Agent Coordination Research Articles

Financial development, income and income inequality

The aim of this paper is twofold. Firstly, we present a model in which both income and income inequality are jointly determined in a counter-cyclical manner via self-fulfilling expectation. We argue that multiple equilibria can arise in the presence of inelastic labor demand, a minimum investment requirement, and imperfections in the credit market. In one equilibrium, the market wage and labor income are both low. Young agents who become entrepreneurs work harder and save more than young agents who become depositors. As a result, the equilibrium is characterized by low-income and high-income inequality. In another equilibrium, the market wage and labor income are both high. Young agents supply the same amount of labor and save the same. As a result, the equilibrium is characterized by high-income and low-income inequality. Secondly, we present different dynamic scenarios predicted by the model and analyze the role of self-fulfilling expectations. The paper ends by providing some policy recommendations on how the coordination of agents’ expectations about labor market conditions and how improvements in financial development may affect the long-run income and income inequality.

AAZTA: The rise of mesocyclic chelating agents for metal coordination in medicine

• AAZTA: an easily synthesised mesocyclic chelating agent. • Paramagnetic AAZTA complexes are studied as efficient MRI contrast agents. • Radioactive AAZTA complexes are evaluated for nuclear medicine applications. • Bifunctional AAZTA derivatives are used to prepare conjugates for diagnostics. The application of metal ions for diagnostic and therapeutic purposes is nowadays a well consolidated practice in current medicine. Gd 3+ for magnetic resonance imaging (MRI), Ga 3+ for positron emission tomography (PET), In 3+ for single photon emission computed tomography (SPECT) and the β-emitter Lu 3+ for tumour treatment are the most representative examples. Metal ions employed in medicine must be administered as stable and inert metal complexes, mainly for safety reasons and targeting purposes. In the last years many new chelating agents have been designed, prepared and studied in order to improve properties such as thermodynamic stability, kinetic inertness, target selectivity and many other characteristics of complexes thereof. The structure of these chelating agents is almost invariably based on linear or macrocyclic backbones. The chelating agent AAZTA (6-amino-6-methylperhydro-1,4-diazepinetetraacetic acid), featuring a mesocyclic 7-membered ring has found steadily increasing attention since its inception in the early 2000s and is now an affirmed alternative for the formation of stable complexes with several metal ions. The scope of this review is to provide the reader with a comprehensive survey on the synthesis and the applications of AAZTA and its derivatives developed in the last two decades.

PRIMAL$_2$: Pathfinding Via Reinforcement and Imitation Multi-Agent Learning - Lifelong

Multi-agent path finding (MAPF) is an indispensable component of large-scale robot deployments in numerous domains ranging from airport management to warehouse automation. In particular, this work addresses lifelong MAPF (LMAPF) - an online variant of the problem where agents are immediately assigned a new goal upon reaching their current one - in dense and highly structured environments, typical of real-world warehouse operations. Effectively solving LMAPF in such environments requires expensive coordination between agents as well as frequent replanning abilities, a daunting task for existing coupled and decoupled approaches alike. With the purpose of achieving considerable agent coordination without any compromise on reactivity and scalability, we introduce PRIMAL2, a distributed reinforcement learning framework for LMAPF where agents learn fully decentralized policies to reactively plan paths online in a partially observable world. We extend our previous work, which was effective in low-density sparsely occupied worlds, to highly structured and constrained worlds by identifying behaviors and conventions which improve implicit agent coordination, and enable their learning through the construction of a novel local agent observation and various training aids. We present extensive results of PRIMAL2 in both MAPF and LMAPF environments and compare its performance to state-of-the-art planners in terms of makespan and throughput. We show that PRIMAL2 significantly surpasses our previous work and performs comparably to these baselines, while allowing real-time re-planning and scaling up to 2048 agents.

A multi agent-based optimal control method for combined cooling and power systems with thermal energy storage

Combined cooling, heating and power (CCHP) systems have been considered as a potential energy saving technology for buildings due to their high energy efficiency and low carbon emission. Thermal energy storage (TES) can improve the energy efficiency of CCHP systems, since they reduce the mismatch between the energy supply and demand. However, it also increases the complexity of operation optimization of CCHP systems. In this study, a multi-agent system (MAS)-based optimal control method is proposed to minimize the operation cost of CCHP systems combined with TES. Four types of agents, i.e., coordinator agents, building agents, energy management agents and optimization agents, are implemented in the MAS to cooperate with each other. The operation optimization problem is solved by the genetic algorithm. A simulated system is utilized to validate the performance of the proposed method. Results show that the operation cost reductions of 10.0% on a typical summer day and 7.7% on a typical spring day are achieved compared with a rule-based control method. A sensitivity analysis is further performed and results show that the optimal operation cost does not change obviously when the rated capacity of TES exceeds a threshold.

Deep neural network based missing data prediction of electrocardiogram signal using multiagent reinforcement learning

ObjectiveClinical morphology of electrocardiogram (ECG) signal is compulsory to analyze the cardiac activity. During long term measurement, missing of data is a common factor, caused by sensor loosening, resulting in devastating feature extraction procedure; hence, prediction of those missing data is indispensable. MethodsIn this work, bidirectional long short-term memory recurrent neural network (LSTM-RNN) based prediction of missing segment of ECG signal is accomplished, governed by reinforcement learning (RL) using multiagent. The LSTM-RNN has internal gate architecture, which can predict the upcoming sequence using the past features of ECG. Each agent of the multiagent system (MAS), which represents various fictitious domain of ECG, has independent learning module to predict respective domain using a ‘critic’ network, which was conducted by a multilayer perceptron neural network (MLPNN). The MAS finally predicted the missing segment by ‘cooperative learning’ algorithm using a ‘coordinator agent’. ResultsThe proposed algorithm was tested on MIMIC-II ECG database, available in Physionet. Experimental result shows that, using the RL, the trained LSTM-RNN was able to predict the missing segments more precisely with correlation coefficient higher than 0.9 and low root mean squared error. ConclusionThe proposed method is applicable to any single channel ECG signal, and the quality of predicted signal ensures a wide application in medical applications as well as telecardiology system. SignificanceA comparative study with previously published works showed an improved performance, related to ECG missing data prediction, implemented on MIMIC-II records.

Adaptive Traffic Signal Control for large-scale scenario with Cooperative Group-based Multi-agent reinforcement learning

Recent research reveals that reinforcement learning can potentially perform optimal decision-making compared to traditional methods like Adaptive Traffic Signal Control (ATSC). With the development of knowledge through trial and error, the Deep Reinforcement Learning (DRL) technique shows its feasibility for the intelligent traffic lights control. However, the general DRL algorithms cannot meet the demands of agents for coordination within large complex road networks. In this article, we introduce a new Cooperative Group-Based Multi-Agent reinforcement learning-ATSC (CGB-MATSC) framework. It is based on Cooperative Vehicle Infrastructure System (CVIS) to realize effective control in the large-scale road network. We propose a CGB-MAQL algorithm that applies k-nearest-neighbor-based state representation, pheromone-based regional green-wave control mode, and spatial discounted reward to stabilize the learning convergence. Extensive experiments and ablation studies of the CGB-MAQL algorithm show its effectiveness and scalability in the synthetic road network, Monaco city and Harbin city scenarios. Results demonstrate that compared with a set of general control methods, our algorithm can better control multiple intersection cases on congestion alleviation and environmental protection.

An agent-based Stackelberg framework for joint expansion planning of privately Owned regional energy systems and sub-transmission grid

Traditionally, a regional electric company (REC) should supply electricity demand of consumers at sub-transmission level in special geographic areas. As a promising alternative for traditional centralized planning, in this paper, a Stackelberg game approach is employed to solve the investment and operational expansion planning problems in the presence of privately owned Regional Energy Systems (RGESs) and multi-agent systems are used to ease the participation of private ownerships in planning model by preventing agents to disclose their private properties. The RGES's and the REC's agents are the active players of the proposed method and in the form of a semi-decentralized framework, the information exchange among them is fully decentralized. Also, an Independent Coordinating Agent (ICA) is centrally responsible for Stackelberg Nash equilibrium point extraction. Further, a bi-stage Stackelberg game approach is proposed for coupled expansion planning program of Sub-transmission network and privately owned RGESs. To validate the feasibility and efficacy of the suggested model, a modified electric sub-transmission system (located in Guilan Province, Islamic Republic of Iran) is employed as a realistic test case system for simulations. And, the obtained results of simulations by the proposed method are compared with conventional expansion planning of sub-transmission systems.

Efficient Electrocatalytic CO2 Reduction to C2+ Alcohols at Defect-Site-Rich Cu Surface

Summary Electrochemical CO2 reduction is a promising approach for upgrading excessive CO2 into value-added chemicals, while the exquisite control of the catalyst atomic structures to obtain high C2+ alcohol selectivity has remained challenging due to the intrinsically favored ethylene pathways at Cu surface. Herein, we demonstrate a rational strategy to achieve ∼70% faradaic efficiency toward C2+ alcohols. We utilized a CO-rich environment to construct Cu catalysts with stepped sites that enabled high surface coverages of ∗CO intermediates and the bridge-bound ∗CO adsorption, which allowed to trigger CO2 reduction pathways toward the formation alcohols. Using this defect-site-rich Cu catalyst, we achieved C2+ alcohols with partial current densities of > 100 mA·cm−2 in both a flow-cell electrolyzer and a membrane electrode assembly (MEA) electrolyzer. A stable alcohol faradaic efficiency of ∼60% was also obtained, with ∼500 mg C2+ alcohol production per cm2 catalyst during a continuous 30-h operation.

On Stars and Galaxies: Optimal Mechanisms to Utilize Social Influence in Networks

We study mechanisms that exploit social influence in networks to coordinate agents to act. Agents’ social benefit from taking action increases with any additional friend who acts. On top of the social benefits, the principal offers external rewards to sustain a unique Nash equilibrium where everyone acts. We first show that in the influence mechanism that minimizes the principal’s expenses, popular agents receive preferential treatment from the principal. We use this observation to identify networks that are most favorable for the principal to induce action. Such networks, “galaxies”, partition nodes into core and periphery, with every core node being linked to all nodes, and every periphery node being linked only to core nodes. We discuss the relevance of this finding to social media platforms (such as Facebook and Twitter) in terms of manipulating the network, as well as to regulators who would attempt to prevent such manipulation.

Stochastic Graph Neural Networks

Graph neural networks (GNNs) model nonlinear representations in graph data with applications in distributed agent coordination, control, and planning among others. Current GNN architectures assume ideal scenarios and ignore link fluctuations that occur due to environment, human factors, or external attacks. In these situations, the GNN fails to address its distributed task if the topological randomness is not considered accordingly. To overcome this issue, we put forth the stochastic graph neural network (SGNN) model: a GNN where the distributed graph convolution module accounts for the random network changes. Since stochasticity brings in a new learning paradigm, we conduct a statistical analysis on the SGNN output variance to identify conditions the learned filters should satisfy for achieving robust transference to perturbed scenarios, ultimately revealing the explicit impact of random link losses. We further develop a stochastic gradient descent (SGD) based learning process for the SGNN and derive conditions on the learning rate under which this learning process converges to a stationary point. Numerical results corroborate our theoretical findings and compare the benefits of SGNN robust transference with a conventional GNN that ignores graph perturbations during learning.

MOSAIK: A Formal Model for Self-Organizing Manufacturing Systems

In this article, we review past and current system architectures displaying self-organization in the domain of manufacturing. Based on a corpus of 84 reference papers, we find that multiagent systems (MAS) play a significant role in self-organization, especially MAS featuring bio-inspired algorithms for agent coordination. The emergence of new classes of cyber-physical systems further strengthens the prevalence of MAS on the subject. As outcome of our review, we devise the MOSAIK model, a generic model synthesizing all system architectures found in our corpus. The MOSAIK model can be used as a reference for formally comparing distinct architectures. We also use it to identify gaps for future research on self-organizing manufacturing systems. The model includes the central concepts of Agent and Artifact, which suggest that the Web is an adequate communication infrastructure for modern manufacturing systems: Agents become (autonomous) Web Agents and Artifacts become resources exposed by Web servers.

Underlying Principles of a Covid-19 Behavioral Vaccine for a Sustainable Cultural Change.

Until pharmacological measures are effective at containing the COVID-19 outbreak, adopting protective behaviors is paramount. In this work, we aim at informing interventions to limit the spread of the contagion and prepare against any future outbreaks by developing a behavioral framework to interpret and prescribe both the individual and large-scale uptake of non-pharmaceutical measures. First, we analyze the barriers and facilitators to adherence to protective behaviors according to a three-term contingency by exploring potential gaps in terms of setting stimuli, motivating operations, delayed consequences, and positive or negative consequences. We explore their roles in the likelihood of individual compliance to protective behaviors, taking physical distancing as an example of functional analysis. Second, we interpret contagion control as the cumulative effect of large-scale adherence to protective behaviors. We explore the interrelations between societal problems caused or amplified by similar behaviors presented by many individuals and the coordination of agents or agencies aiming at promoting large-scale behavioral change. Then, we highlight the potential of developing a behavioral vaccine, and practical steps for applying it to promote sustainable cultural change that may protect against health, social, and economic losses in future outbreaks.

Merging academy and healthcare in the Public Health training of medical students

BackgroundPublic Health remains central to understand health and its determinants, and Public Health teams are essential for an integrated collaborative medical practice. However, current teaching of public health to medical students varies in the European Region though an investment in multidisciplinary workforce is recognised essential to deliver high quality public health services. A recent medical education curricula restructuring in the University of Porto Medical School resulted in the inclusion of a Public Health module linking academic teaching to field practice and provided the opportunity to make an initial appraisal of students’ perceptions.Case studyWe analysed final reports (n = 196), debriefing meetings notes (n = 2), and e-mails sent by students (n = 34) regarding the activities they observed or participated at, their contact with Public Health services’ teams, knowledge and critical appraisal, and opinion about the module. Students gained basic knowledge about how epidemiological surveillance, environmental health, health planning, and health promotion are performed in practice. They reported a better understanding of the roles and importance of Public Health services and its teams. Most considered that this module had an important role in their training. Some activities observed in the field lacked the needed standardisation to provide the students the feeling that core operations were experienced, which needs to be addressed in the future.ConclusionsPublic Health practice-based training within field institutions may bring a better understanding of the discipline and specialty for medical students. It may strengthen interconnectivity and coordination of healthcare agents, which may improve future medical practice with potential improvement of patient-centred care and in terms of public health response, and back their roles as health agents and decision-makers.

Synthesis and characterization of novel poly(HAzPMA-co-SA)/RDX/CdS nanocomposite as a polymer bonded explosive

Cadmium sulfide nanorods with a length of 69 nm have been prepared by using Cd(OAc)2.2H2O and S8 at 125 °C in the presence of triethylenetetramine as the template agent and coordination agent and characterized by using X-ray diffraction, transmission electron microscopy, FTIR, photoluminescence, and UV–vis absorption spectroscopic techniques. Photocopolymerization of glycidyl methacrylate (GMA) and sodium acrylate (SA) was carried out using CdS nanorods as a photocatalyst under UV light exposure at 400 nm in the presence of β-cyclodextrin (β-CD). To optimization of the effective parameters on the synthesis of copolymer nanocomposite, the amounts of initiator, monomers, and β-CD, duration of pre-deoxygenation, and light wavelength were evaluated. Ring opening of poly(GMA-co-SA)/CdS nanocomposite with NaN3 afforded poly(HAzPMA-co-SA)/CdS nanocomposite and subsequent mixing with RDX in DMF led to the formation of poly(HAzPMA-co-SA)/RDX/CdS nanocomposite as a polymer bonded explosive. All of the copolymer nanocomposites were characterized using various tools of instrumental analysis.

A Comparative Analysis of Data-Driven Based Optimization Models for Energy-Efficient Buildings

Energy-efficient and sustainable buildings have become a specialized research theme for energy planning and the green building industry. Rapidly increased energy consumption in the residential area has attracted researchers' attention to minimize energy consumption without affecting occupant comfort. Accordingly, an effective energy management system is required for energy-efficient buildings to utilize energy resources efficiently and maintain desired comfort. In this paper, data-driven based optimization models are compared to maximize the comfort index and minimize energy consumption simultaneously. Heating-cooling systems are used to maintain indoor thermal comfort, which consumes energy according to the environmental temperature and indoor temperature difference. Therefore, the environmental temperature parameter is optimized using optimization techniques Genetic Algorithm (GA), Bat, Neural Network Algorithm (NNA), Particle Swarm Optimization (PSO), and Artificial Bee Colony (ABC). The main objective of the optimization is to reduce the gap/error between the temperature specified by the user and the environmental temperature. Afterward, the discrepancy between optimized temperature and actual temperature is fed to the designed ML (Machine Learning) based controller. The controller's output is further provided to coordinator agents, which give the power to the actuators accordingly. The linearity of the proposed model improves the performance of the ML algorithm. The obtained dataset from fuzzy temperature controller has been used to design an ML controller. The experimental results show that the designed system significantly improves energy efficiency and occupant comfort for energy-efficient buildings. The BAT model has shown effectiveness in achieving a high comfort index with minimum power consumption compared to other considered models.

Quantum-inspired neuro coevolution model applied to coordination problems

In many real-world problems, some coordination between agents is necessary to enable the task to be optimally performed. However, obtaining this coordination can be challenging due to the quantity and characteristics of the agents, the dynamics of the environment and/or the complexity of the task, requiring much computation time. Furthermore, some problems require different types of agent specialization. In this case, it is very difficult for the programmers to define learning strategies and their parameters. Optimization of these parameters using standard evolutionary algorithms is also inadequate due to the high computational cost in these real multi-agent situations. The main objective of this study is therefore to propose a new neuroevolution model to be applied to agent coordination problems, termed the Quantum-Inspired Neuro Coevolution (QNCo) Model. QNCo makes use of paradigms from quantum physics and biological coevolution to evolve sub-populations of quantum individuals aiming convergence gains. The model has the capacity to autonomously obtain the best neural network topology of each agent, eliminating the need for the programmer to set this configuration. New quantum crossover and mutation operators were proposed and compared during function optimization of different dimensions. The proposed model was tested in two simulation problems, prey-predator and multi-rover tasks, and one real problem of mobile telephony coverage. The QNCo model yielded promising results compared to similar algorithms, with good solutions in terms of learning strategies and a great reduction in convergence time.

Fault-Resilient Continuum Deformation Coordination

This article offers a physics-based automation with two operation modes: 1) homogeneous deformation mode (HDM) and 2) containment exclusion mode (CEM), to safely plan a large-scale group coordination and resiliently recover safety in the presence of unpredicted agent failures. HDM becomes active when all agents are healthy, where the group coordination is defined by homogeneous transformation coordination functions. At HDM, a desired n-D homogeneous transformation (n = 2, 3) is uniquely related to the desired trajectories of n + 1 leaders and acquired by the remaining followers in real-time through local communication. At CEM, agent coordination is treated as an ideal fluid flow, where the desired agents' paths are defined along stream lines inspired by fluid flow field theory to circumvent exclusion spaces surrounding failed agent(s). This article formally specifies the transitions between the HDM and CEM by using local proximity and applying the continuum deformation properties. More specifically, local proximity is used by agents to define interagent communication in an unsupervised fashion once they recover safety and HDM is activated. On the other hand, properties of homogeneous transformation are applied to quickly detect each arising anomalous situation and exclude failed agent(s) from group coordination of healthy agents.

Coordination-driven interfacial cross-linked graphene oxide-alginate nacre mesh with underwater superoleophobicity for oil-water separation

Inspired by the seashell nacre and seaweed, a novel GO-Ca2+-SA nacre-inspired hybrid mesh was prepared via an interfacial layer-by-layer self-assembly and cross-linking, using graphene oxide (GO) and sodium alginate (SA) as the building blocks and calcium chloride as the coordination agent, respectively. Hybrid mesh was characterized by FTIR, XPS, XRD, SEM and contact angel instrument, showing superhydrophilic and underwater superoleophobic property and low oil adhesion, due to its wrinkle and rough surface, and high hydration ability of GO-Ca-alginate nanohydrogels. The separation efficiencies of various oil-water mixtures were above 99 %, with a highest flux of 119,426 L m−2 h−1. Hybrid mesh showed an orderly layered “brick and mortar” microstructure with many ultrasmall nanoscaled protuberances. Ca2+ ions could chelate with SA to form the “egg-box” structure, and interact with GO nanosheets. Hybrid mesh possessed high salt/acid/alkaline tolerance, abrasion resistance, mechanical property with Young's modulus of 35.8 ± 4.9 GPa, and excellent cycling stability.

Enhanced thermal stability of Ce0.33Zr0.55(LaNdY)0.12O2 mixed oxides prepared by sulfate-aided coprecipitation method

Ceria-zirconia based mixed oxides (CZs) have been applied in three-way catalysts (TWCs) for their high surface area and oxygen storage capacity (OSC). In this work, enhanced thermal stability of Ce0.33Zr0.55(LaNdY)0.12O2 mixed oxides was realized via a facile and scalable approach, namely, sulfate-aided coprecipitation method was labelled as CZ-S. Sulfate ion (SO42−) was added into raw solution in the form of sulfuric acid and acted as coordination agent. The control sample was prepared by conventional coprecipitation method without sulfuric acid added and labelled as CZ. The promotion effect of sulfate ion was analyzed systematically by X-ray diffraction (XRD), transmission electron microscopy (TEM), scanning electron microscopy (SEM), N2 adsorption-desorption, Fourier transform infrared spectroscopy (FT-IR), X-ray photoelectron spectroscopy (XPS), OSC and hydrogen temperature-programmed reduction (H2-TPR) analysis. XRD and high resolution TEM results reveal that CZ-S have homogeneous distributions of elements. TEM and SEM images show that fresh samples of CZ-S have narrower distributions of grain sizes and larger pore sizes than those of CZ. Through cross analysis of structure and morphology of CZ and CZ-S, we find that the introduction of sulfate ions results in uniform distributions of elements, narrows distributions of grain sizes, and enables the formation of secondary loose packing of sub-particles, which lead to enhanced thermal stability of the samples of CZ-S upon aging process at high temperature. After aging treatment at 1100 °C for 10 h, aged samples of CZ-S present larger specific surface areas and pore volumes than the aged sample prepared by conventional coprecipitation method without sulfate ions added. Furthermore, the aged sample of CZ-S2 (SO42−/Zr = 1) possesses the highest specific surface area of 21.9 m2/g and the biggest pore volume of 0.035 mL/g among all aged samples.

Completability vs (In)completeness

ABSTRACT In everyday conversation, no notion of “complete sentence” is required for syntactic licensing. However, so-called “fragmentary”, “incomplete”, and abandoned utterances are problematic for standard formalisms. When contextualised, such data show that (a) non-sentential utterances are adequate to underpin agent coordination, while (b) all linguistic dependencies can be systematically distributed across participants and turns. Standard models have problems accounting for such data because their notions of ‘constituency’ and ‘syntactic domain’ are independent of performance considerations. Concomitantly, we argue that no notion of “full proposition” or encoded speech act is necessary for successful interaction: strings, contents, and joint actions emerge in conversation without any single participant having envisaged in advance the outcome of their own or their interlocutors’ actions. Nonetheless, morphosyntactic and semantic licensing mechanisms need to apply incrementally and subsententially. We argue that, while a representational level of abstract syntax, divorced from conceptual structure and physical action, impedes natural accounts of subsentential coordination phenomena, a view of grammar as a “skill” employing domain-general mechanisms, rather than fixed form-meaning mappings, is needed instead. We provide a sketch of a predictive and incremental architecture (Dynamic Syntax) within which underspecification and time-relative update of meanings and utterances constitute the sole concept of “syntax”.