Each year, Chaos honors exceptional emerging researchers whose work embodies the creativity, rigor, and boundary-crossing curiosity that define nonlinear science. The 2024 Edward N. Lorenz Early Career Award highlights four such scholars: Rahil Valani, Adam Giammarese, Mirko Goldmann, and Guillaume Pourcel—researchers whose investigations span active matter, climate networks, and the dynamical understanding of neural computation.
As he concludes his term as Editor-in-Chief, Jürgen Kurths reflected on this year’s selection:
“The Lorenz Early Career Award has always represented the spirit of curiosity and boldness that drives nonlinear science forward. This year’s awardees have demonstrated remarkable originality—applying dynamical systems concepts to climate networks, active matter, and neural computation in ways that would have delighted Lorenz himself. As I conclude my term as Editor-in-Chief, it is deeply rewarding to see a new generation carrying this field into exciting new territory.”
Rahil Valani
For Rahil Valani, now a Leverhulme-Peierls Fellow at the University of Oxford, nonlinear science has always been inseparable from the world of physical intuition. His path began in aerospace engineering and science at Monash University, where he discovered a fascination with “walking” and “superwalking” droplets—tiny bouncing particles that move across a vibrating fluid bath while generating waves that guide their own motion. These wave-particle entities exist at the border of order and chaos, tracing paths that mimic quantum-like statistics while remaining entirely classical.
This interplay between internal memory, complex trajectories, and attractor geometry lies at the heart of Valani’s award-winning paper, “Infinite-memory classical wave-particle entities, attractor-driven active particles and the diffusionless Lorenz equations” (DOI: 10.1063/5.0171007). In it, he shows that in the extreme high-memory regime, the dynamics of these droplets reduce to the diffusionless Lorenz equations—one of the simplest algebraic systems capable of generating chaos. By mapping experimental and model droplet trajectories onto the geometry and bifurcations of Lorenz-type attractors, Valani demonstrates how deeply the language of nonlinear dynamics can illuminate physical behavior that appears, at first glance, almost quantum.
“Lorenz’s work on chaos and attractors has been a constant source of inspiration in my research career,” Valani said. “My own research connects Lorenz chaos with the complex motion of memory-driven active particles. It’s an honor to receive this award.”
His future work pushes these ideas toward collective phenomena: not one droplet but many, each carrying its own internal memory and attractor-driven behavior. He is developing a broader framework of “attractor-driven matter,” where each particle’s internal dynamics—not just its external interactions—shape the emergent patterns of the whole.
Adam Giammarese
Where Valani explores the behavior of physical particles in motion, Adam Giammarese explores the climate system as a dynamic, interconnected whole. Giammarese began his academic journey at the Rochester Institute of Technology, earning dual degrees in applied mathematics and mechanical engineering before completing a PhD in mathematical modeling. His doctoral work focused on data-driven, equation-free methods for studying climate phenomena too complex to capture with traditional systems of equations.
Today, as a Staff Engineer at Numerical Advisory Solutions, he applies this expertise to clean-energy systems. But it is his contribution to climate network analysis that earned him the Lorenz Award. In “Reconfiguration of Amazon’s connectivity in the climate system” (DOI: 10.1063/5.0165861), Giammarese uses a suite of complex-network tools—graph diffusion, random walks, Laplacian eigenmaps, and more—to examine how the Amazon rainforest’s climatic interactions are shifting over time. The Amazon is widely considered one of the planet’s most sensitive tipping elements, and his work reveals a striking structural change: local climatic behavior is becoming more homogeneous while long-range connections are strengthening.
Rather than studying only temperature or precipitation, Giammarese studies the relationships between them—how patterns propagate through the climate system like signals in a network. The reconfiguration he identifies offers a new perspective on how tipping behaviors may first manifest.
“I am incredibly honored to receive this award,” he said. “I started my academic career as an engineering student with career-focused aspirations but found a passion in applied mathematics. This award encourages me to continue pursuing my passions.”
He hopes to continue supporting climate network research, even as his current projects focus more on forecasting and reconstruction of climate time series. With climate data becoming ever more abundant, he believes that data-driven and network-based approaches will only grow in significance.
Mirko Goldmann and Guillaume Pourcel
The joint work of Mirko Goldmann of Akhetonics (Germany) and Guillaume Pourcel of the University of Groningen brings together physics, cognitive science, machine learning, and nonlinear dynamics. Their collaboration began within the POST-DIGITAL European Training Network, a program designed to bridge theoretical approaches and hardware for light-based machine learning.
Goldmann’s background spans photonics and physics, while Pourcel’s lies in cognitive science and computational models of intelligence. Together, they investigate neural networks not as abstract computational objects, but as dynamical systems—entities that evolve, adapt, and organize information through time.
Their award-winning paper, “Adaptive control of recurrent neural networks using conceptors” (DOI: 10.1063/5.0211692), demonstrates that recurrent neural networks (RNNs) can remain adaptive even after training, contradicting the standard view that a network’s internal parameters are fixed once learning is complete. The key is the use of conceptors, mathematical objects that describe low-dimensional subspaces of network activity. Goldmann and Pourcel show how implementing an adaptive feedback loop around these conceptors enables an RNN to interpolate smoothly between learned patterns, maintain performance when neurons are removed, and even correct distorted inputs.
Their work brings clarity and interpretability to systems often treated as black boxes, while also suggesting practical strategies for increased robustness in real-world hardware implementations.
“Chaos has always been a melting pot of interdisciplinary ideas, bridging physics, mathematics, and complex systems research,” they said. “This award encourages us to continue exploring such interdisciplinary connections and to contribute to a deeper understanding of complex phenomena.”
Looking forward, they plan to extend conceptor dynamics to more physical and hardware-based systems and investigate how conceptor principles might integrate with gradient-based learning, potentially unifying dynamical and machine-learning approaches.
A Shared Vision for Nonlinear Science
Though their research spans different scales and systems, this year’s Lorenz Award winners share a common trait: they use nonlinear science not merely as a tool, but as a lens through which to reinterpret complex phenomena. From droplets dancing on a fluid surface, to the connectivity of the Amazon rainforest, to adaptive neural networks, each awardee demonstrates how nonlinear thinking can illuminate the hidden structures of the natural and computational world. These winners offer a compelling glimpse of the field’s future, one where interdisciplinary insights, data-driven approaches, and dynamical intuition continue to shape our understanding of complex systems.
To learn more about the Edward N. Lorenz Early Career Award and past winners, visit: https://pubs.aip.org/aip/cha/pages/award.