Biomedical subjects

Ioannis G Kevrekidis

Publications and source records attributed to Ioannis G Kevrekidis.

13 recordsLinked to original sources

PubMed2006

Spatially distributed stochastic systems: Equation-free and equation-assisted preconditioned computations.

Spatially distributed problems are often approximately modeled in terms of partial differential equations (PDEs) for appropriate coarse-grained quantities (e.g., concentrations). The derivation of accurate such PDEs starting from finer scale, atomistic models, and using suitable averaging is often a challenging task; approximate PDEs are typically obtained through mathematical closure procedures (e.g., mean field approximations). In this paper, we show how such approximate macroscopic PDEs can be exploited in constructing preconditioners to accelerate stochastic computations for spatially distributed particle-based process models. We illustrate how such preconditioning can improve the convergence of equation-free coarse-grained methods based on coarse timesteppers. Our model problem is a stochastic reaction-diffusion model capable of exhibiting Turing instabilities.

Algorithms↗

PubMed2006

Equation-free dynamic renormalization of a Kardar-Parisi-Zhang-type equation.

In the context of equation-free computation, we devise and implement a procedure for using short-time direct simulations of a Kardar-Parisi-Zhang-(KPZ-) type equation to calculate the self-similar solution for its ensemble averaged correlation function. The method involves "lifting" from candidate pair-correlation functions to consistent realization ensembles, short bursts of KPZ-type evolution, and appropriate rescaling of the resulting averaged pair correlation functions. Both the self-similar shapes and their similarity exponents are obtained at a computational cost significantly reduced to that required to reach saturation in such systems.

Journal Article↗

PubMed2006

Gene regulatory networks: a coarse-grained, equation-free approach to multiscale computation.

We present computer-assisted methods for analyzing stochastic models of gene regulatory networks. The main idea that underlies this equation-free analysis is the design and execution of appropriately initialized short bursts of stochastic simulations; the results of these are processed to estimate coarse-grained quantities of interest, such as mesoscopic transport coefficients. In particular, using a simple model of a genetic toggle switch, we illustrate the computation of an effective free energy Phi and of a state-dependent effective diffusion coefficient D that characterize an unavailable effective Fokker-Planck equation. Additionally we illustrate the linking of equation-free techniques with continuation methods for performing a form of stochastic "bifurcation analysis"; estimation of mean switching times in the case of a bistable switch is also implemented in this equation-free context. The accuracy of our methods is tested by direct comparison with long-time stochastic simulations. This type of equation-free analysis appears to be a promising approach to computing features of the long-time, coarse-grained behavior of certain classes of complex stochastic models of gene regulatory networks, circumventing the need for long Monte Carlo simulations.

Algorithms↗

PubMed2005

Coarse nonlinear dynamics and metastability of filling-emptying transitions: water in carbon nanotubes.

Using a coarse-grained molecular dynamics (CMD) approach we study the apparent nonlinear dynamics of water molecules filling or emptying carbon nanotubes as a function of system parameters. Different levels of the pore hydrophobicity give rise to tubes that are empty, water-filled, or fluctuate between these two long-lived metastable states. The corresponding coarse-grained free-energy surfaces and their hysteretic parameter dependence are explored by linking MD to continuum fixed point and bifurcation algorithms. The results are validated through equilibrium MD simulations.

Algorithms↗

PubMed2005

Core collapse via coarse dynamic renormalization.

In the context of the recently developed "equation-free" approach to computer-assisted analysis of complex systems, we extract the self-similar solution describing core collapse of a stellar system from numerical experiments. The technique allows us to sidestep the core "bounce" that occurs in direct N-body simulations due to the small-N correlations that develop in the late stages of collapse, and hence to follow the evolution well into the self-similar regime.

Journal Article↗

PubMed2005

Coarse master equation from Bayesian analysis of replica molecular dynamics simulations.

We use Bayesian inference to derive the rate coefficients of a coarse master equation from molecular dynamics simulations. Results from multiple short simulation trajectories are used to estimate propagators. A likelihood function constructed as a product of the propagators provides a posterior distribution of the free coefficients in the rate matrix determining the Markovian master equation. Extensions to non-Markovian dynamics are discussed, using the trajectory "paths" as observations. The Markovian approach is illustrated for the filling and emptying transitions of short carbon nanotubes dissolved in water. We show that accurate thermodynamic and kinetic properties, such as free energy surfaces and kinetic rate coefficients, can be computed from coarse master equations obtained through Bayesian inference.

Bayes Theorem↗

PubMed2005

Coarse-grained computations for a micellar system.

We establish, through coarse-grained computation, a connection between traditional, continuum numerical algorithms (initial value problems as well as fixed point algorithms), and atomistic simulations of the Larson model of micelle formation. The procedure hinges on the (expected) evolution of a few slow, coarse-grained mesoscopic observables of the Monte Carlo simulation, and on (computational) time scale separation between these and the remaining "slaved," fast variables. Short bursts of appropriately initialized atomistic simulation are used to estimate the (coarse grained, deterministic) local dynamics of the evolution of the observables. These estimates are then in turn used to accelerate the evolution to computational stationarity through traditional continuum algorithms (forward Euler integration, Newton-Raphson fixed point computation). This "equation-free" framework, bypassing the derivation of explicit, closed equations for the observables (e.g., equations of state), may provide a computational bridge between direct atomistic/stochastic simulation and the analysis of its macroscopic, system-level consequences.

Journal Article↗

PubMed2005

Coarse-grained kinetic computations for rare events: application to micelle formation.

We discuss a coarse-grained approach to the computation of rare events in the context of grand canonical Monte Carlo (GCMC) simulations of self-assembly of surfactant molecules into micelles. The basic assumption is that the computational system dynamics can be decomposed into two parts-fast (noise) and slow (reaction coordinates) dynamics, so that the system can be described by an effective, coarse-grained Fokker-Planck (FP) equation. While such an assumption may be valid in many circumstances, an explicit form of FP equation is not always available. In our computations we bypass the analytic derivation of such an effective FP equation. The effective free energy gradient and the state-dependent magnitude of the random noise, which are necessary to formulate the effective Fokker-Planck equation, are obtained from ensembles of short bursts of microscopic simulations with judiciously chosen initial conditions. The reaction coordinate in our micelle formation problem is taken to be the size of a cluster of surfactant molecules. We test the validity of the effective FP description in this system and reconstruct a coarse-grained free energy surface in good agreement with full-scale GCMC simulations. We also show that, for very small clusters, the cluster size ceases to be a good reaction coordinate for a one-dimensional effective description. We discuss possible ways to improve the current model and to take higher-dimensional coarse-grained dynamics into account.

Journal Article↗

PubMed2005

Geometric diffusions for the analysis of data from sensor networks.

Harmonic analysis on manifolds and graphs has recently led to mathematical developments in the field of data analysis. The resulting new tools can be used to compress and analyze large and complex data sets, such as those derived from sensor networks or neuronal activity datasets, obtained in the laboratory or through computer modeling. The nature of the algorithms (based on diffusion maps and connectivity strengths on graphs) possesses a certain analogy with neural information processing, and has the potential to provide inspiration for modeling and understanding biological organization in perception and memory formation.

Algorithms↗

PubMed2005

Optical imaging and control of genetically designated neurons in functioning circuits.

Proteins with engineered sensitivities to light are infiltrating the biological mechanisms by which neurons generate and detect electrochemical signals. Encoded in DNA and active only in genetically specified target cells, these proteins provide selective optical interfaces for observing and controlling signaling by defined groups of neurons in functioning circuits, in vitro and in vivo. Light-emitting sensors of neuronal activity (reporting calcium increase, neurotransmitter release, or membrane depolarization) have begun to reveal how information is represented by neuronal assemblies, and how these representations are transformed during the computations that inform behavior. Light-driven actuators control the electrical activities of central neurons in freely moving animals and establish causal connections between the activation of specific neurons and the expression of particular behaviors. Anchored within mathematical systems and control theory, the combination of finely resolved optical field sensing and finely resolved optical field actuation will open new dimensions for the analysis of the connectivity, dynamics, and plasticity of neuronal circuits, and perhaps even for replacing lost--or designing novel--functionalities.

Animals↗

PubMed2004

Apparent hysteresis in a driven system with self-organized drag.

The motion of extended defects in materials is often resisted by interaction with diffusing impurities. The defect undergoes a transition from slow to fast migration, or vice versa. This transition is commonly hysteretic, with the defect jumping back and forth between these kinetic states. We explore such hysteresis within a kinetic Monte Carlo simulation. After identifying the slow variables, we construct an effective potential that quantitatively describes the stable and metastable states of the system (a bifurcation diagram) and the kinetics of the transitions between these. This description provides a means to determine whether hysteresis will be observed in a particular simulation and the detailed nature of this hysteresis (switching times).

Journal Article↗

PubMed2003

Oscillatory thermomechanical instability of an ultrathin catalyst.

Because of the small thermal capacity of ultrathin ( approximately 200 nanometers) metal single crystals, it is possible to explore the coupling of catalytic and thermal action at low pressures. We analyzed a chemothermomechanical instability in this regime, in which catalytic reaction kinetics interact with heat transfer and mechanical buckling to create oscillations. These interacting components are separated and explored through experimentation, mathematical modeling, and scientific computation, and an explanation of the phenomenon emerges from their synthesis.

Journal Article↗

PubMed2003

Bifurcation and stability analysis of rotating chemical spirals in circular domains: boundary-induced meandering and stabilization.

Recent experimental and model studies have revealed that the domain size may strongly influence the dynamics of rotating spirals in two-dimensional pattern forming chemical reactions. Hartmann et al. [Phys. Rev. Lett. 76, 1384 (1996)], report a frequency increase of spirals in circular domains with diameters substantially smaller than the spiral wavelength in a large domain for the catalytic NO+CO reaction on a microstructured platinum surface. Accompanying simulations with a simple reaction-diffusion system reproduced the behavior. Here, we supplement these studies by a numerical bifurcation and stability analysis of rotating spirals in a simple activator-inhibitor model. The problem is solved in a co-rotating frame of reference. No-flux conditions are imposed at the boundary of the circular domain. At large domain sizes, eigenvalues and eigenvectors very close to those corresponding to infinite medium translational invariance are observed. Upon decrease of domain size, we observe a simultaneous change in the rotation frequency and a deviation of these eigenvalues from being neutrally stable (zero real part). The latter phenomenon indicates that the translation symmetry of the spiral solution is appreciably broken due to the interaction with the (now nearby) wall. Various dynamical regimes are found: first, the spiral simply tries to avoid the boundary and its tip moves towards the center of the circular domain corresponding to a negative real part of the "translational" eigenvalues. This effect is noticeable at a domain radius of R<R(cr,1). The spiral subsequently exhibits an oscillatory instability: the tip trajectory displays a meandering motion, which may be characterized as boundary-induced spiral meandering. A systematic study of the spiral rotation as a function of a control parameter and the domain size reveals that the meandering instability in large domains becomes suppressed, and the spiral rotation becomes rigid, at a critical radius R(cr,0). Boundary-induced meandering arises below a second critical radius R(cr,2)<R(cr,0).

Journal Article↗