Earthquake detection and seismic phase picking not only play a crucial role in travel time estimation of body waves(P and S waves) but also in the localisation of the epicenter of the corresponding event. Generally, manual phase picking is a trustworthy and the optimum method to determine the phase arrival time, however, its capacity is restricted by available resources and time. Moreover, noisy seismic data renders an additional critical challenge for fast and accurate phase picking. In this study, a deep learning based model, EPick, is proposed which benefits both from U shaped neural network and attention mechanism, as a strong alternative for seismic event detection and phase picking.

With the detection of gravitational waves from compact binary inspirals by the LIGO-Virgo collaboration, the era of multi-messenger gravitational-wave astronomy has begun. Gravitational-wave spectroscopy opens a new window of some of the most extreme objects in the Universe, such as neutron stars and black holes.  The r-mode instability is one of the mechanisms by which isolated neutron stars might spin down and emit detectable gravitational radiation. In this talk, we present some results of simulations of r-mode oscillations for isolated neutron stars in full General Relativity with an emphasis on their potential observability and on the information that they may provide on neutron star matter.

The putative effects of dark matter are most easily explained by a collisionless fluid on cosmological scales and by Modified Newtonian Dynamics (MOND) on galactic scales. Superfluid dark matter (SFDM) postulates that this differing behavior with scale is caused by a single underlying substance with two phases. SFDM is an example of a general class of models, called hybrid models, which combine a collisionless fluid on cosmological scales with a MOND-like force on galactic scales. I propose a novel Cherenkov radiation constraint specific to hybrid models. This constraint is different from standard modified gravity Cherenkov radiation constraints because hybrid models allow even non-relativistic objects like stars to emit Cherenkov radiation.

Cyanobacteria produce an important part of the oxygen on earth and are able to fix atmospheric nitrogen under nitrogen-poor conditions. To achieve this goal, the genus Anabaena forms filaments in which some cells differentiate into so-called heterocysts and form patterns to effectively supply the colony with nitrogen. Heterocysts lose the ability to divide and are therefore tied to their cell fate. Depending on the stage of development and the number of neighboring cells, however, a regression of the developing proheterocyst could be observed. We present a theory that combines genetic, metabolic and morphological aspects to understand this prokaryotic example of multicellularity.

The detection and rapid characterisation of earthquake parameters such as magnitude are of prime importance in seismology, particularly in applications such as Earthquake Early Warning (EEW). Traditionally, algorithms such as short-term average/long-term average (STA/LTA) are used for event detection, while frequency or amplitude domain parameters calculated from 1-3 seconds of first P-arrival data are sometimes used to provide a first estimate of (body-wave) magnitude. Owing to the extensive involvement of human experts in parameter determination, these approaches are often found to be insufficient.

Moreover, these methods are sensitive to the signal-to-noise ratio and may often lead to false or missed alarms depending on the choice of parameters. We, therefore, propose a multi-tasking deep learning model – the Convolutional Recurrent model for Earthquake Identification and Magnitude Estimation (CREIME) that: (i) detects the first earthquake signal, from background seismic noise, (ii) determines first P-arrival time as well as (iii) estimates the magnitude using the raw 3-component waveform data from a single station as model input. Considering, that speed is of the essence in EEW, we use up to two seconds

of P-wave information which, to the best of our knowledge, is a significantly smaller data window (5-second window with up to 2 seconds of P-wave data) compared to the previous studies. To examine the robustness of CREIME we test it on two independent datasets

and find that it achieves an average accuracy of 98% for event-vs-noise discrimination and is able to estimate first P-arrival time and local magnitude with an average root mean squared errors of 0.13 seconds and 0.65 units, respectively. We also compare CREIME architecture with the architectures of other baseline models, by training them on the same data, and also with traditional algorithms such as STA/LTA, and show that our architecture outperforms these methods.

Blockchain technology has been significantly characterized by Bitcoin. And with this link comes its bad reputation. However, there is more to the technology than a self-declared goal to enable a "world currency." Its decentralized nature provides a way of managing digital identities that complies with data protection regulations. And this can justify its use in public administrations.

I will show what we have already developed and how public life can be positively impacted by the development of digital identity management solutions based on blockchains.

In 2020, 45.4% of the electricity in Germany had been produced from highly volatile renewable sources, and the share is expected to increase.  Energy storage technologies could enhance the flexibility of the energy infrastructure. However, energy storage has a high cost. Within this project, an agent-based algorithm that determines the optimal buying and selling strategy of storage is developed. In our research,  we try to make use of energy price fluctuations by charging the storage at times when there is an excess of energy on the market and therefore the prices are low, and selling the energy at peak load times when the prices are high. We propose a deep reinforcement learning approach to understand the price formation depending on time and thus generate profits using energy arbitrage.

Abstract: In the absence of effective medication or vaccination,
mitigation of an infectious disease relies on so-called non-pharmaceutical interventions. In contrast to population-wide contact restrictions, test-trace-and-isolate (TTI) offers a targeted measure that directly aims at breaking chains of infection. The incorporation of TTI, especially contact tracing (CT), into standard disease spread models is challenging:
information about the timing, proximity, and traceability of contacts of identified index cases has to be lifted to the population-level scale and in-host processes, like the course of infectivity, influence the population-level effect of TTI. In this work, we introduce a mean-field infectious disease model that includes TTI. Through a delay, an approximation of the CT process is directly linked to the success of index case identification, and state-dependent testing and tracing rates are introduced to examine the impact of limited TTI capacities. Using the early spread of SARS-CoV-2 in Germany as an example, we perform numerical simulations to determine the extent to which TTI enables the relaxation of contact restrictions. In addition, we demonstrate how the success of TTI depends on (1) epidemiological characteristics of the disease under consideration and (2) maintaining a low prevalence to prevent TTI capacity from being breached.

Due to the limitations in terrestrial laboratories, the equation of state (EoS) of cold, dense matter remains a challenge. The extreme densities in the cores of neutron stars (NSs) offer the possibility of determining the EoS, through observables like mass and radius. In this work, we present a novel method that uses Deep Learning to reconstruct the NS EoS that is model-independent. The neural network is initially tested on mock observations of mass and radius, and then used to reconstruct an EoS from the limited mass, radius observational data.