arXiv subjects

Arvind Murugan

Publications and source records attributed to Arvind Murugan.

At least 37 records · Page 2Linked to original sources

arXiv2023

Closed ecosystems extract energy through self-organized nutrient cycles

Our planet is roughly closed to matter, but open to energy input from the sun. However, to harness this energy, organisms must transform matter from one chemical (redox) state to another. For example, photosynthetic organisms can capture light energy by carrying out a pair of electron donor and acceptor transformations (e.g., water to oxygen, CO$_2$ to organic carbon). Closure of ecosystems to matter requires that all such transformations are ultimately balanced, i.e., other organisms must carry out corresponding reverse transformations, resulting in cycles that are coupled to each other. A sustainable closed ecosystem thus requires self-organized cycles of matter, in which every transformation has sufficient thermodynamic favorability to maintain an adequate number of organisms carrying out that process. Here, we propose a new conceptual model that explains the self-organization and emergent features of closed ecosystems. We study this model with varying levels of metabolic diversity and energy input, finding that several thermodynamic features converge across ecosystems. Specifically, irrespective of their species composition, large and metabolically diverse communities self-organize to extract roughly 10% of the maximum extractable energy, or 100 fold more than randomized communities. Moreover, distinct communities implement energy extraction in convergent ways, as indicated by strongly correlated fluxes through nutrient cycles. As the driving force from light increases, however, these features -- fluxes and total energy extraction -- become more variable across communities, indicating that energy limitation imposes tight thermodynamic constraints on collective metabolism.

q-bio.PE↗

arXiv2022

Learning to learn: Non-equilibrium design protocols for adaptable materials

Evolution in time-varying environments naturally leads to adaptable biological systems that can easily switch functionalities. Advances in the synthesis of environmentally-responsive materials therefore open up the possibility of creating a wide range of synthetic materials which can also be trained for adaptability. We consider high-dimensional inverse problems for materials where any particular functionality can be realized by numerous equivalent choices of design parameters. By periodically switching targets in a given design algorithm, we can teach a material to perform incompatible functionalities with minimal changes in design parameters. We exhibit this learning strategy for adaptability in two simulated settings: elastic networks that are designed to switch deformation modes with minimal bond changes; and heteropolymers whose folding pathway selections are controlled by a minimal set of monomer affinities. The resulting designs can reveal physical principles, such as nucleation-controlled folding, that enable adaptability.

cond-mat.soft↗

arXiv2022

Non-Convex Optimization by Hamiltonian Alternation

A major obstacle to non-convex optimization is the problem of getting stuck in local minima. We introduce a novel metaheuristic to handle this issue, creating an alternate Hamiltonian that shares minima with the original Hamiltonian only within a chosen energy range. We find that repeatedly minimizing each Hamiltonian in sequence allows an algorithm to escape local minima. This technique is particularly straightforward when the ground state energy is known, and one obtains an improvement even without this knowledge. We demonstrate this technique by using it to find the ground state for instances of a Sherrington-Kirkpatrick spin glass.

cond-mat.dis-nn↗

arXiv2022

Learning to self-fold at a bifurcation

Disordered mechanical systems can deform along a network of pathways that branch and recombine at special configurations called bifurcation points. Multiple pathways are accessible from these bifurcation points; consequently, computer-aided design algorithms have been sought to achieve a specific structure of pathways at bifurcations by rationally designing the geometry and material properties of these systems. Here, we explore an alternative physical training framework in which the topology of folding pathways in a disordered sheet is changed in a desired manner due to changes in crease stiffnesses induced by prior folding. We study the quality and robustness of such training for different `learning rules', that is, different quantitative ways in which local strain changes the local folding stiffness. We experimentally demonstrate these ideas using sheets with epoxy-filled creases whose stiffnesses change due to folding before the epoxy sets. Our work shows how specific forms of plasticity in materials enable them to learn non-linear behaviors through their prior deformation history in a robust manner.

cond-mat.soft↗

arXiv2022

Learning without neurons in physical systems

Learning is traditionally studied in biological or computational systems. The power of learning frameworks in solving hard inverse-problems provides an appealing case for the development of `physical learning' in which physical systems adopt desirable properties on their own without computational design. It was recently realized that large classes of physical systems can physically learn through local learning rules, autonomously adapting their parameters in response to observed examples of use. We review recent work in the emerging field of physical learning, describing theoretical and experimental advances in areas ranging from molecular self-assembly to flow networks and mechanical materials. Physical learning machines provide multiple practical advantages over computer designed ones, in particular by not requiring an accurate model of the system, and their ability to autonomously adapt to changing needs over time. As theoretical constructs, physical learning machines afford a novel perspective on how physical constraints modify abstract learning theory.

cond-mat.dis-nn↗

arXiv2021

Learning to Control Active Matter

The study of active matter has revealed novel non-equilibrium collective behaviors, illustrating their potential as a new materials platform. However, most works treat active matter as unregulated systems with uniform microscopic energy input, which we refer to as activity. In contrast, functionality in biological materials results from regulating and controlling activity locally over space and time, as has only recently become experimentally possible for engineered active matter. Designing functionality requires navigation of the high dimensional space of spatio-temporal activity patterns, but brute force approaches are unlikely to be successful without system-specific intuition. Here, we apply reinforcement learning to the task of inducing net transport in a specific direction for a simulated system of Vicsek-like self-propelled disks using a spotlight that increases activity locally. The resulting time-varying patterns of activity learned exploit the distinct physics of the strong and weak coupling regimes. Our work shows how reinforcement learning can reveal physically interpretable protocols for controlling collective behavior in non-equilibrium systems.

cond-mat.soft↗

arXiv2020

Proofreading through spatial gradients

Key enzymatic processes in biology use the nonequilibrium error correction mechanism called kinetic proofreading to enhance their specificity. Kinetic proofreading typically requires several dedicated structural features in the enzyme, such as a nucleotide hydrolysis site and multiple enzyme-substrate conformations that delay product formation. Such requirements limit the applicability and the adaptability of traditional proofreading schemes. Here, we explore an alternative conceptual mechanism of error correction that achieves delays between substrate binding and subsequent product formation by having these events occur at distinct physical locations. The time taken by the enzyme-substrate complex to diffuse from one location to another is leveraged to discard wrong substrates. This mechanism does not require dedicated structural elements on the enzyme, making it easier to overlook in experiments but also making proofreading tunable on the fly. We discuss how tuning the length scales of enzyme or substrate concentration gradients changes the fidelity, speed and energy dissipation, and quantify the performance limitations imposed by realistic diffusion and reaction rates in the cell. Our work broadens the applicability of kinetic proofreading, and sets the stage for the study of spatial gradients as a possible route to specificity.

q-bio.MN↗

arXiv2019

Physical constraints on epistasis

Living systems evolve one mutation at a time, but a single mutation can alter the effect of subsequent mutations. The underlying mechanistic determinants of such epistasis are unclear. Here, we demonstrate that the physical dynamics of a biological system can generically constrain epistasis. We analyze models and experimental data on proteins and regulatory networks. In each, we find that if the long-time physical dynamics is dominated by a slow, collective mode, then the dimensionality of mutational effects is reduced. Consequently, epistatic coefficients for different combinations of mutations are no longer independent, even if individually strong. Such epistasis can be summarized as resulting from a global non-linearity applied to an underlying linear trait, i.e., as global epistasis. This constraint, in turn, reduces the ruggedness of the sequence-to-function map. By providing a generic mechanistic origin for experimentally observed global epistasis, our work suggests that slow collective physical modes can make biological systems evolvable.

q-bio.PE↗

arXiv2019

Supervised learning in a mechanical system

Mechanical metamaterials are usually designed to show desired responses to prescribed forces. In some applications, the desired force-response relationship might be hard to specify exactly, although examples of forces and corresponding desired responses are easily available. Here we propose a framework for supervised learning in a thin creased sheet that learns the desired force-response behavior from training examples of spatial force patterns and can then respond correctly to previously unseen test forces. During training, we fold the sheet using different training forces and assume a learning rule that changes stiffness of creases in response to their folding strain. We find that this learning process reshapes non-linearities inherent in folding a sheet so as to show the correct response for previously unseen test forces. We study the relationship between training error, test error and sheet size which plays the role of model complexity. Our framework shows how the complex energy landscape of disordered mechanical materials can be reshaped using an iterative local learning rule.

cond-mat.soft↗

arXiv2019

Tuning environmental timescales to evolve and maintain generalists

Natural environments can present diverse challenges, but some genotypes remain fit across many environments. Such `generalists' can be hard to evolve, out-competed by specialists fitter in any particular environment. Here, inspired by the search for broadly-neutralising antibodies during B-cell affinity maturation, we demonstrate that environmental changes on an intermediate timescale can reliably evolve generalists, even when faster or slower environmental changes are unable to do so. We find that changing environments on timescales comparable to evolutionary transients in a population enhances the rate of evolving generalists from specialists, without enhancing the reverse process. The yield of generalists is further increased in more complex dynamic environments, such as a `chirp' of increasing frequency. Our work offers design principles for how non-equilibrium fitness `seascapes' can dynamically funnel populations to genotypes unobtainable in static environments.

q-bio.PE↗

arXiv2019

The self-tuned sensitivity of circadian clocks

Living organisms need to be sensitive to a changing environment while also ignoring uninformative environmental fluctuations. Here, we show that the circadian clock in \textit{Synechococcus elongatus} can naturally tune its environmental sensitivity, through a clock-metabolism coupling quantified in recent experiments. The metabolic coupling can detect mismatch between clock predictions and the day-night light cycle, and temporarily raise the clock's sensitivity to light changes and thus entrain faster. We also analyze analogous behavior in recent experiments on switching between slow and fast osmotic stress response pathways in yeast. In both cases, cells can raise their sensitivity to new external information in epochs of frequent challenging stress, much like a Kalman filter with adaptive gain in signal processing. Our work suggests a new class of experiments that probe the history-dependence of environmental sensitivity in biophysical sensing mechanisms.

q-bio.MN↗

arXiv2019

Learned multi-stability in mechanical networks

We contrast the distinct frameworks of materials design and physical learning in creating elastic networks with desired stable states. In design, the desired states are specified in advance and material parameters can be optimized on a computer with this knowledge. In learning, the material physically experiences the desired stable states in sequence, changing the material so as to stabilize each additional state. We show that while designed states are stable in networks of linear Hookean springs, sequential learning requires specific non-linear elasticity. We find that such non-linearity stabilizes states in which strain is zero in some springs and large in others, thus playing the role of Bayesian priors used in sparse statistical regression. Our model shows how specific material properties allow continuous learning of new functions through deployment of the material itself.

cond-mat.soft↗

arXiv2018

Non-equilibrium statistical mechanics of continuous attractors

Continuous attractors have been used to understand recent neuroscience experiments where persistent activity patterns encode internal representations of external attributes like head direction or spatial location. However, the conditions under which the emergent bump of neural activity in such networks can be manipulated by space and time-dependent external sensory or motor signals are not understood. Here, we find fundamental limits on how rapidly internal representations encoded along continuous attractors can be updated by an external signal. We apply these results to place cell networks to derive a velocity-dependent non-equilibrium memory capacity in neural networks.

cond-mat.dis-nn↗

arXiv2018

Temporal pattern recognition through analog molecular computation

Living cells communicate information about physiological conditions by producing signaling molecules in a specific timed manner. Different conditions can result in the same total amount of a signaling molecule, differing only in the pattern of the molecular concentration over time. Such temporally coded information can be completely invisible to even state-of-the-art molecular sensors with high chemical specificity that respond only to the total amount of the signaling molecule. Here, we demonstrate design principles for circuits with temporal specificity, that is, molecular circuits that respond to specific temporal patterns in a molecular concentration. We consider pulsatile patterns in a molecular concentration characterized by three fundamental temporal features - time period, duty fraction and number of pulses. We develop circuits that respond to each one of these features while being insensitive to the others. We demonstrate our design principles using abstract Chemical Reaction Networks and with explicit simulations of DNA strand displacement reactions. In this way, our work develops building blocks for temporal pattern recognition through molecular computation.

q-bio.MN↗

arXiv2018

Shaping dynamical folding and misfolding pathways in mechanical metamaterials

The design of desired behaviors in mechanical metamaterials has produced remarkable advances but has generally neglected two aspects - the inevitable presence of undesired behaviors and the role of dynamics in avoiding such behaviors. Inspired by similar hurdles in molecular self-assembly and protein folding, we derive design principles to shape dynamical folding and misfolding pathways in disordered mechanical systems. We show that such pathways, i.e., sequences of states traversed at a finite rate, are determined by the bifurcation structure of configuration space which, in turn, can be tuned using imperfections such as stiff joints. We apply these ideas to completely eliminate the exponentially many ways of misfolding a self-folding sheet by making some creases stiffer than others. Our approach also shows how folding at different rates can controllably target different desired behaviors.

cond-mat.soft↗

arXiv2017

Continuous attractor-based clocks are unreliable phase estimators

Statistical estimation theory determines the optimal way of estimating parameters of a fluctuating noisy signal. However, if the estimation is performed on unreliable hardware, a sub-optimal estimation procedure can outperform the previously optimal procedure. Here, we compare classes of circadian clocks by viewing them as phase estimators for the periodic day-night light signal. We find that continuous attractor-based free running clocks, such as those found in the cyanobacterium Synechococcus elongatus and humans, are nearly optimal phase estimators since their flat attractor directions efficiently project out light intensity fluctuations due to weather patterns (`external noise'). However, such flat directions also make these continuous limit cycle attractors highly vulnerable to diffusive 'internal noise'. Given such unreliable biochemical hardware, we find that point attractor-based damped clocks, such as those found in a smaller cyanobacterium with low protein copy number, Prochlorococcus marinus, outperform continuous attractor-based clocks. By interpolating between the two types of clocks found in these organisms, we demonstrate a family of biochemical phase estimation strategies that are best suited to different relative strengths of external and internal noise.

cond-mat.stat-mech↗

arXiv2017

The difficulty of folding self-folding origami

Why is it difficult to refold a previously folded sheet of paper? We show that even crease patterns with only one designed folding motion inevitably contain an exponential number of `distractor' folding branches accessible from a bifurcation at the flat state. Consequently, refolding a sheet requires finding the ground state in a glassy energy landscape with an exponential number of other attractors of higher energy, much like in models of protein folding (Levinthal's paradox) and other NP-hard satisfiability (SAT) problems. As in these problems, we find that refolding a sheet requires actuation at multiple carefully chosen creases. We show that seeding successful folding in this way can be understood in terms of sub-patterns that fold when cut out (`folding islands'). Besides providing guidelines for the placement of active hinges in origami applications, our results point to fundamental limits on the programmability of energy landscapes in sheets.

cond-mat.soft↗

arXiv2017

Associative pattern recognition through macro-molecular self-assembly

We show that macro-molecular self-assembly can recognize and classify high-dimensional patterns in the concentrations of $N$ distinct molecular species. Similar to associative neural networks, the recognition here leverages dynamical attractors to recognize and reconstruct partially corrupted patterns. Traditional parameters of pattern recognition theory, such as sparsity, fidelity, and capacity are related to physical parameters, such as nucleation barriers, interaction range, and non-equilibrium assembly forces. Notably, we find that self-assembly bears greater similarity to continuous attractor neural networks, such as place cell networks that store spatial memories, rather than discrete memory networks. This relationship suggests that features and trade-offs seen here are not tied to details of self-assembly or neural network models but are instead intrinsic to associative pattern recognition carried out through short-ranged interactions.

cond-mat.dis-nn↗