MESA+ Computing & Simulation

11:45 – 13:00h | room 6
Chair: Menno Bokdam

  • 11.45 - 12.00 | Pierre Lechifflart (S&T-CCP) - Computation of Electronic Structure and Dynamics

    Excitons are correlated electron-hole pairs that form in semiconductors upon absorption of photons and are key for understanding light-conversion and designing materials with tailored properties. They are the main research object in our group and we calculate their properties ab initio, without any fitting parameters.

    In this presentation, I will provide an overview of our calculations which allow us to access a broad range of results such as equilibrium and non-equilibrium crystal structures, accurate electronic bandstructures with a many-body description of dielectric screening, and the chemical origin of electronic bands from orbital contributions. These are ground-state quantities that we obtain from Density Functional Theory and the GW approximation. Additionally, we use the Bethe-Salpeter Equation approach to obtain excited-state properties, such as oscillator strengths, wavefunctions and energies of excitons with a resolution up to a few meV, even for very complex, heterogeneous materials.

    Our focus in the last years have been light-harvesting materials such as semiconducting perovskites, which offer tremendous chemical and structural diversity and tunability. For these materials, we examine the effects of dimensionality, alloying, defects, pressure or structural distortions on the electronic and excitonic properties.A future research direction is exploring the lifespan of excitons after their initial creation, such as their interaction with the crystal lattice leading to self-trapped excitons. This requires calculation of the forces created by the photoexcitation, and would eventually unfold the mechanism of photodegradation of solar cells.

    In all of this, we work in close collaboration with experimentalists by providing atomistic  insights for  interpretation of measurements or by using experimental data to validate our calculations.

    The rich variety of material properties available from our calculations paves the way to material synthesis with tunable features and is key for technological applications, for instance in photovoltaics.

  • 12.05 - 12.20 | Zahra Golsanamlou (S&T-CCP) - Computation of Thin Film Emissivity: influence of defects and amorphousness

    Zahra Golsanamlou, Arseniy Baskakov, Silvester Houweling, Giorgio Colombi, Robbert van de Kruijs, Marcelo Ackermann and Menno Bokdam | CCP & XUV, University of Twente

    In recent years the study of emissivity of materials that can withstand at high temperatures has been extensively growing.  is a class of materials with high melting point and is predicted to have high emissivity. In this work we present how to calculate the emissivity of  thin films. The emissivity of the films is related to the absorption spectra according to Kirchhoff low. The absorption spectra is calculated based on dielectric function and consequently electronic structure of the films. Our results indicate that emissivity is related to the metallicity and detailed atomic structure of the films.

  • 12.25 - 12.40 | Davide Selvatici (S&T-PoF) - Impact of stratocumulus clouds on wind farms

    Stratocumulus clouds are the most common type of cloud, as they cover more than 20% of the Earth’s surface on annual average (Wood, 2012). As such, they strongly influence the Earth’s radiative balance, and thus Earth’s climate. However, the effects of these clouds on the Earth’s Boundary Layer are not yet fully understood, and it is still a challenge to study their complex physics experimentally or numerically. In this work, we study Stratocumulus-Topped Boundary Layers (STBL) using a Large Eddy Simulations (LES) code that we developed, with the aim of understanding the effects of these clouds on key properties of the Earth’s Boundary Layer, and in order to draw possible scenarios for wind farms performance in STBLs.

    First, we intend to show that LES is a suitable tool to study STBLs. We will compare our results with data from in situ and radar measurements (Stevens et al., 2005; van der Dussen et al., 2013). Then, we will study the effects of physical parameters, such as the radiation intensity, on the results, in order to understand the sensitivity of the results to such physical phenomena. We will then apply this methodology to study the influence of the different physical mechanisms on the performance of wind farms in STBLs. We will show that the presence of Stratocumulus clouds significantly affects the power output of wind farms, and also that wind farms influence the properties of stratocumulus clouds.

    Overall, this study is a first step towards understanding the effects of Stratocumulus clouds on the Earth’s Boundary Layer, and on wind farms. The results that we intend to show will be useful for the development of new models of Earth’s Boundary Layer, and for the optimization of wind farms in STBLs.

  • 12.45 - 13.00 | Reza Zolfagharinejad (EEMC-NE) - Automatic speech recognition on novel computing architectures

    With the rise of decentralized computing, as in the Internet of Things, autonomous driving, and personalized healthcare, it is increasingly important to process time-dependent signals ‘at the edge’ efficiently: right at the place where the temporal data are collected, avoiding time-consuming, insecure, and costly communication with a centralized computing facility (or ‘cloud’). However, modern-day processors often cannot meet the restrained power and time budgets of edge systems because of intrinsic limitations imposed by their architecture (von Neumann bottleneck) or domain conversions (analogue-to-digital and time-to-frequency). Here, we propose an edge temporal-signal processor based on two in-materia computing systems for both feature extraction and classification, reaching a software-level accuracy of 96.2 ± 0.8% for the TI-46-Word speech-recognition task. First, a nonlinear, room-temperature dopant-network-processing-unit (DNPU) layer realizes analogue, time-domain feature extraction from the raw audio signals,  similar to the human cochlea. Second, an analogue in-memory computing (AIMC) chip, consisting of memristive crossbar arrays, implements a compact neural network trained on the extracted features for classification. With the DNPU feature extraction consuming ~100s nW and AIMC-based classification having the potential for less than 10 fJ per multiply-accumulate operation, our findings offer a promising avenue for advancing the compactness, efficiency, and performance of heterogeneous smart edge processors through in-materia computing hardware.