arXiv subjects

Uttam Bhat

Publications and source records attributed to Uttam Bhat.

11 recordsLinked to original sources

arXiv2022

Recurrent Neural Networks for Partially Observed Dynamical Systems

Complex nonlinear dynamics are ubiquitous in many fields. Moreover, we rarely have access to all of the relevant state variables governing the dynamics. Delay embedding allows us, in principle, to account for unobserved state variables. Here we provide an algebraic approach to delay embedding that permits explicit approximation of error. We also provide the asymptotic dependence of the first order approximation error on the system size. More importantly, this formulation of delay embedding can be directly implemented using a Recurrent Neural Network (RNN). This observation expands the interpretability of both delay embedding and RNN and facilitates principled incorporation of structure and other constraints into these approaches.

cs.LG↗

arXiv2019

Caching in or falling back at the Sevilleta

Foraging in uncertain environments requires balancing the risks associated with finding alternative resources against potential gains. In aridland environments characterized by extreme variation in the amount and seasonal timing of primary production, consumer communities must weigh the risks associated with foraging for preferred seeds that can be cached against fallback foods of low nutritional quality (e.g., leaves) that must be consumed immediately. Here we explore the influence of resource-scarcity, body size, and seasonal uncertainty on the expected foraging behaviors of caching rodents in the northern Chihuahaun Desert by integrating these elements with a Stochastic Dynamic Program (SDP) to determine fitness-maximizing foraging strategies. We demonstrate that resource-limited environments promote dependence on fallback foods, reducing the likelihood of starvation while increasing future risk exposure. Our results point to a qualitative difference in the use of fallback foods and the fitness benefits of caching at the threshold body size of 50 g. Above this threshold the fitness benefits are greater for consumers with smaller caches, affirming empirical observations of cache use among rodents in such dynamic environments. This suggests that larger-bodied consumers with larger caches may be less sensitive to the future uncertainties in monsoonal onset predicted by global climate scenarios.

q-bio.PE↗

arXiv2019

Scaling of the risk landscape drives optimal life history strategies and the evolution of grazing

Consumers face numerous risks that can be minimized by incorporating different life-history strategies. How much and when a consumer adds to its energetic reserves or invests in reproduction are key behavioral and physiological adaptations that structure much of how organisms interact. Here we develop a theoretical framework that explicitly accounts for stochastic fluctuations of an individual consumer's energetic reserves while foraging and reproducing on a landscape with resources that range from uniformly distributed to highly clustered. First, we show that optimal life-history strategies vary in response to changes in the mean productivity of the resource landscape, where depleted environments promote reproduction at lower energetic states, greater investment in each reproduction event, and smaller litter sizes. We then show that if resource variance scales with body size due to landscape clustering, consumers that forage for clustered foods are susceptible to strong Allee effects, increasing extinction risk. Finally, we show that the proposed relationship between consumer body size, resource clustering, and Allee effect-induced population instability offers key ecological insights into the evolution of large-bodied grazing herbivores from small-bodied browsing ancestors.

q-bio.PE↗

arXiv2018

Edge fires drive the shape and stability of tropical forests

In tropical regions, fires propagate readily in grasslands but typically consume only edges of forest patches. Thus forest patches grow due to tree propagation and shrink by fires in surrounding grasslands. The interplay between these competing edge effects is unknown, but critical in determining the shape and stability of individual forest patches, as well the landscape-level spatial distribution and stability of forests. We analyze high-resolution remote-sensing data from protected areas of the Brazilian Cerrado and find that forest shapes obey a robust perimeter-area scaling relation across climatic zones. We explain this scaling by introducing a heterogeneous fire propagation model of tropical forest-grassland ecotones. Deviations from this perimeter-area relation determine the stability of individual forest patches. At a larger scale, our model predicts that the relative rates of tree growth due to propagative expansion and long-distance seed dispersal determine whether collapse of regional-scale tree cover is continuous or discontinuous as fire frequency changes.

q-bio.PE↗

arXiv2017

On the records

World record setting has long attracted public interest and scientific investigation. Extremal records summarize the limits of the space explored by a process, and the historical progression of a record sheds light on the underlying dynamics of the process. Existing analyses of prediction, statistical properties, and ultimate limits of record progressions have focused on particular domains. However, a broad perspective on how record progressions vary across different spheres of activity needs further development. Here we employ cross-cutting metrics to compare records across a variety of domains, including sports, games, biological evolution, and technological development. We find that these domains exhibit characteristic statistical signatures in terms of rates of improvement, "burstiness" of record-breaking time series, and the acceleration of the record breaking process. Specifically, sports and games exhibit the slowest rate of improvement and a wide range of rates of "burstiness." Technology improves at a much faster rate and, unlike other domains, tends to show acceleration in records. Many biological and technological processes are characterized by constant rates of improvement, showing less burstiness than sports and games. It is important to understand how these statistical properties of record progression emerge from the underlying dynamics. Towards this end, we conduct a detailed analysis of a particular record-setting event: elite marathon running. In this domain, we find that studying record-setting data alone can obscure many of the structural properties of the underlying process. The marathon study also illustrates how some of the standard statistical assumptions underlying record progression models may be inappropriate or commonly violated in real-world datasets.

physics.soc-ph↗

arXiv2016

Exotic phase transitions of k-cores in clustered networks

The giant $k$-core --- maximal connected subgraph of a network where each node has at least $k$ neighbors --- is important in the study of phase transitions and in applications of network theory. Unlike Erdős-Rényi graphs and other random networks where $k$-cores emerge discontinuously for $k\ge 3$, we show that transitive linking (or triadic closure) leads to 3-cores emerging through single or double phase transitions of both discontinuous and continuous nature. We also develop a $k$-core calculation that includes clustering and provides insights into how high-level connectivity emerges.

physics.soc-ph↗

arXiv2016

Stochastic Search with Poisson and Deterministic Resetting

We investigate a stochastic search process in one, two, and three dimensions in which $N$ diffusing searchers that all start at $x_0$ seek a target at the origin. Each of the searchers is also reset to its starting point, either with rate $r$, or deterministically, with a reset time $T$. In one dimension and for a small number of searchers, the search time and the search cost are minimized at a non-zero optimal reset rate (or time), while for sufficiently large $N$, resetting always hinders the search. In general, a single searcher leads to the minimum search cost in one, two, and three dimensions. When the resetting is deterministic, several unexpected feature arise for $N$ searchers, including the search time being independent of $T$ for $1/T\to 0$ and the search cost being independent of $N$ over a suitable range of $N$. Moreover, deterministic resetting typically leads to a lower search cost than in stochastic resetting.

cond-mat.stat-mech↗

arXiv2016

Exploring the isotopic niche: isotopic variance, physiological incorporation, and the temporal dynamics of foraging

Consumer foraging behaviors are dynamic, changing in response to prey availability, seasonality, competition, and even the consumer's physiological state. The isotopic composition of a consumer is a product of these factors as well as the isotopic 'landscape' of its prey, i.e. the isotopic mixing space. Here we build a mechanistic framework that links the ecological and physiological processes of an individual consumer to the isotopic distribution that describes its diet, and ultimately to the isotopic composition of its own tissues, defined as its 'isotopic niche'. By coupling these processes, we systematically investigate under what conditions the isotopic niche of a consumer changes as a function of both the geometric properties of its mixing space and foraging strategies that may be static or dynamic over time. Results of our derivations reveal general insight into the conditions impacting isotopic niche width as a function of consumer specialization on prey, as well as the consumer's ability to transition between diets over time. We show analytically that moderate specialization on isotopically unique prey can serve to maximize a consumer's isotopic niche width, while temporally dynamic diets will tend to result in peak isotopic variance during dietary transitions. We demonstrate the relevance of our theoretical findings by examining a marine system composed of nine invertebrate species commonly consumed by sea otters. In general, our analytical framework highlights the complex interplay of mixing space geometry and consumer dietary behavior in driving expansion and contraction of the isotopic niche. Because this approach is established on ecological mechanism, it is well-suited for enhancing the ecological interpretation, and uncovering the root causes, of observed isotopic data.

q-bio.PE↗

arXiv2015

Intermediate-Level Crossings of a First-Passage Path

We investigate some simple and surprising properties of a one-dimensional Brownian trajectory with diffusion coefficient $D$ that starts at the origin and reaches $X$ either: (i) at time $T$ or (ii) for the first time at time $T$. We determine the most likely location of the first-passage trajectory from $(0,0)$ to $(X,T)$ and its distribution at any intermediate time $t<T$. A first-passage path typically starts out by being repelled from its final location when $X^2/DT\ll 1$. We also determine the distribution of times when the trajectory first crosses and last crosses an arbitrary intermediate position $x<X$. The distribution of first-crossing times may be unimodal or bimodal, depending on whether $X^2/DT\ll 1$ or $X^2/DT\gg 1$. The form of the first-crossing probability in the bimodal regime is qualitatively similar to, but more singular than, the well-known arcsine law.

physics.data-an↗

arXiv2014

Emergence of Clustering in an Acquaintance Model without Homophily

We introduce an agent-based acquaintance model in which social links are created by processes in which there is no explicit homophily. In spite of the homogeneous nature of the social interactions, highly-clustered social networks can arise. The crucial feature of our model is that of variable transitive interactions. Namely, when an agent introduces two unconnected friends, the rate at which a connection actually occurs between them depends on the number of their mutual acquaintances. As this transitive interaction rate is varied, the social network undergoes a dramatic clustering transition. Close to the transition, the network consists of a collection of well-defined communities. As a function of time, the network can also undergo an \emph{incomplete} gelation transition, in which the gel, or giant cluster, does not constitute the entire network, even at infinite time. Some of the clustering properties of our model also arise, but in a more gradual manner, in Facebook networks. Finally, we discuss a more realistic variant of our original model in which there is a soft cutoff in the rate of transitive interactions. With this variant, one can construct network realizations that quantitatively match Facebook networks.

physics.soc-ph↗

arXiv2014

Exclusion Processes with Avalanches

In an exclusion process with avalanches, when a particle hops to a neighboring empty site which is adjacent to an island the particle on the other end of the island immediately hops and if it joins another island this triggers another hop. There are no restrictions on the length of islands and the duration of the avalanche. This process is well-defined in the low-density region, ρ<1/2. We describe the nature of steady states (on a ring) and determine all correlation functions. For the asymmetric version of the process, we compute the steady state current, and we describe shock and rarefaction waves which arise in the evolution of the step-function initial profile. For the symmetric version, we determine the diffusion coefficient and examine the evolution of a tagged particle.

cond-mat.stat-mech↗