Micro-macro consistency in multiscale modeling: Score-based model assisted sampling of fast/slow dynamical systems.

A valuable step in the modeling of multiscale dynamical systems in fields such as computational chemistry, biology, and materials science is the representative sampling of the phase space over long time scales of interest; this task is not, however, without challenges. For example, the long term behavior of a system with many degrees of freedom often cannot be efficiently computationally explored by direct dynamical simulation; such systems can often become trapped in local free energy minima. In the study of physics-based multi-time-scale dynamical systems, techniques have been developed for enhancing sampling in order to accelerate exploration beyond free energy barriers. On the other hand, in the field of machine learning (ML), a generic goal of generative models is to sample from a target density, after training on empirical samples from this density. Score-based generative models (SGMs) have demonstrated state-of-the-art capabilities in generating plausible data from target training distributions. Conditional implementations of such generative models have been shown to exhibit significant parallels with long-established-and physics-based-solutions to enhanced sampling. These physics-based methods can then be enhanced through coupling with the ML generative models, complementing the strengths and mitigating the weaknesses of each technique. In this work, we show that SGMs can be used in such a coupling framework to improve sampling in multiscale dynamical systems.

Locating saddle points using gradient extremals on manifolds adaptively revealed as point clouds.

Steady states are invaluable in the study of dynamical systems. High-dimensional dynamical systems, due to separation of time scales, often evolve toward a lower dimensional manifold M. We introduce an approach to locate saddle points (and other fixed points) that utilizes gradient extremals on such a priori unknown (Riemannian) manifolds, defined by adaptively sampled point clouds, with local coordinates discovered on-the-fly through manifold learning. The technique, which efficiently biases the dynamical system along a curve (as opposed to exhaustively exploring the state space), requires knowledge of a single minimum and the ability to sample around an arbitrary point. We demonstrate the effectiveness of the technique on the Müller-Brown potential mapped onto an unknown surface (namely, a sphere). Previous work employed a similar algorithmic framework to find saddle points using Newton trajectories and gentlest ascent dynamics; we, therefore, also offer a brief comparison with these methods.

Dynamics of Fattening and Thinning 2D Sessile Droplets.

We investigate the dynamics of a droplet on a planar substrate as the droplet volume changes dynamically due to liquid being pumped in or out through a pore. We adopt a diffuse-interface formulation which is appropriately modified to account for a localized inflow-outflow boundary condition (the pore) at the bottom of the droplet, hence allowing to dynamically control its volume, as the droplet moves on a flat substrate with a periodic chemical pattern. We find that the droplet undergoes a stick-slip motion as the volume is increased (fattening droplet) which can be monitored by tracking the droplet contact points. If we then switch over to outflow conditions (thinning droplet), the droplet follows a different path (i.e., the distance of the droplet midpoint from the pore location evolves differently), giving rise to a hysteretic behavior. By means of geometrical arguments, we are able to theoretically construct the full bifurcation diagram of the droplet equilibria (positions and droplet shapes) as the droplet volume is changed, finding excellent agreement with time-dependent computations of our diffuse-interface model.

Normal form for the onset of collapse: The prototypical example of the nonlinear Schrödinger equation.

The study of nonlinear waves that collapse in finite time is a theme of universal interest, e.g., within optical, atomic, plasma physics, and nonlinear dynamics. Here we revisit the quintessential example of the nonlinear Schrödinger equation and systematically derive a normal form for the emergence of radially symmetric blowup solutions from stationary ones. While this is an extensively studied problem, such a normal form, based on the methodology of asymptotics beyond all algebraic orders, applies to both the dimension-dependent and power-law-dependent bifurcations previously studied. It yields excellent agreement with numerics in both leading and higher-order effects, it is applicable to both infinite and finite domains, and it is valid in both critical and supercritical regimes.

Multiscale analysis of collective motion and decision-making in swarms: an advection-diffusion equation with memory approach.

We propose a (time) multiscale method for the coarse-grained analysis of collective motion and decision-making in self-propelled particle models of swarms comprising a mixture of 'naïve' and 'informed' individuals. The method is based on projecting the particle configuration onto a single 'meta-particle' that consists of the elongation of the flock together with the mean group velocity and position. We find that the collective states can be associated with the transient and asymptotic transport properties of the random walk followed by the meta-particle, which we assume follows a continuous time random walk (CTRW). These properties can be accurately predicted at the macroscopic level by an advection-diffusion equation with memory (ADEM) whose parameters are obtained from a mean group velocity time series obtained from a single simulation run of the individual-based model.

Spatiotemporal addressing of surface activity.

We have modified surface catalytic activity in real time and space by focusing an addressable laser beam to differentially heat a platinum (110) single-crystal surface. Ellipsomicroscopy imaging of local conditions (such as reactant and product local coverages) enabled us to close the loop between sensing and actuation (both spatiotemporally resolved). Pulses and fronts, the basic building blocks of patterns, could be formed, accelerated, modified, guided, and destroyed at will. Real-time image processing and feedback allow the design and implementation of new classes of nonlocal evolution rules.

Effects of Boundaries on Pattern Formation: Catalytic Oxidation of CO on Platinum.

The effect of boundaries on pattern formation was studied for the catalytic oxidation of carbon monoxide on platinum surfaces. Photolithography was used to create microscopic reacting domains on polycrystalline foils and single-crystal platinum (110) surfaces with inert titanium overlayers. Certain domain geometries give rise to patterns that have not been observed on the untreated catalyst and bring to light surface mechanisms that have no analog in homogeneous reaction-diffusion systems.

Enhancement of surface activity in CO oxidation on Pt(110) through spatiotemporal laser actuation.

We explore the effect of spatiotemporally varying substrate temperature profiles on the dynamics and the resulting reaction rate enhancement of the catalytic oxidation of CO on Pt(110). The catalytic surface is "addressed" by a focused laser beam whose motion is computer controlled. The averaged reaction rate is observed to undergo a characteristic maximum as a function of the speed of this moving laser spot. Experiments as well as modeling are used to explore and rationalize the existence of such an optimal laser speed.

Mass transport of alkali metal with pulses in a surface reaction.

It is shown that the pulses which develop in the NO+H2 reaction on an alkali promoted Rh(110) surface reaction can transport alkali metal. This leads to the accumulation of a substantial alkali-metal concentration in the collision area of pulse trains. Realistic simulations revealed that the effect is due to the strong energetic interactions of the alkali metal with coadsorbates, i.e., the attractive interaction with coadsorbed oxygen and the effectively repulsive interaction with coadsorbed nitrogen.

Coarse-grained computations of demixing in dense gas-fluidized beds.

We use an "equation-free," coarse-grained computational approach to accelerate molecular dynamics-based computations of demixing (segregation) of dissimilar particles subject to an upward gas flow (gas-fluidized beds). We explore the coarse-grained dynamics of these phenomena in gently fluidized beds of solid mixtures of different densities, typically a slow process for which reasonable continuum models are currently unavailable.

Equation-free dynamic renormalization in a glassy compaction model.

Combining dynamic renormalization with equation-free computational tools, we study the apparently asymptotically self-similar evolution of void distribution dynamics in the diffusion-deposition problem proposed by Stinchcombe and Depken [Phys. Rev. Lett. 88, 125701 (2002)]. We illustrate fixed point and dynamic approaches, forward as well as backward in time; these can be used to accelerate simulators of glassy dynamic phenomena.

Geometry-induced pulse instability in microdesigned catalysts: the effect of boundary curvature.

We explore the effect of boundary curvature on the instability of reactive pulses in the catalytic oxidation of on microdesigned Pt catalysts. Using ring-shaped domains of various radii, we find that the pulses disappear (decollate from the inert boundary) at a turning point bifurcation, and we trace this boundary in both physical and geometrical parameter space. These computations corroborate experimental observations of pulse decollation.

Guiding chemical pulses through geometry: Y junctions.

We study computationally and experimentally the propagation of chemical pulses in complex geometries. The reaction of interest, CO oxidation, takes place on single crystal Pt(110) surfaces that are microlithographically patterned; they are also addressable through a focused laser beam, manipulated through galvanometer mirrors, capable of locally altering the crystal temperature and thus affecting pulse propagation. We focus on sudden changes in the domain shape (corners in a Y-junction geometry) that can affect the pulse dynamics; we also show how brief, localized temperature perturbations can be used to control reactive pulse propagation. The computational results are corroborated through experimental studies in which the pulses are visualized using reflection anisotropy microscopy.

Nonlinear resonances and antiresonances of a forced sonic vacuum.

We consider a harmonically driven acoustic medium in the form of a (finite length) highly nonlinear granular crystal with an amplitude- and frequency-dependent boundary drive. Despite the absence of a linear spectrum in the system, we identify resonant periodic propagation whereby the crystal responds at integer multiples of the drive period and observe that this can lead to local maxima of transmitted force at its fixed boundary. In addition, we identify and discuss minima of the transmitted force ("antiresonances") between these resonances. Representative one-parameter complex bifurcation diagrams involve period doublings and Neimark-Sacker bifurcations as well as multiple isolas (e.g., of period-3, -4, or -5 solutions entrained by the forcing). We combine them in a more detailed, two-parameter bifurcation diagram describing the stability of such responses to both frequency and amplitude variations of the drive. This picture supports a notion of a (purely) "nonlinear spectrum" in a system which allows no sound wave propagation (due to zero sound speed: the so-called sonic vacuum). We rationalize this behavior in terms of purely nonlinear building blocks: apparent traveling and standing nonlinear waves.

Heterogeneous versus discrete mapping problem.

We propose a method for mapping a spatially discrete problem, stemming from the spatial discretization of a parabolic or hyperbolic partial differential equation of gradient type, to a heterogeneous one with certain comparable dynamical features pertaining, in particular, to coherent structures. We focus the analysis on a (1+1)-dimensional phi(4) model and confirm the theoretical predictions numerically. We also discuss possible generalizations of the method and the ensuing qualitative analogies between heterogeneous and discrete systems and their dynamics.

Pulse bifurcations and instabilities in an excitable medium: computations in finite ring domains.

We investigate the instabilities and bifurcations of traveling pulses in a model excitable medium; in particular, we discuss three different scenarios involving either the loss of stability or disappearance of stable pulses. In numerical simulations beyond the instabilities we observe replication of pulses ("backfiring") resulting in complex periodic or spatiotemporally chaotic dynamics as well as modulated traveling pulses. We approximate the linear stability of traveling pulses through computations in a finite albeit large domain with periodic boundary conditions. The critical eigenmodes at the onset of the instabilities are related to the resulting spatiotemporal dynamics and "act" upon the back of the pulses. The first scenario has been analyzed earlier [M. G. Zimmermann et al., Physica D 110, 92 (1997)] for high excitability (low excitation threshold): it involves the collision of a stable pulse branch with an unstable pulse branch in a so-called T point. In the framework of traveling wave ordinary differential equations, pulses correspond to homoclinic orbits and the T point to a double heteroclinic loop. We investigate this transition for a pulse in a domain with finite length and periodic boundary conditions. Numerical evidence of the proximity of the infinite-domain T point in this setup appears in the form of two saddle node bifurcations. Alternatively, for intermediate excitation threshold, an entire cascade of saddle nodes causing a "spiraling" of the pulse branch appears near the parameter values corresponding to the infinite-domain T point. Backfiring appears at the first saddle-node bifurcation, which limits the existence region of stable pulses. The third case found in the model for large excitation threshold is an oscillatory instability giving rise to "breathing," traveling pulses that periodically vary in width and speed.

Continuum approach to discreteness.

We study analytically and numerically continuum models derived on the basis of Padé approximations and their effectiveness in modeling spatially discrete systems. We not only analyze features of the temporal dynamics that can be captured through these continuum approaches (e.g., shape oscillations, radiation effects, and trapping) but also point out ones that cannot be captured (such as Peierls-Nabarro barriers and Bloch oscillations). We analyze the role of such methods in providing an effective "homogenization" of spatially discrete, as well as of heterogeneous continuum equations. Finally, we develop numerical methods for solving such equations and use them to establish the range of validity of these continuum approximations, as well as to compare them with other semicontinuum approximations.

Coupling fields and underlying space curvature: an augmented Lagrangian approach.

We demonstrate a systematic implementation of coupling between a scalar field and the geometry of the space which carries the field. This naturally gives rise to a feedback mechanism between the field and the geometry. We develop a systematic model for the feedback in a general form, inspired by a specific implementation in the context of molecular dynamics (the so-called Rahman-Parrinello molecular dynamics, or RP-MD). We use a generalized Lagrangian that allows for the coupling of the space's metric tensor to the scalar field, and add terms motivated by RP-MD. We present two implementations of the scheme: one in which the metric is only time-dependent (which gives rise to an ordinary differential equation for its temporal evolution), and the other with spatiotemporal dependence (wherein the metric's evolution is governed by a partial differential equation). Numerical results are reported for the (1+1)-dimensional model with a nonlinearity of the sine-Gordon type.

Microbial predation in a periodically operated chemostat: a global study of the interaction between natural and externally imposed frequencies.

Predator-prey systems in continuously operated chemostats exhibit sustained oscillations over a wide range of operating conditions. When the chemostat is operated periodically, the interaction of the natural oscillation frequency with the external forcing gives rise to a wealth of dynamic behavior patterns. Using numerical bifurcation techniques, we perform a detailed computational study of these patterns and the transitions (local and especially global) between them as the amplitude and frequency of the forcing vary. The transition from low-forcing-amplitude quasiperiodicity to entrainment of the chemostat behavior by strong forcing (involving the concerted closing of resonance horns) is analyzed. We concentrate on certain strong resonance phenomena between the two frequencies and provide an extensive atlas of computed phase portraits for our model system. Our observations corroborate recent mathematical results and case studies of periodically forced chemical oscillators. In particular, the existence and relative succession of several distinct types of global bifurcations resulting in chaotic transients and multistability are studied in detail. The location in the operating diagram of several key codimension 2 local bifurcations of periodic solutions is computed, and their interaction with an interesting feature we name "real-eigenvalues horns" is examined.

Microbial predation in coupled chemostats: a global study of two coupled nonlinear oscillators.

Predator-prey systems in continuously operated chemostats exhibit sustained oscillations over a wide range of operating conditions. When two such chemostats interact through flow exchange, the interplay of the oscillation frequencies gives rise to a wealth of dynamic behavior patterns. Using numerical bifurcation techniques, we perform a detailed computational study of these patterns and the transitions between them as the coupling strength and relative frequencies of the two chemostats vary. We concentrate on certain strong resonance phenomena between the two frequencies as well as their mutual extinction and provide a representative sampling of possible phase portraits for our model system. Our observations corroborate recent mathematical results and case studies of coupled nonlinear chemical oscillators in which regions of mutual extinction as well as the Arnol'd structure for two-parameter families of maps of the plane have been observed. We highlight certain unexpected features of the operating diagram discovered through our computational study and discuss their implication for the dynamic response of the chemostat system.