural and magnetic properties of the multicomponent R–Fe–H compounds with a high content of Fe and H are reported. The process of synthesis of the hydrides (R,R`)<sub>2</sub>Fe<sub>14</sub>BH<sub>5.5</sub> [where R and R` are light (Nd) and heavy (Dy, Ho, Er, Tm) rare earth metals, respectively] with a maximum hydrogen content is described in detail. The paper also provides insights into the synthesis of single-crystalline hydrides using the example of the R₂(Fe,Co)₁₄BH₃ series. The hydrides (R,Nd)₂Fe₁₄BH_5.5, R₂(Fe,Co)₁₄BH₃, R₂(Fe,Al)₁₇H₃ have a significantly increased volume as compared to the parent materials. High-field magnetization results of both parent and hydrogenated compounds at low temperatures are presented. Spin–reorientation phase transitions induced by an external magnetic field are observed. The parameter of the intersublattice exchange interaction and the influence of hydrogen on it are estimated within the framework of the mean field theory. The magnetocaloric effect of the compounds with a magnetic compensation point is studied with a special emphasis placed on the change of the sign of the effect.

The compound ZrTe₅ has recently been connected to a charge-density-wave (CDW) state with intriguing transport properties. Here, we investigate quantum oscillations in ultrasound measurements that microscopically originate from electron-phonon coupling and analyze how these would be affected by the presence or absence of a CDW. We calculate the phonon self-energy due to electron-phonon coupling, and from there deduce the sound-velocity renormalization and sound attenuation. We find that the theoretical predictions for a metallic Dirac model resemble the experimental data on a quantitative level for magnetic fields up to the quantum-limit regime.

The test for the cancer stem cell (CSC) hypothesis is to find a target expressed on all, and only CSCs in a patient tumor, then eliminate all cells with that target that eliminates the cancer. That test has not yet been achieved, but CSC diagnostics and targets found on CSCs and some other cells have resulted in a few clinically relevant therapies. However, it has become apparent that eliminating the subset of tumor cells characterized by self-renewal properties is essential for long-term tumor control. CSCs are able to regenerate and initiate tumor growth, recapitulating the heterogeneity present in the tumor before treatment. As great progress has been made in identifying and elucidating the biology of CSCs as well as their interactions with the tumor microenvironment, the time seems ripe for novel therapeutic strategies that target CSCs to find clinical applicability. On May 19-21, 2021, researchers in cancer stem cells met virtually for the Keystone eSymposium "Cancer Stem Cells: Advances in Biology and Clinical Translation" to discuss recent advances in the understanding of CSCs as well as clinical efforts to target these populations.

Invited speaker “Particle-based characterization of LIB recycling using automated mineralogy”, 11th Advanced Automotive Battery Conference AABC 2021, online

International talk: “Graphite a critical and strategic mineral for E-Mobility” Sustainable Minerals MEI, Online

The comminution of spent lithium-ion batteries (LIBs) produces a powder containing the active cell components, commonly referred to as “black mass.” Recently, froth flotation has been pro-posed to treat the fine fraction of black mass (< 100µm) as a method to separate anodic graphite particles from cathodic lithium metal oxides (LMOs). So far, pyrolysis has been considered as an effective treatment to remove organic binders in the black mass in preparation for flotation sepa-ration. In this work, the flotation performance of a pyrolyzed black mass obtained from an indus-trial recycling plant was improved by adding a pre-treatment step consisting of mechanical attri-tion. The LMOs recovery in the underflow product increased from 70% to 85% and the graphite recovery remained similar, around 86% recovery in the overflow product. To understand the flo-tation behaviour, the spent black mass was compared to a model black mass, comprising fully lib-erated LMOs and graphite particles. In addition, ultrafine hydrophilic particles were added to the flotation feed as an entrainment tracer, showing that the LMO recovery in overflow products is a combination of entrainment and true flotation mechanisms. This study highlights that adding kerosene during attrition enhance the emulsification of kerosene, simultaneously increasing its (partial) spreading on LMOs, graphite and residual binder, with a subsequent reduction of selectivity.

We demonstrate a straightforward and easily applicable technique for growing the highly improved quality of artificially BaZrO3 doped YBa2Cu3O6+x films on a commercially used buffered metallic template by pulsed laser deposition. Our method relies on reducing the grain size of the target material, which completely prevents the transfer of the harmful grain boundaries or weak links from the substrate through the buffer layers on the deposited film. We have also observed a great improvement in the self-assembly of BaZrO3 dopants and the critical current density is increased in the high temperature range up to 40%. As an extra benefit, our method allows increasing the growth rate of the film by 25%. We have discussed the results comprehensively with the help of the Ginzburg-Landau theory and provided an universal quantitative model of the grain boundary transfer from the substrate to the deposited film. The presented technique can be considered as a groundbreaking advancement for the vastly growing coated conductor industry.

In this work, the radiation responses of Zr/Nb nanostructured metallic multilayers (NMMs) are studied. The nanostructures with different layer thicknesses were deposited on Si (1 0 0)
substrate by using magnetron sputtering and are subjected to heavy-ion irradiation at room temperature with different fluences. Nanoindentation, XRD, SIMS, and Variable Energy
Positron Annihilation Spectroscopy (VEPAS) techniques were used to study the changes in the hardness of the structures as well as the type and distribution of defects, and strain within the material as a function of damage. Our results suggest that the irradiation hardening is independent of the type of implanted ions, and its magnitude decreases with decreasing
individual layer thickness indicating that the number of interfaces has a direct effect on the radiation tolerance enhancement. For thin layers with a periodicity of 27 nm (Zr/Nb27), a
transition from hardening to softening occurs at high fluence, and a saturation point is reached in thick layers with a periodicity of 96 nm (Zr/Nb96). The as-deposited multilayer with the
smaller periodicity experienced a significantly higher atomic-scale disorder which increases with ion irradiation compared to the multilayer with thicker individual layers. VEPAS reveals
the vacancy defects before and after irradiation that contribute to the presented strain. Based on the findings, thin nanostructured Zr/Nb multilayered structures possess excellent radiation
resistance due to the high density of interfaces that act as sinks for radiation-induced point defects.

The talk will give an overview on the numerical simulation of liquid metal batteries, polymer electrolyte fuel cells and electrolysers. After a brief introduction to liquid metal batteries, different exemplary simulations will be presented - with a focus on the interaction of fluid flow and electrochemistry. The presentation will be limited to the continuum mechanical open-source finite-volume library OpenFOAM.

The reaction between 2,3-dimercaptopropane-1-sulfonate (DMPS, unithiol) and four phenylarsonic(V) acids, i.e. phenylarsonic acid (PAA), 4-hydroxy-3-nitrophenylarsonic acid (HNPAA), 2-aminophenylarsonic acid (o-APAA) and 4-aminophenylarsonic acid (p-APAA), is investigated in aqueous solution. The pentavalent arsenic compounds are reduced by DMPS to their trivalent analogs and instantly chelated by the vicinal dithiol, forming covalent As–S bonds within a five-membered chelate ring. The different types and positions of polar substituents at the aromatic ring of the arsonic acids influence the reaction rates in the same way as observed for reaction with glutathione (GSH), as well as the syn : anti molar ratio of the diastereomeric products, which was analyzed using time- and temperature-dependent nuclear magnetic resonance (NMR) spectroscopy. Addition of DMPS to the conjugate formed by a phenylarsonic(V) acid and the biologically relevant tripeptide GSH showed the immediate replacement of GSH by chelating DMPS, underlining the importance of dithiols as detoxifying agent.

It is the intention of this study to elucidate the nested formation of calcium carbonate polymorphs or polyamorphs in the different nanosized compartments. With these observations it can be concluded how the bacteria can survive in a harsh environment with high calcium carbonate supersaturation. The mechanisms of calcium carbonate precipitation at the surface membrane and at the underlying cell wall membrane of the thermophilic soil bacterium Geobacillus stearothermophilus DSM 13240 have been revealed by high-resolution transmission electron microscopy and atomic force microscopy. In this Gram-positive bacterium nanopores in the surface layer (S-layer) and in the supporting cell wall polymers are nucleation sites for metastable calcium carbonate polymorphs and polyamorphs. In order to observe the different metastable forms, various reaction times and a low reaction temperature (4 °C) have been chosen. Calcium carbonate polymorphs nucleate in the confinement of nanosized pores (Ø3-5 nm) of the S-layer. The hydrous crystalline calcium carbonate (ikaite) is formed initially with [110] as the favored growth direction. It transforms into the anhydrous metastable vaterite by a solid-state transition. In a following reaction step calcite is precipitated caused by dissolution of vaterite in the aqueous solution. In the larger pores of the cell wall (Ø 20-50 nm) hydrated amorphous calcium carbonate is grown, which transforms into metastable monohydrocalcite, aragonite or calcite. Due to sequence of reaction steps via various metastable phases the bacteria gain time for chipping the partially mineralized S-layer, and forming a fresh S-layer (characteristic growth time about 20 min). Thus, the bacteria can survive in solutions with high calcium carbonate supersaturation under the conditions of forced biomineralization.

We present an improved method for in-vivo proton range verification by prompt gamma-ray timing based on multivariate statistical modelling.

To this end, prompt gamma-ray timing distributions acquired during pencil beam irradiation of an acrylic glass phantom with air cavities of different thicknesses were analysed. Relevant histogram features were chosen using forward variable selection and the Least Absolute Shrinkage and Selection Operator (LASSO) from a feature assortment based on recommendations of the Image Biomarker Standardisation Initiative. Candidate models were defined by multivariate linear regression and evaluated based on their coefficient of determination \(R^2\) and root mean square error \(RMSE\).

The newly developed models showed a clearly improved predictive power (\(R^2 > 0.7\)) compared to the previously used models (\(R^2 < 0.5\)) and allowed for the identification of introduced air cavities in a scanned treatment field. %The parameter selection showed better predictive power of the energy-specific models (RM SE < 1,8 mm) compared to the energy-independent models (RM SE > 3 mm).
%for counting statistics equivalent to a single spot measured with eight detector units.

These results demonstrate that elaborate statistical models can enhance prompt gamma ray based treatment verification and increase its potential for routine clinical application.

Y-stabilized zirconia (YSZ) phases were found to incorporate Th atoms and the corresponding solubility ranges depend on the relative Y content. For the tetragonal phases with lower Y concentration of 14 at.% a maximal possible Th intake on the Zr/Y metal site reached ca. 10.3 at.%. Cubic phases with higher Y content could dissolve 11 at.% Th in equilibrium state and up to ca. 12.3 at.% Th under non-equilibrium conditions. Larger Th-Zr/Y solubility range for phases with higher Y concentration was found to be related to the associated symmetry increase, as concluded from synchrotron radiation powder diffraction data. Specifically, introduction of Th into tetragonal YSZ induces evolution towards higher cubic symmetry via flattening of the Zr/YO8 polyhedra. In addition, tetragonal YSZ crystal lattice exhibits strongly anisotropic expansion with concomitant decrease in tetragonality upon the intake of Th. This results in easier accommodation of bigger Th atoms via structural stabilization of longer Zr/Y-O bonding distances which yields more symmetrical coordination of central Zr/Y metal ions by surrounding O atoms. Cubic symmetry is, therefore, more favorable to incorporation of large tetravalent Actinide elements.

Difficulties in dating Mississippi Valley-type (MVT) mineralization and the often closely-associated regional dolomitization events have led to considerable controversy regarding the typical tectonic contexts in which they form. This paper presents the first radiometric datings for the Gorno district in Lombardy, Italy, a classic example of the Alpine subclass of MVT deposits. U-Pb dating of hydrothermal carbonate minerals pre- and post-dating the main ore-forming event shows that both regional dolomitization and base-metal mineralization must have occurred within <10 Ma of the deposition of the Carnian host rocks. This implies that the ore deposits of the Gorno district, and probably other Alpine-type deposits, formed at shallow burial, prior to the western Tethys rifting phase. Concurrent Triassic magmatism and extensional tectonics likely contributed to the high geothermal heat fluxes necessary to drive the associated large-scale hydrothermal system. These findings contradict the general paradigm for MVT genesis, i.e., a preferred occurrence in deep-burial settings associated with compressional tectonics.

This paper provides an assessment of fluid transport and mixing processes inside the primary circuit of the test facility ROCOM through the numerical simulation of Test 2.1 with the system code ATHLET. The experiment represents an asymmetric injection of cold and non-borated water into the reactor coolant system (RCS) of a pressurized water reactor (PWR) to restore core cooling, an emergency procedure which may subsequently trigger a core re-criticality. The injection takes place at low velocity under single-phase subcooled conditions and presents a major challenge for the simulation in lumped parameter codes, due to multidimensional effects in horizontal piping and vessel arising from density gradients and gravity forces. Aiming at further validating ATHLET 3-D capabilities against horizontal geometries, the experiment conditions are applied to a ROCOM model, which includes a newly developed horizontal pipe object to enhance code prediction inside coolant loops. The obtained results show code strong simulation capabilities to represent multidimensional flows. Enhanced prediction is observed at the vessel inlet compared to traditional 1-D approach, whereas mixing overprediction from the descending denser plume is observed at the upper-half downcomer region, which leads to eventual deviations at the core inlet.

The exploitation of mine wastes materials as secondary resources requires in-depth mineralogical analyses, with metal deportment being of particular relevance for metal recovery. Using a combination of the Scanning Electron Microscope (SEM)-based Mineral Liberation Analyser (MLA) and Electron Probe Micro-Analyser (EPMA) methods, the deportment of Zn and Pb in the historic Plombières mine site (East Belgium) was investigated. The mine site comprises four different materials: soil, metallurgical waste, brown and yellow tailings. The integration of the MLA and EPMA data allowed the identification and quantification of Pb- and Zn-bearing phases, including minerals present in low abundances as well as slag phases. Slags and Pb oxide are the main source of Pb in different types of mine waste, followed by cerussite and/or anglesite. Zinc is mainly hosted by slags or by fraipontite, with minor amounts contributed by sphalerite, gahnite, willemite and bannisterite. Uncertainties on the metal deportments were estimated by bootstrap resampling and are typically low. The highest uncertainties are observed when the metal-bearing phases are present in low abundances, which is particularly noticeable in the yellow tailings. Mineral deportment studies should become a standard step in the process of maximising the potential of the mine waste deposits. The integrated approach of MLA-EPMA applied in this paper has proven to be effective for assessing the metal deportment between mineral and slag phases in mine wastes and could be successfully applied for other metals and waste deposits which originated from pyrometallurgical processing of the ore.

Since the earliest days of the Industrial Revolution, Europe has been discarding vast quantities of mining residues, commonly referred to as “extractive waste.” Even today, 900 Mton/year of extractive waste is being generated and stored in tailing facilities or ponds. This 900 Mton/year fgure corresponds to 26% of the EU’s current waste output . Without proper management, tailing ponds can lead to environmental problems, ranging from acid mine drainage (AMD) and water contamination to dam bursts and fooding, air pollution, and soil erosion and contamination.

Here we present in-situ U-Pb laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) ages of andradite-grossular garnet from four magmatichydrothermal polymetallic skarn prospects in the Schwarzenberg District, Erzgebirge (Germany), located in the internal zone of the Variscan Orogenic Belt. Within the geochronological framework of igneous rocks and hydrothermal mineralization in the Erzgebirge, the obtained garnet ages define three distinct episodes of Variscan skarn formation: (I) early late-collisional mineralization (338-331 Ma) recording the onset of magmatic-hydrothermal fluid flow shortly after the peak metamorphic event, (II) latecollisional mineralization (~327-310 Ma) related to the emplacement of large peraluminous granites following large-scale extension caused by orogenic collapse and (III) post-collisional mineralization (~310-295 Ma) contemporaneous with widespread volcanism associated with Permian crustal reorganization. Our results demonstrate that the formation of skarns in the Schwarzenberg District occurred episodically in all sub-stages of the Variscan orogenic cycle over a time range of at least 40 Ma. This observation is consistent with the age range of available geochronological data related to magmatic-hydrothermal ore deposits from other internal zones of the Variscan Orogenic Belt in central and western Europe. In analogy to the time-space relationship of major porphyry-Cu belts in South America, the congruent magmatic-hydrothermal evolution in the internal zones and the distinctly later (by ~30 Ma) occurrence of magmatic-hydrothermal ore deposits in the external zones of the Variscan Orogenic Belt may be interpreted as a function of their tectonic position relative to the Variscan collisional front.

We present extensive new ab initio path integral Monte Carlo results for the momentum distribution function n(k) of the uniform electron gas in the warm dense matter regime over a broad range of densities and temperatures. This allows us to study the nontrivial exchange-correlation-induced increase of low-momentum states around the Fermi temperature, and to investigate its connection to the related lowering of the kinetic energy compared to the ideal Fermi gas. In addition, we investigate the impact of quantum statistics on both n(k) and the off-diagonal density matrix in coordinate space, and find that it cannot be neglected even in the strongly coupled electron liquid regime. Our results were derived without any nodal constraints, and thus constitute a benchmark for other methods and approximations.

Nonlinear optical (NLO) properties of materials can be enhanced by assembling them as thin polymer composite films alternating with other polymers and forming dielectric mirrors, 1D photonic crystals (1DPCs), wherein the input light intensity is increased. Based on the poly(vinyl carbazole) (PVK) and poly(vinyl alcohol) (PVA) contrasting polymer pair, variants of such structures, with graphene and fullerene in their high-index layers, have been produced. Their optical switching characteristics have been studied with ns, cw, and quasi-cw fs laser sources in the IR and with a fs laser in the visible range. We have demonstrated slow optical bistability in the polymeric 1DPCs determined by the thermal expansion of polymer composites at intensities over 100W/cm² as well as fast and ultrafast optical switching due to thermo-optic and Kerr nonlinearities, respectively. A subpicosecond fast spectral shift of the 1DPC bandgap has been found. Our results and analysis provide a clear picture of the NLO behavior of 1DPCs at different time scales. The results stimulate the subsequent design of ultrafast switches and bistable memory cells based on polymeric 1DPCs whose micrometer thickness and flexibility offer promise for implementation into fiber and microchip configurations.

Roasting is often required to convert sulfidic minerals into their sulfate and/or oxide
forms with the aim to increase the extractability of targeted metals. In this study, sulfidic
tailings of Neves Corvo Cu-Zn-Pb-mine (Portugal) underwent conventional roasting
(CR) and microwave-assisted roasting (MR) to compare the effect of both heating
technologies on the roasting behavior. Upon roasting in air, transformations of mineral
phases in the tailings were studied by different techniques using quantitative X-ray
diffraction, mineral liberation analysis, Raman spectroscopy and
thermogravimetry−differential scanning calorimetry coupled to mass spectrometry.
Also, the leaching behavior of elements from the roasted tailings was assessed in
water. It was shown that CR and MR induce different reaction pathways for mineral
transformations during roasting. While CR leads to indirect pyrite oxidation through
intermediate sulfate formation, direct pyrite oxidation to oxides is the main
transformation pathway during MR. This change in reaction pathway can be attributed
to the mineral-selective heating induced by microwaves, which in particular accelerates
the oxidation rate of pyrite. In agreement with the proposed roasting mechanism, the
leaching behavior in water showed that the CR tailings resulted in a higher extraction
of the studied metals (Cu, Zn, Pb, As and Fe) – since sulfate phases are more
abundant – compared to the MR tailings at the same roasting temperature. Overall, the
optimal leaching efficiency of the metals was reached after one hour roasting at
500−550 °C for both CR and MR

A short report on experiences with UV photocathode lasers and laser beam line for ELBE SRF Gun in 2020-2021

Malignant melanoma has a steadily increasing incidence, but treatment options are still limited and prognosis for patients, especially for men, is poor. To investigate whether targeting estrogen receptor (ER) signaling is a valid therapeutic approach, we retrospectively analyzed ER gene expression profiles in 448 melanoma patients. High ERα gene expression was associated with improved overall survival (HR=0.881; 95% Cl=0.793-0.979, p=0.018) and increased with tumor stage, while ERβ gene expression did not change with tumor progression. This seemingly protective function of ERα led us to speculate that specific targeting of ERβ has a therapeutic benefit in malignant melanoma. A new type of ERβ-selective ER modulator with a nitric oxide (NO•) releasing moiety (NO-SERM 4d) significantly reduced the pro-metastatic behavior of two melanoma cell lines (A2058 and MelJuso). Epithelial-mesenchymal transition in melanoma is consistent with a switch from E- to N-cadherin expression, mediating the invasive phenotype. NO-SERM 4d reduced N-cadherin expression and impaired spheroid formation in A2058 cells. In addition, growth of A2058 spheroids was significantly reduced, confirming the anti-tumorigenic potential of NO-SERM 4d. Targeting ERβ signaling combined with targeted NO• release represents a promising therapeutic approach in malignant melanoma that has potential to prevent metastatic spread and reduce tumor growth.

Background and Aims
After the rediscovery of the normal tissue sparing FLASH effect of high dose rate radiation, research activities on this topic have been revived. But especially for protons, the portfolio of accelerators capable of performing studies at ultra-high dose rates is limited. Laser-plasma accelerators (LPA) can generate extremely intense proton bunches of many 10 MeV kinetic energy. In combination with dedicated dose delivery systems, LPA proton sources facilitate peak dose rates well above 10^8 Gy/s in a pulse structure regime complementary to conventional accelerators.

Methods
The reliable generation of proton spectra beyond 60 MeV at DRACO-PW [Ziegler et al, SciRep2021], combined with a dedicated energy selective pulsed magnet beam transport system [Brack et al, SciRep2020], allows tailored sample-specific dose distributions. Adapted on-shot dosimetry enables the required spectral monitoring of every proton bunch. Two irradiation series on volumetric biological samples were performed at DRACO-PW, accompanied by reference irradiations at the University Proton Therapy Dresden.

Results
The first small animal pilot study at a laser-driven proton source was conducted successfully. The mouse-ear tumor model’s requirements [Beyreuther et al, PLoS One2018] were fulfilled and verified at high precision (+/- 5%) concerning predefined dose value and conformity. Complementary, a study investigating dose-rate effects such as FLASH was performed irradiating zebrafish embryos with above 10^9 Gy/s.

Conclusions
We present a laser-based irradiation platform at the DRACO-PW facility that enables systematic radiobiological studies, laying the foundations for further studies at LPA sources exploring ultra-high dose-rate effects, such as FLASH, over previously unreachable parameter space.

Resource estimates are crucial to assess the economic potential of tailings storage facilities (TSF) for re-mining and the extraction of critical raw materials. However, a lack of consensus exists on best practices in sampling for this purpose. This study aims to address this gap by assessing different sampling schemes for the resource classification of TSFs. To do so, one layer of a TSF was sampled with regular and nested grids of varying sizes and additional random holes. Systematic spatial trends in the data were fitted with low-order polynomial functions of the coordinates and the grayscale values of a historical aerial photograph of the same layer of the TSF. Variography was performed on the trend residuals, and exponential or Gaussian variogram models were fitted. Universal Sequential Gaussian Simulation was then used to produce 1000 realisations of the potential ground truth. The optimum sampling strategy was investigated by re-sampling these realisations with varying sample densities and configurations and using geostatistical modelling of the re-sampled data to assess uncertainties on the estimated metal tonnages. Robust estimates of metal tonnages can be achieved at relatively low sampling densities, particularly with regular grids. When historical image information is used, spatial variability is better reproduced, and a lower number of samples is required to reach a certain confidence level. Furthermore, an approach to approximate expected errors on grades/tonnages estimated with a given sampling scheme was developed to assess whether further sampling is required. Overall, the findings of this study have clear and transferrable implications for the best-practice sampling and modelling of TSFs and their critical raw materials resource.

Tailings deposits pose a significant threat to the environment and/or contain residual metal contents that may be of economic interest. The mineralogical and textural properties of the tailings dictate both processing behaviour and potential for acid mine drainage. Automated Scanning Electron Microscope-based image analyses enable valuable quantitative mineralogical and textural data to be obtained. Such methods are, however, time and cost intensive. This study investigates the extent to which geochemical data can be related to mineralogical and textural data of tailings materials. Models based on the dominant components (PCs) from principal component analysis were tested for mineral abundance, particle and mineral grain size distributions (PSD, GSD), and degree of liberation. Sedimentary-style deposition of the tailings is represented by the main trend between PCs 1 and 2. This being the dominant process in the tailings, PC 1 is a robust estimator for mineral contents, PSD and mineral GSDs. Degree of liberation is less well constrained by the PCs. This study clearly demonstrates that prediction of mineralogical and textural parameters of tailings with geochemical data is possible. Considering that the dominant processes should be the same in similarly deposited tailings deposits, these findings are widely applicable.

Laser-driven proton pulse provide unique properties in terms of pulse structure (ns) and instantaneous dose rate (10^9 Gy/s) but - inherently broadband and highly divergent - pose a challenge to established beamline concepts on the path to application-adapted irradiation field formation, particularly for three-dimensional cases. We present the successful implementation and characterisation of a highly efficient and tuneable dual pulsed solenoid beamline at the Draco PW facility[1] to generate volumetric dose distribution tailored to specific applications[2].
The vast experimental scope and already successfully performed studies range from systematic volumetric in-vivo tumour irradiations in a dedicated mouse model (with a stable mean dose delivery of ±10 % and a spatial dose homogeneity of ±5 % over a cylindrical volume of 5 mm diameter and height) to high-dose-rate irradiations in the FLASH regime (using proton peak dose rates of up to 10^9 Gy/s with about 20 Gy/shot homogeneously over a cylindrical sample volume of 4.5 mm diameter and 3 mm height) as well as particle diagnostics commissioning (with a multitude of spatial and spectral dose distributions).
The beamline setup is complemented by a complex beam monitoring and dosimetry detector suite adapted to the ultra-high dose rate pulses and is in its unique synergy and redundancy capable of %-level precision dose delivery to samples as required for systematic irradiation studies. In addition to established radiochromic film dosimetry, the detector suite includes saturation-corrected (transmission) ionisation chambers [3] as well as screen and bulk scintillator setups, partly with tomographic reconstruction capabilities for 3D dose distribution retrieval. Moreover, non-invasive, single-shot-capable online time-of-flight-based spectral characterisation of filtered proton pulses has proven a powerful tool for beam monitoring as well as dosimetric purposes.
In this presentation the complex and versatile dose delivery system of laser-driven protons at the Draco PW using pulsed solenoids will be discussed. Its characterisation, technological development and improvement as well as the dosimetry suite as a vital part of the precise dose delivery will be addressed, while the presentation by U. Schramm covers recent experimental activities in detail.
[1] T. Ziegler, et al., Proton beam quality enhancement by spectral phase control of a PW-class laser system, https://arxiv.org/abs/2007.11499 (2020)
[2] Brack, et al., Spectral and spatial shaping of laser-driven proton beams using a pulsed high-field magnet beamline, SciRep, 10:9118, (2020)
[3] Gotz M, et al., A new model for volume recombination in plane‐parallel chambers in pulsed fields of high dose‐per‐pulse. Phys Med Biol., 62: 8634, (2017)

Negative electron affinity (NEA) GaAs- and GaN-based photocathodes are used in modern night vison detectors and light emitting diodes. GaAs semiconductors are already used as electron sources in particle accelerators and well- studied. Like GaAs, GaN belongs to the
III-V semiconductor group. It is assumed that GaN, like GaAs, shows enormous potential as a novel electron source for particle accelerators.

P-type GaN on a substrate material (sapphire, silicon, copper or SiC) is activated by a thin layer of caesium and illuminated by ultra-violet (UV) light to observe the gained photocurrent.
To produce an NEA surface and thus highest QE, it is necessary to remove any residual impurities on the GaN surface. Therefore, GaN is chemical cleaned and vacuum thermal cleaned before it is caesium-activated. The aim of the thermal treatment is to gain an atomically clean surface.
With a comparison of different substrate materials, chemical pre-cleaning, thermal treatment, and activation parameters (e.g. caesium-flux), the quantum efficiency, lifetime and the re-activation of the photocathode is studied. Additionally, the morphology and surface composition of the GaN is examined by XPS, AFM, SEM and EDX.

Non-invasive technologies are key to ensure the sustainable exploration and mining of critical materials such as lithium and rare earth elements (REE), which are needed to fuel battery, magnet, and photovoltaic technologies for the transition towards low-carbon economies (Simandl, 2014; Heredia et al., 2020). Innovative sensing methods such as hyperspectral imaging allow the remote identification and high spatial resolution mapping of characteristic light-material interactions (“spectral fingerprints”) indicative of specific minerals and raw materials. These maps can be produced rapidly to support geological investigations during both exploration and mining. Accurate positioning of the spectral information in 3D space allows creating digital twins of outcrops and mines and integrating the spectral mapping results with other data such as from drill-cores or geochemical sampling. However, the low concentration and generally subtle spectral features associated with Li-bearing minerals and REEs present challenges for spectral mapping at the outcrop scale and, as a result, past approaches tended to use indirect methods based on, for example, mineral associations (Cardoso-Fernandes et al., 2020).

Direct mapping of REE and Li-minerals abundances is possible using high-resolution, high-quality spectral data and appropriate correction techniques. Sensor miniaturisation and development of new platforms now allow rapid data collection across large areas. These advances, combined with novel radiometric and geometric correction workflows (e.g., Thiele et al., 2021), recently allowed the direct detection of REE- and Li-bearing minerals at outcrop scale, using close-range ground-based (Boesche et al., 2015; Booysen et al., 2021) and drone-borne platforms (Booysen et al., 2020).

Two case studies of open pit mines, one from Siilinjärvi, Finland, (Fig. 1), the second from Uis, Namibia (Fig. 2) demonstrate direct 3D mapping of REE and Li mineralization. First, we captured ground-based photographs that were used for structure-from-motion, multi-view-stereo (SfM-MVS) photogrammetry to create digital 3D models/point clouds. We also acquired terrestrial hyperspectral data using a Specim AisaFENIX hyperspectral line scanner (push-broom scanner). The sensor covers the electromagnetic spectrum from 350 to 2500 nm over 384 spatial pixels and creates a datacube by steady rotation on a tripod. The resulting panoramic scan covers the important spectral features for both REE (578 nm, 740 nm and 799 nm for neodymium) and Li-bearing minerals (1593 nm and 1839 nm for cookeite, and 1538 nm, 1766 nm and 1850 nm for montebrasite). However, the results are subject to immense geometric distortion as well as radiometric influences that complicate spectral mapping, spatial localization, and scaling. We applied the processing workflow of Thiele et al. (2021) to perform accurate re-projection in original 3D space, as well as important radiometric and topographic corrections by combining the hyperspectral scans with the photogrammetric 3D point clouds. We subsequently mapped the occurrence of REEs and Li-bearing minerals based on their subtle characteristic spectral absorptions, and validated the occurrences by field sampling and mineralogical and elemental analysis.

Innovative remote sensing approaches allow us to rapidly create accurate digital twins of mine faces and natural geological outcrops. The 3D point clouds provide information on geological structure and can be populated with an arbitrary number of point-specific, high-dimensional spectral attributes to create so-called “hyperclouds”. Accurate processing and correction of the contained spectral information enables us to directly map the subtle spectral features indicative of Li- and REE-bearing minerals. Efficient remote mapping, even in complex terrains, is highly beneficial not only for initial targeting, but also for monitoring ore extraction, thus facilitating both exploration and optimized extraction.

We thank AfriTin and Yara Oy for support during field work, access to the mine site, and geological information and geochemical data.

The alpaka library is a header-only C++14 abstraction library for accelerator development. Its aim is to provide performance portability across accelerators through the abstraction (not hiding!) of the underlying levels of parallelism. In this lecture we will cover the basics of how to program HPC applications with alpaka in a portable yet performant way and have a look at the software ecosystem surrounding alpaka.

The Niederschlag fluorite-barite vein deposit in the Western Erzgebirge, Germany, has been actively mined since 2013. We present the results of a first comprehensive study of the mineralogy, petrography, fluid inclusions, and trace element geochemistry of fluorite related to the Niederschlag deposit. Two different stages of fluorite mineralization are recognized. Stage I fluorite is older, fine-grained, associated with quartz, and forms complex breccia and replacement textures. Conversely, the younger Stage II fluorite is accompanied by barite and often occurs as banded and coarse crystalline open-space infill. Fluid inclusion and REY systematics are distinctly different for these two fluorite stages. Fluid inclusions in fluorite I reveal the presence of a low to medium saline (7–20% eq. w (NaCl+CaCl2)) fluid with homogenization temperatures of 140–180 °C, whereas fluorite II inclusions yield distinctly lower (80–120 °C) homogenization temperatures with at least two high salinity fluids involved (18–27% eq. w (NaCl+CaCl2)). In the absence of geochronological data, the genesis of the earlier generation of fluorite-quartz mineralization remains enigmatic but is tentatively related to Permian magmatism in the Erzgebirge. The younger fluorite-barite mineralization, on the other hand, has similarities to many fluorite-barite-Pb-Zn-Cu vein deposits in Europe that are widely accepted to be related to the Mesozoic opening of the northern Atlantic Ocean.

Deportment data is essential for the planning of mining and minerals processing operations. This need is particularly tangible for deposits of noble metals, such as gold and silver. Therefore, the current paucity of published gold and silver deportment data for individual ore deposits and districts – and the concomitant lack of understanding how this relates to salient geological and mineralogical features of these deposits – is surprising. In the present study, we apply a combination of bulk geochemistry, laser ablation-inductively coupled plasma-mass spectrometry, electron microprobe analysis, and mineral liberation analysis to gold- and silver-rich samples from the epithermal Freiberg Ag-Pb-Zn district, to investigate variability in gold and silver deportments, as well as the corresponding geological/mineralogical controls. The results show that the main carriers of gold are electrum and arsenopyrite, whereas silver is mostly hosted by Ag-sulfosalts (pyrargyrite, miargyrite, polybasite) and fahlore (freibergite). Deportments vary greatly between samples. These variations can be related to the relative abundances of different minerals within the samples, which in turn reflect their spatio-temporal position within the district. Comparisons with other epithermal Ag-Pb-Zn districts similar to Freiberg indicate that the results presented here are of general significance.

With the interest in radiometal-containing diagnostic and therapeutic pharmaceuticals burgeoning, appropriate ligands to coordinate completely and stably said radiometals is essential. Reported here are two novel, bis(amido)bis(oxinate)diamine ligands, H2amidohox and H2amidoC3hox, that combine two 8-hydroxyquinoline with two amide donor groups and differ by one carbon in their 1,2-ethylenediaminevs. 1,3-diaminopropane backbones, respectively. Both ligands have been thoroughly studied via metal complexation, solution thermodynamics and radiolabeling with three radiometal ions: [nat/64Cu]Cu2+,
[nat/111In]In3+, and [nat/203Pb]Pb2+. X-ray crystallography determined the structures of the hexacoordinated Cu2+-ligand complexes, indicating a better fit of Cu2+ to the H2amidohox binding pocket. Concentration dependent radiolabeling with [64Cu]Cu2+ was quantitative as low as 1 μM with H2amidohox and 10 μM
with H2amidoC3hox within 5 minutes at room temperature. However, [64Cu][Cu(amidohox)] maintained higher kinetic inertness against a superoxide dismutase enzyme-challenge assay and ligand challengescompared to the [64Cu][Cu(amidoC3hox)] counterpart. Similarly, H2amidohox had significantly higher
radiochemical conversion with both [111In]In3+ (97% at 1 μM) and [203Pb]Pb2+ (97% at 100 μM) under mild conditions compared to H2amidoC3hox (76% with [111In]In3+ at 1 μM and 0% with [203Pb]Pb2+). By studying non-radioactive and radioactive complexation with both ligands, a comprehensive understanding of the coordination differences between two- and three-carbon diamine backbones is discussed. Overall, the ethylenediamine backbone of H2amidohox proves to be superior with rapid radiolabeling and kinetic inertness towards competing ligands and proteins.

In this work, we introduce a multi-sensor subspace-based clustering algorithm that benefits from fine spectral-resolution hyperspectral images (HSIs) and fine spatial resolution RGB images. In order to extract spatial information, a hidden Markov random field (HMRF) is employed on the fine spatial-resolution RGB image, whereas, spectral information is derived from an HSI using an advanced sparse subspace clustering algorithm. The proposed algorithm is validated on two real geological data sets. The experimental results in this study show that the proposed algorithm outperforms the state-of-the-art clustering algorithms in terms of clustering accuracy.

Mineral dust is an important natural source of aerosols and significantly influences air quality (Querol et al.,
Environ. Int., 2019) and the global radiation budget (Schepanski, Geosci., 2018). Frequent dust intrusions are
observed in the Mediterranean region (Ealo et al., Atmos. Chem. Phys., 2016; Pikridas et al., Atmos. Environ.,
2018) and Central Europe (Collaud Coen et al., Atmos. Chem. Phys., 2004; Schauer et al., Aerosol Air Qual.
Res., 2016), with high potential to cause exceedances of daily PM10 levels. To separate the influence of
anthropogenic and natural contribution to the PM10 levels, the new method was developed within the DNAAP
project (Detection of non-anthropogenic air pollution – http://www.aerosol.si/dnaap/).
Dust weakly absorbs light in the near ultra-violet and short wavelengths of the visible range, while the light
absorption of dust in longer wavelengths from the visible and near infra-red range is negligible. We used
filter-based photometer Aethalometer AE33 (Drinovec et al., Atmos. Meas. Tech., 2015) to measure the light
absorption at seven wavelengths, from 370 to 950 nm. The mineral dust is not the only light-absorbing aerosol
in the air. Black carbon (BC), a unique primary tracer for combustion emissions, strongly absorbs light across
the entire visual, near infra-red and near ultra-violet spectral range. Since optical absorption of mineral dust is
weaker than the optical absorption of black carbon, the coarse mode mineral particles have to be concentrated
using the high-volume virtual impactor (VI). The method is based on the optical absorption measurements of
the two sample streams, sampling particle size below 1 µm and sample stream with the concentrated coarse
mode particles, where mineral dust contribution is substantial. Experimental configuration includes two
Aethalometers AE33 with different size selective inlets: VI inlet for sampling coarse aerosol mode (mostly
mineral dust) and PM1 inlet for sampling fine mode of aerosols (mainly BC). The optical absorption of mineral
dust can be determined by subtracting the absorption of fine aerosol fraction (PM1) from the absorption of
aerosol sampled by the VI, taking into account the enhancement factor of VI setup (Drinovec et al., Atmos.
Meas. Tech., 2020). The mineral dust mass concentration is then calculated using mass absorption cross-section
(MAC) for dust which could be site and source-region specific.
The results from the measurement campaigns performed at six locations in the Mediterranean region will be
presented. The measurements took place in NE Spain (Barcelona – BCN, Montseny – MSY, Montsec – MSA),
on Cyprus (Nicosia – NI, Agia Marina Xyliatou – AMX), and in Slovenia (Ljubljana – LJ). Two year-long
datasets will be presented, focusing on the analyses of aerosol optical properties of PM1 and VI fractions. The
results were validated using low time resolution chemical specification of offline filters and a statistical
approach where dust was extracted from PM10 measurements for dust intrusions periods determined by models
and back-trajectory studies. For better understanding, helium ion microscopy (HIM) was applied to study the
microscopic differences between mineral dust and black carbon captured on the AE33 filter tapes.
This work was supported by SPIRIT Slovenia – Public Agency for Entrepreneurship, Internationalization,
Foreign Investments and Technology, project DNAAP.

Hyperspectral image analysis has considerably evolved during the past decades. The conventional model-based image processing and machine learning techniques are not efficient for hyperspectral image analysis, therefore, other advanced models such as spatial- spectral models were proposed to boost the hyperspectral analysis. Recent advances in ma- chine learning i.e., deep learning (DL) confirm that if an adequate amount of training data is supplied then DL-based algorithms outperform the conventional (shallow) ones. However, the available training data is often limited in hyperspectral imaging, therefore, the advan- tage of DL-based algorithms compared with shallow ones remains an open question. In this paper, we address this issue for two vibrant fields in hyperspectral image analysis i.e., Unmixing and Feature Extraction for Classification.

In this article, we introduce a deep learning-based technique for the linear hyperspectral unmixing problem. The proposed method contains two main steps. First, the endmembers are extracted using a geometric endmember extraction method, i.e., a simplex volume maximization in the subspace of the data set. Then, the abundances are estimated using a deep image prior. The main motivation of this work is to boost the abundance estimation and make the unmixing problem robust to noise. The proposed deep image prior uses a convolutional neural network to estimate the fractional abundances, relying on the extracted endmembers and the observed hyperspectral data set. The proposed method is evaluated on simulated and three real remote sensing data for a range of SNR values (i.e., from 20 to 50 dB). The results show considerable improvements compared to state-of-the-art methods. The proposed method was implemented in Python (3.8) using PyTorch as the platform for the deep network and is available online: https://github.com/BehnoodRasti/UnDIP.

In this letter, we propose a sparse unmixing technique using a convolutional neural network (SUnCNN) for hyperspectral images. SUnCNN is the first deep learning-based technique proposed for sparse unmixing. It uses a deep convolutional encoder-decoder to generate the abundances relying on a spectral library. We reformulate the sparse unmixing into an optimization over the deep network's parameters. Therefore, the deep network learns in an unsupervised manner to map a fixed input into the sparse optimum abundances. Additionally, SUnCNN holds the sum-to-one constraint using a softmax activation layer. The proposed method is compared with the state-of-the-art using two synthetic datasets and one real hyperspectral dataset. The overall results confirm that the proposed method outperforms the other ones in terms of signal to reconstruction error (SRE). Additionally, SUnCNN shows visual superiority for both real and synthetic datasets compared with the competing techniques. The proposed method was implemented in Python (3.8) using PyTorch as the platform for the deep network and is available online: https://github.com/BehnoodRasti/SUnCNN.

We propose a new fusion-based classification technique for optical multi-source remote sensing images called OptFus. OptFus is developed to merge and process optical imagery having different spatial and spectral resolutions. The spatial features are extracted using morphological filters from the RGB data containing high spatial resolution. A feature fusion technique is developed to combine all the sensor data in a subspace using a common set of representative features. Finally, the fused features are classified using a Support Vector Machine to ensure a robust supervised spectral classification. The proposed method is designed to allocate varying weights to the data from various imaging sensors in the fusion process. OptFus is applied to two multi-source optical datasets captured from geological drill-core samples. The classification accuracy demonstrates considerable improvements compared to the state-of-the-art. A Matlab implementation of OptFus is available online: https://github.com/BehnoodRasti/OptFus.

I present some results for an average-atom model derived from first principles. In particular, I compare different choices of boundary condition and exchange-correlation functional and examine their impact on the equation-of-state data and mean ionization state.

In simulations of the warm dense matter regime, it is typical to use a combined finite-temperature Kohn-Sham density-functional theory (KS-DFT) and molecular dynamics approach. However, in KS-DFT, (i) scaling worsens with increasing temperature, and (ii) temperature dependence is usually neglected in the exchange-correlation (XC) functional. We present a derivation from first-principles which reduces the full many-body Hamiltonian to an average-atom model in the dilute gas limit, which significantly reduces the computational cost of the KS-DFT calculation. We also show preliminary results including a comparison of temperature-dependent and zero-temperature XC functionals.

I demonstrate the key steps in a first-principles derivation of a density-functional average-atom model, and show the importance of the boundary condition used to solve the Kohn-Sham equations on results.

In this talk, I present a first-principles derivation of an average-atom model, which is solved using Kohn-Sham density-functional theory. I also demonstrate the impact of various assumptions and approximations of the model, such as the boundary condition and exchange-correlation functional, on various properties such as mean ionization state. I consider some possible causes of error in the model, specifically self-interaction and density-driven errors. Finally, I introduce our new open-source average-atom code for matter under extreme conditions, atoMEC.

We introduce an Öffner stretcher for the diode-pumped Petawatt laser system PENELOPE developed at the HZDR. Spectral and temporal aspects of the stretcher system with large aperture spherical mirrors (600 mm and 250 mm) are analyzed with respect to pulse shape and pulse contrast ratio.

The warm dense matter regime is notoriously challenging from a computational modelling perspective, since a quantum mechanical description is required for large length and time scales, at temperatures well above the ground-state. Average-atom (AA) models reduce the many-body system of interacting electrons and nuclei to an effective ‘average’ atom, with relevant properties typically computed via Kohn–Sham density-functional theory (KS-DFT): this approach has clear computational advantages relative to full ab initio KS-DFT (or alternative) simulations. However, there is some uncertainty regarding the accuracy and predictive capabilities of AA models. In this talk, I will present a first principles derivation of a KS-AA model, carefully analysing the assumptions made and terms neglected in this approach. The results obtained from our model highlight the importance of the choice of boundary conditions and the significance of the self-interaction error. I will discuss the implications of these findings for future developments in AA models.

Remote sensing provides valuable information about objects or areas from a distance in either active (e.g., RADAR and LiDAR) or passive (e.g., multispectral and hyperspectral) modes. The quality of data acquired by remotely sensed imaging sensors (both active and passive) is often degraded by a variety of noise types and artifacts. Image restoration, which is a vibrant field of research in the remote sensing community, is the task of recovering the true unknown image from the degraded observed image. Each imaging sensor induces unique noise types and artifacts into the observed image. This fact has led to the expansion of restoration techniques in different paths according to each sensor type. This review paper brings together the advances of image restoration techniques with particular focuses on synthetic aperture radar and hyperspectral images as the most active sub-fields of image restoration in the remote sensing community. We, therefore, provide a comprehensive, discipline-specific starting point for researchers at different levels (i.e., students, researchers, and senior researchers) willing to investigate the vibrant topic of data restoration by supplying sufficient detail and references. Additionally, this review paper accompanies a toolbox to provide a platform to encourage interested students and researchers in the field to further explore the restoration techniques and fast-forward the community. The toolboxes are provided in https://github.com/ImageRestorationToolbox.

Due to powerful feature extraction capability, convolutional neural networks (CNNs) have been widely used for hyperspectral image (HSI) classification. However, because of a large number of parameters that need to be trained, sufficient training samples are usually required for deep CNN-based methods. Unfortunately, limited training samples are a common issue in the remote sensing community. In this study, a dual graph convolutional network (DGCN) is proposed for the supervised classification of HSI with limited training samples. The first GCN fully extracts features existing in and among HSI samples, while the second GCN utilizes label distribution learning, and thus, it potentially reduces the number of required training samples. The two GCNs are integrated through several iterations to decrease interclass distances, which leads to a more accurate classification step. Moreover, a new idea entitled multiscale feature cutout is proposed as a regularization technique for HSI classification (DGCN-M). Different from the regularization methods (e.g., dropout and DropBlock), the proposed multiscale feature cutout could randomly mask out multiscale region sizes in a feature map, which further reduces the overfitting problem and yields consistent improvement. Experimental results on the four popular hyperspectral data sets (i.e., Salinas, Indian Pines, Pavia, and Houston) indicate that the proposed method obtains good classification performance compared to state-of-the-art methods, which shows the potential of GCN for HSI classification.

Earthquakes and heavy rainfalls are the two leading causes of landslides around the world. Since they
often occur across large areas, landslide detection requires rapid and reliable automatic detection
approaches. Currently, deep learning (DL) approaches, especially different convolutional neural
network and fully convolutional network (FCN) algorithms, are reliably achieving cutting-edge
accuracies in automatic landslide detection. However, these successful applications of various DL
approaches have thus far been based on very high-resolution satellite images (e.g., GeoEye and
WorldView), making it easier to achieve such high detection performances. In this study, we use freely
available Sentinel-2 data and ALOS digital elevation model to investigate the application of two wellknown FCN algorithms, namely the U-Net and residual U-Net (or so-called ResU-Net), for landslide
detection. To our knowledge, this is the first application of FCN for landslide detection only from freely
available data. We adapt the algorithms to the specific aim of landslide detection, then train and test
with data from three different case study areas located in Western Taitung County (Taiwan), Shuzheng
Valley (China), and Eastern Iburi (Japan). We characterize three different window size sample patches
to train the algorithms. Our results also contain a comprehensive transferability assessment achieved
through different training and testing scenarios in the three case studies. The highest f1-score value of
73.32% was obtained by ResU-Net, trained with a dataset from Japan, and tested on China’s holdout
testing area using the sample patch size of 64 × 64 pixels.

atoMEC is a python-based average-atom code for simulations of high energy density phenomena such as in warm dense matter. It is designed as an open-source and modular python package. atoMEC uses Kohn-Sham density functional theory, in combination with an average-atom approximation, to solve the electronic structure problem for single-element materials at finite temperature.

The 2021 Data Fusion Contest, organized by the Image Analysis and Data Fusion Technical Committee (IADF TC) of the IEEE Geoscience and Remote Sensing Society (GRSS), promotes research on geospatial artificial intelligence (AI) for social good. The global objective is to build models for understanding the state of and changes in artificial and natural environments from remote sensing data toward sustainable development.

Electrons in zero external magnetic field can be studied with density functional theory (DFT) or with spin-DFT (SDFT). The latter is normally used for open shell systems because its approximations appear to model better the exchange and correlation (xc) functional, but also because so far the application of DFT implied a closed-shell-like approximation. Correcting this error for open shells allows the approximate DFT xc functionals to become as accurate as those in SDFT. In the limit of zero magnetic field, the Kohn-Sham equations of SDFT emerge as the generalised KS equations of DFT.

This article presents the scientific outcomes of the 2020 Data Fusion Contest (DFC2020) organized by the Image Analysis and Data Fusion Technical Committee of the IEEE Geoscience and Remote Sensing Society. The 2020 Contest addressed the problem of automatic global land-cover mapping with weak supervision, i.e., estimating high-resolution semantic maps while only low-resolution reference data are available during training. Two separate competitions were organized to assess two different scenarios: 1) high-resolution labels are not available at all; and 2) a small amount of high-resolution labels are available additionally to low-resolution reference data. In this article, we describe the DFC2020 dataset that remains available for further evaluation of corresponding approaches and report the results of the best-performing methods during the contest.

Finite-temperature Kohn--Sham density-functional theory (KS-DFT) is a widely-used method in warm dense matter (WDM) simulations and diagnostics. Unfortunately, full KS-DFT-molecular dynamics models scale unfavourably with temperature and there remains uncertainty regarding the performance of existing approximate exchange-correlation (XC) functionals under WDM conditions. Of particular concern is the expected explicit dependence of the XC functional on temperature, which is absent from most approximations. Average-atom (AA) models, which significantly reduce the computational cost of KS-DFT calculations, have therefore become an integral part of WDM modelling. In this paper, we present a derivation of a first-principles AA model from the fully-interacting many-body Hamiltonian, carefully analysing the assumptions made and terms neglected in this reduction. We explore the impact of different choices within this model -- such as boundary conditions and XC functionals -- on common properties in WDM, for example equation-of-state data, ionization degree and the behaviour of the frontier energy levels. Furthermore, drawing upon insights from ground-state KS-DFT, we discuss the likely sources of error in KS-AA models and possible strategies for mitigating such errors.

Recent progress on advanced photocathodes for superconducting RF guns. We will together view the achievements as well as open questions for the reliable gun operation.

Recognizing less salient features is the key for model compression. However, it has not been investigated in the revolutionary attention mechanisms. In this work, we propose a novel normalization-based attention module (NAM), which suppresses less salient weights. It applies a weight sparsity penalty to the attention modules, thus, making them more computational efficient while retaining similar performance. A comparison with three other attention mechanisms on both Resnet and Mobilenet indicates that our method results in higher accuracy. Code for this paper can be publicly accessed at https://github.com/Christian-lyc/NAM.

Kinetics and efficiency of mineral dissolution of multi-phase rocks depend on the properties of the mineral assemblage, the fluid composition and the environment (e.g. temperature, fluid velocity at the surface). Links between the different effects can be studied by reactive transport simulations, although the input of experimentally obtained parameters is not always intuitive and its accuracy depends on the type of experiment. Using gold ore leaching with deep eutectic solvents as an example, we present an in-situ experimental setup to image in 3D the changes of the solid phases during mineral dissolution using laboratory X-Ray computed tomography (CT). Our new approach allows measuring the volume and surface area at different dissolution times, which are used to calculate the dissolution rate spectra for different mineral phases and to study the kinetics of the dissolution process. Additionally, the 3D model of the dissolving solid imaged by CT allows direct surface extraction for fluid flow simulations, which are used to investigate the effect of fluid velocity on the dissolution rate spectra. Altogether, our method can be used to study complex geochemical mineral – fluid interactions and to directly capture the link between different factors that affect dissolution, e.g.rate spectra and flow, which traditionally can only be studied by an indirect combination of experimental methods and modelling.

In this work, we propose a modular multi-modal architecture to automatically detect Alzheimer’s disease using the dataset provided in the ADReSSo challenge. Both acoustic and text-based features are used in this architecture. Since the dataset provides only audio samples of controls and patients, we use Google cloud-based speech-to-text API to automatically transcribe the audio files to extract text-based features. Several kinds of audio features are extracted using standard packages. The proposed approach consists of 4 networks: C-attention-acoustic network (for acoustic features only), C-Attention-FT network (for linguistic features only), C-Attention-Embedding network (for language embeddings and acoustic embeddings), and a unified network (uses all of those features). The architecture combines attention networks and a convolutional neural network (CAttention network) in order to process these features. Experimental results show that the C-Attention-Unified network with Linguistic features and X-Vector embeddings achieves the best accuracy of 80.28% and F1 score of 0.825 on the test dataset.

The Witwatersrand Supergroup in South Africa is not only the best-preserved sequence of Archean sedimentary rocks but also hosts the largest gold deposit on earth, yet discovered. The gold is situated in quartz-pebble conglomerates and is generally associated with a wide variety of minerals, including pyrite (FeS2), uraninite (UO2), pyrobitumen, base metal sulfides and phyllosilicates. Despite extensive research over the past 100 years, the origin of the gold is still debated with two models receiving most of the attention: the modified paleoplacer model and the hydrothermal models. The modified paleoplacer model assumes that detrital gold was transported into the host rock by fluvial processes, followed by a short-range mobilization (micrometer- to meter scale) by hydrothermal fluids that infiltrated the host rock. In the hydrothermal model, gold was introduced into the host rock by postdepositional hydrothermal fluids from an external source.
To find new evidence for the origin of gold, we present new high-resolution 3-dimensional (3D) data based on the combination of X-ray computed micro tomography (micro-CT) and spectral X-ray computed micro tomography (Sp-CT). We combine both imaging techniques as micro-CT is an excellent tool for high resolution structural characterization and Sp-CT has the power to chemically identify a selection of minerals relevant in this study. Sp-CT is based on the analysis of X-ray absorption spectra, in particular specific K-edges, which assists in identification of chemical elements in a sample. Since most of the previous research of the Witwatersrand gold focus on 2-dimensional methods, the outcomes of this study will help to understand the 3D spatial and size distribution of the gold and provide morphological information about individual particles.
This research has received funding from the European Institute of Innovation and Technology (EIT). This body of the European Union receives support from the European Union's Horizon 2020 research and innovation program.