arXiv subjects

Agnese Seminara

Publications and source records attributed to Agnese Seminara.

16 recordsLinked to original sources

arXiv2026

Amphibian water to land transition reveals physical limits of olfaction

The evolutionary transition from water to land required animals to sense and respond to drastically different environments. Chemicals diffuse four orders of magnitude more slowly in water than in air, requiring significant remodeling of the olfactory system. Amphibians provide a powerful model to analyze these adaptations because they transition from an aquatic to terrestrial state during a single lifetime following metamorphosis. Here we exploit laboratory-induced metamorphosis of adult Ambystoma mexicanum to ask how fundamental chemical properties impact aquatic versus terrestrial olfaction. By combining asymptotic theory with numerical simulations using reconstructed olfactory chamber morphologies, we find that odor adsorption takes tens of seconds in water compared to milliseconds in air. Adsorption of an ephemeral odor whiff is maximized at an inhalation speed of $\sim$\SI{20}{\centi\meter\per\second} in air vs near zero in water, matching our measurements of negligible aquatic inhalation. Nevertheless, aquatic axolotls quickly respond to introduced odorant molecules. While morphological differences in the olfactory chamber of aquatic versus terrestrial axolotls are nearly irrelevant, aquatic olfactory cilia may pump water to significantly speed up the rate of odor adsorption. Together with adaptive behavioral responses that reduce proximity to the target, the wait time can reduce to less than one second. Thus, while aquatic olfaction is slower than terrestrial olfaction, aquatic animals exhibit anatomical and behavioral adaptations that support chemical detection as a proximal sense.

physics.bio-ph↗

arXiv2026

Smart strategies to navigate turbulent odor plumes reorienting to local wind

Olfactory search in turbulent environments is a sensorimotor problem that many animals solve with remarkable efficiency, yet replicating this ability in artificial systems is an enduring challenge because detections are intermittent and wind direction fluctuates strongly, rendering standard search strategies unreliable. We introduce a wind-relative reinforcement-learning framework in which an agent navigates a turbulent plume with a single internal variable --- the elapsed time since the last odor detection --- and selects actions relative to a locally estimated wind direction filtered through an exponential memory kernel. Policies are trained and evaluated in direct numerical simulations of turbulence, capturing the multi-scale characteristics of velocity and odor fields in natural environments, both in the presence and absence of a mean wind. In a mild mean wind, the learned policy outperforms well-known biomimetic policies such as cast-and-surge regardless of the wind memory time, yet adapts its movement pattern to wind-estimation quality. In isotropic turbulence, performance peaks at an intermediate wind memory time, identifying temporal wind integration as a regime-dependent resource. Our results highlight the importance of developing and validating olfactory-navigation strategies under realistic turbulent conditions, and offer a compact design principle for minimal robotic olfactory navigation and testable predictions for biological search behavior.

physics.flu-dyn↗

arXiv2026

Clock-state olfactory search in turbulent flows using Q-learning: The geometry of plume recovery

Finding an odor source in a turbulent flow requires effectively leveraging the history of olfactory observations into a robust navigation strategy. In this work, we use tabular Q-learning to train an olfactory search agent with a minimal memory of past observations: only a running clock since the last whiff. This agent learns an interpretable strategy to recover the plume which combines well-known behaviors observed in insects: surging, casting, and a return downwind. While achieving good performance on data from direct numerical simulations of turbulence, the agent is limited by an inability to adapt its strategy to the local intermittency level; we show that providing more flexibility improves robustness.

physics.bio-ph↗

arXiv2026

Zero-information limit of a collective olfactory search model

We address the problem of how individuals can integrate efficiently their private behavior with information provided by others within a group. To this end, we consider the model of collective search introduced in [https://doi.org/10.1103/PhysRevE.102.012402], under a minimal setting with no olfactory information. Agents combine a private exploratory behavior and a social imitation consisting in aligning to their neighbors, and weigh the two contributions with a single ``trust" parameter that controls their relative influence. We find that an optimal trust parameter exists even in the absence of olfactory information, as was observed in the original model. Optimality is dictated by the need to explore the minimal region of space that contains the target. An optimal trust parameter emerges from this constraint because it it tunes imitation, which induces a collective mechanism of inertia affecting the size and path of the swarm. We predict the optimal trust parameter for cohesive groups where all agents interact with one another. We show how optimality depends on the initialization of the agents and the unknown location of the target, in close agreement with numerical simulations. Our results may be leveraged to optimize the design of swarm robotics or to understand information integration in organisms with decentralized nervous systems such as cephalopods.

physics.bio-ph↗

arXiv2026

Noise enhances odor source localization

We address the problem of inferring the location of a target that releases odor in the presence of turbulence. Input for the inference is provided by many sensors scattered within the odor plume. Drawing inspiration from distributed chemosensation in biology, we ask whether the accuracy of the inference is affected by proprioceptive noise, i.e., noise on the perceived location of the sensors. Surprisingly, in the presence of a net fluid flow, proprioceptive noise improves Bayesian inference, rather than degrading it. An optimal noise exists that efficiently leverages additional information hidden within the geometry of the odor plume. Empirical tuning of noise functions well across a range of distances and may be implemented in practice. Other sources of noise also improve accuracy, owing to their ability to break the spatiotemporal correlations of the turbulent plume. These counterintuitive benefits of noise may be leveraged to improve sensory processing in biology and robotics.

physics.bio-ph↗

arXiv2025

Learning to crawl: Benefits and limits of centralized vs distributed control

We present a model of a crawler consisting of several suction units distributed along a straight line and connected by springs. The suction units are rudimentary proprioceptors-actuators, which sense binary states of compression vs elongation of the springs, and can either adhere or remain idle. Muscular contraction is not controlled by the crawler, but follows an endogenous, stereotyped wave. The crawler is tasked to learn patterns of adhesion that generate thrust in response to the wave of contraction. Using tabular Q-learning we demonstrate that crawling can be learned by trial and error and we ask what are the benefits and limitations of distributed vs centralized learning architectures. We find that by centralizing proprioceptive feedback and control, the crawler leverages long range correlations in the dynamics and ride the endogenous wave smoothly. The ensuing benefits are measured in terms of both speed and robustness to failure, although they come at increased computational cost. At the opposite extreme, purely distributed feedback and control only leverages local information and yield a jerkier and slower crawling, although computationally cheap. Intermediate levels of centralization can negotiate fast and robust crawling while avoiding excessive computational burden, demonstrating the computational benefits of a hierarchical organization of crawling. Our model unveils the trade-offs between crawling speed, robustness to failure, computational cost and information exchange that may shape biological solutions for crawling and could inspire the design of robotic crawlers.

physics.bio-ph↗

arXiv2025

Effect of swimming mode on shielding of odor traces in turbulence

Marine organisms manipulate their surrounding flow through their swimming dynamics, which affects the transport of their own odor cues. We demonstrate by direct numerical simulations how a group of swimmers, moving at intermediate Reynolds numbers, immersed in a turbulent flow, alter the shape of the odor plume they release in the water. Odor mixing is enhanced by increased velocity fluctuations and a swimmer-induced flow circulation that widens the odor plume at close range while speeding up dilution of the chemical trace. Beyond a short-range increase in the likelihood of being detected, swimming considerably reduces detections with effects that can persist at distances on the order of ten times the size of the group or more. We find that pullerlike swimmers are more effective at olfactory shielding than pusherlike swimmers. We trace this difference back to the dynamics at the swimmer location, which tends to trap odor at the source for pushers and to dilute it for pullers. Olfactory shielding is robust to changes in the conditions, and is more pronounced for weak turbulent Reynolds numbers and large swimmer Reynolds numbers. Our results suggest that olfactory shielding may play a role in the emergence of different swimming modalities by marine organisms.

physics.flu-dyn↗

arXiv2024

Q-learning with temporal memory to navigate turbulence

We consider the problem of olfactory searches in a turbulent environment. We focus on agents that respond solely to odor stimuli, with no access to spatial perception nor prior information about the odor. We ask whether navigation to a target can be learned robustly within a sequential decision making framework. We develop a reinforcement learning algorithm using a small set of interpretable olfactory states and train it with realistic turbulent odor cues. By introducing a temporal memory, we demonstrate that two salient features of odor traces, discretized in few olfactory states, are sufficient to learn navigation in a realistic odor plume. Performance is dictated by the sparse nature of turbulent odors. An optimal memory exists which ignores blanks within the plume and activates a recovery strategy outside the plume. We obtain the best performance by letting agents learn their recovery strategy and show that it is mostly casting cross wind, similar to behavior observed in flying insects. The optimal strategy is robust to substantial changes in the odor plumes, suggesting minor parameter tuning may be sufficient to adapt to different environments.

physics.bio-ph↗

arXiv2022

Physics Informed Shallow Machine Learning for Wind Speed Prediction

The ability to predict wind is crucial for both energy production and weather forecasting. Mechanistic models that form the basis of traditional forecasting perform poorly near the ground. In this paper, we take an alternative data-driven approach based on supervised learning. We analyze a massive dataset of wind measured from anemometers located at 10 m height in 32 locations in two central and north west regions of Italy (Abruzzo and Liguria). We train supervised learning algorithms using the past history of wind to predict its value at a future time (horizon). Using data from a single location and time horizon we compare systematically several algorithms where we vary the input/output variables, the memory of the input and the linear vs non-linear learning model. We then compare performance of the best algorithms across all locations and forecasting horizons. We find that the optimal design as well as its performance vary with the location. We demonstrate that the presence of a reproducible diurnal cycle provides a rationale to understand this variation. We conclude with a systematic comparison with state of the art algorithms and show that, when the model is accurately designed, shallow algorithms are competitive with more complex deep architectures.

cs.LG↗

arXiv2021

Learning to predict target location with turbulent odor plumes

Animal behavior and neural recordings show that the brain is able to measure both the intensity of an odor and the timing of odor encounters. However, whether intensity or timing of odor detections is more informative for olfactory-driven behavior is not understood. To tackle this question, we consider the problem of locating a target using the odor it releases. We ask whether the position of a target is best predicted by measures of timing intensity of its odor, sampled for a short period of time. To answer this question, we feed data from accurate numerical simulations of odor transport to machine learning algorithms that learn how to connect odor to target location. We find that both intensity and timing can separately predict target location even from a distance of several meters; however their efficacy varies with the dilution of the odor in space. Thus organisms that use olfaction from different ranges may have to switch among different modalities. This has implications on how the brain should represent odors as the target is approached. We demonstrate simple strategies to improve accuracy and robustness of the prediction by modifying odor sampling and appropriately combining distinct measures together. To test the predictions, animal behavior and odor representation should be monitored as the animal moves relative to the target, or in virtual conditions that mimic concentrated dilute environments.

q-bio.QM↗

arXiv2020

Turbulence dictates the fate of virus-containing droplets in violent expiratory events

Violent expiratory events, such as coughing and sneezing, are highly nontrivial examples of a two-phase mixture of liquid droplets dispersed into an unsteady turbulent airflow. Understanding the physical mechanisms determining the dispersion and evaporation process of respiratory droplets has recently become a priority given the global emergency caused by the SARS-CoV-2 infection. By means of high-resolution direct numerical simulations (DNS) of the expiratory airflow and a comprehensive Lagrangian model for the droplet dynamics, we identify the key role of turbulence on the fate of exhaled droplets. Due to the considerable spread in the initial droplet size, we show that the droplet evaporation time is controlled by the combined effect of turbulence and droplet inertia. This mechanism is clearly highlighted when comparing the DNS results with those obtained using coarse-grained descriptions that are employed in the majority of the current state-of-the-art investigations, resulting in errors up to $100\%$ when the turbulent fluctuations are filtered or completely averaged out.

physics.flu-dyn↗

arXiv2019

A precise relationship among Buller's drop, ballistospore and gill morphology enables maximal packing of spores within gilled mushrooms

Basidiomycete fungi eject spores using a surface tension catapult; a fluid drop forms at the base of each spore and after reaching a critical size, coalesces with the spore and launches it from the gill surface. Although basidiomycetes function within ecosystems as both devastating pathogens and mutualists critical to plant growth, an incomplete understanding of ballistospory hinders predictions of spore dispersal and impedes disease forecasting and conservation strategies. Building on a nascent understanding of the physics underpinning the surface tension catapult, we first use the principle of energy conservation to identify ejection velocities resulting from a range of Buller's drop and spore sizes. We next model a spore's trajectory away from a basidium and identify a specific relationship among intergill distances and Buller's drop and spore radii enabling the maximum number of spores to be packaged within a minimum amount of gill tissue. We collect data of spore and gill morphology in wild mushrooms and we find that real species lie in a region where, in order to pack the maximum number of spores with minimum amount of biomass, the volume of Buller's drop should scale as the volume of the spore, and its linear size should be about half of spore size. Previously published data of Buller's drop and spore size confirm this finding. Our results suggest that the radius of Buller's drop is tightly regulated to enable maximum packing of spores.

physics.bio-ph↗

arXiv2009

Summoning the wind: Hydrodynamic cooperation of forcibly ejected fungal spores

The forcibly launched spores of the crop pathogen \emph{Sclerotinia sclerotiorum} must eject through many centimeters of nearly still air to reach the flowers of the plants that the fungus infects. Because of their microscopic size, individually ejected spores are quickly brought to rest by drag. In the accompanying fluid dynamics video we show experimental and numerical simulations that demonstrate how, by coordinating the nearly simultaneous ejection of hundreds of thousands of spores,\emph{Sclerotinia} and other species of apothecial fungus are able to sculpt a flow of air that carries spores across the boundary layer and around intervening obstacles. Many spores are sacrificed to create this flow of air. Although high speed imaging of spore launch in a wild isolate of the dung fungus \emph{Ascobolus} shows that the synchronization of spore ejections is self-organized, which could lead to spores delaying their ejection to avoid being sacrificed, simulations and asymptotic analysis show that, close the fruit body, ejected spores form a sheet-like jet that advances across the fruit body as more spores are ejected. By ejecting on the arrival of the sheet spores maximize \emph{both} their range and their contribution to the cooperative wind.

physics.flu-dyn↗

arXiv2008

Condensation of cloud microdroplets in homogeneous isotropic turbulence

The growth by condensation of small water droplets in a three-dimensional homogeneous isotropic turbulent flow is considered. Within a simple model of advection and condensation, the dynamics and growth of millions of droplets are integrated. A droplet-size spectra broadening is obtained and it is shown to increase with the Reynolds number of turbulence, by means of two series of direct numerical simulations at increasing resolution. This is a key point towards a proper evaluation of the effects of turbulence for condensation in warm clouds, where the Reynolds numbers typically achieve huge values. The obtained droplet-size spectra broadening as a function of the Reynolds number is shown to be consistent with dimensional arguments. A generalization of this expectation to Reynolds numbers not accessible by DNS is proposed, yielding upper and lower bounds to the actual size-spectra broadening. A further DNS matching the large scales of the system suggests consistency of the picture drawn, while additional effort is needed to evaluate the impact of this effect for condensation in more realistic cloud conditions.

nlin.CD↗

arXiv2006

Large-scale anisotropy in scalar turbulence

The effect of anisotropy on the statistics of a passive tracer transported by a turbulent flow is investigated. We show that under broad conditions an arbitrarily small amount of anisotropy propagates to the large scales where it eventually dominates the structure of the concentration field. This result is obtained analytically in the framework of an exactly solvable model and confirmed by numerical simulations of scalar transport in two-dimensional turbulence.

nlin.CD↗

arXiv2005

The large-scale structure of passive scalar turbulence

We investigate the large-scale statistics of a passive scalar transported by a turbulent velocity field. At scales larger than the characteristic lengthscale of scalar injection, yet smaller than the correlation length of the velocity, the advected field displays persistent long-range correlations due to the underlying turbulent velocity. These induce significant deviations from equilibrium statistics for high-order scalar correlations, despite the absence of scalar flux.

nlin.CD↗