67639 research outputs found
Sort by
Response to the Comment on “Mitigating Delamination in Perovskite/Silicon Tandem Solar Modules”
No full textIn response to the comment on article ‘Mitigating Delamination in Perovskite/Silicon Tandem Solar Modules’, we address each specific concern raised, including the assumptions made in the analysis of the peel test data and the methodology used in conducting the peel tests. We believe that this is a good opportunity to clarify some fundamental aspects of peel tests that the community may sometimes find confusing. So we hope this extensive response, formatted as a discussion, will help the community to make better usage of this testing protocol
Randomized Greedy Algorithms for Neural Network Optimization in Solving Partial Differential Equations
No full textGreedy algorithms have been successfully analyzed and applied in training neural networks for solving variational problems, ensuring guaranteed convergence orders. In this paper, we extend the analysis of the orthogonal greedy algorithm (OGA) to convex optimization problems arising from the solution of partial differential equations, establishing its optimal convergence rate. This result broadens the applicability of OGA by generalizing its optimal convergence rate from function approximation to convex optimization problems. In addition, we also address the issue regarding practical applicability of greedy algorithms, which is due to significant computational costs from the subproblems that involve an exhaustive search over a discrete dictionary. We propose to use a more practical approach of randomly discretizing the dictionary at each iteration of the greedy algorithm. We quantify the required size of the randomized discrete dictionary and prove that, with high probability, the proposed algorithm realizes a weak greedy algorithm, achieving optimal convergence orders. Through numerous numerical experiments on function approximation, linear and nonlinear elliptic partial differential equations, we validate our analysis on the optimal convergence rate and demonstrate the advantage of using randomized discrete dictionaries over a deterministic one by showing orders of magnitude reductions in the size of the discrete dictionary, particularly in higher dimensions.The authors would like to thank Dr. Jongho Park for helpful discussions on the paper.Open access publishing provided by King Abdullah University of Science and Technology (KAUST). This work is supported by KAUST Baseline Research Fund
Correlated Quantization for Faster Nonconvex Distributed Optimization
No full textQuantization [Alistarh et al., 2017] is an important (stochastic) compression technique that reduces the volume of transmitted bits during each communication round in distributed model training. Suresh et al. [2022] introduce correlated quantizers and show their advantages over independent counterparts by analyzing distributed SGD communication complexity. We analyze the forefront distributed non-convex optimization algorithm MARINA [Gorbunov et al., 2022] utilizing the proposed correlated quantizers and show that it outperforms the original MARINA and distributed SGD of Suresh et al. [2022] with regard to the communication complexity. We significantly refine the original analysis of MARINA without any additional assumptions using the weighted Hessian variance [Tyurin et al., 2022], and then we expand the theoretical framework of MARINA to accommodate a substantially broader range of potentially correlated and biased compressors, thus dilating the applicability of the method beyond the conventional independent unbiased compressor setup. Extensive experimental results corroborate our theoretical findings.The work was supported by funding from King Abdullah University of Science and Technology (KAUST): i) KAUST Baseline Research Scheme, ii) Center of Excellence for Generative AI, under award number 5940, iii) SDAIA-KAUST Center of Excellence in Artificial Intelligence and Data Science
Terahertz Band UAV Base Stations for Post-Disaster Communication
No full textIn case of emergency, data rate demand significantly increases, whereas the current disaster relief bands are not sufficient to supply this demand. Also, existing terrestrial base station infrastructure may be out of use. On the other hand, terahertz (THz) communication attracts attention due to the available massive bandwidth, promising a very high data rate. Moreover, employing an unmanned aerial vehicle (UAV) base station (BS) is a suitable solution for reliable communication links during search and rescue operations. Hence, in this work, we propose and analyze THz band UAV BS for disasters; consequently, a wideband channel in an unstandardized THz band can be utilized for disaster relief, providing sufficient capacity for high data rate demanding applications. In this study, the capacity performance and the outage probability of THz band UAV BS are evaluated for ground-to-UAV communication links compared to existing disaster relief bands considering beam misalignment fading, turbulence fading, and rain and fog effects, yielding encouraging results. In case a disaster affects a huge area, causing terrestrial base stations to be out-of-use, two alternative scenarios are studied. As a first scenario, the UAV-to-satellite communication link is considered for the THz band, sub-6 GHz disaster relief band, and free space optics, resulting in dramatic capacity reduction because of the high loss in the UAV-to-satellite link. As a second scenario, multi-hop UAV communication is considered by deploying multiple UAVs to connect the nearest available terrestrial BS. Simulation results reveal that the THz band UAV BS stands out as a promising candidate for post-disaster communication since it can provide tens of Gbps ergodic capacity for the ground-to-UAV link, and up to 5 Gbps ergodic capacity is obtained in the multi-hop communication scenario in case of outage in large areas
On the interaction of mode-1 Internal Solitary Wave with higher modes to the west of northern Nicobar Islands
No full textThe synthetic aperture radar (SAR) image acquired on 23 April 2004 at 03:33 UTC by the Envisat shows the separation of short wavelength internal waves from the westward propagating mode-1 ISW generating from a shallow ridge connecting Batti Malv and Chowra Islands (SBM) of the northern Nicobar Islands, India. To investigate the presence of tailless mode-1 ISWs, several other SAR images near the SBM are analyzed. A particular SAR image captured on 12 October 2007 at 15:53 UTC reveals two different higher-mode ISWs in the path of the mode-1 ISW. Fitting the locations of the ISW signatures onto a time-distance curve shows that one of the higher modes is generated locally, while others are generated from SBM. Using simulations of a nonhydrostatic numerical model SUNTANS, we show that the SBM generated westward propagating mode-1 ISW overtakes the locally generated mode-3 ISW and a mode-2 ISW formed over the SBM in the previous tidal cycle. It is shown that the interaction process of the first mode with higher modes results in the formation of short internal waves trailing behind higher mode waves, pertaining to the resonance between the tail of mode-1 ISW and the higher mode solitary wave. Hence, close to a spring tide, with a combination of SAR images and numerical simulations, we show the dynamical process of westward propagating mode-1 ISW from SBM with the higher modes.We thank Theo Gerkema for sharing the codes of the linear internal tide model. We also thank Manikandan Mathur for his discussions. The SUNTANS source code was downloaded from https://github.com/ofringer/suntans. Processing of SUNTANS output is done using the Python package SODA (https://github.com/mrayson/soda). The unstructured grid is generated using the stompy (https://github.com/rustychris/stompy) package
High-friction limit for bipolar Euler-Riesz systems
No full textWe consider a bipolar Euler–Riesz system and rigorously justify the high-friction limit of weak solutions towards a bipolar aggregation-diffusion system with Riesz interactions. The analysis is carried out via the relative entropy method in the regime where smooth solutions of the limiting equations exist. This extends previous results on the high-friction limit of bipolar Euler–Poisson systems to a more general class of interactions, and extends the one-species Euler–Riesz case to the bipolar setting.This research was partially funded by the Austrian Science Fund (FWF), project number 10.55776/F65
A robust Mixed Finite Element model for coupled Thermo-Hydro-Mechanical problems in unsaturated porous media
No full textSimulating Thermo-Hydro-Mechanical (THM) problems in unsaturated porous media presents significant challenges due to the high nonlinearities of the coupled processes, and the potential for numerical instabilities that may lead to unphysical oscillations in pressure, stress, and temperature solutions. The Mixed Finite Element (MFE) method, known for its accurate local mass conservation even in heterogeneous domains and on unstructured meshes, is widely used to discretize fluid flow in both saturated and unsaturated porous media. However, its application in coupled THM processes under variably saturated conditions has not been thoroughly explored. In this work, a robust MFE scheme based on the lowest order Raviart-Thomas space is developed for the fluid flow and heat transport in deformable unsaturated porous media. This method is integrated with the Crouzeix-Raviart finite element method for the displacement field. To prevent unphysical oscillations induced by the hyperbolic convection term in the heat transport equation, the MFE method is combined with an upwind edge/face centered finite volume scheme. The degrees of freedom of the developed formulation are the hydraulic head, the temperature, and the displacement vectors assigned at the mesh edges. To avoid splitting errors, the coupled THM equations are solved simultaneously using a monolithic scheme. The Method of Lines is employed to transform the partial differential equations into a system of nonlinear ordinary differential equations integrated in time with high-order methods using the DASPK time solver. The developed model is validated by comparisons against analytical and finite element solutions for the one-dimensional thermal consolidation problem. Numerical experiments are performed under saturated and unsaturated conditions to show the robustness of the developed model for the simulation of THM problems in variably saturated porous media
Cyber Insurance Design for Load Variation and Load Curtailment in Distribution Grids
No full textA growing number of renewable energy resources (RES) is continuously integrated into power systems contributing towards the green energy and net zero transition. However, the uncertainties of RES generation and load variations often lead to potentially high operational costs. Furthermore, added threats of cyber-attacks such as load-altering attacks (LAAs) via distributed energy resource/load interfaces can lead to substantial load variations. In this paper, we propose a cyber insurance framework to mitigate excessive expenses in load variation conditions while considering a renewable-rich grid. We investigate how operational costs and load curtailments can be influenced due to load variations by solving a bi-level optimization problem and performing Monte Carlo simulations. We use the semi-Markov process (SMP) to estimate the probability of extreme situations that the cyber insurance covers, whose premium and coverage are designed based on its value at risk and tail value at risk. A modified IEEE-118 test bus system considering PV generations, battery storage, and energy market interaction, along with a load curtailment strategy, is used to evaluate the proposed framework. Results show that load variations of up to 30% can lead to doubling the daily operational cost, demonstrating the feasibility of using a cyber insurance policy to hedge against financial risk
An interpretable and adaptive autoencoder for efficient tissue deconvolution
No full textDeconvolution models are a powerful tool for extracting cell-type-specific information from bulk gene expression profiles. Current methods leverage advanced machine learning models and high-resolution sequencing, like single-cell RNA-sequencing, showing promising results across diverse tissues and conditions. However, they still present important limitations: First, many depend on selecting a robust reference, which can strongly affect the deconvolution. Second, pseudobulk data used for training and real bulk RNA-seq samples often exhibit strong distribution shifts, which are currently unaccounted for. Finally, most deconvolution approaches behave as black boxes, which can compromise the reliability of the results. Here, we present Sweetwater, an adaptive and interpretable autoencoder that efficiently deconvolves bulk samples leveraging multiple classes of reference data. Moreover, we propose an improved way of generating training data from a mixture of FACS-sorted FASTQ files, reducing platform-specific biases and outperforming current single-cell-based references. Furthermore, we introduce a gold standard dataset to facilitate fair and accurate evaluation of deconvolution approaches. Finally, we demonstrate that Sweetwater adapts effectively to deconvolved samples during training, uncovering biologically meaningful patterns and enhancing result’s reliability. Sweetwater is available at https://doi.org/10.6084/m9.figshare.29609180, and we anticipate it will expedite the accurate examination of high-throughput clinical data across diverse applications.Author contributions: J.F.: Conceptualization, software, visualization, methodology, and writing. N.L.: Conceptualization, software, visualization, methodology, and writing. A.D.: Data generation (bone marrow dataset). I.M.: Visualization and writing. G.S.: Conceptualization and visualization. M.B.: Benchmarking process (CIBERSORTx). R.K.: Conceptualization. A.G.: Conceptualization. C.F.: Conceptualization and methodology. I.O.: Conceptualization, supervision, methodology, and writing. M.H.: Conceptualization, supervision, methodology, and writing.
This work was supported by the following grants: Instituto de Salud Carlos III (ISCIII) through the project “AC23_2/00016”; TED2021-131300B-I00 funded by MCIN/AEI/10.13039/501100011033; RYC2021-033127-I funded by MCIN/AEI/10.13039/501100011033 and the European Union “NextGenerationEU”/PRTR; PID2023-151980OB-I00 funded by MCIN/AEI/10.13039/501100011033 and FEDER, UE. Fulbright Predoctoral Research Program [PS00342367], a Fundacion Ramon Areces predoctoral grant, MCIN/AEI RYC2019-028578-I, Gipuzkoa Fellows (2022-FELL-000003-01), MCIN/AEI (PID2021-126718OA-I00), Elkartek and RIS3 Grants from the Basque Government [KK-2023/00001, 2023333040], and a grant from the Fundacion Ramon Areces. Funding to pay the Open Access publication charges for this article was provided by Ramon y Cajal Fellowships, Ministry of Science and Innovation, Spain
Electron accumulation across the perovskite layer enhances tandem solar cells with textured silicon
No full textReducing charge carrier transport losses, improving selectivity, and minimizing non-radiative recombination are essential for enhancing the efficiency and stability of perovskite/silicon tandem solar cells. We used a hybrid two-step perovskite deposition method that is compatible with industry-standard textured silicon, incorporating a perovskite surface treatment based on 1,3-diaminopropane dihydroiodide. The interaction of this molecule with the perovskite surface increased the majority charge carrier concentration at the electron-selective contact, which reduced interfacial recombination. Simultaneously, this field-effect passivation increased the electron concentration across the entire intrinsic perovskite absorber, which increased conductivity and reduced transport losses. Combined, this yields high-performance, fully-textured perovskite/silicon tandem solar cells, achieving a 1-sun AM1.5G conversion efficiency of 33.1% with an open-circuit voltage of 2.01 volts, and an extended outdoor stability in the Red Sea Coast.We thank Luis V. T. Merino, Pia Dally, Arsalan Razzaq, Thomas G. Allen, Abdulrahman El Labban, Atteq ur Rehman, Shruti Sarwade, Semen Shikin, and Rey Guadalquiver for fruitful discussions.
This work was funded by the Fraunhofer LIGHTHOUSE PROJECT MaNiTU, the German Federal Ministry for Economic Affairs and Climate Action under Contract Nos. 03EE1086A (PrEsto), 03EE1182A, and 03EE1182B (Perle), the King Abdullah University of Science and Technology (KAUST) Research Funding Office under Award Nos. ORA-CRG10-2021-4669 and ORA-CRG10-2021-4681. H.P. thanks the Finnish Foundation for Technology Promotion, decision 8462, for funding the work. C.E.P. acknowledges support from KAUST Global Fellowship Program under Award No. ORA-2022-5002. E.A. acknowledges financial support from the European Research Council (ERC) under the European Union’s Horizon Europe Research and Innovation Program (INPERSPACE, Grant Agreement No. 101077006). C.M. acknowledges support from Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under SPP 2196: Perovskite Semiconductors: From Fundamental Properties to Devices, project number 402726906. J.B. acknowledges support The Vector Stiftung for funding her research group