arXiv subjects

Nicholas T. Ouellette

Publications and source records attributed to Nicholas T. Ouellette.

At least 19 recordsLinked to original sources

arXiv2026

Effect of sub-critical fluid shear flow on granular bed strength

Interactions between fluids and granular materials are prevalent on the Earth's surface. In the case of fluid flow over a sediment bed, the fluid imparts a shear stress to the granular materials. When the applied shear stress is above a critical value, the grains become entrained in the fluid flow. Prior experimental studies have shown that granular beds subjected to a sub-critical fluid flow can strengthen in the same direction as the sub-critical flow. In contrast, granular beds can become weaker in the direction opposite to the sub-critical fluid flow. To investigate the grain-scale mechanisms that control directional strengthening and weakening, we perform discrete element method (DEM) simulations of granular beds subjected to model fluid flows in two (2D) and three (3D) dimensions with varied inter-particle static friction coefficients and conditioning flow speeds. In these studies, the sub-critical grain motion does not cause significant bed compaction. Instead, we find that the strength of a granular bed in a particular direction is highly correlated with the fraction of {\it surface} grains that can be dislodged by a fluid force applied in that direction. Further, the anisotropic bed strength only persists over a finite time scale that is set by the Shields number. We also show that inter-particle static friction is not required for bed strength anisotropy, but varying the friction affects the magnitude of the anisotropy. This research enhances the grain-scale understanding of erosion of granular beds caused by fluid flows and underscores the importance of tracking the history of the fabric of the bed surface since it couples strongly to bed strength.

cond-mat.soft↗

arXiv2025

Towards mechanistic understanding in a data-driven weather model: internal activations reveal interpretable physical features

Large data-driven physics models like DeepMind's weather model GraphCast have empirically succeeded in parameterizing time operators for complex dynamical systems with an accuracy reaching or in some cases exceeding that of traditional physics-based solvers. Unfortunately, how these data-driven models perform computations is largely unknown and whether their internal representations are interpretable or physically consistent is an open question. Here, we adapt tools from interpretability research in Large Language Models to analyze intermediate computational layers in GraphCast, leveraging sparse autoencoders to discover interpretable features in the neuron space of the model. We uncover distinct features on a wide range of length and time scales that correspond to tropical cyclones, atmospheric rivers, diurnal and seasonal behavior, large-scale precipitation patterns, specific geographical coding, and sea-ice extent, among others. We further demonstrate how the precise abstraction of these features can be probed via interventions on the prediction steps of the model. As a case study, we sparsely modify a feature corresponding to tropical cyclones in GraphCast and observe interpretable and physically consistent modifications to evolving hurricanes. Such methods offer a window into the black-box behavior of data-driven physics models and are a step towards realizing their potential as trustworthy predictors and scientifically valuable tools for discovery.

physics.ao-ph↗

arXiv2020

Shear response of granular packings compressed above jamming onset

We investigate the mechanical response of jammed packings of repulsive, frictionless spherical particles undergoing isotropic compression. Prior simulations of the soft-particle model, where the repulsive interactions scale as a power-law in the interparticle overlap with exponent $α$, have found that the ensemble-averaged shear modulus $\langle G \rangle$ increases with pressure $P$ as $\sim P^{(α-3/2)/(α-1)}$ at large pressures. However, a deep theoretical understanding of this scaling behavior is lacking. We show that the shear modulus of jammed packings of frictionless, spherical particles has two key contributions: 1) continuous variations as a function of pressure along geometrical families, for which the interparticle contact network does not change, and 2) discontinuous jumps during compression that arise from changes in the contact network. We show that the shear modulus of the first geometrical family for jammed packings can be collapsed onto a master curve: $G^{(1)}/G_0 = (P/P_0)^{(α-2)/(α-1)} - P/P_0$, where $P_0 \sim N^{-2(α-1)}$ is a characteristic pressure that separates the two power-law scaling regions and $G_0 \sim N^{-2(α-3/2)}$. Deviations from this form can occur when there is significant non-affine particle motion near changes in the contact network. We further show that $\langle G (P)\rangle$ is not simply a sum of two power-laws, but $\langle G \rangle \sim (P/P_c)^a$, where $a \approx (α-2)/(α-1)$ in the $P \rightarrow 0$ limit and $\langle G \rangle \sim (P/P_c)^b$, where $b \gtrsim (α-3/2)/(α-1)$ above a characteristic pressure $P_c$. In addition, the magnitudes of both contributions to $\langle G\rangle$ from geometrical families and changes in the contact network remain comparable in the large-system limit for $P >P_c$.

cond-mat.soft↗

arXiv2020

Pair Formation in Insect Swarms Driven by Adaptive Long-Range Interactions

In swarms of flying insects, the motions of individuals are largely uncoordinated with those of their neighbors, unlike the highly ordered motion of bird flocks. However, it has been observed that insects may transiently form pairs with synchronized relative motion while moving through the swarm. The origin of this phenomenon remains an open question. In particular, it is not known if pairing is a new behavioral process or whether it is a natural byproduct of typical swarming behavior. Here, using an "adaptive-gravity" model that proposes that insects interact via long-range gravity-like acoustic attractions that are modulated by the total background sound (via "adaptivity" or fold-change detection) and that reproduces measured features of real swarms, we show that pair formation can indeed occur without the introduction of additional behavioral rules. In the model, pairs form robustly whenever two insects happen to move together from the center of the swarm (where the background sound is high) toward the swarm periphery (where the background sound is low). Due to adaptivity, the attraction between the pair increases as the background sound decreases, thereby forming a bound state since their relative kinetic energy is smaller than their pair-potential energy. When the pair moves into regions of high background sound, however, the process is reversed and the pair may break up. Our results suggest that pairing should appear generally in biological systems with long-range attraction and adaptive sensing, such as during chemotaxis-driven cellular swarming.

physics.bio-ph↗

arXiv2019

Similarities between Insect Swarms and Isothermal Globular Clusters

Previous work has suggested that disordered swarms of flying insects can be well modeled as self-gravitating systems, as long as the "gravitational" interaction is adaptive. Motivated by this work we compare the predictions of the classic, mean-field King model for isothermal globular clusters to observations of insect swarms. Detailed numerical simulations of regular and adaptive gravity allow us to expose the features of the swarms' density profiles that are captured by the King model phenomenology, and those that are due to adaptivity and short-range repulsion. Our results provide further support for adaptive gravity as a model for swarms.

physics.bio-ph↗

arXiv2018

Critical scaling near the yielding transition in granular media

We show that the yielding transition in granular media displays second-order critical-point scaling behavior. We carry out discrete element simulations in the low inertial number limit for frictionless, purely repulsive spherical grains undergoing simple shear at fixed nondimensional shear stress $Σ$ in two and three spatial dimensions. To find a mechanically stable (MS) packing that can support the applied $Σ$, isotropically prepared states with size $L$ must undergo a total strain $γ_{\rm ms}(Σ,L)$. The number density of MS packings ($\propto γ_{\rm ms}^{-1}$) vanishes for $Σ> Σ_c \approx 0.11$ according to a critical scaling form with a length scale $ξ\propto |Σ- Σ_c|^{-ν}$, where $ν\approx 1.7-1.8$. Above the yield stress ($Σ>Σ_c$), no MS packings that can support $Σ$ exist in the large system limit, $L/ξ\gg 1$. MS packings generated via shear possess anisotropic force and contact networks, suggesting that $Σ_c$ is associated with an upper limit in the degree to which these networks can be deformed away from those for isotropic packings.

cond-mat.soft↗

arXiv2017

Determining the Onset of Hydrodynamic Erosion in Turbulent Flow

We revisit the longstanding question of the onset of sediment transport driven by a turbulent fluid flow via laboratory measurements. We use particle tracking velocimetry to quantify the fluid flow as well as the motion of individual grains. As we increase the flow speed above the transition to sediment transport, we observe that an increasing fraction of grains are transported downstream, although the average downstream velocity of the transported grains remains roughly constant. However, we find that the fraction of mobilized grains does not vanish sharply at a critical flow rate. Additionally, the distribution of the fluctuating velocities of non-transported grains becomes broader with heavier tails, meaning that unambiguously separating mobile and static grains is not possible. As an alternative approach, we quantify the statistics of grain velocities by using a mixture model consisting of two forms for the grain velocities: a decaying-exponential tail, which represents grains transported downstream, and a peaked distribution centered at zero velocity, which represents grains that fluctuate due to the turbulent flow but remain in place. Our results suggest that more sophisticated statistical measures may be required to quantify grain motion near the onset of sediment transport, particularly in the presence of turbulence.

physics.flu-dyn↗

arXiv2017

The role of grain dynamics in determining the onset of sediment transport

Sediment transport occurs when the nondimensional fluid shear stress $Θ$ at the bed surface exceeds a minimum value $Θ_c$. A large collection of data, known as the Shields curve, shows that $Θ_c$ is primarily a function of the shear Reynolds number ${\rm{Re}}_*$. It is commonly assumed that $Θ>Θ_c({\rm{Re}}_*)$ occurs when the ${\rm Re}_*$-dependent fluid forces are too large to maintain static equilibrium for a typical surface grain. A complimentary approach, which remains relatively unexplored, is to identify $Θ_c({\rm{Re}}_*)$ as the applied shear stress at which grains cannot stop moving. With respect to grain dynamics, ${\rm{Re}}_*$ can be viewed as the viscous time scale for a grain to equilibrate to the fluid flow divided by the typical time for the fluid force to accelerate a grain over the characteristic bed roughness. We performed simulations of granular beds sheared by a model fluid, varying only these two time scales. We find that the critical Shields number $Θ_c({\rm Re}_*)$ obtained from the model mimics the Shields curve and is insensitive to the grain properties, the model fluid flow, and the form of the drag law. Quantitative discrepancies between the model results and the Shields curve are consistent with previous calculations of lift forces at varying ${\rm Re}_*$. Grains at low ${\rm Re}_*$ find more stable configurations than those at high ${\rm{Re}}_*$ due to differences in the grain reorganization dynamics. Thus, instead of focusing on mechanical equilibrium of a typical grain at the bed surface, $Θ_c({\rm{Re}}_*)$ may be better described by the stress at which mobile grains cannot find a stable configuration and stop moving.

cond-mat.soft↗

arXiv2016

Inference of Causal Information Flow in Collective Animal Behavior

Understanding and even defining what constitutes animal interactions remains a challenging problem. Correlational tools may be inappropriate for detecting communication between a set of many agents exhibiting nonlinear behavior. A different approach is to define coordinated motions in terms of an information theoretic channel of direct causal information flow. In this work, we consider time series data obtained by an experimental protocol of optical tracking of the insect species Chironomus riparius. The data constitute reconstructed 3-D spatial trajectories of the insects' flight trajectories and kinematics. We present an application of the optimal causation entropy (oCSE) principle to identify direct causal relationships or information channels among the insects. The collection of channels inferred by oCSE describes a network of information flow within the swarm. We find that information channels with a long spatial range are more common than expected under the assumption that causal information flows should be spatially localized. The tools developed herein are general and applicable to the inference and study of intercommunication networks in a wide variety of natural settings.

q-bio.QM↗

arXiv2015

Long-range Acoustic Interactions in Insect Swarms: An Adaptive Gravity Model

The collective motion of groups of animals emerges from the net effect of the interactions between individual members of the group. In many cases, such as birds, fish, or ungulates, these interactions are mediated by sensory stimuli that predominantly arise from nearby neighbors. But not all stimuli in animal groups are short range. Here, we consider mating swarms of midges, which interact primarily via long-range acoustic stimuli. We exploit the similarity in form between the decay of acoustic and gravitational sources to build a model for swarm behavior. By accounting for the adaptive nature of the midges' acoustic sensing, we show that our "adaptive gravity" model makes mean-field predictions that agree well with experimental observations of laboratory swarms. Our results highlight the role of sensory mechanisms and interaction range in collective animal behavior. The adaptive interactions that we present here open a new class of equations of motion, which may appear in other biological contexts.

physics.bio-ph↗

arXiv2015

Onset and cessation of motion in hydrodynamically sheared granular beds

We performed molecular dynamics simulations of granular beds driven by a model hydrodynamic shear flow to elucidate general grain-scale mechanisms that determine the onset and cessation of sediment transport. By varying the Shields number (the nondimensional shear stress at the top of the bed) and particle Reynolds number (the ratio of particle inertia to viscous damping), we explore how variations of the fluid flow rate, particle inertia, and fluid viscosity affect the onset and cessation of bed motion. For low to moderate particle Reynolds numbers, a critical boundary separates mobile and static states. Transition times between these states diverge as this boundary is approached both from above and below. At high particle Reynolds number, inertial effects become dominant, and particle motion can be sustained well below flow rates at which mobilization of a static bed occurs. We also find that the onset of bed motion (for both low and high particle Reynolds numbers) is described by Weibullian weakest-link statistics, and thus is crucially dependent on the packing structure of the granular bed, even deep beneath the surface.

cond-mat.soft↗

arXiv2014

Measurements of the coupling between the tumbling of rods and the velocity gradient tensor in turbulence

We present simultaneous experimental measurements of the dynamics of anisotropic particles transported by a turbulent flow and the velocity gradient tensor of the flow surrounding them. We track both rod-shaped particles and small spherical flow tracers using stereoscopic particle tracking. By using scanned illumination, we are able to obtain a high enough seeding density of tracers to measure the full velocity gradient tensor near the rod. The alignment of rods with the vorticity and the eigenvectors of the strain rate show agreement with numerical simulations. A full description of the tumbling of rods in turbulence requires specifying a seven-dimensional joint probability density function (PDF) of five scalars characterizing the velocity gradient tensor and two scalars describing the relative orientation of the rod. If these seven parameters are known, then Jeffery's equation specifies the rod tumbling rate and any statistic of rod rotations can be obtained as a weighted average over the joint PDF. To look for a lower-dimensional projection to simplify the problem, we explore conditional averages of the mean-squared tumbling rate. The conditional dependence of the mean-squared tumbling rate on the magnitude of both the vorticity and the strain rate is strong, as expected, and similar. There is also a strong dependence on the orientation between the rod and the vorticity, since a rod aligned with the vorticity vector tumbles due to strain but not vorticity. When conditioned on the alignment of the rod with the eigenvectors of the strain rate, the largest tumbling rate is obtained when the rod is oriented at a certain angle to the eigenvector that corresponds to the smallest eigenvalue, because this particular orientation maximizes the contribution from both the vorticity and strain.

physics.flu-dyn↗

arXiv2013

Alignment of vorticity and rods with Lagrangian fluid stretching in turbulence

Stretching in continuum mechanics is naturally described using the Cauchy-Green strain tensors. These tensors quantify the Lagrangian stretching experienced by a material element, and provide a powerful way to study processes in turbulent fluid flows that involve stretching such as vortex stretching and alignment of anisotropic particles. Analyzing data from a simulation of isotropic turbulence, we observe preferential alignment between anisotropic particles and vorticity. We show that this alignment arises because both of these quantities independently tend to align with the strongest Lagrangian stretching direction, as defined by the maximum eigenvector of the left Cauchy-Green strain tensor. In particular, anisotropic particles approach almost perfect alignment with the strongest stretching direction. The alignment of vorticity with stretching is weaker, but still much stronger than previously observed alignment of vorticity with the eigenvectors of the Eulerian strain rate tensor. The alignment of strong vorticity is almost the same as that of rods that have experienced the same stretching.

physics.flu-dyn↗

arXiv2013

Quantifying stretching and rearrangement in epithelial sheet migration

Although understanding the collective migration of cells, such as that seen in epithelial sheets, is essential for understanding diseases such as metastatic cancer, this motion is not yet as well characterized as individual cell migration. Here we adapt quantitative metrics used to characterize the flow and deformation of soft matter to contrast different types of motion within a migrating sheet of cells. Using a Finite-Time Lyapunov Exponent (FTLE) analysis, we find that - in spite of large fluctuations - the flow field of an epithelial cell sheet is not chaotic. Stretching of a sheet of cells (i.e., positive FTLE) is localized at the leading edge of migration. By decomposing the motion of the cells into affine and non-affine components using the metric D$^{2}_{min}$, we quantify local plastic rearrangements and describe the motion of a group of cells in a novel way. We find an increase in plastic rearrangements with increasing cell densities, whereas inanimate systems tend to exhibit less non-affine rearrangements with increasing density.

physics.bio-ph↗

arXiv2012

Visualization of Kelvin waves on quantum vortices

In superfluid helium, vorticity is quantized and constrained on line-like phase singularities, called quantum or quantized vortices. By visualizing the motion of sub-micron frozen particles in superfluid $^{4}$He, we directly observe for the first time the helical Kelvin waves excited after quantized vortex reconnections. We compare the data with self-similar solutions of vortex filament models. We report the results in a fluid dynamic video.

physics.flu-dyn↗

arXiv2011

Path lengths in turbulence

By tracking tracer particles at high speeds and for long times, we study the geometric statistics of Lagrangian trajectories in an intensely turbulent laboratory flow. In particular, we consider the distinction between the displacement of particles from their initial positions and the total distance they travel. The difference of these two quantities shows power-law scaling in the inertial range. By comparing them with simulations of a chaotic but non-turbulent flow and a Lagrangian Stochastic model, we suggest that our results are a signature of turbulence.

physics.data-an↗

arXiv2010

Direct multi-scale reconstruction of velocity fields from measurements of particle tracks

We present a method for reconstructing two-dimensional velocity fields at specified length scales using observational data from tracer particles in a flow, without the need for interpolation or smoothing. The algorithm, adapted from techniques proposed for oceanography, involves a least-squares projection of the measurements onto a set of two-dimensional, incompressible basis modes with known length scales. Those modes are constructed from components of the velocity potential function, which accounts for inflow and outflow at the open boundaries of the measurement region; and components of the streamfunction, which accounts for the remainder of the flow. All calculations are evaluated at particle locations, without interpolation onto an arbitrary grid. Since the modes have a well-defined length scales, scale-local flow properties are available directly. The technique eliminates outlier particles automatically and reduces the apparent compressibility of the data. Moreover the technique can be used to produce spatial power spectra and to evaluate the spatial effects of open boundaries; it also holds promise for direct calculation of scale-to-scale transfer of enstrophy and energy.

physics.flu-dyn↗

arXiv2007

Modification of the turbulent energy cascade by polymer additives

By tracking small particles in the bulk of an intensely turbulent flow, we show that even a very small concentration of long-chain polymers disrupts the usual turbulent energy cascade. The polymers affect scales much larger than their physical size, from the dissipation range to the inertial range. The effect depends strongly on the polymer concentration. While the dissipative-scale statistics change continuously as the polymer concentration is increased, the inertial-range energy transfer rate is only altered by the polymer additives when the concentration is above a threshold (approximately 5 parts per million by weight for the polymer we used).

physics.flu-dyn↗