In this paper, two framelet based deconvolution algorithms are proposed. The basic idea of framelet based approach is to convert the deconvolution problem to the problem of inpainting in a frame domain by constructing a framelet system with one of the masks being the given (discrete) convolution kernel via the unitary extension principle of [26], as introduced in [6,9]. The first algorithm unifies our previous works in high resolution image reconstruction and infra-red chopped and nodded image restoration, and the second one is a combination of our previous frame-based deconvolution algorithm and the iterative thresholding algorithm given by [14, 16]. The strong convergence of the algorithms in infinite dimensional settings is given by employing proximal forward-backward splitting (PFBS) method. Consequently, it unifies iterative algorithms of infinite and finite dimensional setting and simplifies the proof of the convergence of the algorithms of [6].

We compare 13 different a posteriori error estimators for the Poisson problem with lowest-order finite element discretization. Residual-based error estimators compete with a wide range of averaging estimators and estimators based on local problems. Among our five benchmark problems we also look on two examples with discontinuous isotropic diffusion and their impact on the performance of the estimators. (Supported by DFG Research Center MATHEON.)

In this paper we develop two conforming finite element methods for a fourth order bi-wave equation arising as a simplified Ginzburg-Landau-type model for $d$-wave superconductors in absence of applied magnetic field. Unlike the biharmonic operator $\Delta^2$, the bi-wave operator $\Box^2$ is not an elliptic operator, so the energy space for the bi-wave equation is much larger than the energy space for the biharmonic equation. This then makes it possible to construct low order conforming finite elements for the bi-wave equation. However, the existence and construction of such finite elements strongly depends on the mesh. In the paper, we first characterize mesh conditions which allow and not allow construction of low order conforming finite elements for approximating the bi-wave equation. We then construct a cubic and a quartic conforming finite element. It is proved that both elements have the desired approximation properties, and give optimal order error estimates in the energy norm, suboptimal (and optimal in some cases) order error estimates in the $H^1$ and $L^2$ norm. Finally, numerical experiments are presented to guage the efficiency of the proposed finite element methods and to validate the theoretical error bounds.

In magnetic resonance elastography, one seeks to reconstruct the shear modulus from measurements of the displacement field in the whole body. In this paper, we present an optimization approach which solves the problem by simply minimizing a discrepancy functional. In order to recover a complex anomaly in a homogenous medium, we first observe that the information contained in the wavefield should be decomposed into two parts, a "near-field" part in the region around the anomaly and a "far-field" part in the region away from the anomaly. As will be justified both theoretically and numerically, separating these scales provides a local and precise reconstruction.

We study a time domain decorrelation method of source signal separation from convolutive sound mixtures based on an infinite impulse response (IIR) model. The IIR model uses fewer parameters to capture the physical mixing process and is useful for finding low dimensional separating solutions. We present inversion formulas to decorrelate the mixture signals and derive filter equations involving second order time lagged statistics of mixtures. We then formulate an $l_1$ constrained minimization problem and solve it by an iterative method. Numerical experiments on recorded sound mixtures show that our method is capable of sound separation in low dimensional parameter spaces with good perceptual quality and low correlation coefficient comparable to the known infomax method.

We consider the approximation of systems of reaction-diffusion equations, with the finite element method. The highest derivative in each equation is multiplied by a parameter $\varepsilon \in (0,1]$, and as $\varepsilon \rightarrow 0$ the solution of the system will contain boundary layers. We extend the analysis of the corresponding scalar problem from [Melenk, IMA J. Numer. Anal. 17(1997), pp. 577-601], to construct a finite element scheme which includes elements of size $\mathcal{O}(\varepsilon p)$ near the boundary, where $p$ is the degree of the approximating polynomials. We show that, under the assumption of analytic input data, the method yields exponential rates of convergence, independently of $\varepsilon $, when the error is measured in the energy norm associated with the problem. Numerical computations supporting the theory are also presented, which also show that the method yields robust exponential convergence rates when the error in the maximum norm is used.

We present a new iterative reconstruction algorithm to improve the algebraic reconstruction technique (ART) for the Single-Photon Emission Computed Tomography. Our method is a generalization of the Kaczmarz iterative algorithm for solving linear systems of equations and introduces exact and implicit attenuation correction derived from the attenuated Radon transform operator at each step of the algorithm. The performances of the presented algorithm have been tested upon various numerical experiments in presence of both strongly non-uniform attenuation and incomplete measurements data. We also tested the ability of our algorithm to handle moderate noisy data. Simulation studies demonstrate that the proposed method has a significant improvement in the quality of reconstructed images over ART. Moreover, convergence speed was improved and stability was established, facing noisy data, once we incorporate filtration procedure in our algorithm.

We propose and analyze a $C^0$ spectral element method for a model eigenvalue problem with discontinuous coefficients in the one dimensional setting. A super-geometric rate of convergence is proved for the piecewise constant coefficients case and verified by numerical tests. Furthermore, the asymptotical equivalence between a Gauss-Lobatto collocation method and a spectral Galerkin method is established for a simplified model.