Search

Electricity generated from renewable energy sources has been established as an efficient remedy for both energy shortages and the environmental pollution stemming from conventional energy production methods. Solar and wind power are two of the most dominant renewable energy sources. The accurate forecasting of the energy generation of those sources facilitates their integration into electric grids, by minimizing the negative impact of uncertainty regarding their management and operation. This paper proposes a novel methodology for deterministic wind and solar energy generation forecasting for multiple generation sites, utilizing multi-location weather forecasts. The method employs a U-shaped Temporal Convolutional Auto-Encoder (UTCAE) architecture for temporal processing of weather-related and energy-related time-series across each site. The Multi-sized Kernels convolutional Spatio-Temporal Attention (MKST-Attention), inspired by the multi-head scaled-dot product attention mechanism, is also proposed aiming to efficiently transfer temporal patterns from weather data to energy data, without a priori knowledge of the locations of the power stations and the locations of provided weather data. The conducted experimental evaluation on a day-ahead solar and wind energy forecasting scenario on five datasets demonstrated that the proposed method achieves top results, outperforming all competitive time-series forecasting state-of-the-art methods.

Context: In the realm of software development, maintaining high software quality is a persistent challenge. However, this challenge is often impeded by the lack of comprehensive understanding of how specific code modifications influence quality metrics. Objective: This study ventures to bridge this gap through an approach that aspires to assess and interpret the impact of code modifications. The underlying hypothesis posits that code modifications inducing similar changes in software quality metrics can be grouped into distinct clusters, which can be effectively described using an AI language model, thus providing a simple understanding of code changes and their quality implications. Method: To validate this hypothesis, we built and analyzed a dataset from popular GitHub repositories, segmented into individual code modifications. Each project was evaluated against software quality metrics pre and post-application. Machine learning techniques were utilized to cluster these modifications based on the induced changes in the metrics. Simultaneously, an AI language model was employed to generate descriptions of each modification's function. Results: The results reveal distinct clusters of code modifications, each accompanied by a concise description, revealing their collective impact on software quality metrics. Conclusions: The findings suggest that this research is a significant step towards a comprehensive understanding of the complex relationship between code changes and software quality, which has the potential to transform software maintenance strategies and enable the development of more accurate quality prediction models.

On-chip optical isolators, functioning as unidirectional gates for light, play a crucial role in maintaining signal integrity, preventing laser destabilization, and fortifying the overall performance of optical systems. In this paper, we propose a five-layered heterostructure consisting of a magneto-optic material sandwiched between parallel dielectric slab waveguides. Under TMOKE configuration, the coupled optical modes undergo an electromagnetic profile transformation that can be harnessed to confine the input mode in one waveguide during forward propagation and in the other during backward propagation. Together with radiative subwavelength gratings, such a system can provide a 20dB isolation ratio with negligible insertion losses.

Far-infrared observations from the \textit{Herschel Space Observatory} are used to estimate the infrared (IR) properties of ultraviolet-selected galaxies. We stack the PACS (100, 160 $\mu \mathrm{m}$) and SPIRE (250, 350 and 500$\mu \mathrm{m}$) maps of the Chandra deep field south (CDFS) on a source list of galaxies selected in the rest-frame ultraviolet (UV) in a redshift range of $0.6-1.2$. This source list is created using observations from the XMM-OM telescope survey in the CDFS using the UVW1 (2910 {\AA}) filter. The stacked data are binned according to the UV luminosity function of these sources, and the average photometry of the UV-selected galaxies is estimated. By fitting modified black bodies and IR model templates to the stacked photometry, average dust temperatures and total IR luminosity are determined. The luminosity-weighted average temperatures are consistent with a weak trend of increasing temperature with redshift found by previous studies. Infrared excess, unobscured, and obscured star formation rate (SFR) values are obtained from the UV and IR luminosities. We see a trend in which dust attenuation increases as UV luminosity decreases. It remains constant as a function of IR luminosities at fixed redshift across the luminosity range of our sources. In comparison to local luminous infrared galaxies with similar SFRs, the higher redshift star-forming galaxies in the sample show a lesser degree of dust attenuation. Finally, the inferred dust attenuation is used to correct the unobscured SFR density in the redshift range $0.6-1.2$. The dust-corrected SFR density is consistent with measurements from IR-selected samples at similar redshifts., Comment: 19 pages, 15 figure, accepted for publication in MNRAS

An increasing number of models and frameworks for Virtual Assistant (VA) development exist nowadays, following the progress in the Natural Language Processing (NLP) and Natural Language Understanding (NLU) fields. Regardless of their performance, popularity, and ease of use, these frameworks require at least basic expertise in NLP and software engineering, even for simple and repetitive processes, limiting their use only to the domain and programming experts. However, since the current state of practice of VA development is a straightforward process, Model-Driven Engineering approaches can be utilized to achieve automation and rapid development in a more convenient manner. To this end, we present \textit{dFlow}, a textual Domain-Specific Language (DSL) that offers a simplified, reusable, and framework-agnostic language for creating task-specific VAs in a low-code manner. We describe a system-agnostic VA meta-model, the developed grammar, and all essential processes for developing and deploying smart VAs. For further convenience, we create a cloud-native architecture and expose it through the Discord platform. We conducted a large-scale empirical evaluation with more than 200 junior software developers and collected positive feedback, indicating that dFlow can accelerate the entire VA development process, while also enabling citizen and software developers with minimum experience to participate.

Aims: We leverage the largest available Atacama Large Millimetre/submillimetre Array (ALMA) survey from the archive (A$^3$COSMOS) to study to study infrared luminosity function and dust-obscured star formation rate density of sub-millimeter/millimeter (sub-mm/mm) galaxies from $z=0.5\,-\,6$. Methods: The A$^3$COSMOS survey utilizes all publicly available ALMA data in the COSMOS field, therefore having inhomogeneous coverage in terms of observing wavelength and depth. In order to derive the luminosity functions and star formation rate densities, we apply a newly developed method that corrects the statistics of an inhomogeously sampled survey of individual pointings to those representing an unbiased blind survey. Results: We find our sample to mostly consist of massive ($M_{\star} \sim 10^{10} - 10^{12}$ $\rm M_{\odot}$), IR-bright ($L_* \sim 10^{11}-10^{13.5} \rm L_{\odot}$), highly star-forming (SFR $\sim 100-1000$ $\rm M_{\odot}$ $\rm yr^{-1}$) galaxies. We find an evolutionary trend in the typical density ($\Phi^*$) and luminosity ($L^*$) of the galaxy population, which decrease and increase with redshift, respectively. Our IR LF is in agreement with previous literature results and we are able to extend to high redshift ($z > 3$) the constraints on the knee and bright-end of the LF, derived by using the Herschel data. Finally, we obtain the SFRD up to $z\sim 6$ by integrating the IR LF, finding a broad peak from $z \sim 1$ to $z \sim 3$ and a decline towards higher redshifts, in agreement with recent IR/mm-based studies, within the uncertainties, thus implying the presence of larger quantities of dust than what is expected by optical/UV studies., Comment: 17 pages, 15 figures, 5 tables, accepted for publication on A&A

Energy time-series analysis describes the process of analyzing past energy observations and possibly external factors so as to predict the future. Different tasks are involved in the general field of energy time-series analysis and forecasting, with electric load demand forecasting, personalized energy consumption forecasting, as well as renewable energy generation forecasting being among the most common ones. Following the exceptional performance of Deep Learning (DL) in a broad area of vision tasks, DL models have successfully been utilized in time-series forecasting tasks. This paper aims to provide insight into various DL methods geared towards improving the performance in energy time-series forecasting tasks, with special emphasis in Greek Energy Market, and equip the reader with the necessary knowledge to apply these methods in practice., Comment: 13 papges, 4 figures

Clinical trials are a critical component of evaluating the effectiveness of new medical interventions and driving advancements in medical research. Therefore, timely enrollment of patients is crucial to prevent delays or premature termination of trials. In this context, Electronic Health Records (EHRs) have emerged as a valuable tool for identifying and enrolling eligible participants. In this study, we propose an automated approach that leverages ChatGPT, a large language model, to extract patient-related information from unstructured clinical notes and generate search queries for retrieving potentially eligible clinical trials. Our empirical evaluation, conducted on two benchmark retrieval collections, shows improved retrieval performance compared to existing approaches when several general-purposed and task-specific prompts are used. Notably, ChatGPT-generated queries also outperform human-generated queries in terms of retrieval performance. These findings highlight the potential use of ChatGPT to enhance clinical trial enrollment while ensuring the quality of medical service and minimizing direct risks to patients., Comment: Under Review

We examine the far-IR properties of a sample of 5391 optically selected QSOs in the 0.544.7, using SPIRE data from Herschel-ATLAS. We split the sample in a grid of 74 luminosity-redshift bins and compute the average optical-infrared spectral energy distribution (SED) in each bin. By normalising an intrinsic AGN template to the AGN optical power (at 5100A) we decompose the total infrared emission (L_IR; 8-1000um) into an AGN (L_IR,AGN) and star-forming component (L_IR,SF). We find that the AGN contribution to L_IR increases as a function of AGN power which manifests as a reduction of the `far-IR bump' in the average QSO SEDs. We note that L_IR,SF does not correlate with AGN power; the mean star formation rates (SFRs) of AGN host galaxies are a function of redshift only and they range from ~6 Msun/yr at z~0 to a plateau of <200 Msun/yr at z~2.6. Our results indicate that the accuracy of far-IR emission as a proxy for SFR decreases with increasing AGN luminosity. We show that, at any given redshift, observed trends between infrared luminosity (whether monochromatic or total) and AGN power (in the optical or X-rays) can be explained by a simple model which is the sum of two components: (A) the infrared emission from star-formation, uncorrelated with AGN power and (B) the infrared emission from AGN, directly proportional to AGN power in the optical or X-rays., Comment: 20 pages, 10 figures, accepted for publication in MNRAS

This paper is an overview of the Machine Learning Operations (MLOps) area. Our aim is to define the operation and the components of such systems by highlighting the current problems and trends. In this context, we present the different tools and their usefulness in order to provide the corresponding guidelines. Moreover, the connection between MLOps and AutoML (Automated Machine Learning) is identified and how this combination could work is proposed., Comment: 8 pages, figure 5, table 4

Far-infrared (far-IR)/sub-mm emission linked to AGN-heated dust has been a topic of contention for many years. Results have been diverse and various views have been presented. The empirical AGN SED derived by Symeonidis et al. (2016, hereafter S16) has more far-IR/sub-mm emission than other SEDs in the literature, and thus it is contested by other works which argue that its luminosity in that part of the spectrum is overestimated. Here, I investigate this topic and the concerns raised over the S16 AGN SED. I also examine the differences between the S16 AGN SED and other commonly-used empirical AGN SEDs. My findings show that the reasons proposed by other works as to why the S16 AGN SED is not a reasonable representation of AGN emission in the far-IR/sub-mm, do not hold., Comment: 10 pages, 9 figures, accepted for publication in MNRAS

We introduce pygrank, an open source Python package to define, run and evaluate node ranking algorithms. We provide object-oriented and extensively unit-tested algorithm components, such as graph filters, post-processors, measures, benchmarks and online tuning. Computations can be delegated to numpy, tensorflow or pytorch backends and fit in back-propagation pipelines. Classes can be combined to define interoperable complex algorithms. Within the context of this paper we compare the package with related alternatives and demonstrate its flexibility and ease of use with code examples., Comment: 6 pages, 1 figure, 2 tables, 3 code snippets