arXiv subjects

Andrea J. Liu

Publications and source records attributed to Andrea J. Liu.

At least 19 recordsLinked to original sources

arXiv2026

Thermally Robust Cooperative Function in Mechanical Networks

Elastic networks can be tuned to exhibit complex mechanical responses and have been extensively used to study protein allostery, where a localized strain regulates the conformation at a distant site. We show that cooperative binding, where one site enhances the other's ability to function, can be improved by tuning for \textit{bidirectional} allostery: a symmetric coupling where the strain propagated between the two sites is independent of which site is perturbed. We identify a crossover temperature above which functionality breaks down due to thermal fluctuations. We introduce a modified tuning process to increase this crossover temperature, showing that function can be robust at biologically relevant temperatures if thermal stability is important.

cond-mat.soft↗

arXiv2026

Contrastive learning in tunable dynamical systems

We generalize the theory of supervised contrastive learning, previously applied to physical systems at equilibrium or steady state, to systems following any dynamics described by coupled ordinary differential equations. We show that if physical dynamics break time reversal symmetry, gradient descent on a cost function embodying the desired behavior cannot be achieved with a scalable process, even in principle. We therefore introduce Probably Approximately Right (PAR) learning processes, composed of a local contrastive learning rule and a scalable supervision protocol. We show that approximate, local supervision with forward propagation of the error signal can be used to successfully train several tunable models of physical dynamics inspired by examples in biological and machine learning.

cond-mat.dis-nn↗

arXiv2026

Understanding and Embracing Imperfection in Physical Learning Networks

Performing machine learning with analog signals offers advantages in speed and energy efficiency, but sensitivity to component and measurement imperfections often foils training without a system-specific companion digital model. Here we take a different perspective, accepting and characterizing these inherent imperfections and ultimately overcoming them without digital models. We train an analog network of self-adjusting resistors -- a contrastive local learning network -- for multiple tasks, and observe limit cycles and scaling behaviors that limit precision, erase memory of previous tasks, and are absent in `perfect' systems. We develop an analytical model capturing these phenomena as a consequence of an uncontrolled learning bias continuously modifying the underlying representation of learned tasks, reminiscent of representational drift in the brain. Finally, we introduce and demonstrate a system-agnostic training method that greatly suppresses these effects. Our work points to a new, scalable analog approach that eschews precise modeling and instead thrives in the mess of real systems.

cond-mat.dis-nn↗

arXiv2025

Remembrance of Tasks Past in Tunable Physical Networks

Sequential learning in physical networks is hindered by catastrophic forgetting, where training a new task erases solutions to earlier ones. We show that we can significantly enhance memory of previous tasks by introducing a hard threshold in the learning rule, allowing only edges with sufficiently large training signals to be altered. Thresholding confines tuning to the spatial vicinity of inputs and outputs for each task, effectively partitioning the network into weakly overlapping functional regions. Using simulations of tunable resistor networks, we demonstrate that this strategy enables robust memory of multiple sequential tasks while reducing the number of edges and the overall tuning cost. Our results hint at constrained training as a simple, local, and scalable mechanism to overcome catastrophic forgetting in tunable matter.

cond-mat.dis-nn↗

arXiv2025

Unsupervised and probabilistic learning with Contrastive Local Learning Networks: The Restricted Kirchhoff Machine

Autonomous physical learning systems modify their internal parameters and solve computational tasks without relying on external computation. Compared to traditional computers, they enjoy distributed and energy-efficient learning due to their physical dynamics. In this paper, we introduce a self-learning resistor network, the Restricted Kirchhoff Machine, capable of solving unsupervised learning tasks akin to the Restricted Boltzmann Machine algorithm. The circuit relies on existing technology based on Contrastive Local Learning Networks, in which two identical networks compare different physical states to implement a contrastive local learning rule. We simulate the training of the machine on the binarized MNIST dataset, providing a proof of concept of its learning capabilities. Finally, we compare the scaling behavior of the time, power, and energy consumed per operation as more nodes are included in the machine to their Restricted Boltzmann Machine counterpart operated on CPU and GPU platforms.

cond-mat.dis-nn↗

arXiv2025

Solving the compute crisis with physics-based ASICs

Escalating artificial intelligence (AI) demands expose a critical "compute crisis" characterized by unsustainable energy consumption, prohibitive training costs, and the approaching limits of conventional CMOS scaling. Physics-based Application-Specific Integrated Circuits (ASICs) present a transformative paradigm by directly harnessing intrinsic physical dynamics for computation rather than expending resources to enforce idealized digital abstractions. By relaxing the constraints needed for traditional ASICs, like enforced statelessness, unidirectionality, determinism, and synchronization, these devices aim to operate as exact realizations of physical processes, offering substantial gains in energy efficiency and computational throughput. This approach enables novel co-design strategies, aligning algorithmic requirements with the inherent computational primitives of physical systems. Physics-based ASICs could accelerate critical AI applications like diffusion models, sampling, optimization, and neural network inference as well as traditional computational workloads like scientific simulation of materials and molecules. Ultimately, this vision points towards a future of heterogeneous, highly-specialized computing platforms capable of overcoming current scaling bottlenecks and unlocking new frontiers in computational power and efficiency.

cs.ET↗

arXiv2025

Rigidity of Epithelial Tissues as a Double Optimization Problem

How do cells tune emergent properties at the scale of tissues? One class of such emergent behaviors are rigidity transitions, in which a tissue changes from a solid-like to a fluid-like state or vice versa. Here, we introduce a new way for a tissue described by a vertex model to tune its rigidity, by using ``tunable degrees of freedom." We use the vertex model elastic energy as a cost function and the cell stiffnesses, target shapes, and target areas as different sets of degrees of freedom describing cell-cell interactions that can be tuned to minimize the cost function. We show that the rigidity transition is unaffected when cell stiffnesses are treated as tunable degrees of freedom. When preferred shapes or areas are treated as tunable degrees of freedom, however, induced spatial correlations in target cell shapes or areas shift the rigidity transition. These observations suggest that tissues can coordinate changes in cell-scale properties, treated here as tunable degrees of freedom, to achieve desired tissue-scale behaviors.

cond-mat.soft↗

arXiv2024

Microscopic imprints of learned solutions in adaptive resistor networks

In physical networks trained using supervised learning, physical parameters are adjusted to produce desired responses to inputs. An example is electrical contrastive local learning networks of nodes connected by edges that are resistors that adjust their conductances during training. When an edge conductance changes, it upsets the current balance of every node. In response, physics adjusts the node voltages to minimize the dissipated power. Learning in these systems is therefore a coupled double-optimization process, in which the network descends both a cost landscape in the high-dimensional space of edge conductances, and a physical landscape -- the power -- in the high-dimensional space of node voltages. Because of this coupling, the physical landscape of a trained network contains information about the learned task. Here we demonstrate that all the physical information relevant to the trained input-output relation can be captured by a susceptibility, an experimentally measurable quantity. We supplement our theoretical results with simulations to show that the susceptibility is positively correlated with functional importance and that we can extract physical insight into how the system performs the task from the conductances of highly susceptible edges.

cond-mat.dis-nn↗

arXiv2024

Building rigid networks with prestress and selective pruning

Biopolymer networks from the intracellular to tissue scale display high rigidity and tensile stress while having coordinations well below the normal threshold for mechanical rigidity. The elastic filaments in these networks are often severed by enzymes in a tension-inhibited manner. The effects of such pruning on the mechanics of prestressed networks have not been studied. We show that networks pruned by a tension-inhibited method remain rigid at much lower coordinations than randomly pruned ones. These findings suggest a possible reason for the repeated evolution of tension-inhibited filament-severing proteins.

cond-mat.soft↗

arXiv2024

Insight: A Multi-Modal Diagnostic Pipeline using LLMs for Ocular Surface Disease Diagnosis

Accurate diagnosis of ocular surface diseases is critical in optometry and ophthalmology, which hinge on integrating clinical data sources (e.g., meibography imaging and clinical metadata). Traditional human assessments lack precision in quantifying clinical observations, while current machine-based methods often treat diagnoses as multi-class classification problems, limiting the diagnoses to a predefined closed-set of curated answers without reasoning the clinical relevance of each variable to the diagnosis. To tackle these challenges, we introduce an innovative multi-modal diagnostic pipeline (MDPipe) by employing large language models (LLMs) for ocular surface disease diagnosis. We first employ a visual translator to interpret meibography images by converting them into quantifiable morphology data, facilitating their integration with clinical metadata and enabling the communication of nuanced medical insight to LLMs. To further advance this communication, we introduce a LLM-based summarizer to contextualize the insight from the combined morphology and clinical metadata, and generate clinical report summaries. Finally, we refine the LLMs' reasoning ability with domain-specific insight from real-life clinician diagnoses. Our evaluation across diverse ocular surface disease diagnosis benchmarks demonstrates that MDPipe outperforms existing standards, including GPT-4, and provides clinically sound rationales for diagnoses.

cs.CL↗

arXiv2024

Experimental Demonstration of Coupled Learning in Elastic Networks

Coupled learning is a contrastive scheme for tuning the properties of individual elements within a network in order to achieve desired functionality of the system. It takes advantage of physics both to learn using local rules and to "compute" the output response to input data, thus enabling the system to perform decentralized computation without the need for a processor or external memory. We demonstrate a proof-of-concept mechanical network that can learn simple tasks such as self-symmetrizing via iterative tuning of individual spring rest lengths. These mechanical networks could feasibly be scaled and automated to solve increasingly complex tasks, hinting at a new class of smart metamaterials.

cond-mat.soft↗

arXiv2024

Cornerstones are the Key Stones: Using Interpretable Machine Learning to Probe the Clogging Process in 2D Granular Hoppers

The sudden arrest of flow by formation of a stable arch over an outlet is a unique and characteristic feature of granular materials. Previous work suggests that grains near the outlet randomly sample configurational flow microstates until a clog-causing flow microstate is reached. However, factors that lead to clogging remain elusive. Here we experimentally observe over 50,000 clogging events for a tridisperse mixture of quasi-2D circular grains, and utilize a variety of machine learning (ML) methods to search for predictive signatures of clogging microstates. This approach fares just modestly better than chance. Nevertheless, our analysis using linear Support Vector Machines (SVMs) highlights the position of potential arch cornerstones as a key factor in clogging likelihood. We verify this experimentally by varying the position of a fixed (cornerstone) grain, and show that such a grain dictates the size of feasible flow-ending arches, and thus the time and mass of each flow. Positioning this grain correctly can even increase the ejected mass by over 50%. Our findings demonstrate that interpretable ML algorithms like SVMs can uncover meaningful physics even when their predictive power is below the standards of conventional ML practice.

cond-mat.soft↗

arXiv2024

The dynamics of machine-learned "softness" in supercooled liquids describe dynamical heterogeneity

The dynamics of supercooled liquids slow down and become increasingly heterogeneous as they are cooled. Recently, local structural variables identified using machine learning, such as "softness", have emerged as predictors of local dynamics. Here we construct a model using softness to describe the structural origins of dynamical heterogeneity in supercooled liquids. In our model, the probability of particles to rearrange is determined by their softness, and each rearrangement induces changes in the softness of nearby particles, describing facilitation. We show how to ensure that these changes respect the underlying time-reversal symmetry of the liquid's dynamics. The model reproduces the salient features of dynamical heterogeneity, and demonstrates how long-ranged dynamical correlations can emerge at long time scales from a relatively short softness correlation length.

cond-mat.soft↗

arXiv2024

Training of Physical Neural Networks

Physical neural networks (PNNs) are a class of neural-like networks that leverage the properties of physical systems to perform computation. While PNNs are so far a niche research area with small-scale laboratory demonstrations, they are arguably one of the most underappreciated important opportunities in modern AI. Could we train AI models 1000x larger than current ones? Could we do this and also have them perform inference locally and privately on edge devices, such as smartphones or sensors? Research over the past few years has shown that the answer to all these questions is likely "yes, with enough research": PNNs could one day radically change what is possible and practical for AI systems. To do this will however require rethinking both how AI models work, and how they are trained - primarily by considering the problems through the constraints of the underlying hardware physics. To train PNNs at large scale, many methods including backpropagation-based and backpropagation-free approaches are now being explored. These methods have various trade-offs, and so far no method has been shown to scale to the same scale and performance as the backpropagation algorithm widely used in deep learning today. However, this is rapidly changing, and a diverse ecosystem of training techniques provides clues for how PNNs may one day be utilized to create both more efficient realizations of current-scale AI models, and to enable unprecedented-scale models.

physics.app-ph↗

arXiv2024

Dynamical Facilitation Governs the Equilibration Dynamics of Glasses

Convincing evidence of domain growth in the heating of ultrastable glasses suggests that the equilibration dynamics of super-cooled liquids could be driven by a nucleation and growth mechanism. We investigate this possibility by simulating the equilibration dynamics of a model glass during both heating and cooling between poorly and well-annealed states. Though we do observe the growth of domains during heating, we find that domains are absent during cooling. This absence is inconsistent with classical nucleation theory. By comparing the equilibration dynamics of our glass with that of two models with kinetic constraints, we demonstrate that dynamical facilitation generically leads to heating driven by domain growth and cooling without domains. Our results provide strong evidence that dynamical facilitation, not nucleation and interfacial-tension-driven domain growth, is the driving mechanism for the equilibration dynamics of glass-formers.

cond-mat.soft↗

arXiv2024

Universal scaling function ansatz for finite-temperature jamming

We cast a nonzero-temperature analysis of the jamming transition into the framework of a scaling ansatz. We show that four distinct regimes for scaling exponents of thermodynamic derivatives of the free energy such as pressure, bulk and shear moduli, can be consolidated by introducing a universal scaling function with two branches. Both the original analysis and the scaling theory assume that the system always resides in a single basis in the energy landscape. The two branches are separated by a line $T^*(Δϕ)$ in the $T-Δϕ$ plane, where $Δϕ=ϕ-ϕ_c^Λ$ is the deviation of the packing fraction from its critical, jamming value, $ϕ_c^Λ$, for that basin. The branch for $T T^*(Δϕ)$ reproduces exponents observed for thermal hard spheres. In contrast to the usual scenario for critical phenomena, the two branches are characterized by different exponents. We suggest that this unusual feature can be resolved by the existence of a dangerous irrelevant variable $u$, which can appear to modify exponents if the leading $u=0$ term is sufficiently small in the regime described by one of the two branches of the scaling function.

cond-mat.soft↗

arXiv2024

The Physical Effects of Learning

Interacting many-body physical systems ranging from neural networks in the brain to folding proteins to self-modifying electrical circuits can learn to perform diverse tasks. This learning, both in nature and in engineered systems, can occur through evolutionary selection or through dynamical rules that drive active learning from experience. Here, we show that \added{learning in linear physical networks with weak input signals} leaves architectural imprints on the Hessian of a physical system. Compared to a generic organization of the system components, (a) the effective physical dimension of the response to inputs decreases, (b) the response of physical degrees of freedom to random perturbations (or system ``susceptibility'') increases, and (c) the low-eigenvalue eigenvectors of the Hessian align with the task. Overall, these effects embody the typical scenario for learning processes in physical systems in the weak input regime, suggesting ways of discovering whether a physical network may have been trained.

cond-mat.dis-nn↗

arXiv2024

Topological characterization of the continuum of allosteric response

Allosteric regulation in proteins is often accompanied by conformational changes that facilitate transmission of mechanical signals between distant ligand binding sites. Typically, these deformations are classified in terms of specific archetypes, including various types of hinge mechanisms or allosteric pathways localized to sequences of amino acids.However, many allosteric deformations resist such strict categorization. Here, we introduce a quantitative topological description of allosteric deformation, unifying all archetypal mechanisms into a single framework. The topological description aligns with two key structural features often associated with allosteric deformations, namely hinge domains and allosteric pathways, enabling us to quantify the significance of each of these features. To develop the analysis, we tune computer-generated mechanical networks to perform allostery-like functions, obtaining an ensemble of networks that establish a range of possible allosteric deformations. The analysis shows that these networks' allosteric mechanisms cannot be described in terms of discrete archetypes - they fall on a continuum. We then apply the same analysis to a collection of allosteric proteins with similar results, showing that our framework encompasses these proteins as well as designed allosteric networks. Our results provide a new picture for allostery, demonstrating not only how it can be described quantitatively, but also giving insight into how it emerges as a collective property.

cond-mat.soft↗