Publications (167)195.67 Total impact

[Show abstract] [Hide abstract]
ABSTRACT: Polynomial chaos expansions (PCE) allow us to propagate uncertainties in the coefficients of differential equations to the statistics of their solutions. Their main advantage is that they replace stochastic equations by systems of deterministic equations. Their main challenge is that the computational cost becomes prohibitive when the dimension of the parameters modeling the stochasticity is even moderately large. We propose a generalization of the PCE framework that allows us to keep this dimension as small as possible in favorable situations. For instance, in the setting of stochastic differential equations (SDE) with Markov random forcing, we expect the future evolution to depend on the present solution and the future stochastic variables. We present a restart procedure that precisely allows PCE to depend only on that information. The computational difficulty then becomes the construction of orthogonal polynomials for dynamically evolving measures. We present theoretical results of convergence for our Dynamically Orthogonal generalized Polynomial Chaos (DOgPC) method. Numerical simulations for linear and nonlinear SDEs show that it adequately captures the longtime behavior of their solutions as well as their invariant measures when the latter exist. 
[Show abstract] [Hide abstract]
ABSTRACT: This paper concerns the imaging of a complexvalued anisotropic tensor {\gamma} = {\sigma}+{\iota}{\omega}{\epsilon} from knowledge of several inter magnetic fields H where H satisfies the anisotropic Maxwell system on a bounded domain with prescribed boundary conditions. We show that {\gamma} can be uniquely reconstructed with a loss of two derivatives from errors in the acquisition H. A minimum number of five wellchosen functionals guaranties a local reconstruction of {\gamma} in dimension two. The explicit inversion procedure is presented in several numerical simulations, which demonstrate the influence of the choice boundary conditions on the stability of the reconstruction. This problem finds applications in the medical imaging modalities Current Density Imaging and Magnetic Resonance Electrical Impedance Tomography. 
Article: Reconstruction of a fully anisotropic elasticity tensor from knowledge of displacement fields
[Show abstract] [Hide abstract]
ABSTRACT: We present explicit reconstruction algorithms for fully anisotropic unknown elasticity tensors from knowledge of a finite number of internal displacement fields, with applications to transient elastography. Under certain rankmaximality assumptions satified by the strain fields, explicit algebraic reconstruction formulas are provided. A discussion ensues on how to fulfill these assumptions, describing the range of validity of the approach. We also show how the general method can be applied to more specific cases such as the transversely isotropic one. 
SIAM Journal on Imaging Sciences 01/2015; 8(2):10701089. DOI:10.1137/140988504 · 2.87 Impact Factor

Communications in mathematical sciences 01/2015; 13(3):729748. DOI:10.4310/CMS.2015.v13.n3.a7 · 1.00 Impact Factor

[Show abstract] [Hide abstract]
ABSTRACT: This work is devoted to the stability/resolution analysis of several imaging functionals in complex environments. We consider both linear functionals in the wavefield as well as quadratic functionals based on wavefield correlations. Using simplified measurement settings and reduced functionals that retain the main features of functionals used in practice, we obtain optimal asymptotic estimates of the signaltonoise ratios depending on the main physical parameters of the problem. We consider random media with possibly longrange dependence and with a correlation length that is less than or equal to the central wavelength of the source we aim to reconstruct. This corresponds to the wave propagation regimes of radiative transfer or homogenization. 
[Show abstract] [Hide abstract]
ABSTRACT: This paper derives an efficient procedure for using the threedimensional (3D) vector radiative transfer equation (VRTE) to adjust atmosphere and surface properties and improve their fit with multiangle/multipixel radiometric and polarimetric measurements of scattered sunlight. The proposed adjoint method uses the 3D VRTE to compute the measurement misfit function and the adjoint 3D VRTE to compute its gradient with respect to all unknown parameters. In the remote sensing problems of interest, the scalarvalued misfit function quantifies agreement with data as a function of atmosphere and surface properties, and its gradient guides the search through this parameter space. Remote sensing of the atmosphere and surface in a threedimensional region may require thousands of unknown parameters and millions of data points. Many approaches would require calls to the 3D VRTE solver in proportion to the number of unknown parameters or measurements. To avoid this issue of scale, we focus on computing the gradient of the misfit function as an alternative to the Jacobian of the measurement operator. The resulting adjoint method provides a way to adjust 3D atmosphere and surface properties with only two calls to the 3D VRTE solver for each spectral channel, regardless of the number of retrieval parameters, measurement view angles or pixels. This gives a procedure for adjusting atmosphere and surface parameters that will scale to the large problems of 3D remote sensing. For certain types of multiangle/multipixel polarimetric measurements, this encourages the development of a new class of threedimensional retrieval algorithms with more flexible parameterizations of spatial heterogeneity, less reliance on data screening procedures, and improved coverage in terms of the resolved physical processes in the Earth's atmosphere.Journal of Quantitative Spectroscopy and Radiative Transfer 09/2014; 144. DOI:10.1016/j.jqsrt.2014.03.030 · 2.29 Impact Factor 
Conference Paper: A New NASA Initiative in ThreeDimensional Tomography of the AerosolCloud Environment (3DTRACE): Outcome of a Pilot Study
[Show abstract] [Hide abstract]
ABSTRACT: Remote sensing is a key tool for sorting cloud ensembles by dynamical state, aerosol environments by source region, and establishing causal relationships between aerosol amounts, type, and cloud microphysicsthe socalled indirect aerosol climate impacts, which is identified as one of the main sources of error in current climate models. Current satellite imagers use data processing approaches that invariably start with cloud detection/masking to isolate aerosol airmasses from clouds, and then rely on onedimensional (1D) radiative transfer (RT) to interpret the aerosol and cloud measurements in isolation. Not only does this lead to welldocumented biases for the estimates of aerosol radiative forcing and cloud optical depths in current missions, but it is fundamentally inadequate for future missions such as ESA's Earth Cloud, Aerosol and Radiation Experiment (EarthCARE), NASA's PreACE (PACE) and AerosolCloudEcosystem mission (ACE, a NASA Tier2 Decadal Survey mission). All of these missions are intent on capturing the complex, threedimensional (3D) interactions between clouds and aerosols. In order to advance the stateoftheart, the next generation of satellite information processing systems must incorporate technologies that will enable the treatment of the atmosphere as a fully 3D environment, represented more realistically as a continuum. At one end, there is an optically thin background dominated by aerosols and molecular scattering that is strongly stratified and relatively homogeneous in the horizontal. At the other end, there are optically thick embedded elements, clouds and aerosol plumes, which can be more or less uniform and quasiplanar or else highly 3D with boundaries in all directions; in both cases, strong internal variability may be present. To make this paradigm shift possible, we propose to combine the standard models for satellite signal prediction physically grounded in 1D and 3D RT, both scalar and vector, with technologies adapted from biomedical imaging, digital image processing, and computer vision. This will enable us to demonstrate how the 3D distribution of atmospheric constituents, and their associated microphysical properties, can be reconstructed from multiangle/multispectral imaging radiometry and, more and more, polarimetry. Specific technologies of interest are computed tomography (reconstruction from projections), optical tomography (using crosspixel radiation transport in the diffusion limit), stereoscopy (depth/height retrievals), blind source and scale separation (signal unmixing), and disocclusion (information recovery in the presence of obstructions). In time, these potentially powerful inverse problem solutions will be fully integrated in a versatile satellite data analysis toolbox. At present, we will report on substantial progress at the component level achieved in the course of a oneyear pilot study sponsored by NASA's Earth Science and Technology office (ESTO). We focused specifically on the most elementary problems in atmospheric tomography: * One basic problem is to infer the outer shape and mean extinction of optically thick cumulustype 3D clouds, along with a bulk measure of cloud particle size. Two independent approaches were tested at JPL and Columbia University. * Another is to reconstruct the 3D spatial distribution of aerosol particle density in a plume, or crystal density in a cirrus layer, using only passive imaging. Here again, two independent approaches were tested at JPL and Technion  IIT. See Aides et al. [2013] for a description and demonstration of the latter tomographic reconstruction method. * Yet another is to separate high (cirrus) and low (broken cumulus) cloud layers based on their characteristically different spatial textures. See Yanovsky et al. [2014] for a description and demonstration of the image processing methodology that was used to solve this basic problem. Across all of these efforts, the wideopen frontier of multiangle/multipixel algorithms was explored. The suite of five independent feasibility studies will amount to a compelling proofofconcept for the ambitious 3DTomographic Reconstruction of the AerosolCloud Environment (3DTRACE) project as a whole. Finally, a notable spinoff of the 3DTRACE project is the development of a highperformance computing framework for generating highfidelity synthetic multiangle imagery to test new algorithms in a setting where the truth is known at every level of detail. First, the JPL LargeEddy Simulation (LES) code [Matheou and Chung, 2014] is used to obtain very realistic clouds or aerosol plumes (5 to 20 m gridcells over 5 to 20 km domains). Then LorentzMie code is used to convert the LES's bulk or bin microphysical quantities into optical ones. Finally, the stateoftheart MYSTIC 3D vector radiative transfer code [Emde et al., 2010] is applied to this large gridded scene using highly optimized backward Monte Carlo methods [Buras and Mayer, 2011] to deliver the imagery just as a remote sensing instrument would record it. This standalone capability at JPL will also be used to test operational algorithms in new ways. References: Aides, A., Y. Y. Schechner, V. Holodovsky, M. J. Garay, and A. B. Davis (2013). Multi SkyView 3D Aerosol Distribution Recovery, Opt. Express, 21, 2582025833. Emde, C., R. Buras, B. Mayer, and M. Blumthaler (2010). The impact of aerosols on polarized sky radiance: Model development, validation, and applications. Atmos. Chem. Phys., 10, 383396. Buras, R., and B. Mayer (2011). Efficient unbiased variance reduction techniques for Monte Carlo simulations of radiative transfer in cloudy atmospheres: The solution. J. Quant. Spectrosc. Radiat. Transfer, 112, 434447. Matheou, G., and D. Chung (2014). Largeeddy simulation of stratified turbulence. Part II: Application of the stretchedvortex model to the atmospheric boundary layer. J. Atmos. Sci. (under revision). Yanovsky, I., A. B. Davis, and V. M. Jovanovic (2014). Separation of radiances from a cirrus layer and broken cumulus clouds in multispectral images. IEEE Trans. Geosc. and Remote Sens. (submitted).14th Conference on Cloud Physics/14th Conference on Atmospheric Radiation/Anthony Slingo Symposium 2014 American Meteorological Society; 07/2014 
[Show abstract] [Hide abstract]
ABSTRACT: We consider the imaging of anisotropic conductivity tensors $\gamma=(\gamma_{ij})_{1\leq i,j\leq 2}$ from knowledge of several internal current densities $\mathcal{J}=\gamma\nabla u$ where $u$ satisfies a second order elliptic equation $\nabla\cdot(\gamma\nabla u)=0$ on a bounded domain $X\subset\mathbb{R}^2$ with prescribed boundary conditions on $\partial X$. We show that $\gamma$ can be uniquely reconstructed from four {\em wellchosen} functionals $\mathcal{J}$ and that noise in the data is differentiated once during the reconstruction. The inversion procedure is local in the sense that (most of) the tensor $\gamma(x)$ can be reconstructed from knowledge of the functionals $\mathcal{J}$ in the vicinity of $x$. We obtain the existence of an open set of boundary conditions on $\partial X$ that guaranty stable reconstructions by using the technique of complex geometric optics (CGO) solutions. The explicit inversion procedure is presented in several numerical simulations, which demonstrate the influence of the choice boundary conditions on the stability of the reconstruction. This problem finds applications in the medical imaging modality called Current Density Imaging or Magnetic Resonance Electrical Impedance Tomography.SIAM Journal on Imaging Sciences 03/2014; 7(4). DOI:10.1137/140961754 · 2.87 Impact Factor 
[Show abstract] [Hide abstract]
ABSTRACT: This paper analyzes the random fluctuations obtained by a heterogeneous multiscale firstorder finite element method applied to solve elliptic equations with a random potential. We show that the random fluctuations of such solutions are correctly estimated by the heterogeneous multiscale algorithm when appropriate finescale problems are solved on subsets that cover the whole computational domain. However, when the finescale problems are solved over patches that do not cover the entire domain, the random fluctuations may or may not be estimated accurately. In the case of random potentials with shortrange interactions, the variance of the random fluctuations is amplified as the inverse of the fraction of the medium covered by the patches. In the case of random potentials with longrange interactions, however, such an amplification does not occur and random fluctuations are correctly captured independent of the (macroscopic) size of the patches. These results are consistent with those obtained by the authors for more general equations in the onedimensional setting and provide indications on the loss in accuracy that results from using coarser, and hence less computationally intensive, algorithms.ESAIM Mathematical Modelling and Numerical Analysis 03/2014; 48(2):387409. DOI:10.1051/m2an/2013112 · 1.63 Impact Factor 
[Show abstract] [Hide abstract]
ABSTRACT: In this paper, we present a fluctuation analysis of a type of parabolic equations with large, highly oscillatory, random potentials around the homogenization limit. With a FeynmanKac representation, the KipnisVaradhan’s method, and a quantitative martingale central limit theorem, we derive the asymptotic distribution of the rescaled error between heterogeneous and homogenized solutions under different assumptions in dimension \(d\ge 3\) . The results depend highly on whether a stationary corrector exits.03/2014; 3(1):151. DOI:10.1007/s4007201400408 
[Show abstract] [Hide abstract]
ABSTRACT: We propose a method to reconstruct the density of a luminescent source in a highly scattering medium from ultrasoundmodulated optical measurements. Our approach is based on the solution to a hybrid inverse source problem for the diffusion equation.Physical Review E 03/2014; 89(31):031201. DOI:10.1103/PhysRevE.89.031201 · 2.33 Impact Factor 
[Show abstract] [Hide abstract]
ABSTRACT: This paper concerns the homogenization problem of heat equation with large, timedependent, random potentials in high dimensions $d\geq 3$. Depending on the competition between temporal and spatial mixing of the randomness, the homogenization procedure turns to be different. We characterize the difference by proving the corresponding weak convergence of Brownian motion in random scenery. When the potential depends on the spatial variable macroscopically, we prove a convergence to SPDE.Stochastic Processes and their Applications 01/2014; DOI:10.1016/j.spa.2014.07.024 · 1.05 Impact Factor 
Article: A Note on Central Limit Theorem for Heat Equation with Large, Highly Oscillatory, Random Potential
[Show abstract] [Hide abstract]
ABSTRACT: In this note, we prove a central limit result for heat equation with large, highly oscillatory, random potential. With a FeynmanKac representation, the KipnisVaradhan's method and a refined quantitative martingale central limit theorem, we derive the asymptotic Gaussian distribution of the rescaled corrector when the random potential is Gaussian or Poissonian and the dimension $d=3$. 
[Show abstract] [Hide abstract]
ABSTRACT: We consider an elliptic equation with purely imaginary, highly heterogeneous, and large random potential with a sufficiently rapidly decaying correlation function. We show that its solution is well approximated by the solution to a homogeneous equation with a realvalued homogenized potential as the correlation length of the random medium $\varepsilon\rightarrow 0$ and estimate the size of the random fluctuations in the setting $d\geq3$. 
[Show abstract] [Hide abstract]
ABSTRACT: Assuming the availability of internal fullfield measurements of the continuum deformations associated with a nonhomogeneous isotropic linear elastic solid, this article focuses on the quantitative reconstruction of its constitutive parameters. Starting from the governing momentum equation, algebraic manipulations are employed to construct a simple gradient system for the quantities of interest in which the featured coefficients are expressed in terms of the measured displacement fields and their spatial derivatives. A direct integration of this system is discussed to finally demonstrate the inexpediency of such an approach when dealing with polluted measurements. Upon using noisy data, an alternative variational formulation is deployed to invert for the unknown physical parameters. Analysis of this latter inversion procedure provides existence and uniqueness results while the reconstruction stability with respect to the measurements is investigated. As the inversion procedure requires differentiating the measurements twice, a numerical differentiation scheme based on ad hoc regularization then allows an optimally stable reconstruction of the sought moduli. Numerical results are included to illustrate and assess the performance of the overall approach.Inverse Problems 10/2013; 30(12). DOI:10.1088/02665611/30/12/125004 · 1.80 Impact Factor 
[Show abstract] [Hide abstract]
ABSTRACT: This paper concerns the reconstruction of possibly complexvalued coefficients in a secondorder scalar elliptic equation posed on a bounded domain from knowledge of several solutions of that equation. We show that for a sufficiently large number of solutions and for an open set of corresponding boundary conditions, all coefficients can be uniquely and stably reconstructed up to a well characterized gauge transformation. We also show that in some specific situations, a minimum number of such available solutions equal to $I_n=\frac12n(n+3)$ is sufficient to uniquely and globally reconstruct the unknown coefficients. This theory finds applications in several coupledphysics medical imaging modalities including photoacoustic tomography, transient elastography, and magnetic resonance elastography.Communications on Pure and Applied Mathematics 10/2013; 66(10). DOI:10.1002/cpa.21453 · 3.08 Impact Factor 
[Show abstract] [Hide abstract]
ABSTRACT: This paper concerns the reconstruction of a complexvalued anisotropic tensor $\gamma=\sigma+\i\omega\varepsilon$ from knowledge of several internal magnetic fields $H$, where $H$ satisfies the anisotropic Maxwell system on a bounded domain with prescribed boundary conditions. We show that $\gamma$ can be uniquely reconstructed with a loss of two derivatives from errors in the acquisition of $H$. A minimum number of 6 such functionals is sufficient to obtain a local reconstruction of $\gamma$. In the special case where $\gamma$ is close to a scalar tensor, boundary conditions are chosen by means of complex geometric optics (CGO) solutions. For arbitrary symmetric tensors $\gamma$, a Runge approximation property is used to obtain partial results. This problem finds applications in the medical imaging modalities Current Density Imaging and Magnetic Resonance Electrical Impedance Tomography.Inverse Problems and Imaging 08/2013; 8(4). DOI:10.3934/ipi.2014.8.1033 · 1.39 Impact Factor 
[Show abstract] [Hide abstract]
ABSTRACT: This paper concerns the quantitative step of the medical imaging modality Thermoacoustic Tomography (TAT). We model the radiation propagation by a system of Maxwell's equations. We show that the index of refraction of light and the absorption coefficient (conductivity) can be uniquely and stably reconstructed from a sufficiently large number of TAT measurements. Our method is based on verifying that the linearization of the inverse problem forms a redundant elliptic system of equations. We also observe that the reconstructions are qualitatively quite different from the setting where radiation is modeled by a scalar Helmholtz equation as in [10].Inverse Problems 08/2013; 30(5). DOI:10.1088/02665611/30/5/055013 · 1.80 Impact Factor 
[Show abstract] [Hide abstract]
ABSTRACT: An invariance principle is proved for Brownian motion in random scenery, which is chosen to be either a Gaussian or Poissonian random field.ELECTRONIC JOURNAL OF PROBABILITY 06/2013; DOI:10.1214/EJP.v192894 · 0.77 Impact Factor
Publication Stats
2k  Citations  
195.67  Total Impact Points  
Top Journals
Institutions

2014

California Institute of Technology
Pasadena, California, United States


2001–2014

Columbia University
 Department of Applied Physics and Applied Mathematics
New York, New York, United States


2000–2002

University of Chicago
 Department of Mathematics
Chicago, Illinois, United States


1998–2001

Stanford University
 Department of Mathematics
Palo Alto, California, United States
