This study examines the transition of small-scale producer organizations (SPOs) of fair-trade coffee toward the domestic Mexican market, articulating the Localized Agri-Food Systems (SIAL) approach and complex adaptive systems theory to identify elements of competitive advantage. The methodology employed a sequential mixed-methods design incorporating data mining (12,253 reviews from 148 coffee shops across 8 cities) and participant observation (39 site visits and 4 cupping sessions with 4 SPOs). The main findings indicate that territorial embeddedness constitutes the most relevant factor (up to 61% in Oaxaca), with large effect sizes (η2p ranging from 0.17 to 0.22). Cooperation is not statistically significant, operating instead as an invisible underlying process. Oaxaca exhibits the most consolidated SIAL, with the strongest correlation between collective resource activation and the emergence of innovations (CRA-EI), whereas Tijuana constructs origin narratives through specialized demand. Intermediate projects functioned as living laboratories, generating innovations in quality criteria, service practices, and narratives. Self-organization manifested in the spontaneous reconfiguration of roles (producers as educators, baristas as evaluators). The most relevant second-order learning consisted in recognizing market heterogeneity, identifying differentiated niches (specialty, intermediate, traditional) and new strategic roles. It is concluded that SPO competitiveness is associated with the articulation of an integrated system of actors, resources, narratives, and innovations that anchor coffee to its territory, thereby humanizing the value chain to establish an emotional connection with the consumer. This research contributes theoretically by integrating the SIAL approach with complex adaptive systems theory and introducing the notion of a “value ceiling”; methodologically, by providing a replicable protocol combining data mining, participant observation, and co-construction workshops; and empirically, by documenting the territorial heterogeneity of the Mexican coffee market.
Smart cities rely on Cyber Physical Human Systems (CPHS) for efficiently delivering critical services (water, electricity, weather, traffic, etc.) to the community. These community-scale CHPS typically rely on an edge-fog-cloud computing continuum spanning multiple abstraction levels, and are typically purpose-built to deliver these services. This article makes the case that these siloed CPHS can be repurposed to deliver additional services in the face of disasters and unexpected events, raising the overall resilience of these community-scale infrastructures. We use two motivating CPHS case studies–information exchange for healthcare resilience (NSF project CAREDEX) and community-scale urban mobility through end-to-end orchestration (EU project MYRTUS) – as exemplars to tease out the opportunities and challenges faced in repurposing these CHPS in extreme events such as disasters. We highlight some of the key technologies required to facilitate repurposing, summarize their current status, and discuss ongoing efforts and future challenges for achieving community-scale CPHS resilience.
Every winter for the past 11 years, young researchers from across disciplines and around the world have gathered to explore complexity science and to collaborate intensively on self-organised projects. What began in 2015 as a gathering of colleagues who had first met at a previous workshop at the Santa Fe Institute has since consolidated into a self-organised, resilient workshop for early-career scientists in the complexity community. The Winter Workshop on Complex Systems is characterised by its format: participants form interdisciplinary teams and spend a week developing projects of their own design. By maintaining an economical model that serves as a low-cost alternative to expensive traditional conferences, the workshop prioritises inclusion, ensuring that financial constraints do not prevent talented early-career researchers from participating. This approach not only promotes collaboration and the exchange of ideas but also helps to build a strong, supportive community of young researchers. Many of these collaborations have endured well beyond the workshop, leading to successful publications and long-term partnerships and friendships.
Experimentation on distributed, heterogeneous computing environments—from edge devices to large-scale cloud platforms—demands orchestration technologies that are both flexible and extensible. Kiso is an open-source framework designed to provision resources and manage complex scientific workflows across the edge-to-cloud continuum. Its architecture unifies infrastructure provisioning, experiment configuration, and reproducible execution, enabling researchers to compose and monitor experiments that span geographically dispersed sites and variable network conditions. Although Kiso was conceived for workflow management—coordinating data-intensive tasks and ensuring reproducibility across dynamic infrastructures—its modular design makes it equally promising for providing reproducible environments for deploying and studying emerging agentic frameworks, where autonomous AI agents require consistent resource provisioning, cross-site communication, and result collection. We describe Kiso’s core capabilities for resource orchestration, experiment lifecycle management, and integration with containerized services, and we outline how these capabilities can support distributed multi-agent systems. In particular, we discuss how its declarative provisioning, extensible task abstractions, and built-in monitoring and output collection provide a natural foundation for experiments in which reasoning agents plan, negotiate, and adapt in real time. This study situates Kiso at the intersection of scientific workflow management and complex, agent-based computing, highlighting its potential to accelerate research on adaptive, self-organizing cyber-physical systems—an emerging frontier in complex systems science.
This paper develops a dynamic framework for analyzing markets in which decision-making is increasingly delegated to adaptive algorithms. In contrast to classical approaches that treat equilibrium as the primitive object of analysis, the framework models markets as systems of interacting learning processes operating within endogenous informational environments. Market outcomes arise from the interaction between adaptive decision rules, endogenous data generation, and feedback amplification. The analysis shows that market stability depends jointly on learning responsiveness and feedback intensity. When adaptive updating is sufficiently regular and feedback effects remain weak, the system converges toward stable equilibrium-like behavior, nesting standard equilibrium models as limiting benchmark cases. As learning responsiveness and endogenous feedback increase jointly, however, the adaptive process may generate persistent fluctuations, endogenous volatility, oscillatory dynamics, and local instability. Equilibrium therefore emerges not as a maintained assumption, but as a possible limiting outcome of the underlying learning dynamics. The framework further implies an adaptation-stability trade-off: faster responsiveness may improve information processing and adjustment while simultaneously amplifying endogenous instability. The analysis also shows how decentralized adaptive interaction may generate synchronized or coordinated patterns of behavior without explicit communication. A stylized quantitative illustration demonstrates how the interaction between learning and feedback shapes distinct dynamic regimes, while an illustrative empirical application to algorithmic trading environments documents patterns broadly consistent with the amplification mechanisms emphasized by the theory. The empirical evidence is interpreted as suggestive rather than definitive causal confirmation. More broadly, the paper suggests that understanding algorithmic markets may require analytical approaches that place adaptive learning, endogenous information generation, and feedback dynamics at the center of economic analysis.
We present SWEET, an end-to-end, accuracy-aware inference serving system that tailors execution to heterogeneous edge devices under diverse hardware constraints. Rather than deploying a single fixed pre-trained model for all future queries, serving workload-balanced end-to-end efficient and tailored edge inference (SWEET) plans a request-specific inference pattern by co-optimizing the model configuration and the device–server workload split according to the device’s compute capacity, accuracy requirements, and time constraints. At its core, SWEET integrates joint model quantization and inference partitioning: upon receiving an inference request, the server dynamically provides a quantized model and adaptively shares computation with the edge device, while the system determines the partition point and quantization settings based on the available device computation, channel capacity, and target accuracy. To enable principled decision making, we introduce a unified optimization framework that jointly selects layer-wise quantization bitwidths and partition points to minimize overall time and cost while explicitly accounting for task-dependent accuracy requirements through an accuracy degradation metric in the optimization model. To our knowledge, SWEET is the first inference serving system to optimize layer-wise quantization bitwidth using a theoretical measurement of accuracy degradation. Simulation results demonstrate substantial reductions in end-to-end time and power consumption, achieving a greater than 80% reduction in communication payload while keeping accuracy degradation below 1%.
The Compute Continuum—spanning IoT, Edge, Cloud, and HPC resources—is reshaping how hyper-distributed applications are designed and orchestrated. Traditional service orchestrators and workload management systems rely on centralized runtimes; however, the emerging paradigm requires decentralized coordination, where autonomous agents cooperate to achieve common goals and dynamically distribute workloads. Consensus algorithms play a crucial role in multi-agent systems (MAS), as they enable agents to reach agreement on how to coordinate and execute functionalities in a cooperative manner. While consensus has previously been applied to distributed job selection, here we extend its use to swarm environments. In this setting, agents autonomously decide which service functionalities (i.e., roles) to execute based on their capabilities and the real-time quality of service (QoS). Functionalities can be elastically activated or terminated as application needs evolve. To support this model, we leverage the COLMENA framework, a programming environment for defining and managing such dynamic services. We apply a greedy consensus-based approach to modern power systems, which are increasingly decentralized due to the large-scale integration of renewable energy sources. Centralized power plants are giving way to distributed, intermittent resources that require decentralized control paradigms. To demonstrate this, we simulate the Northeastern Power Coordinating Council’s (NPCC) 140-bus grid using the ANDES simulator in conjunction with the COLMENA middleware. We deploy this use case across six different sites in the FABRIC testbed, using up to 60 different nodes. Our results show that, under contingency scenarios such as load and generator disconnections, agents self-organize, elect local leaders, and execute optimization algorithms to stabilize grid frequency. Detection and organization times remain below 10s across all experiments, even as the number of agents per area scales from 3 to 10. Stability is restored within approximately 27s and 40s for the respective cases. Resource overhead is minimal, with CPU and memory usage remaining below 7.5% and 2%, respectively. Experiment automation and reproducibility are ensured through Kiso. These findings indicate that role-based programming models complement traditional workflows and that consensus-driven coordination can effectively decentralize decision-making in swarm environments. This approach represents a step toward enabling resilient, decentralized power systems.
Transatlantic scientific collaborations require computational testbeds that can be provisioned on demand and reconfigured rapidly while spanning institutions in different regulatory and operational domains. During the DISCOVER-US exchange program, we integrated the Infrastructure Manager (IM), a TOSCA-based orchestrator for the computing continuum, with the Chameleon cloud infrastructure. The workflow combined federated identity management, delegated project administration, and an IM plugin that targets Chameleon’s OpenStack endpoints through application credentials. We validated the approach by deploying single virtual machines, a production-ready Galaxy environment, distributed OSCAR-based serverless clusters that offload an AI-based fish detection pipeline, a workflow for flood impact modeling, and a hybrid SLURM cluster. Transatlantic computational testbeds included dynamically provisioned computational resources from EGI Federated Cloud and Chameleon. The study also documents operational constraints encountered with lease automation, bare-metal introspection, and the exposure of Kubernetes services across wide-area networks. The resulting blueprint demonstrates a reproducible path to deploy secure, elastic, and scientifically useful transatlantic computational testbeds.
The perennial debate about the possible directionality of evolution, as indicated by the apparent increase in the complexity of living systems over time, has recently witnessed renewed arguments in favor of the growth of complexity being “entropic,” that is, consistent with the growth of entropy as it is construed in thermodynamics. Here, I offer a brief review of formal treatments of complexity and of evolutionary mechanisms that are capable of causing it to increase. I then propose that both the evolutionary emergence and the individual learning of basic phenomenal awareness, a type of consciousness, are characterized by the same time-asymmetrical dynamics. Like life itself, biological consciousness arguably evolves towards greater complexity, and for the same reasons.
Previous work has shown that the dynamical regime of Recurrent Neural Networks (RNNs) - ranging from oscillatory to chaotic and fixpoint behavior - can be controlled by the global distribution of weights in connection matrices with statistically independent elements. However, it remains unclear how network dynamics respond to organizational regularities in the weight matrix, as often observed in biological neural networks. Here, we investigate three such regularities: (1) monopolar output weights per neuron, in accordance with Dale's principle, (2) reciprocal symmetry between neuron pairs, as in Hopfield networks, and (3) modular structure, where strongly connected blocks are embedded in a background of weaker connectivity. We construct weight matrices in which the strength of each regularity can be continuously tuned via control parameters, and analyze how key dynamical signatures of the RNN evolve as a function of these parameters. Moreover, using the RNN for actual information processing in a reservoir computing framework, we study how each regularity affects performance. We find that Dale monopolarity and modularity significantly enhance task accuracy, while Hopfield reciprocity tends to reduce it by promoting early saturation, limiting reservoir flexibility.
Although this Special Issue calls for a theory of emergence, the present paper argues that the breadth of the phenomenon\a requires a science, within which various theories can be explored and tested. To identify a structure for such a science of emergence, I pursued an in-depth cross-disciplinary analysis of emergence and its emergents. The result was identifying 9 emergence Prototypes, each of which reflects a unique aspect or context of emergence. Further, within some Prototypes, decades of scientific research has led to one or more Principles that its scholars ascribe to. Finally, the potential of an emergence science is explored by introducing applications of emergence to Leadership, Entrepreneurship, and Sustainability.
Power-law behavior frequently emerges in physical, biological, and social systems, particularly near continuous phase transitions characterized by diverging correlation lengths and universal scaling. The contact process is a prototypical model for studying absorbing-state phase transitions, typically belonging to the directed percolation (DP) universality class in its clean form. In this study, we investigate how quenched disorder influences the absorbing-state transition of the contact process on a one-dimensional comb lattice, a minimal geometry that incorporates structural inhomogeneity while remaining analytically and computationally tractable. In our model, activity spreads over a fraction of the branches q and is blocked in the rest. Without disorder, the system belongs to the directed percolation (DP) universality class. Introducing quenched disorder leads to significant changes in the critical dynamics. For q≤0.15, the system develops a Griffiths phase characterized by algebraic decay away from the critical point and logarithmic scaling at criticality, indicating a transition to the activated scaling universality class. In contrast, for q>0.15, the contact process on the comb lattice shows power-law decay of the order parameter only at the critical point, demonstrating a clean transition with standard critical dynamics and no extended Griffiths region. The results show that quenched disorder induces non-universal slow dynamics for small q, while larger values of q suppress the disorder-driven effects, restoring standard DP-like criticality. This transition underscores the role of lattice geometry and disorder strength in shaping nonequilibrium phase transitions.
IntroductionFutsal has become increasingly popular over the years; hence, studies focusing on the integration of drills to improve the performance of athletes are relevant. This study investigates the impact of modified training drills in improving the dribbling agility of futsal athletes.MethodsA total of 13 athletes participated in the research. Their dribbling agility was classified based on the Test of Agility in Dribbling before and after the training duration. The training program, which follows the FITT principle, consisted of three sessions per week (following a TThS schedule) for 2 weeks, adapting their normal training days. Descriptive statistics (frequency counts, percentages, mean, and standard deviation) summarized the demographic characteristics and agility levels. A Wilcoxon signed-rank test was utilized to determine the significant difference between the pre-test and post-test results, as the participants were not randomly sampled. A Spearman's rank correlation test was used to test the association of demographic profiles with the levels of dribbling agility.ResultsDemographic analysis revealed that most participants were 14 years old, with a majority having a height between 140-149 cm and a weight of 40-49 kg. Descriptive statistics showed a significant improvement in agility performance, as the average agility time decreased from 24.51 s in the pre-test to 20.50 s in the post-test. After training, the participants’ dribbling agility levels shifted from predominantly ‘poor’ classifications to ‘average’, ‘good’, and ‘excellent’. Statistical analysis confirmed that this difference was statistically significant (p < 0.05). Further analysis revealed that weight has a significant association with agility performance, while age and height did not.DiscussionThe results support the hypothesis that modified training drills positively impact agility. The findings suggest that weight can be considered an important factor in evaluating the impact of agility training.ConclusionThe modified agility training program effectively enhanced dribbling agility among futsal athletes. The study suggests that future researchers may extend the training duration and control external factors. It is also recommended that weight be considered in planning and evaluating agility training programs.
This paper presents a methodology combining Network Science (NS) and Explainable Machine Learning (XML) that could hypothetically uncover shared principles across seemingly disparate scientific domains. As an example, it presents how the approach could be applied to four fields: materials science, neuroscience, social science, and cosmology. The study focuses on criticality, a phenomenon associated with the transition of complex systems between states, characterized by sudden and significant behavioral shifts. By proposing a five-step methodology—ranging from relational data collection to cross-domain analysis with XML—the paper offers a hypothetical framework for potentially identifying criticality-related features in these fields and transferring insights across disciplines. The results of domains cross-fertilization could support practical applications, such as improving neuroprosthetics and brain-machine interfaces by leveraging criticality in materials science and neuroscience or developing advanced materials for space exploration. The parallels between neural and social networks could deepen our understanding of human behavior, while studying cosmic and social systems may reveal shared dynamics in large-scale, interconnected structures. A key benefit could be the possibility of using transfer learning, that is XML models trained in one domain might be adapted for use in another with limited data. For instance, if common aspects of criticality in neuroscience and cosmology are identified, an algorithm trained on brain data could be repurposed to detect critical states in cosmic systems, even with limited cosmic data. This interdisciplinary approach advances theoretical frameworks and fosters practical innovations, laying the groundwork for future research that could transform our understanding of complex systems across diverse scientific fields.
Our revisit of fundamental issues of climate challenges the notion and term of the “greenhouse effect”, and attempts a scientific reevaluation using minimal assumptions, such as Newton’s laws, maximum entropy and gas spectroscopy. It replaces terms like “greenhouse gas” with “radiatively active gas” (RAG) and “greenhouse effect” with “atmospheric radiative effect” (ARE). While ARE exists in several planets’ atmospheres, on Earth it is primarily driven by water vapor and clouds, with CO2 playing a minor role (especially anthropogenic CO2 which represents 4% of total emissions). Equilibrium thermodynamics, via entropy maximization or molecular collision simulation, leads to an isothermal atmosphere at about 250 K (the average temperature of the troposphere and stratosphere) irrespective of RAG presence or not. It is the troposphere’s 6.5 K/km temperature gradient (lapse rate), partly shaped by moist adiabatic processes, that drives the atmosphere away from this equilibrium and warms the surface to about 288 K on average, with ARE (mainly water vapor and clouds) contributing to the warming, but only when this gradient exists. The temperature gradient varies spatially and temporally and, since 1950, has weakened in the tropics and grown in the polar areas, resulting in a decrease of the surface equator-to-pole gradient, as expected in global warming conditions.
In finite-size thermal systems that exhibit second-order phase transition, the fluctuations of the order parameter ϕn obey type I intermittent dynamics at their pseudocritical temperature Tpc. Moreover, as recently demonstrated, spontaneous symmetry breaking (SSB) is gradually completed as temperature is reduced until reaching an SSB completion temperature, TSSB. Within this temperature zone, ϕn obey the dynamics of critical intermittency. This behavior has also been observed in pre-seismic fracture-induced electromagnetic emissions (FEME) of the MHz band—a real-world finite-size system undergoing a second-order phase transition. Interestingly, MHz FEME has recently been found to consistently present indications of tricritical dynamics after the SSB. We examine here whether this could also be true for a finite-size thermal system. We conduct a numerical experiment for the 3D Ising model at different temperatures by gradually reducing temperature beyond SSB and analyze order parameter fluctuations using the method of critical fluctuations (MCF) and a recently introduced wavelet-based method for detecting scaling behavior in noisy experimental data. Our results reveal that power-laws still exist within a very narrow zone of temperatures right after SSB completion for the 3D Ising model. These power-laws are shown to be compatible with another form of intermittency that determines the dynamics of the order parameter fluctuations close to the Griffiths tricritical point. As a possible interpretation of this finding, we suggest that our results imply that 3D Ising presents, just below TSSB, an imprint approaching the Griffiths tricritical point from the second-order phase transition line.
Traditional evolutionary theory explains adaptation and diversification through random mutation and natural selection. While effective in accounting for trait variation and fitness optimization, this framework provides limited insight into the physical principles underlying the spontaneous emergence of complex, ordered systems. A complementary theory is proposed: that evolution is fundamentally driven by the reduction of informational entropy. Grounded in non-equilibrium thermodynamics, systems theory, and information theory, this perspective posits that living systems emerge as self-organizing structures that reduce internal uncertainty by extracting and compressing meaningful information from environmental noise. These systems increase in complexity by dissipating energy and exporting entropy, while constructing coherent, predictive internal architectures, fully in accordance with the second law of thermodynamics. Informational entropy reduction is conceptualized as operating in synergy with Darwinian mechanisms. It generates the structural and informational complexity upon which natural selection acts, whereas mutation and selection refine and stabilize those configurations that most effectively manage energy and information. This framework extends previous thermodynamic models by identifying informational coherence, not energy efficiency, as the primary evolutionary driver. Recently formalized metrics, Information Entropy Gradient (IEG), Entropy Reduction Rate (ERR), Compression Efficiency (CE), Normalized Information Compression Ratio (NICR), and Structural Entropy Reduction (SER), provide testable tools to evaluate entropy-reducing dynamics across biological and artificial systems. Empirical support is drawn from diverse domains, including autocatalytic networks in prebiotic chemistry, genome streamlining in microbial evolution, predictive coding in neural systems, and ecosystem-level energy-information coupling. Together, these examples demonstrate that informational entropy reduction is a pervasive, measurable feature of evolving systems. While this article presents a theoretical perspective rather than empirical results, it offers a unifying explanation for major evolutionary transitions, the emergence of cognition and consciousness, the rise of artificial intelligence, and the potential universality of life. By embedding evolution within general physical laws that couple energy dissipation to informational compression, this framework provides a generative foundation for interdisciplinary research on the origin and trajectory of complexity.
