Delay match to sample experiments have inspired much of the modeling work on attractor neural networks. The basic experiment involves showing a target image, removing it, and after a delay showing a cue image: either the original image or a different one. The subject needs to indicate if the cue is the same or different than the target. Electrophysiological recordings have shown that if the target is a learned one (has been observed multiple times) neurons selective for it maintain activity during the delay between target and cue presentation. This persistent activity is hypothesized to represent `working’ or `short term’ memory. The attractor model posits that learning creates modifications in the synaptic connections such that stimulation with learned patterns leads to sustained activity correlated with these learned patterns. There are a number of variations on the basic DMS paradigm involving distractors in between target and cue, or repetition detection experiments where a sequence of images is shown and one of them chosen at random is repeated. I will present a parsimonious network model of binary neurons and binary synapses and show how many of the phenomena observed in these different experiments can be handled within this framework using simple adjustments of certain global parameters.

Stemming from visualization studies in superfluid helium, I'll review some basic phenomenology of quantized vortices, reconnection, and Kelvin waves. Some observations about the untangling of vortices lead to predictions regarding the helicity, and some puzzles and questions about the role of invariants like the helicity in the Gross-Pitaevskii (nonlinear Schrodinger) equation.

With the rise of online social media, scientists can communicate their work to the public in ways that were unimaginable fifteen years ago. In this talk, I will discuss my outreach efforts through FYFD, a fluid dynamics blog and YouTube channel with an audience of around a quarter of a million followers. The talk will also present recent work with Harvard University undergraduates to integrate science communication into their fluid dynamics curriculum, and how these undergraduate projects kicked off a joint research effort with Harvard to investigate the physics of the Boston Molasses Flood, an industrial accident from 1919 that flooded Boston’s North End neighborhood with nearly 9000 cubic meters of molasses.

The metal-insulator transition driven by strong electronic correlations – generically called the “Mott” transition – is usually described entirely by electronic Hamiltonians, with models designed to exhibit related emergent phenomena such as magnetism and superconductivity. In real solids, the electronic localization also couples to the crystal lattice, and it turns out that these elastic degrees of freedom insert important new entropic phenomena more familiar in soft matter physics.

The coupling to the lattice induces elastic strain fields, which have intrinsic long-range interactions that cannot be screened. When strain fields are produced as a secondary order parameter in phase transitions - as for example in ferroelectrics - this produces unexpected consequences for the dynamics of order parameter fluctuations, including the generation of a gap in what would otherwise have been expected to be Goldstone modes.

A very important class of transition metal oxides – the perovskites – can be thought of as an array of tethered octahedra where the Mott transition produces a shape-change in the unit cell. Coupling of the fundamental order parameter to octahedral rotations gives rise to large entropic effects that can shift the transition temperature by hundreds of degrees K , essentially by exploiting the physics of jammed solids. The insight might offer ways to make better refrigerators by enhancing electro-caloric and magneto-electric effects. I will also speculate on how this might be relevant for theories of the quantum critical point.

I will overview the Dark Energy Survey (DES) project, highlight its early science results, and discuss its on-going activities and plans. The DES collaboration built the 570-megapixel Dark Energy Camera for the Blanco 4-meter telescope at Cerro Tololo Inter-American Observatory in Chile to carry out a 5-year, deep, multi-band, optical survey over one eighth of the sky and a time-domain survey that will discover several thousand supernovae. The survey started in Aug. 2013 and is now nearing completion of its fourth observing season. DES was designed to address the questions: why is the expansion of the Universe speeding up? Is cosmic acceleration due to dark energy or does it require a modification of General Relativity? If dark energy, is it the energy density of the vacuum (Einstein's cosmological constant) or something else? DES is addressing these questions by measuring the history of cosmic expansion and the growth of structure through four complementary techniques: galaxy clusters, the large-scale galaxy distribution, gravitational lensing, and supernovae, as well as through cross-correlation with other data sets. I will also discuss how the data are being used to make a variety of other astronomical discoveries, from our Solar System to the most distant quasars.

An inclusion of particles in a Newtonian liquid can fundamentally change the interfacial dynamics and even cause interfacial instabilities. For instance, viscous fingering can arise even in the absence of the destabilizing viscosity ratio between invading and defending phases, when particles are added to the viscous invading fluid inside a Hele-Shaw cell. Our experimental results demonstrate that the onset and characteristics of fingering are most directly affected by the particle volume fraction but also depend on the ratio of the particle to gap size. In particular, the formation and destabilization of a particle band are observed on the interface only when the particle diameter is comparable to the channel gap thickness. The physical mechanism behind the instability and a quantitative model will also be discussed.

Buckling of slender structures is typically regarded as a first step towards failure that is to be avoided ('Buckliphobia'). Instead, we envision mechanical instabilities in soft structures as opportunities for scalable, reversible, and robust mechanisms that are first to be predictively understood, and then harvested for function ('Buckliphilia'). A series of examples with a focus on thin elastic shells will be provided. I will first revisit the canonical Mechanics problem of sensitivity of shell-buckling to geometric imperfections. I shall then move on to the post-buckling regime of shells where periodic dimpled patterns are observed; for the cases when i) the shell is constrained from within by a rigid mandrel or ii) bound to an equally curved soft substrate (curved wrinkling). This periodic dimpled patterns will be used as a model system to study fundamental properties of curved surface crystals. Finally, taking inspiration from the resemblance of our dimpled wrinkling patterns to golf balls, I will introduce a new class of smart morphable surfaces for switchable and tunable aerodynamic drag.

Turbulent flows of incompressible fluid in two dimensions are comprised of dense systems of vortices. In 1949 Onsager suggested to treat vortices as a macroscopical system whose statistical properties are described by Gibbsian statistical ensemble [1]. In the talk I address hydrodynamics of the vortex fluid. The hydrodynamics of the vortex fluid is different from Euler hydrodynamics of the original fluid. It features the anomalous stress absent in Euler's hydrodynamics, which yields a number of interesting effects. Some of them are: a deflection of stream lines, a correction to the Bernoulli law, accumulation of vortices in regions with high curvature in the curved space [1] L. Onsager, Nuovo Cimento, Suppl. 6, 249, 279 (1949)

Soft materials are rapidly changing our expectations of what machines can do, but mechanically activating these systems remains challenging. We investigate a new physical effect that serves as a powerful strategy to inject mechanical energy into hydrogels, a widely-utilized class of soft materials. By dropping hydrogel spheres onto a hot substrate, we achieve vigorous energy injection in the form of persistent bouncing and intense screeching. This effect is robust, with spheres bouncing several times their height for minutes at a time. The underlying mechanism arises from a synergistic combination of elasticity and the well-known Leidenfrost effect: vaporization couples with gel deformations to create rapid pressure oscillations that do mechanical work. With the fuel, mechanism, and mechanical output embedded into a single object made from a single material, our results introduce the concept of a soft engine and promise practical ramifications in fields such as active matter, metamaterials, and soft robotics.

During development, tissue maintenance, and in diseases, cells proliferate and transition between physiologically and functionally distinct states. Despite the centrality of these transitions for diverse biological functions, it has remained challenging to determine which transitions can occur and at what rates without perturbations or cell engineering. I will discuss how quantitative cell state transition dynamics can be inferred from a static snapshot of gene expression in individual cells combined with their lineage history. We have been using synthetic biology, single-cell time-lapse microscopy, and single-molecule imaging to apply this framework to determine the dynamics of embryonic stem cells in culture. I will discuss these efforts and some counterintuitive, but beautiful, mathematical structures that have helped us interpret our experimental observations in stem cells and other biological systems.

Could we significantly reduce U.S. unemployment by helping job seekers move closer to jobs? Using data from the leading employment board CareerBuilder.com, we show that, indeed, workers dislike applying to distant jobs: job seekers are 35% less likely to apply to a job 10 miles away from their ZIP code of residence. However, because job seekers are close enough to vacancies on average, this distaste for distance is fairly inconsequential: our search and matching model predicts that relocating job seekers to minimize unemployment would decrease unemployment by only 5.3%. Geographic mismatch is thus a minor driver of aggregate unemployment. We will discuss how we have measured mismatch more broadly, including the skills dimension. Finally, we will discuss the National Center for Opportunity Engineering and Analysis, and how we may build a recommendation engine linking education, skills and career opportunities.

Thermal soaring by birds and olfactory searches by insects are biological examples of navigation in the presence of complex orientation cues. The two problems also have technological applications, namely for extending the autonomy of flying vehicles/gliders, as well as for the development of olfactory robots (sniffers). I shall first review the animal behavior, then present the physics of the orientation cues, and finally discuss the computational aspects of navigation.

I will present recent work realizing topological phases of photons, both in curved space, and in lattices. The talk will focus on our recent exploration of Landau levels on a conical surface, generated using a non-planar (twisted) optical resonator to induce a synthetic magnetic field for optical photons, and employed to validate the famous Wen-Zee action. I will then discuss recent results demonstrating strong photon-photon interactions mediated by resonator Rydberg-electromagnetically induced transparency (EIT), and techniques we are developing to assemble topological few-body states both photon-by-photon, and through microscopic devices engineered for photon thermalization. I will conclude with our recent observation of time-resolved helical edge dynamics in Z_2 topological circuit lattices, and a T-broken extension in the microwave domain using arrays of 3D cavities and circuit quantum electrodynamics techniques. This work showcases the unique possibilities for Hamiltonian engineering and control in the photonic sector, a provides a taste of upcoming breakthroughs in engineering quantum materials from photons.

Flexible slender structures in flow are everywhere. While a great deal is known about individual flexible fibers interacting with fluids, considerably less work has been done on fiber ensembles, such as fur or hair, in flow. These hairy surfaces are abundant in nature and perform multiple functions from thermal regulation to water harvesting to sensing. Motivated by these biological systems, we consider two examples of hairy surfaces interacting with flow: (1) air entrainment in the fur of diving mammals and (2) symmetry breaking in hairy micro-channels.

In the first example, we take inspiration from semi-aquatic mammals (such as fur seals, otters, and beavers) which have specially adapted fur that serves as an effective insulator both above and below water. Many of these animals have evolved pelts that naturally entrap air when they dive. This air: (1) provides additional insulation under water, (2) provides added buoyancy, and (3) facilitates water shedding when the animals resurface. In this study we investigate diving conditions and fur properties which amplify air entrainment in fur. In the second example, we consider a fundamental component in hydraulic systems, the flow rectifier. One of the simplest ways to generate asymmetry in these devices is with a ball valve in which flow is completely obstructed in one direction and free to flow in the other. In this work we seek a variation that: (1) allows the designer to modulate the relative resistances in the rectifier and (2) can be achieved with solid state components (i.e. no moving parts).

Bayesian inference and statistical physics are formally closely related. Therefore methodology and concepts developed in statistical physics to understand disordered materials such as glasses and spin glasses can be elevated to analyze models of in statistical inference. We will present this approach in a rather general setting that covers analysis of compressed sensing, generalized linear regression, and the perceptron - a kind of a single layer neural network. At the one hand, this approach leads to the approximate message passing algorithm that is gaining its place among other widely used regression and classification algorithms. At the other hand, the related analyses leads to identification of phase transitions in the performance of Bayes-optimal estimators. We will discuss relation between these phase transitions and algorithmic hardness, and in the case of compressed sensing we will show how this understanding leads to a design of optimal measurement protocols.

Based partly on "Statistical-physics-based reconstruction in compressed sensing" PRX 2012 and reviewed in "Statistical physics of inference: Thresholds and algorithms" Advances of Physics 2016.

Astrophysicists use standard candles, objects which have roughly the same luminosity, to infer distances to far-away parts of the universe. Standard candles of variable stars called ‘cepheids’ were used to discover the expanding universe, and standard candles of exploding stars called ‘supernovae’ were used to discover the accelerating universe. Together, these two standard candles can be used to measure the size of the universe. Interestingly, this measurement of the size of the universe recovered conflicts with measurements of the size of the universe from extrapolations of data from the Cosmic Microwave Background. I will go over how we make our measurement, from soup to nuts, and discuss how we can be confident in the accuracy of our values. I will then discuss different ways too explain the tension we see in the different sets of measurements, and possible new physics that may be on the horizon.

One often hears that human vision is “sensitive to single photons,” when in fact the faintest flash of light that can reliably be reported by human subjects is closer to 100 photons. Nevertheless, there is a sense in which the familiar claim is true. Experiments conducted long after the seminal work of Hecht, Shlaer, and Pirenne now allow a more precise, and in some ways even more remarkable, conclusion to be drawn about our visual apparatus. A simple model that incorporates both old news (response of single rod cells) and newer news (loss at the first synapse) can account in detail for both old and new psychophysical data.

The equilibrium steady state of an undriven mixture of reacting chemical species is uniquely determined by the free energy. Once external environmental drives are introduced, however, steady-state concentrations may deviate from these equilibrium values via sustained absorption and dissipation of work. From a physical standpoint, the living cell is a particularly intriguing example of such a nonequilibrium system, because the environmental work sources that power it are relatively difficult to access – only the proper orchestration of many distinct catalytic actors leads to a collective behavior that is competent to harvest and exploit available metabolites. Here, we study the dynamics of an in silico chemical network with random connectivity in a driven environment that only makes strong chemical forcing available to rare combinations of concentrations of different molecular species. We find that the long-time dynamics of such systems biased towards the spontaneous extremization of forcing, so that the molecular composition converges on states that exhibit exquisite fine-tuning to available work sources.

The dynamic bond can be defined as any class of bond that selectively undergoes reversible breaking and reformation, usually under equilibrium conditions. The incorporation of dynamic bonds (which can be either covalent or non-covalent) allows access to structurally dynamic polymers. Such polymers can exhibit macroscopic responses upon exposure to an environmental stimulus, on account of a rearrangement of the polymeric architecture. In such systems, the nature of the dynamic bond not only dictates which stimulus the material will be responsive to but also plays a role in the response itself. Thus, such a design concept represents a molecular level approach to the development of new stimuli-responsive materials. We have been interested in the potential of such systems to access new material platforms and have developed a range of new mechanically stable, structurally dynamic polymer films that change their properties in response to a given stimulus, such as temperature, light or specific chemicals. Such adaptive materials have been targeted toward applications that include healable plastics, responsive liquid crystalline polymers, chemical sensors, thermally responsive hydrogels, shape memory materials and mechanically dynamic biomedical implants. Our latest results in this area will be discussed.

I will describe the observation of slow relaxations, aging and memory effects - hallmarks of glassy dynamics – in the mechanical response of two disordered systems: thin sheets crumpled into a ball and elastic foams. In particular, I’ll report the observation of a non-monotonic aging response that can last many hours. I will then describe ongoing experiments that exploit the macroscopic nature of these systems to try and uncover the underlying mechanisms. The experimental results are in good agreement with a phenomenological framework recently used to describe observations of monotonic aging in several glassy systems. This suggests not only a general mechanism, but also that the non-monotonic behavior may be generic and that a-thermal, macroscopic systems can exhibit glassy behaviors.

Diverse multi-cellular animals encode a breathtaking diversity of natural behaviors. Non local interactions in traditional nervous systems make the study of underlying origins of behavior in animals difficult (and fascinating). It is a well known fact that simple dynamical systems can also encode perplexing complexity with purely local update rules. In this talk, using a variety of toy models and systems, we will explore how complex behavior can arise in non-neuronal ensembles; or in short "how do animals with no brains (neurons), decide, compute or think?

How do fluids become turbulent as their flow velocity is increased? During the last ten years, exquisite experiments, numerical simulations and pure theory have uncovered a remarkable series of connections between transitional turbulence, phase transitions and renormalization group theory, high energy hadron scattering, the statistics of extreme events, and even population biology. In this talk, I will outline how these developments and strange connections imply that a fluid at the boundary between turbulence and laminar flow behaves precisely like an ecosystem at the verge of extinction, a prediction that is supported by recent experiments.

There is considerable interest in controlling the assembly of polymeric material in order to create highly ordered materials for applications. Such materials are often trapped in metastable, non-equilibrium states, and the processes through which they assemble become an important aspect of the materials design strategy. An example is provided by di-block copolymer directed self-assembly, where a decade of work has shown that, through careful choice of process variables, it is possible to create ordered structures whose degree of perfection meets the constraints of commercial semiconductor manufacturing. As impactful as that work has been, it has focused on relatively simple materials – neutral polymers, consisting of two or at most three blocks. Furthermore, the samples that have been produced have been limited to relatively thin films, and the assembly has been carried out on ideal, two-dimensional substrates. The question that arises now is whether one can translate those achievements to polymeric materials having a richer sequence, to monomers that include charges, to three-dimensional substrates, or to active systems that are in a permanent non-equilibrium state. Building on discoveries from the biophysics literature, this presentation will review recent work from our group and others that explains how nature has evolved to (1) direct the assembly of nucleic acids into intricate, fully three-dimensional macroscopic functional materials that are not only active, but also responsive to external cues, and (2) to direct the assembly of polymeric actin and tubulin filaments into liquid crystalline phases, where the interplay between elasticity and activity can be used to manipulate the dynamics of the system. The results presented in this talk will then be used to discuss how one might design a new generation of synthetic active systems capable of performing specific, engineered functions.

Hyperuniformity of a many-body system means anomalously small density fluctuations. In an ordinary liquid or solution the mean-squared fluctuations E(N - E(N))^2 of particle number N in a given region are proportional to the average particle number E(N). By contrast, in a hyperuniform system E(N - E(N))^2/E(N) goes to 0 as N goes to infinity. The classic example is the one-component plasma, a gas of charged particles with a uniform neutralizing background charge. Recently granular materials[1] and periodically sheared colloidal dispersions[2] were shown to be hyperuniform. We investigate the question of uniformity in a dilute colloidal dispersion in which particles are settling under gravity. When the particles are of generic shapes, their asymmetry strongly affects their interaction via the coupling between orientation, drag and fluid velocity gradients. Two isolated objects generally separate over time. In one regime hyperuniform and anisotropic behavior of the density correlation function probed by light scattering is inevitable.

[1] S.Torquato, Rev. Mod. Phys, vol.82, 2633-2640 (2010).

[2] D. Hexner and D. Levine, Phys. Rev. Lett., vol.114, 110602 (2015).

Two thousand years ago, the mass-manufacturing of paper simplified all aspects of information technology: generation, processing, communication, delivery and storage. Similarly powerful changes have been seen in the last century through the development of integrated circuits based on silicon. Monolayers of 2D crystals provide an ideal material platform for realizing these integrated circuits thin and free-standing, which were the key advantages of paper over other medium two thousand years ago. Once realized, these atomically thin circuits will be foldable and actuatable, which will further increase the device density and functionality, allowing them to be used tether-free (or wirelessly) in environments not previously accessible to conventional circuits, such as water, air or in space. In this talk, we will discuss our recent progresses toward building atomically-thin integrated circuits using wafer-scale 2D crystals. In order for this, we developed a series of approaches that are scalable, precise, and modular. We developed wafer-scale synthesis of three atom thick semiconductors, reported a wafer-scale patterning method for one-atom-thick lateral heterojunctions, and showed how atomically thin films and devices can be vertically stacked to form more complicated 3D circuitry. Then we will discuss our most recent efforts to turn these 2D circuits into 3D structures.

From automatic speech recognition to discovering unusual stars, underlying almost all automated discovery tasks is the ability to compare and contrast data streams with each other, to identify connections and spot outliers. Despite the prevalence of data, however, automated methods are not keeping pace. A key bottleneck is that most data comparison algorithms today either rely on a human expert to specify what ‘features’ of the data are relevant for comparison, or require copious amounts of data for machine learning. Data Smashing is a new principle for estimating the similarity between the sources of arbitrary data streams, using neither domain knowledge nor learning. We demonstrate the application of this principle to the analysis of data from a number of real-world challenging problems, including the disambiguation of electro-encephalograph patterns pertaining to epileptic seizures, detection of anomalous cardiac activity from heart sound recordings and classification of astronomical objects from raw photometry. In all these cases and without access to any domain knowledge, performance is on a par with the accuracy achieved by specialized algorithms and heuristics devised by domain experts. Work done with Ishanu Chattopadhyay.

Time-dependent and process-dependent properties of soft jammed solids like gelled networks, compressed emulsions or colloidal glasses, stem from stress heterogeneities frozen-in during solidification and their coupling with an imposed deformation. I’ll discuss recent novel insights gained through numerical simulations of statistical microscopic models that suggest how to control the persistence of flow inhomogeneities upon yielding, and provide new cues to design softening, hardening and brittleness in soft solids.

Arguably, science' goal of understanding nature can be formulated as inferring mathematical laws that govern natural systems from experimental data. With the fast growth of power of modern computers and of artificial intelligence algorithms, there has been a recent surge in attempts to automate this goal and to design, to some extent, an “artificial scientist.” I will discuss this emerging field, but will focus primarily on our own approach to it. I will introduce an algorithm that we have recently developed, which allows one to infer the underlying dynamical equations behind a noisy time series, even if the dynamics are nonlinear, and only a few of the relevant variables are measured. I will illustrate the method on applications to toy problems, including inferring the iconic Newton’s law of universal gravitation, as well as a few biochemical reaction networks. I will end with applications to experimental biological data: modeling the landscape of possible behavioral states underlying reflexive escape from pain in a roundworm and (if time permits) modeling insulin secretion in pancreatic beta cells.