Unifying pairwise interactions in complex dynamics

Nature Computational Science, Journal Year: 2023, Volume and Issue: 3(10), P. 883 - 893

Published: Sept. 25, 2023

Language: Английский

---

A new approach for estimating effective connectivity from activity in neural networks

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Feb. 6, 2024

Abstract Inferring and understanding the underlying connectivity structure of a system solely from observed activity its constituent components is challenge in many areas science. In neuroscience, techniques for estimating are paramount when attempting to understand network neural systems their recorded patterns. To date, no universally accepted method exists inference effective connectivity, which describes how node mechanistically affects other nodes. Here, focussing on purely excitatory networks small intermediate size continuous dynamics, we provide systematic comparison different approaches connectivity. Starting with Hopf neuron model conjunction known ground truth structural reconstruct system’s matrix using variety algorithms. We show that, sparse non-linear delays, combining lagged-cross-correlation (LCC) approach recently published derivative-based covariance analysis provides most reliable estimation matrix. also that linear networks, LCC has comparable performance based transfer entropy, at drastically lower computational cost. highlight works best decreases larger less networks. Applying dynamics without time find it does not outperform methods. Employing model, then use estimated as basis forward simulation order recreate under certain conditions, method, LCC, results higher trace-to-trace correlations than methods noise-driven systems. Finally, apply empirical biological data. subset nervous nematode C. Elegans . computationally simple performs better another published, more expensive reservoir computing-based method. Our comparatively can be used reliably estimate directed presence spatio-temporal delays noise. concrete suggestions scenario common research, where only neuronal set neurons known.

Language: Английский

---

Attractor reconstruction with reservoir computers: The effect of the reservoir’s conditional Lyapunov exponents on faithful attractor reconstruction

Chaos An Interdisciplinary Journal of Nonlinear Science, Journal Year: 2024, Volume and Issue: 34(4)

Published: April 1, 2024

Reservoir computing is a machine learning framework that has been shown to be able replicate the chaotic attractor, including fractal dimension and entire Lyapunov spectrum, of dynamical system on which it trained. We quantitatively relate generalized synchronization dynamics driven reservoir during training stage performance trained computer at attractor reconstruction task. show that, in order obtain successful spectrum estimation, maximal conditional exponent must significantly more negative than most target system. also find depends strongly spectral radius adjacency matrix; therefore, for small computers perform better general. Our arguments are supported by numerical examples well-known systems.

Language: Английский

---

Can system dynamics explain long-term hydrological behaviors? The role of endogenous linking structure

Hydrology and earth system sciences, Journal Year: 2025, Volume and Issue: 29(1), P. 159 - 177

Published: Jan. 14, 2025

Abstract. Hydrological models with conceptual tipping bucket and process-based evapotranspiration formulations are the most common tools in hydrology. However, these consistently fail to replicate long-term slow dynamics of a hydrological system, indicating need for model augmentation shift formulation approach. This study employed an entirely different approach – system towards more realistic replication observed behaviors at inter-annual inter-decadal scales. Using headwaters Baiyang Lake China as case study, endogenous linking structure was gradually unraveled from 1982 2015 through wavelet analysis, Granger's causality test, model. The analysis test identified negatively correlated bidirectional causal relationship between actual catchment water storage change across distinct climatic periodicities, suggested combined vegetation reinforcing feedback soil water–vegetation balancing system. dynamics' successfully captured under both natural human-intervention scenarios, demonstrating self-sustained oscillation arising within system's boundary. Our results showed that interaction soil-bound dominates process scale, while soil-water-holding capacity scale. Conventional models, which typically employ physiological-based assume invariable characteristics, ignore scale leading failure predicting behaviors. is its early stage applications primarily confined water-stressed regions novel insights proposed our including hierarchies corresponding mechanisms timescales among stocks being important driver behaviors, offer potential solutions better understanding guidelines improving configuration performance conventional models.

Language: Английский

---

Detecting Attacks and Estimating States of Power Grids from Partial Observations with Machine Learning

PRX Energy, Journal Year: 2025, Volume and Issue: 4(1)

Published: Feb. 4, 2025

The ever-increasing complexity of modern power grids makes them vulnerable to cyber and/or physical attacks. To protect them, accurate attack detection is essential. A challenging scenario that a localized has occurred on specific transmission line but only small number lines elsewhere can be monitored. That is, full state observation the whole grid not feasible, so and estimation need done with limited, partial observations. We articulate machine-learning framework address this problem, where necessity deal sequential time-series data dynamical memories avoid vanishing gradient led us choose long short-term memory (LSTM) architecture. Leveraging inherent capabilities LSTM handle capture temporal dependencies, we demonstrate, using three benchmark power-grid networks, complete faithfully reconstructed accurately from observations even in presence noise. performance improves as more become available. Further justification for provided by our comparing its alternative architectures such feedforward neural networks random forest. Despite gigantic existing literature applications grids, knowledge, problem locating an estimating limited had been addressed before work. method developed potentially generalized broad complex cyber-physical systems. Published American Physical Society 2025

Language: Английский

---

Comparison of derivative-based and correlation-based methods to estimate effective connectivity in neural networks

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Feb. 13, 2025

Language: Английский

---

Detecting Directional Coupling in Network Dynamical Systems via Kalman’s Observability

Physical Review Letters, Journal Year: 2025, Volume and Issue: 134(7)

Published: Feb. 18, 2025

Detecting coupling in network dynamical systems from time series is an open problem the physics of complex systems. In this Letter, we tackle issue a control-theoretic perspective. Drawing inspiration Kalman's notion observability, argue presence directional between two units, X→Y, when X detected as internal state measurement Y. We illustrate approach on analytically tractable systems, showcasing how it overcomes limitations state-of-the-art methods for inference.

Language: Английский

---

Causal discovery from city data, where urban scaling meets information theory

Cities, Journal Year: 2025, Volume and Issue: 162, P. 105980 - 105980

Published: April 15, 2025

Language: Английский

---

Longitudinal host-microbiome dynamics of metatranscription identify hallmarks of progression in periodontitis

Microbiome, Journal Year: 2025, Volume and Issue: 13(1)

Published: May 14, 2025

Abstract Background In periodontitis, the interplay between host and microbiome generates a self-perpetuating cycle of inflammation tooth-supporting tissues, potentially leading to tooth loss. Despite increasing knowledge phylogenetic compositional changes periodontal microbiome, current understanding in situ activities oral interactions among community members with is still limited. Prior studies on subgingival plaque metatranscriptome have been cross-sectional, allowing for only snapshot highly variable do not include transcriptome profiles from host, critical element progression disease. Results To identify host-microbiome milieu that lead periodontitis progression, we conducted longitudinal analysis clinically stable progressing sites 15 participants over 1 year. Our research uncovered distinct timeline microbial responses linked disease revealing significant clinical metabolic change point (the moment time when statistical properties series change) at 6-month mark study, 1722 genes differentially expressed (DE) 111,705 microbiome. Genes associated immune response, especially antigen presentation genes, were up-regulated before but sites. Activation cobalamin, porphyrin, motility contribute Conversely, inhibition lipopolysaccharide glycosphingolipid biosynthesis coincided increased response. Correlation delay revealed positive feedback loop consists regulation response activation leads an increase potassium ion transport cobalamin which turn induces Causality identified two clusters whose can accurately predict outcomes specific high confidence (AUC = 0.98095 0.97619). Conclusions A characterizes The dysbiotic are responsible reciprocally reinforced tissue destruction.

Language: Английский

---

Conditional cross-map-based technique: From pairwise dynamical causality to causal network reconstruction

Chaos An Interdisciplinary Journal of Nonlinear Science, Journal Year: 2023, Volume and Issue: 33(6)

Published: June 1, 2023

Causality detection methods based on mutual cross mapping have been fruitfully developed and applied to data originating from nonlinear dynamical systems, where the causes effects are non-separable. However, these pairwise still shortcomings in discriminating typical network structures, including common drivers, indirect dependencies, facing curse of dimensionality, when they stepping causal reconstruction. A few endeavors devoted conquer shortcomings. Here, we propose a novel method that could be regarded as one endeavors. Our method, named conditional cross-map-based technique, can eliminate third-party information successfully detect direct causality, results exactly categorized into four standard normal forms by designed criterion. To demonstrate practical usefulness our model-free, data-driven generated different representative models covering all kinds motifs measured real-world systems investigated. Because correct identification links is essential successful modeling, predicting, controlling underlying complex does shed light uncovering inner working mechanisms only using experimentally obtained variety disciplines.

Language: Английский

---