Accurate ocean dynamics modeling is crucial for enhancing understanding of ocean circulation, predicting climate variability, and tackling challenges posed by climate change. Despite improvements in traditional numerical models, predicting global ocean variability over multi-year scales remains challenging. Here, we propose ORCA-DL (Oceanic Reliable foreCAst via Deep Learning), the first data-driven 3D ocean model for seasonal to decadal prediction of global ocean circulation. ORCA-DL accurately simulates three-dimensional ocean dynamics and outperforms state-of-the-art dynamical models in capturing extreme events, including El Ni\~no-Southern Oscillation and upper ocean heatwaves. This demonstrates the high potential of data-driven models for efficient and accurate global ocean forecasting. Moreover, ORCA-DL stably emulates ocean dynamics at decadal timescales, demonstrating its potential even for skillful decadal predictions and climate projections.

Molecular foundation models are rapidly advancing scientific discovery, but their unreliability on out-of-distribution (OOD) samples severely limits their application in high-stakes domains such as drug discovery and protein design. A critical failure mode is chemical hallucination, where models make high-confidence yet entirely incorrect predictions for unknown molecules. To address this challenge, we introduce Molecular Preference-Aligned Instance Ranking (Mole-PAIR), a simple, plug-and-play module that can be flexibly integrated with existing foundation models to improve their reliability on OOD data through cost-effective post-training. Specifically, our method formulates the OOD detection problem as a preference optimization over the estimated OOD affinity between in-distribution (ID) and OOD samples, achieving this goal through a pairwise learning objective. We show that this objective essentially optimizes AUROC, which measures how consistently ID and OOD samples are ranked by the model. Extensive experiments across five real-world molecular datasets demonstrate that our approach significantly improves the OOD detection capabilities of existing molecular foundation models, achieving up to 45.8%, 43.9%, and 24.3% improvements in AUROC under distribution shifts of size, scaffold, and assay, respectively.

Nonlinear dynamical systems exposed to changing forcing can exhibit catastrophic transitions between alternative and often markedly different states. The phenomenon of critical slowing down (CSD) can be used to anticipate such transitions if caused by a bifurcation and if the change in forcing is slow compared to the internal time scale of the system. However, in many real-world situations, these assumptions are not met and transitions can be triggered because the forcing exceeds a critical rate. For example, given the pace of anthropogenic climate change in comparison to the internal time scales of key Earth system components, such as the polar ice sheets or the Atlantic Meridional Overturning Circulation, such rate-induced tipping poses a severe risk. Moreover, depending on the realisation of random perturbations, some trajectories may transition across an unstable boundary, while others do not, even under the same forcing. CSD-based indicators generally cannot distinguish these cases of noise-induced tipping versus no tipping. This severely limits our ability to assess the risks of tipping, and to predict individual trajectories. To address this, we make a first attempt to develop a deep learning framework to predict transition probabilities of dynamical systems ahead of rate-induced transitions. Our method issues early warnings, as demonstrated on three prototypical systems for rate-induced tipping, subjected to time-varying equilibrium drift and noise perturbations. Exploiting explainable artificial intelligence methods, our framework captures the fingerprints necessary for early detection of rate-induced tipping, even in cases of long lead times. Our findings demonstrate the predictability of rate-induced and noise-induced tipping, advancing our ability to determine safe operating spaces for a broader class of dynamical systems than possible so far.

The languages of mathematical physics and modelling are endowed with a rich "grammar of dimensions" that common abstractions of programming languages fail to represent. We propose a dependently typed domain-specific language (embedded in Idris) that captures this grammar. We apply it to explain basic notions of dimensional analysis and Buckingham's Pi theorem. We hope that the language makes mathematical physics more accessible to computer scientists and functional programming more palatable to modelers and physicists.

Foundation models such as GPT-4 are fine-tuned to avoid unsafe or otherwise problematic behavior, such as helping to commit crimes or producing racist text. One approach to fine-tuning, called reinforcement learning from human feedback, learns from humans' expressed preferences over multiple outputs. Another approach is constitutional AI, in which the input from humans is a list of high-level principles. But how do we deal with potentially diverging input from humans? How can we aggregate the input into consistent data about "collective" preferences or otherwise use it to make collective choices about model behavior? In this paper, we argue that the field of social choice is well positioned to address these questions, and we discuss ways forward for this agenda, drawing on discussions in a recent workshop on Social Choice for AI Ethics and Safety held in Berkeley, CA, USA in December 2023.

Tropical monsoons play a critical role in shaping regional and global climate systems, with profound ecological and socio-economic impacts. However, their long-term prediction remains challenging due to the complex interplay of regional dynamics, global climate drivers, and large-scale teleconnections. Here, we introduce a unified network-based framework for predicting monsoon precipitation across diverse tropical regions. By leveraging global 2-meter air temperature fields, this approach captures large-scale climate teleconnections, such as the El Nino-Southern Oscillation (ENSO) and Rossby waves, enabling accurate forecasts for four key monsoon systems: the South American, East Asian, African, and Indian monsoons. Our framework achieves remarkable forecasting accuracy with lead times of 4-10 months, outperforming traditional systems such as SEAS5 and CFSv2. Beyond its predictive capabilities, the framework offers flexibility for application to other regions and climate phenomena, advancing our understanding of global climate dynamics. These findings have far-reaching implications for disaster preparedness, resource management, and sustainable development.

Climate change exacerbates extreme weather events like heavy rainfall and flooding. As these events cause severe losses of property and lives, accurate high-resolution simulation of precipitation is imperative. However, existing Earth System Models (ESMs) struggle with resolving small-scale dynamics and suffer from biases, especially for extreme events. Traditional statistical bias correction and downscaling methods fall short in improving spatial structure, while recent deep learning methods lack controllability over the output and suffer from unstable training. Here, we propose a novel machine learning framework for simultaneous bias correction and downscaling. We train a generative diffusion model in a supervised way purely on observational data. We map observational and ESM data to a shared embedding space, where both are unbiased towards each other and train a conditional diffusion model to reverse the mapping. Our method can be used to correct any ESM field, as the training is independent of the ESM. Our approach ensures statistical fidelity, preserves large-scale spatial patterns and outperforms existing methods especially regarding extreme events and small-scale spatial features that are crucial for impact assessments.

The rapid expansion of low-cost renewable electricity combined with end-use electrification in transport, industry, and buildings offers a promising path to deep decarbonisation. However, aligning variable supply with demand requires strategies for daily and seasonal balancing. Existing models either lack the wide scope required for long-term transition pathways or the spatio-temporal detail to capture power system variability and flexibility. Here, we combine the complementary strengths of REMIND, a long-term integrated assessment model, and PyPSA-Eur, an hourly energy system model, through a bi-directional, price-based and iterative soft coupling. REMIND provides pathway variables such as sectoral electricity demand, installed capacities, and costs to PyPSA-Eur, which returns optimised operational variables such as capacity factors, storage requirements, and relative prices. After sufficient convergence, this integrated approach jointly optimises long-term investment and short-term operation. We demonstrate the coupling for two Germany-focused scenarios, with and without demand-side flexibility, reaching climate neutrality by 2045. Our results confirm that a sector-coupled energy system with nearly 100\% renewable electricity is technically possible and economically viable. Power system flexibility influences long-term pathways through price differentiation: supply-side market values vary by generation technology, while demand-side prices vary by end-use sector. Flexible electrolysers and smart-charging electric vehicles benefit from below-average prices, whereas less flexible heat pumps face almost twice the average price due to winter peak loads. Without demand-side flexibility, electricity prices increase across all end-users, though battery deployment partially compensates. Our approach therefore fully integrates power system dynamics into multi-decadal energy transition pathways.

Complex networks serve as abstract models for understanding real-world complex systems and provide frameworks for studying structured dynamical systems. This article addresses limitations in current studies on the exploration of individual birth-death and the development of community structures within dynamic systems. To bridge this gap, we propose a networked evolution model that includes the birth and death of individuals, incorporating reinforcement learning through games among individuals. Each individual has a lifespan following an arbitrary distribution, engages in games with network neighbors, selects actions using Q-learning in reinforcement learning, and moves within a two-dimensional space. The developed theories are validated through extensive experiments. Besides, we observe the evolution of cooperative behaviors and community structures in systems both with and without the birth-death process. The fitting of real-world populations and networks demonstrates the practicality of our model. Furthermore, comprehensive analyses of the model reveal that exploitation rates and payoff parameters determine the emergence of communities, learning rates affect the speed of community formation, discount factors influence stability, and two-dimensional space dimensions dictate community size. Our model offers a novel perspective on real-world community development and provides a valuable framework for studying population dynamics behaviors.

Beneficial societal outcomes cannot be guaranteed by aligning individual AI systems with the intentions of their operators or users. Even an AI system that is perfectly aligned to the intentions of its operating organization can lead to bad outcomes if the goals of that organization are misaligned with those of other institutions and individuals. For this reason, we need full-stack alignment, the concurrent alignment of AI systems and the institutions that shape them with what people value. This can be done without imposing a particular vision of individual or collective flourishing. We argue that current approaches for representing values, such as utility functions, preference orderings, or unstructured text, struggle to address these and other issues effectively. They struggle to distinguish values from other signals, to support principled normative reasoning, and to model collective goods. We propose thick models of value will be needed. These structure the way values and norms are represented, enabling systems to distinguish enduring values from fleeting preferences, to model the social embedding of individual choices, and to reason normatively, applying values in new domains. We demonstrate this approach in five areas: AI value stewardship, normatively competent agents, win-win negotiation systems, meaning-preserving economic mechanisms, and democratic regulatory institutions.

The topological Kuramoto model generalizes classical synchronization models by including higher-order interactions, with oscillator dynamics defined on cells of arbitrary dimension within simplicial or cell complexes. In this article, we demonstrate multistability in the topological Kuramoto model and develop the topological nonlinear Kirchhoff conditions algorithm to identify all phase-locked states on arbitrary cell complexes. The algorithm is based on a generalization of Kirchhoff's laws to cell complexes of arbitrary dimension and nonlinear interactions between cells. By applying this framework to rings, Platonic solids, and simplexes, as minimal representative motifs of larger networks, we derive explicit bounds (based on winding number constraints) that determine the number of coexisting stable states. We uncover structural cascades of multistability, inherited from both lower and higher dimensions and demonstrate that cell complexes can generate richer multistability patterns than simplicial complexes of the same dimension. Moreover, we find that multistability patterns in cell complexes appear to be determined by the number of boundary cells, hinting a possible universal pattern.

The Amazon rainforest (ARF) is threatened by deforestation and climate change, which could trigger a regime shift to a savanna-like state. Previous work suggesting declining resilience in recent decades was based only on local resilience indicators. Moreover, previous results are potentially biased by the employed multi-sensor and optical satellite data and undetected anthropogenic land-use change. Here, we show that the spatial correlation provides a more robust resilience indicator than local estimators and employ it to measure resilience changes in the ARF, based on single-sensor Vegetation Optical Depth data under conservative exclusion of human activity. Our results show an overall loss of resilience until around 2019, which is especially pronounced in the southwestern and northern Amazon for the time period from 2002 to 2011. The demonstrated reliability of spatial correlation in coupled systems suggests that in particular the southwest of the ARF has experienced pronounced resilience loss over the last two decades.

In bistable dynamical systems driven by Wiener processes, the widely used Kramers' law relates the strength of the noise forcing to the average time it takes to see a noise-induced transition from one attractor to the other. We extend this law to bistable systems forced with fast chaotic dynamics, which we argue to be a more realistic modeling approach than unbounded noise forcing in some cases. We test our results numerically in a reduced-order model of the Atlantic Meridional Overturning Circulation (AMOC) and discuss the limits of the chaotic Kramers' law as well as its surprisingly wide parameter range of applicability. Hereby, we show in detail how to apply Hasselmann's program in practice and give a possible explanation for recent findings of AMOC collapse and recovery in complex climate models.

Synchronization in various complex networks is significantly influenced by higher-order interactions combined with non-Gaussian stochastic perturbations, yet their mechanisms remain mainly unclear. In this paper, we systematically investigate the synchronization and spike dynamics in a higher-order Kuramoto model subjected to non-Gaussian Lévy noise. Using the mean order parameter, mean first-passage time, and basin stability, we identify clear boundaries distingusihing synchronization and incoherent states under different stability indexes and scale parameters of Lévy noise. For fixed coupling parameters, synchronization weakens as the stability index decreases, and even completely disappears when the scale parameter exceeds a critical threshold. By varying coupling parameters, we find significant dynamical phenomena including bifurcations and hysteresis. Lévy noise smooths the synchronization transitions and requires stronger coupling compared to Gaussian white noise. Furthermore, we define spikes and systematically investigate their statistical and spectral characteristics. The maximum number of spikes is observed at small scale parameter. A generalized spectral analysis further reveals burst-like structure via an edit distance algorithm. Moreover, a power-law distribution is observed in the large inter-window intervals of spikes, showing great memory effects. These findings deepen the understanding of synchronization and extreme events in complex networks driven by non-Gaussian Lévy noise.

Outlier detection refers to the identification of data points that deviate from a general data distribution. Existing unsupervised approaches often suffer from high computational cost, complex hyperparameter tuning, and limited interpretability, especially when working with large, high-dimensional datasets. To address these issues, we present a simple yet effective algorithm called ECOD (Empirical-Cumulative-distribution-based Outlier Detection), which is inspired by the fact that outliers are often the "rare events" that appear in the tails of a distribution. In a nutshell, ECOD first estimates the underlying distribution of the input data in a nonparametric fashion by computing the empirical cumulative distribution per dimension of the data. ECOD then uses these empirical distributions to estimate tail probabilities per dimension for each data point. Finally, ECOD computes an outlier score of each data point by aggregating estimated tail probabilities across dimensions. Our contributions are as follows: (1) we propose a novel outlier detection method called ECOD, which is both parameter-free and easy to interpret; (2) we perform extensive experiments on 30 benchmark datasets, where we find that ECOD outperforms 11 state-of-the-art baselines in terms of accuracy, efficiency, and scalability; and (3) we release an easy-to-use and scalable (with distributed support) Python implementation for accessibility and reproducibility.

This review synthesizes recent advancements in understanding tipping points and cascading transitions within the Earth system, framing them through the lens of nonlinear dynamics and complexity science. It outlines the fundamental concepts of tipping elements, large-scale subsystems like the Atlantic Meridional Overturning Circulation and the Amazon rainforest, and classifies tipping mechanisms into bifurcation-, noise-, and rate-induced types. The article critically evaluates methods for detecting early-warning signals, particularly those based on critical slowing down, while also acknowledging their limitations and the promise of non-conventional indicators. Furthermore, we explore the significant risk of cascading failures between interacting tipping elements, often modeled using conceptual network models. This shows that such interactions can substantially increase systemic risk under global warming. The review concludes by outlining key challenges related to data limitations and methodological robustness, and emphasizes the promising role of artificial intelligence and complex network science in advancing prediction and risk assessment of Earth system tipping dynamics.

The phenomenon of critical slowing down (CSD) has played a key role in the search for reliable precursors of catastrophic regime shifts. This is caused by its presence in a generic class of bifurcating dynamical systems. Simple time-series statistics such as variance or autocorrelation can be taken as proxies for the phenomenon, making their increase a useful early warning signal (EWS) for catastrophic regime shifts. However, the modelling basis justifying the use of these EWSs is usually a finite-dimensional stochastic ordinary differential equation, where a mathematical proof for the aptness is possible. Only recently has the phenomenon of CSD been proven to exist in infinite-dimensional stochastic partial differential equations (SPDEs), which are more appropriate to model real-world spatial systems. In this context, we provide an essential extension of the results for SPDEs under a specific noise forcing, often referred to as red noise. This type of time-correlated noise is omnipresent in many physical systems, such as climate and ecology. We approach the question with a mathematical proof and a numerical analysis for the linearised problem. We find that also under red noise forcing, the aptness of EWSs persists, supporting their employment in a wide range of applications. However, we also find that false or muted warnings are possible if the noise correlations are non-stationary. We thereby extend a previously known complication with respect to red noise and EWSs from finite-dimensional dynamics to the more complex and realistic setting of SPDEs.

To mitigate climate change, the share of renewable energies in power production needs to be increased. Renewables introduce new challenges to power grids regarding the dynamic stability due to decentralization, reduced inertia, and volatility in production. Since dynamic stability simulations are intractable and exceedingly expensive for large grids, graph neural networks (GNNs) are a promising method to reduce the computational effort of analyzing the dynamic stability of power grids. As a testbed for GNN models, we generate new, large datasets of dynamic stability of synthetic power grids, and provide them as an open-source resource to the research community. We find that GNNs are surprisingly effective at predicting the highly non-linear targets from topological information only. For the first time, performance that is suitable for practical use cases is achieved. Furthermore, we demonstrate the ability of these models to accurately identify particular vulnerable nodes in power grids, so-called troublemakers. Last, we find that GNNs trained on small grids generate accurate predictions on a large synthetic model of the Texan power grid, which illustrates the potential for real-world applications.

Predicting chaotic dynamical systems is critical in many scientific fields, such as weather forecasting, but challenging due to the characteristic sensitive dependence on initial conditions. Traditional modeling approaches require extensive domain knowledge, often leading to a shift towards data-driven methods using machine learning. However, existing research provides inconclusive results on which machine learning methods are best suited for predicting chaotic systems. In this paper, we compare different lightweight and heavyweight machine learning architectures using extensive existing benchmark databases, as well as a newly introduced database that allows for uncertainty quantification in the benchmark results. In addition to state-of-the-art methods from the literature, we also present new advantageous variants of established methods. Hyperparameter tuning is adjusted based on computational cost, with more tuning allocated to less costly methods. Furthermore, we introduce the cumulative maximum error, a novel metric that combines desirable properties of traditional metrics and is tailored for chaotic systems. Our results show that well-tuned simple methods, as well as untuned baseline methods, often outperform state-of-the-art deep learning models, but their performance can vary significantly with different experimental setups. These findings highlight the importance of aligning prediction methods with data characteristics and caution against the indiscriminate use of overly complex models.