For the 2022 March meeting, the Topical Group on Statistical and Nonlinear Physics is pleased to present six invited sessions and to serve as the primary sponsor of eleven focus topics. A big thanks goes to the members of GSNP for making this program possible! We encourage you to get involved in the excitement by submitting contributed abstracts to the GSNP focus topics. Each invited session is paired with a focus topic to help facilitate interaction. As a reminder, abstracts are due by October 22.
This year’s invited sessions include granular and particulate systems, network theory in a variety of contexts, the deeper statistical underpinnings of machine learning, and the intersection of quantum mechanics and thermodynamics. The session titles and their paired focus topic numbers are:
Flow and Packing of Dense Granular Materials [paired with 03.01.01]
Statistical Physics Meets Machine Learning [paired with 03.01.06]
Network Physics of Particulate Systems [paired with 03.01.03]
Network Theory and Applications to Complex Systems [paired with 03.01.04]
Quantum Heat Engines [paired with 03.01.09]
Thermodynamics of Quantum Information Processing [paired with 03.01.09]
In addition to the above areas, this year’s focus topics with primary sponsorship by GSNP will address the statistical physics of disease propagation, machine learning applied to nonlinear and complex systems, noise-driven dynamics far from equilibrium, thermodynamics of information processing in biological and artificial systems, higher-order interactions, and methods of manipulating steerable particles. A list of the focus topic titles and numbers is below, followed by a list of descriptions of each focus topic. (Note: GSNP is co-sponsoring many other focus topics with a variety of units, but for the topics listed below, GSNP is the primary sponsor.)
03.01.01 Nonlinear Response of Complex Granular Materials
03.01.02 Statistical Physics of Disease Propagation
03.01.03 Network Physics of Particulate Systems
03.01.04 Network Theory and Applications to Complex Systems
03.01.05 Predicting Nonlinear and Complex Systems with Machine Learning
03.01.06 Statistical Physics Meets Machine Learning
03.01.07 Noise-Driven Dynamics in Far-From-Equilibrium Systems
03.01.08 Stochastic Thermodynamics of Biological and Artificial Information Processing
03.01.09 Thermodynamics in Quantum Information
03.01.10 Higher-Order Interactions: The Next Frontier of Complex Systems
03.01.11 Steerable Particles: New Ways to Manipulate Fluid-Mediated Forces
03.01.01 Nonlinear Response of Complex Granular Materials
The mechanics and rheology of complex granular materials are routinely observed in industry and in nature such as in various geophysical phenomena. Though most studies of granular materials have analyzed moderately size-disperse spheres with simple contact interactions, real granular materials exhibit complex frictional, cohesive, electrostatic and brittle plastic interactions that are intricately coupled to their mechanics and rheology, both in dry states and suspensions. Furthermore, dispersity in particle sizes and shapes, along with the ability of the particles to fracture, can lead to complex, nonlinear phenomena such as segregation and stratification.
Organizers: Gary Grest, Jeremy Lechman, Ishan Srivastava
03.01.02 Statistical Physics of Disease Propagation
Long before COVID-19 became a household name, a variety of statistical mechanics approaches were developed to tackle the problem of disease propagation in systems with complex transmission environments. This session focuses on the application of such techniques to the evolution, prediction, and control of the COVID-19 outbreak and to other disease outbreaks. The models may range from homogeneously mixing (mean-field) populations to studies with behavioral feedback and/or spatial or social structure in the population. Techniques could include network analysis, compartmental modeling, epidemiology, percolation concepts, and general nonequilibrium approaches.
Organizers: Tim Germann, Cynthia Reichhardt
03.01.03 Network Physics of Particulate Systems
Over the past few years, experimental, theoretical and computational research on physics and rheology of particulate systems have made considerable progress in better understanding the connections between different length scales - from microscopic origins of different forces to a continuum-level constitutive description of the entire system. These developments have brought together researchers from tribology, granular physics, fluid mechanics and rheology, with much of the recent literature focusing on the micro and macro scales. These have also resulted in a consensus that we need to bridge the gap between the microscopic physics at the contact and particle scales to macroscopic rheology. Of particular interest are studies that reveal mechanisms of force and stress propagation and involve mesoscale force networks and their evolution under flowing conditions. With respect to specific systems of interest, shear-thickening dense suspensions, fluidization and strain localization in granular systems, polymers and hydrogels, and complex rheology of colloidal gels are widely accepted to be controlled directly by the force and contact networks formed within the material's microstructure. These investigations should include developments of both computational and experimental techniques to resolve details of contact and force networks that ultimately control the behavior of the system. As such, bringing network scientists, and a deeper understanding of the mathematics of networks, to this area will potentially open up new areas of research and explorations.
Organizers: Jeffrey Morris, Safa Jamali
03.01.04 Network Theory and Applications to Complex Systems
Many real systems of interest in physics, engineering, social and computer sciences, biology and finance can be represented as networks: the Internet, the World Wide Web, social networks, transportation systems, scientific citation graphs, biochemical and ecological networks, and financial credit graphs are just a few examples of real systems that can be modelled as networks, and thinking of them in this way can lead to useful insights. Network science has produced many fundamental results for the description of critical phenomena in real-world networks. These represent essential tools for the control and design of real systems with concrete and practical applications. For example, thanks to the results of network science, we know how to design a computer network that is less vulnerable under intentional or random attacks, what is the best strategy for the immunization of a social network subjected to the spreading of a virus, how to identify the most influential nodes in a network, and how to drive a dynamical system living on a complex network towards the desired state. The number of real-world applications of network science is literally exploding due to the growing availability of data and computational resources. The goal of this Focus Topic is to address recent advances in the network theory of real-world complex systems.
Organizer: Filippo Radicchi
03.01.05 Predicting Nonlinear and Complex Systems with Machine Learning
In recent years, machine learning techniques have been applied to nonlinear and complex dynamical systems to solve previously unsolved problems in the field. For example, a research area based on reservoir computing, a class of recurrent neural networks, has gained considerable momentum and emerged as a powerful paradigm for model-free prediction of the state evolution of nonlinear and chaotic dynamical systems. Machine learning methods have also been developed to predict critical transitions, transient chaos, and Hamiltonian chaos. This Focus Session will bring together a number of active researchers to discuss current issues in this emergent interdisciplinary field.
Organizer: Yin-Cheng Lai
03.01.06 Statistical Physics Meets Machine Learning
In the past 15 years, there has been tremendous development in machine learning based on deep neural networks (DNNs). However, despite their many successful applications, there is no theory regarding the underlying principles of DNNs, i.e., why they work and how they work. Historically, statistical physics played an important role in the initial development of artificial neural networks, such as the Hopfield model, the Boltzmann machine, and applications of spin-glass theory to neural networks. We believe the time is ripe to develop a solid theoretical foundation for DNN algorithms based on concepts and methods from statistical physics. This focus session will bring experts from the statistical physics and machine learning community together to discuss fundamental issues and possible directions for understanding and advancing artificial intelligence research based on ideas and tools from statistical physics.
Organizers: Yuhai Tu, David Schwab
03.01.07 Noise-Driven Dynamics in Far-From-Equilibrium Systems
The last few years have witnessed impressive experimental and theoretical progress to quantitatively characterize noise-driven dynamics in far-from-equilibrium systems. Examples abound in diverse systems such as active biological matter, optically levitated nanoparticles, electronic transport circuits, climate dynamics, as well as voting models and financial markets. Common features include the observation and characterization of non-vanishing probability currents in steady state, the development of novel metrics to quantify how far systems are from equilibrium behavior, the characterization of detailed balance violation - an essential feature in the functioning of many non-equilibrium systems, and the generalization of fluctuation-dissipation relations. The proposed session will bring together theoretical and experimental researchers from a range of traditional fields including biophysics, nonlinear and statistical physics, climate science, and condensed matter physics, for whom it will be stimulating to explore common sets of new and emerging analytical tools and techniques for understanding the noisy dynamics of far-from-equilibrium systems.
Organizer: Stephen Teitsworth
03.01.08 Stochastic Thermodynamics of Biological and Artificial Information Processing
The recent revolution in non-equilibrium statistical physics has allowed researchers to significantly advance and generalize Landauer's original results concerning the thermodynamics of bit-erasure into a full-fledged "thermodynamics of information processing," which analyzes computational systems ranging from information ratchets to digital circuits, all the way up to Turing machines. At the same time, our understanding of the information processing within biological cells has greatly expanded, driving an explosion of work on the thermodynamics of biological information processing, including processes such as polymerization, kinetic proofreading, and cellular sensing. Recently researchers have started to consolidate these research thrusts into a unifying thermodynamics of information processing/computation in both biological and artificial (human-engineered) systems. In addition, the community is now beginning to apply stochastic thermodynamics more broadly, to everything from social opinion networks to replicator dynamics to electronic circuits to neurobiology. The aim of this session is to continue to deepen our understanding of this important set of issues.
Organizers: David Wolpert, Jan Korbel
03.01.09 Thermodynamics in Quantum Information
Ever since its inception in the 1950s the notion of a "quantum heat engine" has been at the forefront of research in quantum thermodynamics; however, only over the last five years or so have experimentalists started to succeed in realizing genuinely "quantum" engines. In particular, the global push for the development of quantum technologies has sparked concentrated efforts to produce new devices. At the same time, with the rise of quantum technologies, understanding the laws governing energy, entropy and information flows between quantum systems is of crucial importance. Practical motivations are to keep a fair account of the resources to process information in the quantum realm, and optimize them with respect to well-defined performances. The question of the physical resources potentially consumed by quantum computing (whether deterministic or reservoir-assisted) has been largely overlooked so far, while it may become a bottleneck for scalability. This session aims to open an interdisciplinary dialog and lay the ground of new research addressing these questions.
Organizer: Nicole Halpern; organizers of associated invited sessions: Obinna Abah, Sebastian Deffner, Alexia Auffeves, Benjamin Huard, and Andrew Jordan
03.01.10 Higher-Order Interactions: The Next Frontier of Complex Systems
The complexity of many biological, social and technological systems stems from the interaction richness among their units. Over the last few decades, many complex systems have been successfully described as networks whose links connect interacting pairs of nodes; however, it is not always possible to describe group interactions as sums of pairwise interactions only. Indeed, recent works are revealing new types of behaviors. Despite the mounting evidence showing that considering the high-order structure of these systems can significantly enhance our capacity to understand and predict their emerging dynamical behaviors, little attention was given to their higher-order architecture, and just recently, the topic has begun to attract considerable attention. Capitalizing on algebraic topology, statistical mechanics, and hypergraph theory, we aim to stimulate the discussion about how, when, and why current descriptions of complex systems can be extended to higher-order formalisms.
Organizers: Guilherme Ferraz de Arruda, Giovanni Petri, Yamir Moreno
03.01.11 Steerable Particles: New Ways to Manipulate Fluid-Mediated Forces
The transport of fluid-suspended particles by various driving forces occurs in natural systems and various technologies. Well-known examples include the transport of charged particles by an electric field (electrophoresis), dielectric particles by an electric field gradient (dielectrophoresis), magnetic particles by a magnetic field gradient (magnetophoresis), buoyant or heavy particles by gravity (sedimentation), and particles experiencing a temperature gradient (thermophoresis) or a concentration gradient (diffusophoresis). The past several years have seen a body of work focused on new means to drive mesoscopic particles in more intricate and programmable ways using their shape, charge distribution, magnetic response, optical properties, or through tuned environments. Importantly, sophisticated driving of colloidal particles has been used to examine fundamental issues of nonequilibrium statistical physics. This category of externally-driven systems should be distinguished from that of active, self-propelled particles, which has already attracted a lot of attention.
Organizer: Haim Diamant