Large Space Research Articles

Bayesian Optimization with Experience for Fast Development of Monolithic Tandem Solar Cells: Simulation Case Study

AbstractMachine learning (ML) is gaining more attention in photovoltaic research and will be a vital tool in reaching record‐high power conversion efficiencies (PCE) in the near future. One area, where ML is significantly beneficial is reducing the number of experiments needed to find the optimum combination of parameters in solar cell fabrication. Bayesian optimization (BO) provides routes for quickly identifying optimum parameters in problems with large parameter space. In this work, BO algorithms utilizing previous knowledge are demonstrated to result in faster optimization of tandem solar cells, a technology rapidly gaining more interest due to its potential to deliver record‐high performance. Namely, it is shown that in the space of all possible parameter combinations that take ≈88 years to evaluate, optimum PCE can be obtained in 20 min. Moreover, it is demonstrated that methods utilizing previous knowledge outperform those that do not by yielding an increase of ≈8.9%abs. in the 1st iteration and requiring 5× less time to reach a target PCE of 38.5% across 20 different trials. Results from this work help accelerate the development of tandem solar cells by removing the need for large numbers of experiments in identifying optimum parameters.

Learning Logic Specifications for Policy Guidance in POMDPs: an Inductive Logic Programming Approach

Partially Observable Markov Decision Processes (POMDPs) are a powerful framework for planning under uncertainty. They allow to model state uncertainty as a belief probability distribution. Approximate solvers based on Monte Carlo sampling show great success to relax the computational demand and perform online planning. However, scaling to complex realistic domains with many actions and long planning horizons is still a major challenge, and a key point to achieve good performance is guiding the action-selection process with domain-dependent policy heuristics which are tailored for the specific application domain. We propose to learn high-quality heuristics from POMDP traces of executions generated by any solver. We convert the belief-action pairs to a logical semantics, and exploit data- and time-efficient Inductive Logic Programming (ILP) to generate interpretable belief-based policy specifications, which are then used as online heuristics. We evaluate thoroughly our methodology on two notoriously challenging POMDP problems, involving large action spaces and long planning horizons, namely, rocksample and pocman. Considering different state-of-the-art online POMDP solvers, including POMCP, DESPOT and AdaOPS, we show that learned heuristics expressed in Answer Set Programming (ASP) yield performance superior to neural networks and similar to optimal handcrafted task-specific heuristics within lower computational time. Moreover, they well generalize to more challenging scenarios not experienced in the training phase (e.g., increasing rocks and grid size in rocksample, incrementing the size of the map and the aggressivity of ghosts in pocman).

Flexural behaviors of a novel precast hollow UHPC composite beam reinforced with inverted T-shaped steel: Experimental investigation and theoretical analysis

To fully utilize the excellent mechanical properties and durability of ultra-high performance concrete (UHPC) as well as reduce the production cost, a novel UHPC encasing inverted T-steel (PUES) composite beam with hollow section is proposed in this paper. This paper aims to study the flexural behaviors of PUES beams, a bending test program consisted of one solid PUES beam and seven hollow PUES beams is carried out. The test parameters include thickness of inverted T-shape steel flange, tensile reinforcement ratio and stud spacing. The results showed that the PUES beams failed in the typical flexural failure mode and exhibited excellent ductility without obvious drop in load. Compared to composite beam with solid section, the hollow section had negligible influence on the flexural performance of PUES beams, while it saved 26 % UHPC used compared with the one with solid section. The increasing of the thickness of inverted T-shape steel flange and the reinforcement ratio effectively improved the stiffness and the flexural capacity of PUES beams, while the larger diameter of reinforcement incurred an obvious reduction of ductility. An excessively large stud spacing resulted in a poor composite behavior and a reduction of flexural capacity. Based on the analysis of the experimental results, calculating methods considering the shear connection degree between inverted T-steel and peripheral reinforced UHPC were developed to predict the ultimate flexural capacity of the hollow PUES beams. This study provides a reference for further research and practical engineering applications of this innovation composite beam.

Differential Fault and Algebraic Equation Combined Analysis on PICO

In modern information technology, research on block cipher security is imperative. Concerning the ultra lightweight block cipher PICO, there has been only one study focused on recovering its complete master key, with a large search space of 264, and no fault analysis yet. This paper proposes a new fault analysis approach, combining differential fault and algebraic equation techniques. It achieved the recovery of PICO’s entire master key with 40 faults in an average time of 0.57 h. S-box decomposition was utilized to optimize our approach, reducing the time by a remarkable 75.83% under the identical 40-fault condition. Furthermore, PICO’s complete master key could be recovered with 28 faults in an average time of 0.78 h, indicating a significant 237 reduction in its search space compared to the previous study. This marks the first fault analysis on PICO. Compared to conventional fault analysis methods DFA (differential fault analysis) and AFA (algebraic fault analysis), our approach outperforms in recovering PICO’s entire master key, highlighting the cruciality of key expansion complexity in block cipher security. Therefore, our approach could serve to recover master keys of block ciphers with comparably complicated key expansions, and production of more secure block ciphers could result.

An integrally bilayered flexible cathode woven from CNTs for homogeneous hosting and fast electrocatalytic conversion of Li2O2 in Li-O2 battery

Due to the ultrahigh theoretical energy density, Li-O2 batteries are considered as potential candidates for power accessories for wearable electronic devices. An effective oxygen-breathing cathode is essential to enable high cyclic stability in Li-O2 batteries. Herein, an integrally bilayered, freestanding flexible Li-O2 battery cathode is constructed by in-situ growth of a layer of microflowers (which is an assembly of nitrogen-doped carbon nanotubes embedded with cobalt nanoparticles) on the surface of a CNT-interwoven film. Owing to the integrally bilayered structural feature and its compositional merits, such three-dimensionally conductive network provides a large space and surface area for homogeneous hosting and also enables efficient electrocatalytic conversion of the insulating Li2O2 solids. Thus obtained cathode delivers a large full capacity of 8.03 mAh cm−2; it also exhibits a low charge potential of approximately 3.6 V and excellent cyclic stability for stable operation up to nearly 2 months at a capacity of 0.1 and 1 mAh cm−2. Moreover, being flexible and binder-free, the cathode is highly robust, which can withstand repetitive folding and unfolding. A co-axial Li-air battery was fabricated accordingly, which exhibits excellent tolerance to bending deformation, showing great potential for application in flexible and wearable electronics.

Voices from the field: exploring practitioners' experiences and perceptions of the voices of athletes (VOA) in the pacific islands.

This exploratory study aims to gain insights into practitioners' experiences and perceptions of the "Voices of Athletes" (VOA), a specialized athlete support program, developed and implemented in the Pacific Islands. Semi-structured interviews enabled participants to share detailed experiences, with 14 practitioners participating. Thematic analysis of the data identified five key themes: "Fostering Athlete Empowerment", "Impact of Sport on Education and Social Change", "Expanding VOA's Reach and Impact", "VOA's Role in Preparing Athletes for Post-Sport Life", and "Optimizing VOA Implementation and Experience". The findings demonstrated that the VOA plays a critical role in empowering athletes and assisting them in becoming leaders within their societies. Practitioners emphasized the power of sport as an effective channel for education and inspiration, and the potential for the VOA framework to be applied in various contexts and regions. The study also revealed that the VOA helps athletes prepare for life after sport, contributing to increased self-esteem, development of transferable skills, and awareness of their social roles. The findings also emphasized the need for VOA improvements, including enhanced interactivity, larger spaces, and financial support. Recognition from sport governing bodies and coaches could broaden the program's reach and impact. Incorporating internship schemes within the VOA or related programs could address life after sport more effectively. Developing written VOA guidelines would ensure consistent, sustainable delivery, supporting its potential for wider implementation and adaptation, contributing to holistic development for athletes and young people globally.

RM-GPT: Enhance the comprehensive generative ability of molecular GPT model via LocalRNN and RealFormer

Due to the surging of cost, artificial intelligence-assisted de novo drug design has supplanted conventional methods and become an emerging option for drug discovery. Although there have arisen many successful examples of applying generative models to the molecular field, these methods struggle to deal with conditional generation that meet chemists’ practical requirements which ask for a controllable process to generate new molecules or optimize basic molecules with appointed conditions. To address this problem, a Recurrent Molecular-Generative Pretrained Transformer model is proposed, supplemented by LocalRNN and Residual Attention Layer Transformer, referred to as RM-GPT. RM-GPT rebuilds GPT model’s architecture by incorporating LocalRNN and Residual Attention Layer Transformer so that it is able to extract local information and build connectivity between attention blocks. The incorporation of Transformer in these two modules enables leveraging the parallel computing advantages of multi-head attention mechanisms while extracting local structural information effectively. Through exploring and learning in a large chemical space, RM-GPT absorbs the ability to generate drug-like molecules with conditions in demand, such as desired properties and scaffolds, precisely and stably. RM-GPT achieved better results than SOTA methods on conditional generation.

Nanopatterned Monolayers of Bioinspired, Sequence-Defined Polypeptoid Brushes for Semiconductor/Bio Interfaces.

The ability to control and manipulate semiconductor/bio interfaces is essential to enable biological nanofabrication pathways and bioelectronic devices. Traditional surface functionalization methods, such as self-assembled monolayers (SAMs), provide limited customization for these interfaces. Polymer brushes offer a wider range of chemistries, but choices that maintain compatibility with both lithographic patterning and biological systems are scarce. Here, we developed a class of bioinspired, sequence-defined polymers, i.e., polypeptoids, as tailored polymer brushes for surface modification of semiconductor substrates. Polypeptoids featuring a terminal hydroxyl (-OH) group are designed and synthesized for efficient melt grafting onto the native oxide layer of Si substrates, forming ultrathin (∼1 nm) monolayers. By programming monomer chemistry, our polypeptoid brush platform offers versatile surface modification, including adjustments to surface energy, passivation, preferential biomolecule attachment, and specific biomolecule binding. Importantly, the polypeptoid brush monolayers remain compatible with electron-beam lithographic patterning and retain their chemical characteristics even under harsh lithographic conditions. Electron-beam lithography is used over polypeptoid brushes to generate highly precise, binary nanoscale patterns with localized functionality for the selective immobilization (or passivation) of biomacromolecules, such as DNA origami or streptavidin, onto addressable arrays. This surface modification strategy with bioinspired, sequence-defined polypeptoid brushes enables monomer-level control over surface properties with a large parameter space of monomer chemistry and sequence and therefore is a highly versatile platform to precisely engineer semiconductor/bio interfaces for bioelectronics applications.

Robust training of machine learning interatomic potentials with dimensionality reduction and stratified sampling

Machine learning interatomic potentials (MLIPs) enable accurate simulations of materials at scales beyond that accessible by ab initio methods and play an increasingly important role in the study and design of materials. However, MLIPs are only as accurate and robust as the data on which they are trained. Here, we present DImensionality-Reduced Encoded Clusters with sTratified (DIRECT) sampling as an approach to select a robust training set of structures from a large and complex configuration space. By applying DIRECT sampling on the Materials Project relaxation trajectories dataset with over one million structures and 89 elements, we develop an improved materials 3-body graph network (M3GNet) universal potential that extrapolates more reliably to unseen structures. We further show that molecular dynamics (MD) simulations with the M3GNet universal potential can be used instead of expensive ab initio MD to rapidly create a large configuration space for target systems. We combined this scheme with DIRECT sampling to develop a reliable moment tensor potential for titanium hydrides without the need for iterative augmentation of training structures. This work paves the way for robust high-throughput development of MLIPs across any compositional complexity.

Coal Tar Pitch Derived sp2 Configuration‐Dominated Vacancy‐Rich Carbon with Expand Interlayer Spacing for Low‐Voltage, Durable, and Fast Potassium Storage

AbstractThe low‐voltage charging capacity and rate performance are two crucial but mutually restricted properties of K‐ion batteries. Herein, a sulfonamide and amide linkages‐triggered N doping high‐temperature removal strategy is proposed to obtain porous carbon with rich carbon vacancies, high sp2 hybridized C content (65.3%), and large interlayer spacing using coal tar pitch as carbon precursor. Carbon vacancies and sp2 hybridized C are proved to be not only active sites for low‐voltage potassium storage but also conducive to the fast transfer of ions and electrons, endowing reversible and fast potassium storage at low voltage. Moreover, abundant carbon vacancies and large carbon interlayer spacing alleviate volume variation, improving the cycling stability for 8500 cycles (about 3700 h). This work sheds new light on the design of low‐voltage, durable, and fast potassium storage carbon anodes, and promotes the development of practical carbon anodes for high energy density K‐ion batteries in the future.

Ranking‐based architecture generation for surrogate‐assisted neural architecture search

AbstractArchitectures generation optimization has been received a lot of attention in neural architecture search (NAS) since its efficiency in generating architecture. By learning the architecture representation through unsupervised learning and constructing a latent space, the prediction process of predictors is simplified, leading to improved efficiency in architecture search. However, searching for architectures with top performance in complex and large NAS search spaces remains challenging. In this paper, an approach that combined a ranker and generative model is proposed to address this challenge through regularizing the latent space and identifying architectures with top rankings. We introduce the ranking error to gradually regulate the training of the generative model, making it easier to identify architecture representations in the latent space. Additionally, a surrogate‐assisted evolutionary search method that utilized neural network Bayesian optimization is proposed to efficiently explore promising architectures in the latent space. We demonstrate the benefits of our approach in optimizing architectures with top rankings, and our method outperforms state‐of‐the‐art techniques on various NAS benchmarks. The code is available at https://github.com/outofstyle/RAGS‐NAS.

Accelerated Lignocellulosic Molecule Adsorption Structure Determination.

Here, we present a study combining Bayesian optimization structural inference with the machine learning interatomic potential Neural Equivariant Interatomic Potential (NequIP) to accelerate and enable the study of the adsorption of the conformationally flexible lignocellulosic molecules β-d-xylose and 1,4-β-d-xylotetraose on a copper surface. The number of structure evaluations needed to map out the relevant potential energy surfaces are reduced by Bayesian optimization, while NequIP minimizes the time spent on each evaluation, ultimately resulting in cost-efficient and reliable sampling of large systems and configurational spaces. Although the applicability of Bayesian optimization for the conformational analysis of the more flexible xylotetraose molecule is restricted by the sample complexity bottleneck, the latter can be effectively bypassed with external conformer search tools, such as the Conformer-Rotamer Ensemble Sampling Tool, facilitating the subsequent lower-dimensional global minimum adsorption structure determination. Finally, we demonstrate the applicability of the described approach to find adsorption structures practically equivalent to the density functional theory counterparts at a fraction of the computational cost.

Mechanism of the influence of sand on the energy dissipation inside the hydraulic turbine under sediment erosion condition

Sediment erosion can lead to a general decrease in hydraulic performance of hydraulic turbines and an increase in energy dissipation. In this paper, aiming the characteristics of SE at the clearance the erosion model was modified. The purpose is to study the influence of sediment erosion on the energy dissipation inside the hydraulic turbine. Research shows that the blade is most severely worn near the lower ring of the runner, which can lead to structural loss and hydraulic model changes. The impact of particle size on clearance erosion is much greater than that in a large space flow channel. The particle size affects the impact angle and impact velocity of particles, and is more pronounced at the clearance. The larger the particle size, the greater the acceleration of the jet within the guide vane clearance, exacerbating the SE and leakage of the guide vane clearance. The jet from the guide vane clearance propagates into the runner blade channel, exacerbating energy dissipation. SE can cause the local structure undergoes changes, leading to an increase in local jet flow and deterioration of flow patterns, increasing local energy dissipation. Among various components, the runner is most severely worn, and energy dissipation dominates.

Optimization of Redundant Degrees of Freedom in Robotic Flat-End Milling Based on Dynamic Response

With the advantages of large working space, low cost and more flexibility, industrial robots have become an important carrier in intelligent manufacturing. Due to the low rigidity of robotic milling systems, cutting vibrations are inevitable and have a significant impact on surface quality and machining accuracy. To improve the machining performance of the robot, a posture optimization approach based on the dynamic response index is proposed, which combines posture-dependent dynamic characteristics with surface quality for robotic milling. First, modal tests are conducted at sampled points to estimate the posture-dependent dynamic parameters of the robotic milling system. The modal parameters at the unsampled points are further predicted using the inverse distance weighted method. By combining posture-independent modal parameters with calibrating the cutting forces, a dynamic model of a robotic milling system is established and solved with a semi-discretization method. A dynamic response index is then introduced, calculated based on the extraction of the vibration signal peaks. The optimization model is validated through milling experiments, demonstrating that optimizing redundant angles significantly enhances milling stability and quality.

Acoustic characteristics of bedrooms in two types of long-term care facilities in China

The aim of this study was to quantitatively determine the acoustic characteristics of bedrooms in two types of long-term care facilities in China. Objective acoustic conditions, including noise levels and reverberation times, were measured through a series of acoustic measurements in twelve bedrooms in two nursing homes and three adult care homes in Kunming city, China. The impacts of noise and sound preferences were evaluated through a questionnaire survey administered to residents and nursing staff. In terms of the sound field, the background noise levels in most measured bedrooms exceeded the WHO's recommended value (30 dBA) by approximately 10–15 dBA. Compared to those in adult care homes, the noise levels in nursing homes were approximately 5–7 dBA higher during the daytime and 2–3 dBA higher during the nighttime due to frequent nursing activities. Moreover, noise levels were 5–15 dBA higher in roadside bedrooms. The reverberation time of five bedrooms reached 0.8 s at low frequency (125 Hz) due to their large space and absence of sound-absorbing materials. The questionnaire showed that noise sources were mainly perceived as coming from corridors and out-of-windows by residents and nursing staff. Traffic noise, residents' yelling in pain (just in nursing homes) and footsteps were considered the most noticeable noises, which may have had negative effects on participants' sleep quality, health, and emotional state. Moreover, the residents in roadside bedrooms reported that noise had a greater impact on their sleep (p < 0.01). Compared to artificial and mechanical sounds, participants preferred nature sounds, such as streams and birds, which were significantly (p < 0.01) positively correlated with age.

A novel thermal bus architecture for large cryogenic space telescopes utilizing helium pulsating heat pipes

Future space telescopes, such as those proposed for the Far-infrared Surveyor Mission, are expected to employ actively cooled optical arrays with a similar overall surface area compared to the James Webb Space Telescope. Therefore, there is a need for a cryogenic cooling system with a thermal bus architecture that can distribute cooling to these large optical arrays. Recent experimental research of helium Pulsating Heat Pipes (PHPs) has shown that helium PHPs can transfer heat over long distances (on the order of 2 m) with high efficiency, and also have the ability to act as a passive thermal switch upon the removal of the cooling source. PHPs’ high thermal performance, passive switching capability, low mass, and ease of manufacturing make them an appealing option compared to high-purity metal straps for a thermal bus architecture on large cryogenic space telescopes. A novel architecture for the thermal control of optical arrays is proposed utilizing unique configurations of helium pulsating heat pipes that minimize mass, maximize thermal performance, and reduce the risk of mission failure.

Cooperative design of feeder bus and bike-sharing systems

Feeder services, which provide travellers effective access from local areas to trunk-line systems, are essential, especially when the long-haul trunk routes are designed with large spacing between stations. Apart from the short-haul bus, the development of bike-sharing systems has led to a powerful supplementary feeder service. Most research on feeder services focuses on the separate design of bike and bus systems, but their joint design deserves further study. This research proposes a bi-level mixed-integer programming model to simultaneously design the locations of feeder buses and shared bike stations as well as the bus routes. A modified genetic algorithm that encapsulates the fixed-point iteration and Frank-Wolfe algorithms is developed to solve the model for large network applications. Numerical examples are given to compare the joint design with the separate design. The key factors that affect the design outcomes are investigated. The effects of joint design on operators and travellers are also discussed.

Research on Optical Mutual Injection to Generate Tunable Microwave Frequency Combs

In this study, a scheme for generating tunable microwave frequency combs (MFCs) based on optical mutual injection is proposed and experimentally investigated. The scheme is based on the optical injection of lasers to generate MFCs, and constitutes a feedback loop by using dual-laser mutual injection to obtain MFCs with a large continuous bandwidth and tunable comb spacing. The experimental setup analyzes the effects of injected optical power, modulation frequency and amplitude, and wavelength detuning on the generated MFC signals. The experimental results indicate that when the single-frequency electrical signal is set to 2 GHz, flat MFCs with amplitude variations within 10 dB can be obtained by optimizing the injected power and the frequency detuning between the two semiconductor lasers. Furthermore, the comb spacing of the MFCs can be made tunable by varying the modulation frequency and selecting the matched operating parameters to adapt to different application scenarios.

Transformer-based anomaly detection in P-LEO constellations: A dynamic graph approach

Successful management of large space systems, such as the most recent Proliferated Low Earth Orbit (P-LEO) constellations, demands an increased level of autonomy and irregular behavior detection capability. Existing techniques for satellite network management rely on monitoring satellites individually, potentially failing to capture events shared across multiple units. In this work, we propose a pipeline for identifying anomalous connections in proliferated satellite networks. Our pipeline relies on representing a satellite constellation as a dynamic graph. Dynamic graphs can be conveniently exploited to represent P-LEO networks, as they allow to capture meaningful structural and temporal correlations. Such spatial and temporal information are first projected into an embedding space; next, this encoded representation is processed by a transformer-encoder network for the anomaly detection task. We conduct extensive analyses of the main problem parameters, including the temporal horizon, the structure of the graph, and the size of the constellation. To assess the general performance of the method, we perform a Monte Carlo analysis on 960 ground node pair connections, providing empirical evidence of the algorithm’s generalization capability. The obtained results encourage the extension of the method to more complex, realistic modeling and kindred applications such as the on-edge detection problem.

Efficient encoding of large antigenic spaces by epitope prioritization with Dolphyn

We investigate a relatively underexplored component of the gut-immune axis by profiling the antibody response to gut phages using Phage Immunoprecipitation Sequencing (PhIP-Seq). To cover large antigenic spaces, we develop Dolphyn, a method that uses machine learning to select peptides from protein sets and compresses the proteome through epitope-stitching. Dolphyn compresses the size of a peptide library by 78% compared to traditional tiling, increasing the antibody-reactive peptides from 10% to 31%. We find that the immune system develops antibodies to human gut bacteria-infecting viruses, particularly E.coli-infecting Myoviridae. Cost-effective PhIP-Seq libraries designed with Dolphyn enable the assessment of a wider range of proteins in a single experiment, thus facilitating the study of the gut-immune axis.