Wind farm placement arranges the size and the location of multiple wind farms within a given region. The power output is highly related to the wind speed on spatial and temporal levels, which can be modeled by advanced data-driven approaches. To this end, we use a probabilistic neural network as a surrogate that accounts for the spatiotemporal correlations of wind speed. This neural network uses ReLU activation functions so that it can be reformulated as mixed-integer linear set of constraints (constraint learning). We embed these constraints into the placement decision problem, formulated as a two-stage stochastic optimization problem. Specifically, conditional quantiles of the total electricity production are regarded as recursive decisions in the second stage. We use real high-resolution regional data from a northern region in Spain. We validate that the constraint learning approach outperforms the classical bilinear interpolation method. Numerical experiments are implemented on risk-averse investors. The results indicate that risk-averse investors concentrate on dominant sites with strong wind, while exhibiting spatial diversification and sensitive capacity spread in non-dominant sites. Furthermore, we show that if we introduce transmission line costs in the problem, risk-averse investors favor locations closer to the substations. On the contrary, risk-neutral investors are willing to move to further locations to achieve higher expected profits. Our results conclude that the proposed novel approach is able to tackle a portfolio of regional wind farm placements and further provide guidance for risk-averse investors.

The Schrödinger Bridge provides a principled framework for modeling stochastic processes between distributions; however, existing methods are limited by energy-conservation assumptions, which constrains the bridge's shape preventing it from model varying-energy phenomena. To overcome this, we introduce the non-conservative generalized Schrödinger bridge (NCGSB), a novel, energy-varying reformulation based on contact Hamiltonian mechanics. By allowing energy to change over time, the NCGSB provides a broader class of real-world stochastic processes, capturing richer and more faithful intermediate dynamics. By parameterizing the Wasserstein manifold, we lift the bridge problem to a tractable geodesic computation in a finite-dimensional space. Unlike computationally expensive iterative solutions, our contact Wasserstein geodesic (CWG) is naturally implemented via a ResNet architecture and relies on a non-iterative solver with near-linear complexity. Furthermore, CWG supports guided generation by modulating a task-specific distance metric. We validate our framework on tasks including manifold navigation, molecular dynamics predictions, and image generation, demonstrating its practical benefits and versatility.

This paper puts forward the vision of creating a library of neural-network-based models for power system simulations. Traditional numerical solvers struggle with the growing complexity of modern power systems, necessitating faster and more scalable alternatives. Physics-Informed Neural Networks (PINNs) offer promise to solve fast the ordinary differential equations (ODEs) governing power system dynamics. This is vital for the reliability, cost optimization, and real-time decision-making in the electricity grid. Despite their potential, standardized frameworks to train PINNs remain scarce. This poses a barrier for the broader adoption and reproducibility of PINNs; it also does not allow the streamlined creation of a PINN-based model library. This paper addresses these gaps. It introduces a Python-based toolbox for developing PINNs tailored to power system components, available on GitHub this https URL com/radiakos/PowerPINN. Using this framework, we capture the dynamic characteristics of a 9th-order system, which is probably the most complex power system component trained with a PINN to date, demonstrating the toolbox capabilities, limitations, and potential improvements. The toolbox is open and free to use by anyone interested in creating PINN-based models for power system components.

Topology optimization of permanent magnet systems consisting of permanent magnets, high permeability iron and air is presented. An implementation of topology optimization for magnetostatics is discussed and three examples are considered. First, the Halbach cylinder is topology optimized with iron and an increase of 15% in magnetic efficiency is shown, albeit with an increase of 3.8 pp. in field inhomogeneity - a value compared to the inhomogeneity in a 16 segmented Halbach cylinder. Following this a topology optimized structure to concentrate a homogeneous field is shown to increase the magnitude of the field by 111% for the chosen dimensions. Finally, a permanent magnet with alternating high and low field regions is considered. Here a $\Lambda_\mathrm{cool}$ figure of merit of 0.472 is reached, which is an increase of 100% compared to a previous optimized design.

Machine Learning (ML) models have, in contrast to their usefulness in molecular dynamics studies, had limited success as surrogate potentials for reaction barrier search. It is due to the scarcity of training data in relevant transition state regions of chemical space. Currently, available datasets for training ML models on small molecular systems almost exclusively contain configurations at or near equilibrium. In this work, we present the dataset Transition1x containing 9.6 million Density Functional Theory (DFT) calculations of forces and energies of molecular configurations on and around reaction pathways at the wB97x/6-31G(d) level of theory. The data was generated by running Nudged Elastic Band (NEB) calculations with DFT on 10k reactions while saving intermediate calculations. We train state-of-the-art equivariant graph message-passing neural network models on Transition1x and cross-validate on the popular ANI1x and QM9 datasets. We show that ML models cannot learn features in transition-state regions solely by training on hitherto popular benchmark datasets. Transition1x is a new challenging benchmark that will provide an important step towards developing next-generation ML force fields that also work far away from equilibrium configurations and reactive systems.

The performance of a combined solar photovoltaic (PV) and thermoelectric generator (TEG) system is examined using an analytical model for four different types of commercial PVs and a commercial bismuth telluride TEG. The TEG is applied directly on the back of the PV, so that the two devices have the same temperature. The PVs considered are crystalline Si (c-Si), amorphous Si (a-Si), copper indium gallium (di)selenide (CIGS) and cadmium telluride (CdTe) cells. The degradation of PV performance with temperature is shown to dominate the increase in power produced by the TEG, due to the low efficiency of the TEG. For c-Si, CIGS and CdTe PV cells the combined system produces a lower power and has a lower efficiency than the PV alone, whereas for an a-Si cell the total system performance may be slightly increased by the TEG.

Masked pre-training removes random input dimensions and learns a model that can predict the missing values. Empirical results indicate that this intuitive form of self-supervised learning yields models that generalize very well to new domains. A theoretical understanding is, however, lacking. This paper shows that masked pre-training with a suitable cumulative scoring function corresponds to maximizing the model's marginal likelihood, which is de facto the Bayesian model selection measure of generalization. Beyond shedding light on the success of masked pre-training, this insight also suggests that Bayesian models can be trained with appropriately designed self-supervision. Empirically, we confirm the developed theory and explore the main learning principles of masked pre-training in large language models.

Accurately modeling and predicting complex dynamical systems, particularly those involving force exchange and dissipation, is crucial for applications ranging from fluid dynamics to robotics, but presents significant challenges due to the intricate interplay of geometric constraints and energy transfer. This paper introduces Geometric Contact Flows (GFC), a novel framework leveraging Riemannian and Contact geometry as inductive biases to learn such systems. GCF constructs a latent contact Hamiltonian model encoding desirable properties like stability or energy conservation. An ensemble of contactomorphisms then adapts this model to the target dynamics while preserving these properties. This ensemble allows for uncertainty-aware geodesics that attract the system's behavior toward the data support, enabling robust generalization and adaptation to unseen scenarios. Experiments on learning dynamics for physical systems and for controlling robots on interaction tasks demonstrate the effectiveness of our approach.

This paper introduces quantum computing methods for Monte Carlo simulations in power systems which are expected to be exponentially faster than their classical computing counterparts. Monte Carlo simulations is a fundamental method, widely used in power systems to estimate key parameters of unknown probability distributions, such as the mean value, the standard deviation, or the value at risk. It is, however, very computationally intensive. Approaches based on Quantum Amplitude Estimation can offer a quadratic speedup, requiring orders of magnitude less samples to achieve the same accuracy. This paper explains three Quantum Amplitude Estimation methods to replace the Classical Monte Carlo method, namely the Iterative Quantum Amplitude Estimation (IQAE), Maximum Likelihood Amplitude Estimation (MLAE), and Faster Amplitude Estimation (FAE), and compares their performance for three different types of probability distributions for power systems.

Ride-sourcing services offered by companies like Uber and Didi have grown rapidly in the last decade. Understanding the demand for these services is essential for planning and managing modern transportation systems. Existing studies develop statistical models for ride-sourcing demand estimation at an aggregate level due to limited data availability. These models lack foundations in microeconomic theory, ignore competition of ride-sourcing with other travel modes, and cannot be seamlessly integrated into existing individual-level (disaggregate) activity-based models to evaluate system-level impacts of ride-sourcing services. In this paper, we present and apply an approach for estimating ride-sourcing demand at a disaggregate level using discrete choice models and multiple data sources. We first construct a sample of trip-based mode choices in Chicago, USA by enriching household travel survey with publicly available ride-sourcing and taxi trip records. We then formulate a multivariate extreme value-based discrete choice with sampling and endogeneity corrections to account for the construction of the estimation sample from multiple data sources and endogeneity biases arising from supply-side constraints and surge pricing mechanisms in ride-sourcing systems. Our analysis of the constructed dataset reveals insights into the influence of various socio-economic, land use and built environment features on ride-sourcing demand. We also derive elasticities of ride-sourcing demand relative to travel cost and time. Finally, we illustrate how the developed model can be employed to quantify the welfare implications of ride-sourcing policies and regulations such as terminating certain types of services and introducing ride-sourcing taxes.

This paper evaluates the short and medium-term effectiveness of hiring incentives aimed at promoting the permanent conversion of temporary contracts through social contribution exemptions. Using rich administrative data from Tuscany, providing detailed employment histories, we use difference in differences and regression discontinuity designs to exploit a unique change in eligibility criteria in 2018. We find that the incentives immediately increased the probability of conversion, with no evidence of substitution against non-eligible cohorts. However, these positive effects were short-lived and appear to reflect anticipated conversions, as we find null longer-term effects on permanent hirings.

Continuous Glucose Monitoring (CGM) has enabled important opportunities for diabetes management. This study explores the use of CGM data as input for digital decision support tools. We investigate how Recurrent Neural Networks (RNNs) can be used for Short Term Blood Glucose (STBG) prediction and compare the RNNs to conventional time-series forecasting using Autoregressive Integrated Moving Average (ARIMA). A prediction horizon up to 90 min into the future is considered. In this context, we evaluate both population-based and patient-specific RNNs and contrast them to patient-specific ARIMA models and a simple baseline predicting future observations as the last observed. We find that the population-based RNN model is the best performing model across the considered prediction horizons without the need of patient-specific data. This demonstrates the potential of RNNs for STBG prediction in diabetes patients towards detecting/mitigating severe events in the STBG, in particular hypoglycemic events. However, further studies are needed in regards to the robustness and practical use of the investigated STBG prediction models.

Electro-optic modulation is an attractive approach for generating flat, stable, and low-noise optical frequency combs with relatively high power per comb line. However, a key limitation of electro-optic combs is the restricted number of comb lines imposed by the available RF source power. To overcome this limitation, a nonlinear spectral broadening stage is typically employed. The phase noise characteristics of an electro-optic comb are well described by the standard phase noise model, which depends on two parameters: the seed laser and the RF source phase noise. A fundamental question that arises is how nonlinear broadening processes affect the phase noise properties of the expanded comb. To address this, we employ coherent detection, digital signal processing, and subspace tracking. Our experimental results show that the nonlinearly broadened comb preserves the standard phase noise model of the input electro-optic comb. In other words, the nonlinear processes neither introduce additional phase noise terms nor amplify the existing contributions from the seed laser and RF source. Hence, nonlinear broadening can be viewed as equivalent to driving the electro-optic comb with a much higher RF modulation power.

Hybrid fast-charging stations with battery storage and local renewable generation can facilitate low-carbon electric vehicle (EV) charging, while reducing the stress on the distribution grid. This paper proposes energy management strategies for a novel multi-battery design that directly connects its strings to other DC components through a busbar matrix without the need for interfacing power converters. Hence, the energy management system has two degrees of control: (i) allocating strings to other DC microgrid components, in this case a photovoltaic system, two EV fast chargers, and a grid-tie inverter, and (ii) managing the energy exchange with the local distribution grid. For the grid exchange, a basic droop control is compared to an enhanced control including forecasts in the decision making. To this end, this paper evaluates results from multiple Monte Carlo simulations capturing the uncertainty of EV charging. For a realistic charging behaviour in each simulation run, random fast-charging profiles were created based on probability distributions of actual fast-charging data for arrival time, charging duration, and requested energy. The impact of different utilisation levels of the chargers was assessed by varying the average charging instances from 1 to 30 EVs per day. Using actual photovoltaic measurements from different months, the numerical analyses show that the enhanced control increases self-sufficiency by reducing grid exchange, and decreases the number of battery cycles. However, the enhanced control operates the battery closer to its charge limits, which may accelerate calendar ageing.

Accurate time-series forecasting is vital for numerous areas of application such as transportation, energy, finance, economics, etc. However, while modern techniques are able to explore large sets of temporal data to build forecasting models, they typically neglect valuable information that is often available under the form of unstructured text. Although this data is in a radically different format, it often contains contextual explanations for many of the patterns that are observed in the temporal data. In this paper, we propose two deep learning architectures that leverage word embeddings, convolutional layers and attention mechanisms for combining text information with time-series data. We apply these approaches for the problem of taxi demand forecasting in event areas. Using publicly available taxi data from New York, we empirically show that by fusing these two complementary cross-modal sources of information, the proposed models are able to significantly reduce the error in the forecasts.

A metamaterial magnetic flux concentrator is investigated in detail in combination with a Halbach cylinder of infinite length. A general analytical solution to the field is determined and the magnetic figure of merit is determined for a Halbach cylinder with a flux concentrator. It is shown that an ideal flux concentrator will not change the figure of merit of a given magnet design, while the non-ideal will always lower it. The geometric parameters producing maximum figure of merit, i.e. the most efficient devices, are determined. The force and torque between two concentric Halbach cylinders with flux concentrators is determined and the maximum torque is found. Finally, the effect of non-ideal flux concentrators and the practical use of flux concentrators, as well as demagnetization issues, is discussed.

The successful integration of machine learning models into decision support tools for grid operation hinges on effectively capturing the topological changes in daily operations. Frequent grid reconfigurations and N-k security analyses have to be conducted to ensure a reliable and secure power grid, leading to a vast combinatorial space of possible topologies and operating states. This combinatorial complexity, which increases with grid size, poses a significant computational challenge for traditional solvers. In this paper, we combine Physics-Informed Neural Networks with graph-aware neural network architectures, i.e., a Guided-Dropout (GD) and an Edge-Varying Graph Neural Network (GNN) architecture to learn the set points for a grid that considers all probable single-line reconfigurations (all critical N-1 scenarios) and subsequently apply the trained models to N-k scenarios. We demonstrate how incorporating the underlying physical equations for the network equations within the training procedure of the GD and the GNN architectures performs with N-1, N-2, and N-3 case studies. Using the AC Power Flow as a guiding application, we test our methods on the 6-bus, 24-bus, 57-bus, and 118-bus systems. We find that GNN not only achieves the task of contingency screening with satisfactory accuracy but does this up to 400 times faster than the Newton-Raphson power flow solver. Moreover, our results provide a comparison of the GD and GNN models in terms of accuracy and computational speed and provide recommendations on their adoption for contingency analysis of power systems.

Involving residential actors in the energy transition is crucial for its success. Local energy generation, consumption and trading are identified as desirable forms of involvement, especially in energy communities. The potentials for energy communities in the residential building stock are high but are largely untapped in multi-family buildings. In many countries, rapidly evolving legal frameworks aim at overcoming related barriers, e.g. ownership structures, principal-agent problems and system complexity. But academic literature is scarce regarding the techno-economic and environmental implications of such complex frameworks. This paper develops a mixed-integer linear program (MILP) optimisation model for assessing the implementation of multi-energy systems in an energy community in multi-family buildings with a special distinction between investor and user. The model is applied to the German Tenant Electricity Law. Based on hourly demands from appliances, heating and electric vehicles, the optimal energy system layout and dispatch are determined. The results contain a rich set of performance indicators that demonstrate how the legal framework affects the technologies' interdependencies and economic viability of multi-energy system energy communities. Certain economic technology combinations may fail to support national emissions mitigation goals and lead to lock-ins in Europe's largest residential building stock. The subsidies do not lead to the utilisation of a battery storage. Despite this, self-sufficiency ratios of more than 90% are observable for systems with combined heat and power plants and heat pumps. Public CO2 mitigation costs range between 147.5-272.8 EUR/tCO2. Finally, the results show the strong influence of the heat demand on the system layout.